Mechanism of Action of Ribosomally Synthesized and Post-Translationally Modified Peptides

Abstract

Ribosomally synthesized and post-translationally modified peptides (RiPPs) are a natural product class that has undergone significant expansion due to the rapid growth in genome sequencing data and recognition that they are made by biosynthetic pathways that share many characteristic features. Their mode of actions cover a wide range of biological processes and include binding to membranes, receptors, enzymes, lipids, RNA, and metals as well as use as cofactors and signaling molecules. This review covers the currently known modes of action (MOA) of RiPPs. In turn, the mechanisms by which these molecules interact with their natural targets provide a rich set of molecular paradigms that can be used for the design or evolution of new or improved activities given the relative ease of engineering RiPPs. In this review, coverage is limited to RiPPs originating from bacteria.

Graphical Abstract

1. Introduction

Natural products account for a remarkably rich reservoir of bioactive pharmaceutical leads, as nearly half of the drugs introduced over the past four decades were derived from molecules produced by bacteria, fungi, or plants.^1–3 Historic approaches towards discovery of novel therapeutics typically focused on using target- or phenotype-based screening.⁴ While these strategies have been successful, each is bereft with significant limitations. Recent advances in analytical instrumentation, genome sequencing, and metabolomics have facilitated the discovery of new natural products with therapeutic potential at an exponential rate.⁵ However, the elucidation of biological targets and mode of action (MOA) of these newly discovered natural products have lagged. Decades-long advances in various methodologies have set the framework for detailed MOA studies and avails many future opportunities for such efforts.^6–8

The ribosomally synthesized and post-translationally modified peptides (RiPPs) are a natural product class that has undergone significant expansion as a consequence of rapid growth in genome sequencing data.^9,10 Biosynthetic gene clusters encoding RiPPs typically consist of one or more genes encoding a precursor peptide, which is modified by biosynthetic enzymes to yield the mature natural product or products. The co-localization of genes encoding the biosynthetic proteins with those encoding the substrate peptide facilitates prediction of the entire biosynthetic pathway. Growing interest in the potential of RiPP therapeutics has been spurred by the discovery of compounds with antibacterial,¹¹ antifungal,¹² antinociceptive,^13,14 antiviral,¹⁵ and antitumor^16–20 activities. The myriad activities of RiPPs serve to emphasize how post-translational modifications have allowed molecular evolution of structures that exploit a wide range of biological targets in different domains of life. These targets include the cell membrane, small molecule metabolites and biosynthetic intermediates, enzymes, receptors, RNA, and metal ions. In turn, the wide variety of natural targets and activities suggests that this class of compounds offers much promise for discovery and engineering of new activities. This review will provide a comprehensive discussion of the MOA of bacterial RiPPs for which such information is available, to complement earlier reviews that focused on specific compound classes or a subset of RiPPs.^21–38 The review covers studies reported up to and including June 2022.

Despite a simplistic biosynthetic logic, RiPPs display an expansive diversity of molecular scaffolds. Based on the still limited information on RiPP MOA, the bioactivities associated with even closely related structural classes of RiPPs are rarely predictable. In some instances, elucidation of MOA and validation of biological targets for several compounds predate their characterization as members of a RiPP class.^39,40 As both the substrate peptide and the enzymes that catalyze the post-translational modifications are direct gene products, RiPP biosynthetic gene clusters are excellent candidates for genome mining exercises, and several available robust bioinformatics platforms harness this capability.^41–46 Likewise, the synteny of genes necessary for the production of the bioactive RiPP final product has enabled refactoring efforts for heterologous production. In several instances, an affinity tag was appended to the precursor peptide, allowing for facile purification of the modified precursor peptide prior to removal of the leader sequence (Figure 1) to yield high levels of purified compound as well as analogs.¹⁰ These advances in technologies that have enabled computational identification, high-level production in homologous/heterologous hosts, and relative ease of purification using affinity tags can empower studies on RiPPs for which MOA data have not yet been obtained.

Figure 1.

General biosynthetic scheme for RiPP maturation. The follower peptide is shown as dashes because it is not ubiquitous across known RiPP classes.

1.1. Preface - Organizational approach

In this review, we cover the MOA, identification, and validation of biological targets for all classes of currently known bacterial RiPPs for which such data are available. Reported values for minimal inhibitory concentration (MIC) have been converted to molar values to both maintain internal consistency across different studies and to allow for comparisons of potency against non-RiPP antibiotics. Generally, bioactivity and MOA cannot be predicted based on the RiPP class, which is defined by biosynthetic pathway and not activity. For example, many lanthipeptides share common biosynthetic origins but demonstrate widely varying bioactivities. The diversity of targets likely reflects the ability to rapidly evolve new RiPP structures and associated bioactivities. We organized this review based on structural and biosynthetic similarities rather than by MOA because of the well-established framework within the field.¹⁰ To ensure that similar MOAs across RiPP classes are recognized, we provide cross references when different RiPP classes target related processes and collated all activities in abbreviated format in Table 1.

Table 1. Summary of MOAs.

The table is organized to correspond to the order in which the RiPP classes are presented in this review.

Bioactivity / Target or mechanism	Compound name	RiPP class
Antibacterial / lipid II; pore formation Autoinduction of biosynthesis	nisin	lanthipeptide
Antibacterial / lipid II; pore formation Autoinduction of biosynthesis	subtilin	lanthipeptide
Antibacterial / lipid II	microbisporicin (NAI-107)	lanthipeptide
Antibacterial / lipid II	mutacin 1140	lanthipeptide
Antibacterial / lipid II; pore formation	epidermin	lanthipeptide
Antibacterial / lipid II; pore formation	geobacillin I	lanthipeptide
Antibacterial / lipid II; pore formation	gallidermin	lanthipeptide
Antibacterial / pore formation	Pep5	lanthipeptide
Antibacterial / pore formation	epilancin 15X	lanthipeptide
Antifungal	pinensin A and B	lanthipeptide
Morphogenic peptide / facilitates aerial hyphae formation	SapT	lanthipeptide
Morphogenic peptide / facilitates aerial hyphae formation	SapB	lanthipeptide
Antibacterial / lipid II Autoinduction of biosynthesis	mersacidin	lanthipeptide
Antibacterial / lipid II	lacticin 481	lanthipeptide
Antibacterial / lipid II	nukacin ISK-1	lanthipeptide
Antibacterial / lipid II; pore formation Autoinduction of biosynthesis	bovicin HJ50	lanthipeptide
Antibacterial / lipid II; pore formation	plantaricin C	lanthipeptide
Antibacterial / lipid II; pore formation	lacticin 3147	lanthipeptide
Antibacterial / lipid II; pore formation	staphylococcin C55	lanthipeptide
Antibacterial / lipid II; pore formation	haloduracin	lanthipeptide
Antibacterial Hemolytic Autoinduction of biosynthesis	cytolysin	lanthipeptide
Antibacterial / Phosphatidylethanolamine (PE) binding and membrane disruption Antiviral / PE binding	cinnamycin	lanthipeptide
Antibacterial / PE binding and membrane permeabilization. Antiviral / PE binding	duramycin	lanthipeptide
ACE inhibition	ancovenin	lanthipeptide
Antiviral / PE binding	divamide A and B	lanthipeptide
Antiviral / PE and gp120 binding	labyrinthopeptin A1	lanthipeptide
Antiviral / PE binding Antiallodynic	labyrinthopeptin A2	lanthipeptide
Antibacterial / cell wall biosynthesis Antiallodynic	NAI-112	lanthipeptide
Antibacterial	avermipeptin B	lanthipeptide
Morphogenic peptide / facilitates aerial hyphae formation	AmfS	lanthipeptide
No identified activity	catenulipeptin	lanthipeptide
Anticancer Antibacterial	ammosamide C	lanthipeptide
Anticancer / myosin binding and QR2 inhibition	ammosamide A and B	lanthipeptide
Anticancer / myosin binding	ammosamide 272	lanthipeptide
Anticancer / kinase inhibition	lymphostin	lanthipeptide
No identified activity	3-thiaglutamate	lanthipeptide
Antibacterial / cell wall biosynthesis	cacaoidin	lanthipeptide/lanthidin
Antibacterial	lexapeptide	lanthipeptide/lanthidin
Antibacterial	microvionin	lipolanthine
Antibacterial	goadvionin	lipolanthine
Cytolytic	streptolysin S (SLS)	LAP
Hemolytic	clostridiolysin S (CLS)	LAP
Bactericidal (mildly hemolytic)	listeriolysin S (LLS)	LAP
Hemolytic	stapholysin S (StsA)	LAP
Antibacterial	plantazolicin	LAP
Antibacterial / DNA gyrase	microcin B17 (MccB17)	LAP
Antibacterial / block of 50S ribosomal subunit exit tunnel	klebsazolicin	LAP
Antibacterial / block of 50S ribosomal subunit exit tunnel	phazolicin	LAP
Antibacterial Induction of sporulation and secondary metabolite production	goadsporin	LAP
Antibacterial / protein synthesis (translocation) Antimalarial Anticancer / cytotoxic Reactivates latent HIV reservoirs	thiostrepton	pyritide
Antibacterial / protein synthesis (translocation) Antimalarial Anticancer / cytotoxic	micrococcin P1	pyritide
Antibacterial / protein synthesis (EF-Tu)	GE2270A	pyritide
Antibacterial / protein synthesis (EF-Tu)	GE37468	pyritide
Antibacterial / protein synthesis (EF-Tu)	amythiamicin A	pyritide
Antibacterial / protein synthesis (EF-Tu)	thiomuracin A	pyritide
Bacterial MDR activation	promothiocins A & B	pyritide
Antibacterial Bacterial MDR activation / TipA	nosiheptide	pyritide
MDR activation / TipA Anticancer/cytotoxic	berninamycins A & B	pyritide
Bacterial MDR activation / TipA	thioxamycin	pyritide
Bacterial MDR activation / TipA	thiotipin	pyritide
Anticancer / cytotoxic	siomycin A	pyritide
Anticancer / cytotoxic	thiocillin I	pyritide
Anticancer / cytotoxic	YM-266183	pyritide
Bacteriophage RNA polymerase and human renin inhibitor	cyclothiazomycin	pyritide
Antibacterial	nocathiacins	pyritide
Antineoplastic / apoptosis inducer	thioviridamide	thioamitide
Antineoplastic / F1 subunit of ATP synthase; induces apoptosis	prethioviridamide	thioamitide
Antineoplastic / apoptosis inducer	thioalbamide	thioamitide
Antibacterial / protein synthesis; block of tRNA entry into ribosomal A-site.	bottromycin A2	bottromycin
Antineoplastic / protein and RNA synthesis	ulithiacyclamide	cyanobactin
Antineoplastic / predicted to affect cytokinesis	cycloxazoline	cyanobactin
Antineoplastic	trunkamide A	cyanobactin
Reverses MDR in breast carcinoma model	dendroamide A	cyanobactin
Proposed chalkophore or siderophore	patellamide A	cyanobactin
Proposed chalkophore or siderophore MDR reversal in leukemic lymphoblasts	patellamide B	cyanobactin
Proposed chalkophore or siderophore MDR reversal in leukemic lymphoblasts	patellamide C	cyanobactin
MDR reversal in leukemic lymphoblasts	patellamide D	cyanobactin
Proposed chalkophore or siderophore	patellamide E	cyanobactin
Plasmin inhibitor	agardhipeptin A	cyanobactin
Antibacterial	sphaerocyclamide	cyanobactin
Antibacterial	kawaguchipeptin B	cyanobactin
Cytotoxic	microcyclamide	cyanobactin
Cytotoxic	lissoclinamide	cyanobactin
Cytotoxic	ascidiacyclamide	cyanobactin
Antibacterial / L-histidinol phosphate aminotransferase	pantocin A	pantocin
Antibacterial / Asp-tRNA synthetase	microcin C7 (McC)	microcin
Antibacterial / Asp-tRNA synthetase	McCYPs	microcin
Antibacterial / mannose PTS and pore formation	MccE492/MccE492m	microcin
Antibacterial / ATP synthase (proposed)	MccH47/MccH47m	microcin
Unknown	MccM/MccMm	microcin
Chalkophore	methanobactin	methanobactin
Signaling	AIPs (I, II, III, IV).	autoinducing peptide
Signaling	GBAP	autoinducing peptide
Competence activation	ComX168	ComX
Competence activation	ComXRO-E-2	ComX
Atrial natriuretic factor receptor antagonist	anantin	lasso peptide
Endothelin type B receptor antagonist	RES-701–1, 2, 3, and 4	lasso peptide
Human glucagon receptor inhibitor	BI-32169	lasso peptide
Antibacterial / prolyl endopeptidases	propeptin and propeptin-2	lasso peptide
Antibacterial / ClpC1 protease	lassomycin	lasso peptide
Viral fusion inhibition Antibacterial / cell wall biosynthesis / Quorum sensing inhibition	siamycin I	lasso peptide
Viral fusion inhibition	siamyin II	lasso peptide
Quorum sensing inhibition	sviceucin	lasso peptide
Antibacterial / cell wall	streptomonomicin	lasso peptide
Antibacterial / RNA polymerase (RNAP); dissipates membrane potential Superoxide production inducer Antimitochondrial activity	microcin J25 (MccJ25)	lasso peptide
Unknown	microcin Y	lasso peptide
Antibacterial / RNAP	capistruin	lasso peptide
Antibacterial / RNAP	klebsidin	lasso peptide
Antibacterial / RNAP	citrocin	lasso peptide
Antibacterial / RNAP	ubonodin	lasso peptide
Antibacterial	lariatin A and B	lasso peptide
Antifungal / SakA kinase (proposed)	humidimycin	lasso peptide
Unknown	tryptorubin A	atroptide
Human tyrosinase and serine protease inhibition	microviridin A	graspetide
Serine protease inhibition	microviridin B-N	graspetide
Serine protease inhibition	plesiocin	graspetide
Serine protease inhibition Metalloprotease inhibition	marinostatin	graspetide
Trypsin-like protease inhibition Daphnid protease inhibition	microviridin J	graspetide
Subtilisin inhibition	microviridin K and L	graspetide
Antibacterial / pore formation	AS-48	circular bacteriocin
Antibacterial / MalEFG implicated	garvicin ML	circular bacteriocin
Antibacterial / pore formation	carnocyclin A	circular bacteriocin
Antibacterial	circularin A	circular bacteriocin
Antibacterial	uberolysin A	circular bacteriocin
Antibacterial	lactocyclicin Q	circular bacteriocin
Antibacterial	amylocyclicin	circular bacteriocin
Antibacterial	enterocin NKR-5–3B	circular bacteriocin
Antibacterial	pumilarin	circular bacteriocin
Antibacterial	gassericin A	circular bacteriocin
Antibacterial	acidocin B	circular bacteriocin
Antibacterial	butyrivibriocin AR10	circular bacteriocin
Antibacterial	plantaricyclin A	circular bacteriocin
Antibacterial / cell wall implicated	pheganomycins (including deoxypheganomocyin D)	amidinotide
Inhibitor of interaction of CsrA with its inhibitory RNA	crocagin A and B	crocagin
Antibacterial / glucose PTS implicated	sublancin 168	glycocin
Bacteriostatic / GlcNAc PTS implicated	glycocin F	glycocin
Bacteriostatic	ASM1	glycocin
Bacteriostatic	enterocin 96 and F4–9	glycocin
Antibacterial	pallidocin	glycocin
Bacterial blight in rice crop / mimic of PSY1 plant hormone	RaxX	sulfatide
Cytotoxic / pore formation and lysosome neutralization	polytheonamide B	polytheonamide
Antibacterial / proteobacterial BamA; impedes protein insertion and assembly in the outer membrane	darobactin	darobactin
Antibacterial / membrane permeabilization Spermicidal Antiviral Biofilm formation inhibition	subtilosin A	sactipeptide
Antibacterial Biofilm formation inhibition	hyicin	sactipeptide
Antibacterial / pore formation	thuricin CD	sactipeptide
Antibacterial / membrane permeabilization	huazacin (thuricin Z)	Sactipeptide
Antibacterial Morphogenesis	thurincin H	sactipeptide
Antibacterial (lytic)	sporulation killing factor (SKF)	sactipeptide
Antibacterial	ruminococcin C	sactipeptide
Antibacterial (bactericidal)	streptosactin	sactipeptide
Antibacterial	EpeX*	epipeptide
Redox cofactor	mycofactocin	mycofactocin
Redox cofactor	pyrroloquinoline pyrrole	pyrroloquinoline pyrrole

During RiPP biosynthesis, a portion of the precursor peptide is post-translationally modified in what is referred to as the “core” region. The biosynthetic enzymes necessary for modification are directed to the correct substrate by specific motifs known as recognition sequences present in a region known as the “leader peptide” if located N-terminal to the core peptide,⁴⁷ or “follower peptide” if located C-terminal to the core region.^9,10 The vast majority of characterized bacterial RiPPs utilize a leader-core paradigm, as opposed to a core-follower or a tripartite leader-core-follower paradigm.⁴⁸ In all RiPPs, the recognition and modification sites are physically separated, thus allowing the biosynthetic pathways to maximize substrate selectivity via recognition of conserved leader/follower sequences, while tolerating variation in the core region. In this review, we use the standardized nomenclature for the precursor peptide⁹ where residues in the core peptide are indicated with a positive number starting from the junction with the leader sequence, while residues in the leader peptide are indicated with negative numbers counting back from this junction (Figure 1). Hence, the residue numbering of several RiPPs has been adapted to be consistent with this numbering scheme (which may differ from the numbering used in the primary literature). We do not discuss in further detail the biosynthetic pathways for the different classes of RiPPs. A number of excellent, comprehensive reviews detail the biosynthetic mechanisms for various RiPPs.^9,10,29,33

2. Lanthipeptides and compounds made via related pathways

Lanthipeptides are characterized by the thioether crosslinked bis amino acids lanthionine (Lan) and methyllanthionine (MeLan; Figure 2). At present, these macrocyclic peptides can be generated by different sets of enzymes that are the basis for their classification scheme (class I-V). Lanthipeptides with antimicrobial activities have historically been termed lantibiotics.⁴⁹ A salient feature of lantibiotics with a known MOA is that the targets are metabolites rather than macromolecules like proteins or RNA as discussed in the subsequent sections. In most currently characterized lanthipeptides, the (Me)Lan have DL stereochemistry (Figure 2), but compounds with LL- and D-allo-L-stereochemistry have been increasingly reported. In this review, the shorthand notations shown in Figure 2 will be used for lanthipeptides.

Figure 2.

Structures of thioether crosslinks and dehydrated amino acids commonly found in lanthipeptides. Shorthand notations are presented below the corresponding structure. Structures containing non-canonical LL-(methyl)lanthionine and D-allo-L-methyllanthionine stereochemistry are indicated with either a single or double asterisk in the shorthand notation, respectively.

2.1. Class I lanthipeptides

Nisin, a class I lanthipeptide produced by Lactococcus lactis, is the most extensively characterized family member and exhibits antimicrobial activity against many non-Proteobacteria often with submicromolar minimal inhibitory concentrations (MICs).^49,50 Many nisin variants have been reported over the years, differing in a few amino acids but retaining the overall ring pattern. Unless specified otherwise, when discussing nisin in this review we refer to nisin Z (1). For more than 50 years, nisin has been used in the food industry to combat food-borne pathogens.⁵¹ The molecule exerts its bioactivity through binding of the cell wall precursor lipid II (2, Figure 3A) and the formation of relatively stable pores in the cell membrane that are composed of nisin and lipid II.^52,53 Nisin binding is facilitated by the amide backbone of the A and B rings interacting with the pyrophosphate moiety of lipid II, as characterized by NMR spectroscopy using a soluble lipid II analog with a farnesyl chain instead of the full C55 undecaprenol tail (Figure 3C).⁵⁴ The binding of lipid II is believed to act as an anchor for the formation of pores in the membrane involving a complex comprised of eight nisin and four lipid II molecules (Figure 3F).^54,55 The 2:1 nisin:lipid II stoichiometry in the pore indicates that the 1:1 NMR structure (Figure 3B and C) only reveals part of the molecular interactions and that a second nisin molecule must bind to lipid II, nisin, or both. This aspect of pore formation is still poorly understood.

Figure 3.

(A) Chemical structure of nisin Z (1) and lipid II (2). The A and B rings of nisin interact with the pyrophosphate moiety of lipid II (green) and the C-E rings of nisin (orange) are involved in pore formation. A hinge region (purple) is critical for pore formation. (B) NMR characterization of nisin (space filling) binding to a lipid II analog (sticks). (C) Zoom-in view showing the interactions between the pyrophosphate of lipid II (green and red) with the amide backbone of the nisin A ring (carbon atoms in grey). (D-E) Proposed model for lipid II mediated pore formation wherein nisin first interacts with lipid II through the above-described interaction, followed by pore formation through a complex of eight nisin molecules and four lipid II molecules. The molecular details of the arrangement of nisin and lipid II in the pore remain unknown.

Lipid II and its variants have been used to investigate the molecular determinants of selectivity and affinity. Nisin binds to lipid I (lacking the GlcNAc, Figure 3A) and lipid II with affinities in the 10–50 nM range as determined by isothermal titration calorimetry (ITC), radiolabeling, and dye leakage experiments from vesicles.^56–58 Hence, the GlcNAc moiety is not important for binding. Similarly, a lipid II analog in which the pentapeptide was removed still bound tightly to nisin.⁵⁶ Thus, the pyrophosphate group appears the main recognition site on both lipid I and lipid II. However, pores are not formed with undecyl pyrophosphate and hence the MurNAc group is important for this activity.^56,57 Several studies focusing on the length of the prenyl chain and the anchoring of this chain in the membrane have demonstrated that the interaction of nisin with lipid II is strongly dependent on the membrane environment.^56,59 These experiments typically report on both the binding of nisin to lipid II and pore complex formation in the membrane, and deconvolution into specific molecular interactions is therefore difficult. The pores have been shown to be relatively stable (lifetime of seconds as opposed to milliseconds in the absence of lipid II) by electrophysiology, and once the complex is established in vesicles in model studies, nisin does not exchange when exposed to new vesicles containing lipid II and the complex appears stable for hours.^56,60

The biosynthetic machinery of nisin is tolerant to changes in the precursor peptide and this feature has been leveraged to generate many variants. In-depth structure-activity relationships (SAR) for nisin have been extensively reviewed and will not be covered in detail here.^{10,51,61–68} These studies have identified a critical hinge region between the C and D-rings (Figure 3A) that is thought to allow the D and E rings to insert into the membrane to form the pore.^55,69–74 Furthermore the N-terminus cannot be extended but the D and E rings can be removed without abolishing antibacterial activity.^51,75–77 The latter truncates do, however, loose pore-forming activity. Several variants have been identified that are more potent than nisin against a subset of Gram-positive and/or Gram-negative bacteria.^72,77–83

Many other class I lanthipeptides such as microbisporicin (NAI-107, 3), mutacin 1140 (4), epidermin (5), epilancin 15X (6) and geobacillin I (7) produced by Microbispora corallina, Streptococcus mutans, Staphylococcus epidermidis (5 and 6), and Geobacillus thermodinitrificans, respectively, contain variants of the lipid II-binding domain of nisin (i.e., the A and B rings; Figure 4).^52,84–87 These compounds have a wide variety of C-terminal ring patterns and sequences and while all bind lipid II, several (e.g., microbisporicin and mutacin 1140) do not form pores.^{79,84,88–90} Studies on epidermin and the closely related gallidermin, which have shorter C-terminal tails compared to nisin, showed that pore formation depends on membrane thickness.⁹¹ Collectively, these studies showed that pore formation was not required for antibacterial activity and that lipid II binding was sufficient. In addition to inhibition of the transglycosylation step of cell wall biosynthesis (see also class II two-component lanthipeptides; section 2.2.1), nisin disrupts the localization of lipid II, taking it away from the well-controlled localization of cell wall biosynthetic enzyme complexes.⁹²

Figure 4.

Shorthand notations for the structures of microbisporicin (3), mutacin 1140 (4), epidermin (5), epilancin 15X (6), geobacillin I (7), and subtilin (8). The nisin-like lipid II binding domains are indicated in blue, nisin-like pore forming domains are indicated in orange, and the hinge region, present in (8), is indicated in purple. The stereochemistry of the dihydroxyproline in (3) is not known. Throughout this review, in the shorthand notations the N-terminus is indicated by H- and the C-terminus by –OH.

Class I lanthipeptides often also display quorum-sensing activities. The biosynthetic gene clusters typically encode two-component signaling systems made up of a receptor histidine kinase and a response regulator.⁹³ The best-studied examples are the NisRK autoinduction system that regulates nisin production and the corresponding SpaRK system involved in subtilin (8) autoinduction in Bacillus subtilis.^51,94,95 Although the molecular details of lanthipeptide recognition are not known, NisK and SpaK are highly specific for their cognate ligand.^96–98 Several SAR studies have demonstrated that residues important for signaling are different from those essential for antibacterial activity.⁹⁹ For instance, the B-ring of nisin is critical for lipid II binding and antibacterial activity but is dispensable for autoinduction.⁷⁷ Extensive mutagenesis studies along with investigation of crosstalk between the nisin and subtilin systems and construction of hybrid molecules identified the N-terminal Trp and Phe20 as critical residues in subtilin for SpaK recognition.⁹⁷ Furthermore, whereas the B-ring and the native hinge region between C- and D-rings were shown to be required for subtilin signaling, the D- and E-rings were expendable.⁹⁶

Many class I lanthipeptides do not contain the A/B ring lipid II-binding motif. Examples such as the structurally related Pep5 and epilancin 15X (6, Figure 4), produced by S. epidermis, induce pore formation, but do not appear to bind to lipid II.⁵² Given the low minimum inhibitory concentration (MIC) values of these compounds (nM against many Staphylococcus strains), they may also have a specific molecular target that remains undiscovered.⁵² Epilancin 15X is of particular interest because it lacks the lipid II-binding A and B rings of nisin but has a similar C-E ring pattern (Figure 4).^100,101 Total synthesis and structural variants of epilancin 15X yielded insight into the importance of the C-terminal ring and the N-terminal region for activity,¹⁰² but without a known molecular target, these SAR studies are not fully interpretable.

Nisin and subtilin also inhibit the outgrowth of bacterial spores at sub-nanomolar concentrations, which are lower than their antibacterial activities.^103–105 Initially Dha5 was reported to be important for this activity, but later studies refuted this finding.⁷⁷ Mechanistic investigations suggest that inhibition of spore outgrowth is also mediated by lipid II binding. Nisin variants in the hinge region (N20P/M21P and M21P/K22P) capable of binding lipid II but deficient in pore formation retained antimicrobial activity against vegetative Bacillus anthracis cells but did not inhibit spore outgrowth. This observation suggests that pore formation is critical for the latter activity.¹⁰⁶ Furthermore, nisin did not prevent spore germination and required germination to inhibit spore outgrowth.¹⁰⁷

Several class I lanthipeptides exhibit weak or no antibacterial activity, but instead act as antifungal or morphogenetic peptides. Pinensin A and B (10 and 11) are the first characterized lanthipeptides produced by the Bacteroidetes Chitinophaga pinensis (Figure 5).¹⁰⁹ The pinensins exhibit weak antibacterial activity but display antifungal activity with MICs in the micromolar range against yeast and filamentous fungi.¹⁰⁹ Although the pinensin MOA remains unknown, it is hypothesized to have a novel biological target (for RiPPs) due to the specificity for fungi.¹⁰⁹

Figure 5.

Chemical structures of SapT (9) and pinensin A (10) and B (11). The stereochemistry of the MeLan rings in SapT (9) was determined to be D-allo-L-methyllanthionine, the first lanthipeptide with such stereochemistry.¹⁰⁸

SapT (9, Figure 5) produced by Streptomyces tendae, is another example of a lanthipeptide lacking robust antibacterial activity.¹¹⁰ The compound has morphogenetic activity that helps facilitate the formation of aerial hyphae in its producing organism and is required for spore formation.¹¹⁰ SapT is amphipathic, was the first lanthipeptide identified to contain D-allo-L-methyllanthionine,¹⁰⁸ and acts as a biosurfactant. This activity is shared by a number of class III lanthipeptides including SapB (section 2.3), and SapT is able to restore hyphae formation and sporulation in Streptomyces coelicolor bld mutants that are unable to produce SapB.^110,111

2.2. Class II lanthipeptides

Similar to the nisin group of class I lanthipeptides, a structurally unrelated subset of class II lanthipeptides target the cell wall biosynthetic intermediate lipid II. Class II lanthipeptide binding to lipid II can be predicted by the presence of a sequence motif that was first identified in the C-ring of mersacidin (12) produced by Bacillus sp HIL Y-85,54728 (Figure 6).¹¹² The highly conserved Glu in this ring (Asp in some structurally related compounds) is critical for antibacterial activity, but unlike the NMR structure of nisin bound to lipid II, currently high-resolution structural information is not available for the binding mode of mersacidin and related compounds to lipid II.¹¹³ Using NMR spectroscopy, mersacidin was shown to undergo a conformational rearrangement upon lipid II binding, with the carboxyl group of Glu17 thought to interact with lipid II through a Ca²⁺ mediated bridge.¹¹⁴ This hypothesis is supported by the loss of activity upon mutagenesis of Glu17 to alanine as well as by the requirement of Ca²⁺ for activity.^114,115 Unlike nisin, mersacidin requires the GlcNAc of lipid II for binding and does not bind to lipid I, and mersacidin lacks the ability to form pores.^112,116

Figure 6.

Structures of the class II lanthipeptides mersacidin (12), lacticin 481 (13), bovicin HJ50 (14), nukacin ISK-1 (15), plantaricin C (16), and an alternative representation of plantaricin C (17). The 6-amino acid ring containing an Asp/Glu (orange) that is proposed to be the lipid II binding motif is shown in blue. The report on the structure of plantaricin C 16 deduced by NMR spectroscopy acknowledges that the distances between the β-carbons of the lanthionines in the proposed A and D-rings are ~8 Å, whereas these corresponding distances for the B and C rings are ~4 Å.¹²⁷ We suggest an alternative possible ring pattern (17) based on the structures of mersacidin, lacticin 481, and nukacin ISK-1.^128,129

Analogs of the C-ring of mersacidin are also found in other class II lantibiotics such as the lacticin 481 (13) group of compounds that include bovicin HJ50 (14) and nukacin ISK-1 (15) (Figure 6), and the α-peptides of many two-component lantibiotics (section 2.2.1).^115,117,118 This conserved 6-amino acid containing ring with a Glu residue is also present in plantaricin C (16) in its originally reported structure. An alternative ring pattern would also contain a lipid II-binding ring (17, Figure 6). Although the overall ring patterns are varied, these compounds have all been shown to bind to lipid II using various biophysical techniques.^119–124 As discussed for the nisin group of lantibiotics, lipid II binding by class II lantibiotics is often sufficient for antimicrobial activity with a subset of compounds also inducing pore formation and others not affecting the membrane potential.^{118,121,125,126}

Variants of nukacin ISK-1, lacticin 481, and bovicin HJ50 have been investigated for bioactivity (Figure 6).^{121,130–136} As first shown for mersacidin, the highly conserved Glu/Asp in the ring C lipid II binding motif (A-ring for lacticin/nukacin/bovicin) is essential for bioactivity in all investigated compounds.^118,119,136 In addition, Lys residues in these compounds are important for bioactivity, possibly by increasing the affinity for the negatively charged lipids in the bacterial membrane.^{118,130,137,138} Disruption of any of the rings in bovicin HJ50, including the macrocycle generated by disulfide formation, resulted in complete loss of bioactivity.¹³⁶ The importance of these rings was also reported for lacticin 481,¹³⁹ and synthetic studies showed that the stereochemistry of the three (Me)Lan crosslinks is also critical for bioactivity. Diastereomers in which each of the three lacticin 481 rings was individually changed from the DL to the LL-stereochemistry were inactive.¹⁰² Use of a fluorescently labeled lacticin 481 analog showed localization in rod-shaped bacteria that is consistent with the reported localization of lipid II.¹⁴⁰

The MOA of nukacin ISK-1 (15) has been investigated by using the two-component antibiotic-sensing system LiaRS from B. subtilis. LiaRS is activated by cell wall-active antibiotics that interfere with the lipid II cycle, including nisin, nukacin ISK-1, and lacticin 481 (see also section 15.4 on epipeptides).^137,141 Nukacin ISK-1 variants were tested for interference with the lipid II cycle using this reporter system, demonstrating the importance of the A-ring, the conserved Asp in this ring, and the positively charged amino acids at the N-terminus.¹³⁷ NMR studies on nukacin ISK-1 showed the compound exists in two, interconverting conformations that differ in the relative orientation of the A and C rings (Figure 7).^123,142 Only one of the two conformers binds to lipid II and a model was proposed in which amino acid residues in ring A are involved in lipid II binding via hydrogen bonding and that residues in ring C then engage in hydrophobic interactions, possibly with the prenyl chain of lipid II. Disruption of the C-ring abolished lipid II binding, and individual replacement of the three Phe in the C-ring with Ala also eliminated lipid II binding as suggested by the LiaRS reporter assay.¹³⁷ It was hypothesized that similar movement of ring C could also explain the dynamic behavior reported for lacticin 481 and bovicin HJ50.^118,128 Independently, incorporation of non-canonical amino acids in lacticin 481 resulted in variants with improved bioactivity and inhibition of the transglycosylation reaction by the penicillin-binding protein (PBP) 1b. These studies showed that the increased antibacterial activity was caused by improved binding to lipid II, the substrate for PBP1b.^121,132 The positions at which improvements were observed were again the aromatic amino acids in the C-ring that were also identified in nukacin ISK-1s as important for lipid II binding, and the non-canonical amino acids that were incorporated at these sites had aromatic side chains, like the Phe residues in nukacin ISK-1.

Figure 7.

Two different conformations of nukacin ISK-1 that interconvert on the timescale of seconds. Both structures were solved by NMR spectroscopy (PDB IDs: 5Z5Q and 5Z5R).¹²³

Like the nisin group of class I lanthipeptides, select class II lanthipeptides also have autoinduction activity as shown for mersacidin, bovicin HJ50, and other members of the lacticin 481 group, but not nukacin ISK-1.^143–147 At present, the molecular details of mersacidin induction of its own biosynthesis are not known.¹⁴³ Saturation mutagenesis on bovicin HJ50 (14) identified several charged and hydrophobic amino acids in ring B as well as two Gly residues at positions 4 and 23 as critical for recognition by its autoinducing receptor kinase BovK.¹³⁶ Surprisingly, the A-ring, which is critical for bioactivity (see above), is dispensable for autoinduction, but the linear N-terminal sequence is required. Experiments with biotinylated bovicin HJ50 showed that the lanthipeptide interacts with the membrane domain of BovK and not the cytosolic domain. Additional mutagenesis studies suggested that a conserved hydrophobic region in the sixth transmembrane segment of BovK may be responsible for binding the lanthipeptide.¹³⁶

2.2.1. Two-component class II lanthipeptides

A subset of class II lanthipeptides are two-component systems wherein two peptides work synergistically to elicit bioactivity. For most examples, these peptide pairs have been termed the α and β-peptides. The best-studied examples in terms of MOA are the enterococcal cytolysin (hereafter cytolysin), lacticin 3147 (18 and 19, Figure 8), and haloduracin (20 and 21). Although cytolysin is the longest-known two-component lanthipeptide,¹⁴⁸ we will first discuss lacticin 3147 and haloduracin because their MOAs have similarities with the class II lanthipeptides discussed thus far.

Figure 8.

Structures of select two-component lanthipeptides. Lacticin 3147 composed of Ltnα (18, also called LtnA1) and Ltnβ (19, also called LtnA2), haloduracin composed of Halα (20) and Halβ (21), and staphylococcin C55 consisting of SacAα (22) and SacAβ (23). The proposed lipid II binding motifs are colored blue with the Glu that is critical for antibacterial activity in orange.

Lacticin 3147 (Figure 8) produced by L. lactis has antimicrobial activity against Gram-positive bacteria including other L. lactis strains, Listeria monocytogenes, and B. subtilis.^149–151 Like many other lanthipeptides, lacticin 3147 disrupts the cell membrane potential.¹⁵⁰ The α-peptide (Ltnα, 18) possesses the lipid II binding motif discussed above for mersacidin and has been shown to interact with lipid II.^120,152 Although the α-peptide binds lipid II, it is not able to inhibit cell wall biosynthesis on its own. Upon the addition of the β-peptide (Ltnβ, 19), pores with 0.6 nm diameter are formed.¹²⁰ Stoichiometry studies suggest that two-component lanthipeptides act in a 1:1 stoichiometry wherein the α peptide binds first followed by recruitment of the β-peptide. The structure of the pore forming complex has yet to be characterized.^120,153 A model has been proposed in which Ltnα binds to the pyrophosphate moiety of lipid II via hydrogen bonds to the backbone amide groups in the lipid II-binding motif.¹⁵² An analog of Ltnβ in which the MeLan were replaced by Lan using chemical synthesis retained the ability to synergistically kill bacteria in the presence of wild type Ltnα, albeit with 100-fold reduced potency, but a synthetic analog in which the thioether bridges were replaced by ether linkages in Ltnβ lost the synergistic activity with wild type Ltnα.^154,155 Extensive SAR (structure activity relationship) data has also been reported using variants generated by mutagenesis including Ala scanning and saturation mutagenesis of Ltnα and Ltnβ. Unfortunately, most of these studies did not distinguish between mutations that impacted biosynthesis and those that impacted antimicrobial activity.^156,157 Nevertheless, mutations that disrupted the ring structures resulted in loss of activity with the exception of the A-ring of Ltnα.¹⁵⁸ However, the A-ring does endow the peptide with enhanced resistance to thermal and proteolytic degradation.¹⁵⁹ Furthermore, Glu24 in the lipid II-binding motif of Ltnα is critical for bioactivity.¹⁵⁷ As for the single component mersacidin-like molecules discussed above, a molecular explanation for the importance of the Glu is currently not available.

Both peptides of lacticin 3147 contain D-Ala residues that are generated by post-translational modification of L-Ser. The importance of the three D-Ala in lacticin 3147 was confirmed by systematic substitution of these residues by L-Ala, which diminished both the production and the bioactivity. For Ltnα, the reduced production prevented activity determination. In Ltnβ, a single D to L conversion resulted in a 4-fold loss in activity, and a double mutation resulted in a 16-fold loss in activity.¹⁶⁰ Additionally, combination of the individual peptides of lacticin 3147 with those from another two-component lanthipeptide, staphylococcin C55 (22 and 23), was examined. The lacticin 3147 and staphylococcin C55 peptides contain α-peptides with 86% identity and β-peptides with 55% identity (Figure 8). Peptides from both compounds displayed synergism within native pairs as well as cross synergism when a hybrid pair was spotted with similar, low nanomolar activity as the original pairs.¹⁶¹ Given the differences in sequence between Ltnβ (19) and SacAβ (23) (Figure 8), these data suggest that the synergistic activity has some flexibility with respect to the β-peptide.

Another well-studied, two-component lanthipeptide is haloduracin, the first RiPP discovered by genome mining produced by Bacillus halodurans C-125.^162,163 Haloduracin has sequence homology with the lacticin 3147 peptides including the mersacidin lipid II-binding motif in Halα (20). The haloduracin α and β-peptides have optimal bioactivity when used in a 1:1 ratio and like the proposed lacticin 3147 model, the α-peptide binds first to a target with the β-peptide required for inducing pore formation.^120,164 In vitro inhibition studies of the transglycosylation reaction catalyzed by PBP2b provided direct evidence for haloduracin α binding to lipid II and demonstrated that the binding ratio for pore formation is 1:2:2 lipid II:Halα:Halβ.^122,164 Moreover, microscopy studies with fluorescently labeled Halα or Halβ resulted in localization of the α peptide primarily at sites of cell division and in punctate patterns along the long axis of bacilli.¹⁴⁰ This localization is similar to that reported for lipid II. Conversely, Halβ was localized nonspecifically to the cell surface in the absence of Halα but formed the same specific patterns discussed above when co-administered with its partner. Using two-color labeling, colocalization of both components of the two-component lantibiotic was observed. These in vivo data support a model in which the α component first recognizes lipid II, followed by the recruitment of the β-component, which aids in pore formation.

Extensive SAR experiments have been reported for both the Halα and Halβ peptides.¹⁶⁵ The C ring and Glu22 of Halα are essential for activity, whereas the A ring of Halα and the C and D rings of Halβ were determined to be important for activity but not essential. The B ring of Halα is not important despite a high level of conservation. Finally, unlike the findings with bovicin HJ50 described above, the disulfide of Halα was not important for activity but was important for stability.¹⁶⁵

Cytolysin, produced by Enterococcus faecalis, was the first reported example of a two-component lanthipeptide.¹⁶⁶ Cytolysin is comprised of a large and small subunit denoted as CylL_L” (24) and CylL_S” (25), respectively (Figure 9), which display synergistic bioactivity against both bacterial and eukaryotic cells.¹⁶⁷ In addition, the CylL_S” peptide serves as a quorum-sensing molecule that activates the cytolysin operon, leading to increased production of the cytolysin peptides.¹⁶⁸ Cytolysin has been known to be a virulence factor since the 1930s and the presence of the biosynthetic locus contributes greatly to the lethality of E. faecalis infections.¹⁴⁸ Recently, the toxin was directly correlated to the deleterious outcome of alcoholic liver disease.¹⁶⁹ While the details of the mechanism of cell lysis remain unknown, SAR studies have provided insight into the features important for cytotoxicity. The rings of both components of cytolysin are required for activity.¹⁷⁰ As noted previously, the canonical stereochemistry of (methyl)lanthionines is DL, which refers to D-stereochemistry at the α-carbon of the former Ser/Thr residues and L-stereochemistry at the former Cys residues (Figure 2). The cytolysin peptides were the first lanthipeptides reported to deviate from this canonical stereochemistry as the A ring of CylL_S” and the A and B-rings of CylL_L” were shown to contain LL-(Me)Lan.¹⁷¹ The stereochemistry of the A ring of the CylL_S” peptide is important for bioactivity, as the epimer with DL-stereochemistry in the A-ring decreased its antimicrobial activity 20-fold when combined with CylL_L”.¹⁷² However, its hemolytic activity was unaltered providing support for a model in which the two activities have different SAR and may involve different targets.¹⁷³ A recent Ala-scanning study in which all residues in both peptides were individually substituted also suggests that the two activities of cytolysin have different SAR.¹⁷⁰

Figure 9.

(A) Shorthand structures of cytolysin composed of CylLL” (24) and CylLS” (25). Sites with non-canonical LL-stereochemistry are annotated with asterisks (see Figure 2). (B) Two conformations of CylLL” in the bent and extended positions elucidated by solution NMR 28 spectroscopy (PDB ID: 6VGT; BMRB 30710). (C) NMR solution structure of CylLS” (PDB ID: 6VGT; BMRB 30702).

The large subunit CylL_L” binds to erythrocytes with seven-fold higher affinity than the small subunit CylL_S”; this may either suggest it recognizes a different molecular target or reflects a higher affinity for the hydrophobic membrane environment.^170,174 Recently the NMR structures of CylL_L” and CylL_S” were determined in methanol showing that the large subunit forms two helices that are probably held in place by the lanthionine crosslinks.¹⁷⁵ The two helices are separated by a flexible, Gly-rich sequence that results in a series of structures that differ in the relative orientation of the two helices (Figure 9B), with the colinear arrangement sufficient in length to form a membrane-spanning pore. This hinge arrangement is reminiscent of the hinge region between the C and D rings of nisin that is critical for pore formation (Figure 3A).

2.2.2. Phosphatidylethanolamine-binding class II lanthipeptides

A distinct group of lanthipeptides bind phosphatidylethanolamine (PE) in a 1:1 ratio.^176,177 These compounds contain a rare lysinoalanine crosslink and usually also a hydroxylated Asp (Figure 10). One example is cinnamycin (26, previously Ro 09–0198), a class II lanthipeptide produced by various Streptomyces sp. Cinnamycin induces transbilayer phospholipid movement in a PE-concentration dependent manner, leading to membrane permeabilization and cytotoxicity.^177,178 The mechanism by which cinnamycin accesses PE that is located in the inner leaflet of the membrane remains incompletely resolved. The interaction between PE and cinnamycin was characterized by NMR spectroscopy demonstrating that the hydrophobic pocket created by residues Phe7 through Cys14 as well as the erythro-3-hydroxy-L-aspartic acid at position 15 are important for the high selectivity of ligand recognition (Figure 11)^179,180 and the high observed affinity (K_d = 20 nM by ITC).¹⁸¹ Experiments with other lipids showed that an acylated glycerol backbone and a primary amine group are necessary.^176,182 The length of the acyl chains are not critical for affinity but at least one acyl chain is needed as no binding was observed with glycerophosphorylethanolamine by ITC.¹⁸¹ Recent modeling experiments suggest that more hydrogen bonds may be present than assigned in the NMR structure.¹⁸³ Binding to PE also results in antiviral activity of cinnamycin and the structurally related duramycin (28) against herpes simplex virus type 1 (HSV-1) by inhibiting viral proliferation.¹⁸⁴

Figure 10.

Chemical structure of cinnamycin (26), and its shorthand rendition (27). Structurally related molecules are shown with lysinoalanine crosslink shown in orange, (methyl)lanthionine rings shown in blue and hydroxyaspartate residues indicated in purple: duramycin (28), ancovenin (29), divamide A (30), and divamide B (31). Asp-OH, (3R)-hydroxy-aspartate.

Figure 11.

NMR structure of cinnamycin (26) bound to C12-lysophosphatidylethanolamine (labeled as PE). The carbons of hydroxyaspartate 15 (HyAsp15) are indicated in yellow and the carbon atoms of PE in pink (PDB ID: 2DDE). For clarity, only one carbon is shown of the C12 acyl chain of PE.¹⁷⁹

Other lanthipeptides with similar ring patterns as cinnamycin have been reported to possess a wide array of bioactivities. These include the duramycins (e.g. 28), ancovenin (29), and the divamides (30, 31) (Figure 10). Duramycin was discovered as an inhibitor of phospholipase A₂ with an IC₅₀ around 1 μM. This activity is likely an indirect effect caused by the binding of the substrate PE rather than direct interaction with the protein itself.^185–188 PE binding is also likely responsible for efflux of chloride from airway epithelium cells, which has led to investigation of duramycin as a potential treatment for cystic fibrosis.^189–193 Similar to cinnamycin, the divamides and duramycin display antiviral activity. Duramycin efficiently inhibits cellular entry of West Nile, Dengue, and Ebola viruses in a mechanism mediated by phosphatidylserine (PS) receptors like T-cell immunoglobulin mucin domain protein 1 (TIM1). These various antiviral activities are likely directly related to PE binding because the virions of flaviviruses and filoviruses utilize PE for cell entry by TIM1.¹⁹⁴ Disrupting PE association with PS receptors may serve as a promising broad-spectrum antiviral strategy. Divamides exhibit anti-human immunodeficiency virus (HIV) activity, thought to be mediated through lipid binding.¹⁹⁵

The specific binding of PE by duramycin/cinnamycin and their analogs with high affinity has been used extensively to detect PE on cultured cells and in animal models.^196,197 Fluorescently labeled analogs have been shown to be able to detect apoptotic cells,¹⁹⁸ and analogs labeled with (99m)Tc have been used to detect various types of cell death,^199–201 tumors,^202,203 ionizing irradiation-induced tissue injuries,^204,205 atherosclerotic plaques,²⁰⁶ and myocardial ischemic/reperfusion injury²⁰⁷ in various animal models.

Ancovenin is an inhibitor of the angiotensin 1-converting enzyme (ACE). Such inhibitors are commonly used in the treatment of high blood pressure, and the compound inhibits rat lung ACE with an IC₅₀ of 870 nM.²⁰⁸ Ancovenin was not growth suppressive against B. subtilis ATCC 6633 or Staphylococcus aureus IFO 12732.²⁰⁹ The mechanism of this divergent activity within the cinnamycin family of peptides has not been investigated.

2.3. Class III lanthipeptides

In contrast to class I and II lanthipeptides, most known class III lanthipeptides exhibit weak or no antibacterial activity. Labyrinthopeptins are class III lanthipeptides that contain a labionin moiety first identified in 2010 (Figures 2 and 12).²¹⁰ Labyrinthopeptin (Laby) A1 (32) and A2 (33) demonstrate antiviral activity against a variety of viruses, including Herpes simplex, Zika, and hepatitis C.^210–212 The broad spectrum antiviral activity is thought to be in part exerted through the binding of PE leading to viral membrane disruption.²¹¹ N-terminally hexynoyl-derivatized LabyA1 and A2 fluorescently labeled using click chemistry was used to demonstrate PE binding. These modified peptides also displayed weak affinity for phosphatidylcholine (PC) and sphingomyelin, two molecules containing an ethanolamine-derived head group.²¹¹ Evaluation of hexynoyl-LabyA1 activity in the presence of PE-containing vesicles led to an 8-fold decrease in antiviral activity whereas vesicles containing PC had no effect on potency. Hexynoyl-LabyA1 and A2 peptides induced leakage in a PE-dependent and concentration dependent manner for vesicles of varying lipid compositions, monitored by release of fluorescent molecules within the vesicles, supporting the hypothesis that PE is a molecular target.²¹¹

Figure 12.

Structures of labyrinthopeptin A1 (32) and labyrinthopeptin A2 (33). For the structure of labionin, see Figure 2.

Hexynoyl-LabyA1 and A2 act weakly synergistically in a 1:1 ratio, featuring up to 2-fold higher PE-binding affinities than the individual peptides.²¹¹ While shown to increase activity when together, the molecular interaction between the Laby peptides has yet to be structurally characterized and it is unclear if these compounds interact directly or via PE in a sandwich-like manner. While it has been proposed that the virolytic activity of the Laby peptides could be exerted through PE binding, other PE-binding peptides such as duramycin, cinnamycin, and the divamides (section 2.2.2) are believed to exhibit antiviral activity by preventing PE-mediated infection.

Labyrinthopeptin A1 demonstrated anti-HIV and -HSV activity.^213,214 In HIV transmission models, LabyA1 inhibited cell-free, cell-to-cell, and dendritic cell-specific intercellular adhesion molecule 3-grabbing non-integrin (DC-SIGN)-mediated viral infection. Time-of-drug addition studies and the lack of inhibition of viral binding to CD4⁺ T cells suggest that LabyA1 targets viral entry,²¹³ like duramycin.¹⁹⁴ Surface plasmon resonance experiments identified an interaction between LabyA1 and the gp120 protein of the HIV envelope, but the peptide did not interact with the HIV cellular receptors CXCR4 and CCR5. LabyA1 exhibits synergistic effects when given alongside a number of antiviral therapeutics such as tenovir or acyclovir and also maintains activity against acyclovir-resistant strains.²¹³ The interaction between LabyA1 and gp120 has not yet been characterized; however, LabyA1 is active against HIV strains resistant to carbohydrate-binding agents, suggesting that it is not targeting the N-linked glycans of gp120.²¹³ Time-of-drug addition studies indicate that it is also an entry inhibitor for HSV,^213,214 most likely again mediated by PE binding.

In addition to antiviral activities, LabyA2 is effective in the treatment of neuropathic pain in animal models, making it the first lanthipeptide reported to have antiallodynic activity.²¹⁰ NAI-112 (34), a glycosylated class III lanthipeptide produced in Actinoplanes sp. (Figure 13), also showed efficacy in the treatment of neuropathic pain.¹³ In mouse model studies, both compounds displayed a decrease in allodynia while NAI-112 also reduced hyperalgesia.^13,210 While neither LabyA2 nor NAI-112 have a known MOA for these activities, it has been proposed that NAI-112 may interact with the vanilloid pathway,¹³ and that the molecule blocks pain sensation by decreasing the levels of lysophosphatidic acid (LPA) and increasing the levels of phosphatidic acid.²¹⁵ LPA activates the vanilloid receptor 1 (also known as capsaicin receptor and TRPV1) triggering pain. NAI-112 also exhibits modest antibiotic activity with micromolar MIC values against various staphylococci and streptococci.^13,216 NAI-112 was initially identified as a cell wall inhibitor,²¹⁶ with resistance mutants suggesting that the peptidoglycan intermediate lipid II (Figure 3) may be a target.²¹⁵

Figure 13.

The structure of NAI-112 (34). The sugar stereochemistry has not yet been determined.

Avermipeptin B from Streptomyces actuosus was identified by genome mining and is one of the few class III lanthipeptides exhibiting potent antibiotic activity.²¹⁷ Avermipeptin B displays activity against S. aureus, E. faecalis and B. subtilis with MICs in the nanomolar range.²¹⁷ To date, the mechanism by which this compound exerts its antibacterial activity has not been examined.

A subset of class III lanthipeptides exhibit morphogenic activity similar to the class I lanthipeptide SapT (see section 2.1). SapB (35) produced by S. coelicolor (Figure 14) from the bld locus facilitates aerial hyphae formation in the producing organism through reduction of surface tension.^218–221 SapB-deficient mutants cannot form aerial hyphae and are unable to form spores.²²⁰ Restoration of aerial growth can be achieved with the addition of purified SapB; however, bld mutants fail to sporulate even with added SapB, indicating the timing of SapB production for spore development is critical.^221,222 Molecular modeling of SapB suggests that the structure may be amphiphilic, supporting proposed biosurfactant functionality. Several attempts at structural characterization using NMR spectroscopy have been unsuccessful due to aggregation and insolubility of the compound.²¹⁸

Figure 14.

Class III lanthipeptides. (A) Structures of SapB (35), AmfS (36), and catenulipeptin (37). (B) Sequence alignment of the core peptides of catenulipeptin and SapB illustrating the similar constellation of Cys and Ser residues, but different outcome of the cyclization process, resulting in labionins for the former and lanthionines for the latter.

Several other peptides with sequence homology to SapB have been identified including AmfS (36), produced in Streptomyces griseus (Figure 14).^223,224 AmfS is also important for aerial hyphae formation and sporulation.^223,225 Mutagenesis studies of AmfS revealed that all conserved post-translationally modified residues selected for replacement (Ser3, Cys10, Ser13, Ser16, and Cys20) were essential for activity.²²⁵ Ser6, although conserved and converted to Dha, was not targeted by mutagenesis in this study. Additional variants involving select non-conserved residues within the rings (L7A, L8V, V9S, and L19V) and the removal of the tail region (residues 21–22) still facilitated aerial hyphae formation, indicating that these residues do not play an important role in this process.²²⁵ However, SapB with V9L and L19T substitutions lost the ability to restore aerial mycelium formation in neighboring bld mutant colonies.

Catenulipeptin (37) has sequence homology with SapB but contains labionins instead of the lanthionines in SapB (Figure 14). Unlike SapB, catenulipeptin did not induce acceleration of aerial hyphae formation or restore formation in bld mutants in S. coelicolor, suggesting it may not function like SapB.²²⁴ Alternatively, catenulipeptin may be inactive in S. coelicolor as it is not its producing organism. Some biosurfactants are specific to the producing organism and are not cross compatible, with some even acting antagonistically to suppress aerial growth in other organisms,²²⁶ suggesting that a molecular target may be involved. It remains to be investigated if catenulipeptin shows morphogenetic activity in its producing organism, Catenulispora acidiphilia.

2.4. Pearlins

Pearlins are a recently defined RiPP class characterized by the post-translational nonribosomal addition of amino acids to the C-terminus of a precursor peptide in an aminoacyl-tRNA dependent process.²²⁷ These amino acids are then further enzymatically modified and eventually cleaved from the peptide to produce a variety of small molecule natural products including pyrroloquinoline alkaloids (38–41) and 3-thiaglutamate (42, Figure 15). Although pearlins do not contain (Me)Lan residues, they are discussed in the lanthipeptide section because their class-defining biosynthetic enzymes are related to class I lanthipeptide synthetases.

Figure 15.

Chemical structures of ammosamides A (38), B (39) and C (40), lymphostin (41), and 3-thiaglutamate (42).

The first pyrroloquinoline alkaloid discorhabdin C was isolated from the sea sponge Lantrunculia apicalis in the early 1980s.^228,229 Since then, pyrroloquinoline alkaloids have been isolated from a wide variety of organisms including bacteria (ammosamides), marine sponges (damirone B), and terrestrial fungi (sanguinone A). The bacterial biosynthesis remained enigmatic until recent investigations determined that many of these natural products belong to the pearlin RiPP class.^230–232 The two currently known bacterial pyrroloquinoline alkaloid families are the ammosamides (38-40) and lymphostin (41) (Figure 15).²²⁹

While a variety of ammosamide derivatives have been isolated (from Streptomyces CNR-698) they are believed to be all derived from ammosamide C (40), which is the naturally produced structure. All other ammosamides have been reported to be artifacts of isolation methods during which ammosamide C was exposed to oxidative conditions and various nucleophiles.²³³ While these compounds are derived from a single compound, the ammosamides exhibit diverse bioactivities.²³³ Ammosamides A and B (38 and 39) display cytotoxicity to a number of human cancer cell lines including colon carcinoma.^234,235 Ammosamide B conjugated to a fluorescent probe underwent efficient cellular uptake and its molecular interaction with one or more targets was suggested to be either through strong noncovalent interactions or a covalent mechanism owing to the inability to remove the labeled compound through washing procedures. The labeled compound was then used for affinity co-purification from cell lysates, and protein MS/MS analysis identified the target as a member of the myosin family, a motor protein involved in muscle contraction and cytoskeletal structure.²³⁴ This finding was later confirmed through in vitro myosin II labeling.²³⁴ Further experiments indicated that ammosamides A and B interact with several other myosin family members.^234,236 A crystal structure of ammosamide 272 (43) co-crystalized with the myosin 2 heavy chain (PDB 4AE3) from Dictyostelium discoideum revealed the molecular mechanism of binding as noncovalent in an allosteric region of the motor domain (Figure 16).

Figure 16.

(A) Chemical structure of ammosamide 272 (43). (B) Crystal structure showing the interaction between ammosamide 272 (carbons labeled in green) and the myosin 2 heavy chain (PDB ID: 4AE3). Side chains of myosin residues interacting with ammosamide 272 are shown in yellow.

Ammosamides A and B also inhibit quinone reductase 2 (QR2) with sub-micromolar IC₅₀ values.²³⁷ QR2 is a flavin-dependent enzyme involved in cellular protection from quinone species. A co-crystal structure of ammosamide B and QR2 was obtained (PDB ID: 3UXE) (Figure 17), demonstrating that the compound engages in a stacking interaction with the bound flavin and forms a hydrogen bond network to Asn161 and Thr71 of QR2.²³⁸ Derivatives of ammosamide B have been synthesized and are potent inhibitors of human QR2 with IC₅₀s as low as 4.1 nM.²³⁸

Figure 17.

Crystal structure of ammosamide B (carbon atoms in cyan) bound to QR2 (purple and green; PDB 3UXE). Asn161 is located on monomer A (purple), while Thr71 is located on monomer B (green). The FAD carbons are shown in orange.

In addition to the reported cytotoxicity of ammosamide C against mammalian cells, it also exhibits antimicrobial activity against Bacillus oceanisediminis. No antibacterial activity was detected for ammosamide A or B against this strain.²³³ Neither the biological target nor the MOA has been determined.

Lymphostin (41, Figure 15) produced by Streptomyces sp. KY11783, is another pyrroloquinoline alkaloid identified by activity-based screening for inhibitors of lymphocyte kinase (Lck), a tyrosine kinase produced in lymphoid cells involved in T-cell activation.²³⁹ Lymphostin inhibited Lck with an IC₅₀ of 50 nM with the IC₅₀ value against a murine myeloid leukemia cell line of 0.16 μM.²³⁹ In a mixed lymphocyte reaction measuring T-cell activation, lymphostin inhibited Lck with an IC₅₀ of 9 nM. The compound also inhibited phosphatidylinositol 3-kinase with an IC₅₀ of 1 nM.²⁴⁰ During efforts to characterize the lymphostin biosynthetic gene cluster, analogs were generated and investigated for inhibition of another prominent kinase, mammalian target of rapamycin (mTOR). Lymphostin and derivatives showed 0.8–1.8 nM inhibitory activity against mTOR as well as 14–700 nM cytotoxicity against the human prostate and breast cancer cell lines, LNCap and MDA-468, respectively.²⁴¹

3-Thiaglutamate (42) is the only other currently reported pearlin (Figure 15).²³¹ The biosynthetic gene cluster was discovered in the plant pathogen Pseudomonas syringae as well as other pathogenic pseudomonads. The compound is unstable, and it has not been determined to be the active product. Glutamate was recently shown to be a messenger that is involved in systemic defense responses in plants that involves the glutamate receptor,²⁴² and hence 3-thiaglutamate or a derivative thereof could be an antimetabolite to interfere with plant defenses.^231,243

2.5. Lanthidins (Class V lanthipeptides)

Lanthidins contain post-translational modifications commonly present in linaridins and lanthipeptides giving rise to their name. The first reported member, cacaoidin (44) isolated from Streptomyces cacaoi, contains an N-terminal N,N-dimethyl lanthionine, O-glycosylation of a Tyr, and several D-amino acids (Figure 18). While linaridins commonly contain N-terminal bis-N-methylation, they are defined by the presence of dehydrobutyrine residues, which are not present in cacaoidin.²⁴⁴ This observation combined with the presence of lanthionine and isolation of additional compounds that share these structural features led to the renaming of lanthidins as class V lanthipeptides.^10,245–247

Figure 18.

Class V lanthipeptides cacaoidin (44) and lexapeptide (45). The D-Ala that was demonstrated to be important for antibacterial activity of lexapeptide is highlighted in red.

Cacaoidin was isolated via bioactivity-guided screening and showed activity against Firmicutes including methicillin-resistant S. aureus (MRSA) and Clostridium difficile. Investigation of the MOA of cacaoidin utilizing a lacZ reporter assay suggested the compound targets cell wall biosynthesis due to the induction of the LiaRS bioreporter (sections 2.2 and 15.4). This hypothesis was supported by experiments showing that the response was mitigated by supplementation with exogenous lipid II. The authors used these studies to suggest a lipid II binding mechanism, similar to other known lanthipeptides (see sections 2.1 and 2.2).²⁴⁴ Structural studies have yet to be performed on the proposed binding of cacaoidin to lipid II.

Another class V lanthipeptide, lexapeptide (45, Figure 18), was isolated from Streptomyces rochei, and exhibits activity against MRSA and methicillin-resistant S. epidermidis (MRSE).²⁴⁶ Mutational studies identified the D-Ala in the compound as important for activity with the L-isomer having reduced activity against a subset of tested bacterial strains. No further MOA studies have been published to date.

2.6. Lipolanthines

Lipolanthines are lipopeptides in which the peptide portion is ribosomally synthesized. This class is characterized by the presence of labionin/avionin(s) (Figure 2) and an N-terminal lipid moiety. Microvionin (46), the first lipolanthine isolated from Microbacterium arborescens (Figure 19), has activity against MRSA and Streptococcus pneumoniae.²⁴⁸ Goadvionin is another lipidated class III lanthipeptide and shows similar antibacterial activity.²⁴⁹ The precise MOA of lipolanthines requires further study.

Figure 19.

Structure of microvionin (46).

3. RiPPs generated by YcaO superfamily enzymes

YcaO enzymes activate the amide backbone of peptides/proteins to catalyze diverse reactions through the intermediacy of a phosphorylated hemiorthoamide.²⁵⁰ These enzymes induce an intra- or inter-molecular nucleophilic attack at an amide carbonyl to generate an oxyanion, which is then O-phosphorylated in an ATP-dependent manner.²⁵¹ The nature of the final product depends on the identity of the nucleophile. Intramolecular attack by the side chains of Ser/Thr/Cys results in azoline formation and is involved in the biosynthesis of the compounds discussed in sections 3.1, 3.2 and 4. Intramolecular attack by the N-terminal amine results in lactamidine formation as observed in the bottromycins discussed in section 3.3. Finally, intermolecular attack, by an external sulfur donor, results in thioamide formation, exemplified by the thioamitides (section 3.4).

3.1. Linear azoline/azole-containing peptides (LAPs)

Linear azoline/azole-containing peptides (LAPs) are a class of non-macrocyclic RiPPs characterized by the presence of thiazol(in)e and/or (methyl)oxazol(in)e heterocycles.²⁵⁰ The cognate YcaO enzymes catalyze the cyclodehydration of Cys, Ser, and Thr residues to the corresponding azoline heterocycle, some of which undergo oxidation to the corresponding azole by an FMN-dependent dehydrogenase (Figure 20). As detailed in section 1.1, the peptide numbering scheme used for all RiPPs in this section will denote the first residue in the core peptide as residue “1”.

Figure 20.

Generic scheme for the biosynthesis of azol(in)e on a peptide substrate by a trifunctional heterocyclase. In LAP biosynthesis, the C-D complex is composed of an E1-like protein (yellow) and an ATP-dependent YcaO protein (green), which together perform cyclodehydration of Cys, Ser, and/or Thr residues. The B protein (orange) is an FMN-dependent dehydrogenase that generates the azole. Abbreviations for (methyl)azol(in)es heterocycles used in this review are shown.

The toxic virulence factor of Streptococcus pyogenes, streptolysin S (SLS), is a seminal LAP. The ability of certain streptococci to hemolyze erythrocytes was first observed in the 1890s, and four decades later, SLS was identified as one of two toxins from Group A Streptococcus (GAS) that could lyse mammalian erythrocytes.²⁵² The other toxin, streptolysin O (SLO), is a larger protein that forms pores in cholesterol-containing membranes. The cytolytic activity of SLS is broad and includes erythrocytes, leukocytes, thrombocytes, as well as membrane-defined organelles such as the mitochondria and lysosomes. SLS is bactericidal against streptococci, staphylococci, and bacilli, but does not affect bacteria that possess an outer membrane (e.g. Pseudomonas, Vibrio, Rhodospirillum).^253–257 This broad but nuanced cytolytic spectrum suggests SLS activity depends on membrane composition and/or access to the membrane.^256,257

Though the detailed MOA of SLS is not known, it has been proposed to form pores in target membranes. Early studies demonstrated that treatment of artificial liposomes with SLS made these membranes permeable to cations.²⁵⁸ SLS was later shown to induce the formation of large pores in sheep erythrocyte membranes,²⁵⁹ suggesting that cytolysis could be caused by shredding of the membrane. Beta-hemolysis was shown to be temperature dependent, as erythrocytes treated with SLS at 17 °C did not lyse, while treatment at warmer temperatures resulted in lysis.²⁶⁰ Additionally, complexes of SLS pre-mixed with erythrocytes were insensitive to protease degradation at elevated temperatures. However, at lower temperatures, the SLS-erythrocyte complex was sensitive to proteolytic degradation. These results suggest a MOA in which SLS localizes to the membrane surface, but only inserts into the membrane and forms pores at permissive temperatures.

Three mechanisms have been proposed to account for SLS-mediated virulence: (1) soft tissue damage, (2) host phagocyte damage, and (3) GAS translocation across the epithelial barrier. A transposon insertion variant GAS was unable to induce hemolysis in a mouse model of infection.²⁶¹ Further analysis mapped the insertion to the sagA promoter sequence, suggesting that expression of the SagA precursor peptide was essential for virulence. Subsequently, the role of SLS in tissue damage was shown using clinical isolates with group G Streptococcus (GGS) infections, whose SagA precursor peptide highly resembles that of GAS (Figure 21). In a mouse model of infection, injection with the GGS isolate resulted in necrotic ulcers, while injection of strains with a sagA deletion did not show this phenotype.²⁶²

Figure 21.

Amino acid sequences of the precursor peptides encoding SLS-like cytolysins. Shown are the predicted leader and core regions for SLS, clostridiolysin S (CLS), listeriolyin S (LLS), stapholysin S (StsA), and the group G Streptococcus (GGS) SLS-like cytolysin. Residues in the SLS sequence that when replaced with Ala abolished cytolytic activity are shown in blue. The exact structure of any of these RiPPs has yet to be determined.

Beyond hemolysis and tissue damage, S. pyogenes producing both SLS and SLO can induce macrophage cell-death through depolarization of the mitochondrial membrane and accumulation of reactive oxygen species.²⁶³ However, the individual contributions of each compound to these effects have not been characterized. Lastly, SLS has been implicated in more complex bioactivities observed during infection including paracellular invasion across the epithelial barrier. In vitro studies utilizing a Caco-2 cell permeability assay showed that SLS-producing GAS disrupted tight junctions and epithelial translocation, while SLS-deficient GAS did not demonstrate this translocation activity.²⁶⁴ These studies further demonstrated that intercellular junctions were not disrupted as a consequence of SLS-induced cell lysis or inflammatory cytokine production. The intercellular junctions were in fact degraded by the SLS-mediated recruitment of the endogenous cysteine protease calpain to the plasma membrane. The molecular mechanism for calpain recruitment to the plasma membrane by SLS has not yet been characterized. Given its role in promoting paracellular invasion, there may be some future promise in exploiting SLS scaffolds to develop oral therapeutics that would otherwise be poorly absorbed.

The mechanism of hemolysis has also been partially elucidated.²⁶⁵ In hypotonic phosphate buffered solution, introduction of SLS induced cellular swelling, indicating significant influx of water preceding lysis. Use of 6-methoxy-N-ethylquinolinium iodide as a fluorescent indicator also demonstrated that SLS triggered an influx of chloride ions. Subsequent experiments demonstrated that SLS-induced hemolysis occurred in a membrane cation transporter-dependent fashion. Anion exchanger 1 (AE1), also called band 3, is the most abundant protein in the erythrocyte membrane and mediates the exchange of chloride and bicarbonate anions. Treatment of erythrocytes with the AE1 inhibitors 4,4’-diisothiocyanatostilbene-2,2’-disulphonate (DIDS) and 4-acetamido-4’-isothiocyanato-2,2’-stilbenedisulphonic acid significantly reduced the hemolytic activity of SLS. Pretreatment of erythrocytes with antibodies specific to the exposed C-terminal epitope of AE1 also protected the cells from hemolysis. Synthetic peptides corresponding to this C-terminal fragment were also shown to block SLS activity in a dose-dependent fashion, supporting the hypothesis that SLS directly interacts with this region of AE1 to induce hemolysis. Furthermore, in mouse models of SLS-producing GAS bacteria, administration of DIDS significantly reduced the size of necrotic lesions. These findings suggest that SLS likely interacts with homologous ion channels to promote invasion of other cell types.²⁶⁵

The SLS core peptide contains 15 residues that can potentially undergo heterocyclization into (methyl)azol(in)es, and mutational studies have been used to dissect the SAR of the SLS core (Figure 21). The site-directed variants C1A, C4A and K30A of SLS did not demonstrate cytolytic activity (positions have been renumbered using standardized RiPP nomenclature; see Introduction).²⁶⁶ Follow-up studies identified additional residues that contributed to lytic activity.²⁶⁷ Ala substitutions at Cys8 or Ser16 abolished in vitro or in vivo bioactivity. Notably, substitutions with Pro (which like azol(in)es also possesses a five-membered ring) at these two positions rescued lytic activity in vitro. Activity was retained in an SLS variant in which all Cys residues were replaced with Pro, highlighting the importance of five-membered rings in imparting cytolytic activity. Truncation analysis of the SagA precursor demonstrated that the last 17 residues of the SLS core peptide were dispensable for hemolytic activity, but removal of the last 20 residues abolished activity.²⁶⁸ Thus, the minimal bioactive fragment of SLS consists of core residues Cys1– Ser11.

SLS-like virulence factors have also been identified in other bacteria (Figure 21) and each of these compounds show similarity in core sequences, exhibit similar bioactivities, and are likely to possess the same MOA. Clostridiolysin S (CLS), isolated from Clostridium botulinum showed in vitro hemolytic activity.^269,270 S. aureus RF112 is predicted to produce a LAP termed stapholysin S (StsA).²⁶⁹ Though the compound has yet to be isolated from the producing organism, it has been produced using a chimeric substrate approach and was found to be cytolytic.²⁶⁷ More recently, SLS-like virulence factors were identified in certain Borrelia burgdorferi sensu lato strains, which are responsible for Lyme disease.²⁶⁸ The precursor peptides identified in Borrelia (Spirochaetes) are shorter than their Firmicute homologs by lacking ~15 residues of the most C-terminal portion of the core peptide. The N-terminal portion of the core peptide, consisting of contiguous residues capable of azol(in)e formation are retained.

A subset of L. monocytogenes lineage I strains produce a hemolytic compound termed listeriolysin S (LLS).²⁷¹ LLS is expressed solely in the intestines of orally infected mice, and was shown to affect the growth of gut-associated bacteria and alter the composition of the gut microbiome.²⁷² Cellular fractionation experiments demonstrated that LLS was contained with the cell envelope of producing bacteria.²⁷³ Further evidence of cell-envelope association was provided using immunostaining of producing cells expressing LLS tagged with a human influenza hemagglutinin-derived epitope. Treatment of L. lactis with LLS-containing membrane fractions did not induce lysis, suggesting further metabolic activity is necessary for bactericidal activity. Co-culture experiments were performed wherein LLS-producers and L. lactis were separated by semipermeable membranes. Membranes with a pore size of 0.4 μm prevented L. lactis killing, while a pore size of 8 μm allowed passage of LLS and induced lysis of L. lactis cells. Fluorescent reporter strains further demonstrated that L. lactis cells in contact with LLS-producing cells had drastically increased doubling times, suggesting a general decrease in metabolic activity. Imaging studies with SYTOX blue and DiBAC₄ dyes demonstrated that LLS-treated cells lost membrane potential in a time dependent manner. Mass spectrometry-based proteomics on LLS-treated cells further demonstrated up-regulation of proteins involved in ATP synthesis, indicating that these cells increase ATP production to counteract membrane depolarization and ATP efflux.

The N^α,N^α-dimethylated LAP plantazolicin (47, Figure 22) also contains heterocyclic residues and similarly targets membranes. First isolated from Bacillus velezensis (reclassified from Bacillus amyloliquefaciens), plantazolicin was initially described as an antibiotic that inhibited the growth of B. subtilis, Bacillus cereus, and Bacillus megaterium at an extremely high dosage (1 mg in a spot-on-lawn assay).^274,275 The spectrum of activity was later reevaluated using precise microbroth dilution assays, which demonstrated plantazolicin is specific towards B. anthracis (MIC = 0.88–14.1 μM), while possessing no or weak activity against B. subtilis or B. cereus (> 57 μM).^276,277 Bioactivity was dependent on the presence of the N^α,N^α-dimethylation and the most N-terminal azole heterocycle. Acid hydrolysis of the lone MeOxH (residue 13) of plantazolicin led to a ~8-fold increase in MIC against B. anthracis Sterne, while an otherwise fully modified peptide lacking the N-terminal methyl groups negated all bioactivity.²⁷⁶ In liquid culture, the producing organism produces a plantazolicin side-product containing two azolines, which lacks antibiotic activity. The position of the azoline residues was not determined. The minimal fragment with a comparable MIC to plantazolicin is an N-terminal fragment (48) with N^α,N^α-dimethylation and five heterocycles (Figure 22). Surprisingly, this derivative demonstrated a wider growth inhibition spectrum that included additional strains of Bacilli and MRSA.²⁷⁸

Figure 22.

(A) Chemical structure of plantazolicin (47). The plantazolicin precursor peptide is shown above, and the core peptide is shown in bold. The numbering scheme for (47) is reflected in the labels above the core peptide. The sole methyloxazoline (blue) can be selectively acid hydrolyzed, which results in an 8-fold reduction in MIC against B. anthracis. The N^α-dimethylations (orange) are essential for activity. (B) Minimal synthetic plantazolicin fragment (48) (the C-terminal methyl ester was utilized to facilitate chemical synthesis and to better serve as a mimic of the amino acid that follows).

Observation of B. anthracis cells by differential interference contrast microscopy demonstrated that plantazolicin-treated cells underwent lysis, while the few remaining cells displayed compromised cell walls.²⁷⁶ Assays utilizing radiolabeled precursors revealed that plantazolicin treatment induced general perturbation of protein, fatty acid, and nucleic acid biosynthesis, but did not specifically inhibit cell wall biosynthesis.²⁷⁷ RNA-seq analysis of plantazolicin-treated cells showed significant upregulation of two genes, homologous to B. subtilis genes liaI and liaH, which are involved in the cell-envelope stress response (section 2.2 and 15.4). Two homologs of the B. subtilis membrane fluidity sensors DesK and DesR were also upregulated.^277,279 The identity of the upregulated genes implicate the cell membrane as the target of plantazolicin, and this hypothesis was supported by observation of a marked loss in membrane potential in B. anthracis treated with plantazolicin, even at concentrations 100-fold below the MIC.²⁷⁷ The localization of fluorescently modified plantazolicin was determined to be the cellular envelope by sub-diffraction limit microscopy.²⁷⁷

Spontaneous mutations in the cardiolipin synthase gene conferred resistance to plantazolicin in B. anthracis Sterne.²⁷⁷ Cardiolipin is a cone-shaped phospholipid that is found in bacterial membranes and the mitochondrial inner membrane.²⁸⁰ The role of cardiolipin in the plantazolicin MOA was further supported by findings that pre-treatment of B. anthracis with cardiolipin increased the MIC by up to 16-fold and that plantazolicin co-localized with cardiolipin in the cell membrane.²⁷⁷ However, exogenously supplied cardiolipin did not render B. subtilis or B. cereus susceptible to plantazolicin. Atomic level details of plantazolicin-mediated membrane perturbation remain unknown, but these data strongly implicate cardiolipin in the plantazolicin MOA.

Other LAPs are known to inhibit bacterial growth through membrane-independent MOAs. For instance, microcin B17 (49, MccB17, Figure 23) inhibits DNA synthesis and induces the bacterial SOS response in E. coli.²⁸¹ More specifically, treatment of target cells with MccB17 elicits rapid termination of DNA synthesis, and results in significant DNA degradation within 30 minutes of treatment. A resistance-conferring mutation was identified in the E. coli gyrB gene (corresponding to GyrB W751R), implicating GyrB as the molecular target of MccB17.⁴⁰ E. coli DNA gyrase is a tetramer composed of two copies of both GyrA and GyrB. GyrA is responsible for DNA binding and handling, while GyrB contains the site of ATP hydrolysis and binds both GyrA and DNA.

Figure 23.

Chemical structure of microcin B17 (MccB17) (49). The MccB17 precursor peptide is shown above with the core peptide shown in bold. The numbering scheme for (49) is reflected in the labels above the core peptide.

Subsequently, in vitro and in vivo studies revealed that MccB17 induces DNA cleavage in a DNA gyrase-dependent fashion.⁴⁰ Like other known gyrase inhibitors (e.g., fluoroquinolones), MccB17 stabilizes a transient gyrase-DNA-cleavage complex that blocks DNA polymerase movement.²⁸² However, MccB17 has a unique MOA compared to the fluoroquinolones, given that MccB17 does not influence the ATPase activity or DNA supercoiling/relaxation activities of DNA gyrase in vitro. Moreover, MccB17 requires multiple turnovers before DNA cleavage occurs, and unlike the fluoroquinolones, MccB17 slows DNA supercoiling and relaxation, but does not completely halt both processes.^282,283 Lastly, inhibition of gyrase cleavage and re-ligation reactions by MccB17 was found to be ATP dependent. MccB17 cleaves only 8% of closed-circular DNA in the absence of ATP, while DNA cleavage levels with ATP are comparable to those observed for ciprofloxacin (~30% vs. 40%).²⁸³ Kinetic simulations of the cleavage-ligation equilibrium suggest that MccB17 increases the rate of cleavage of the second strand of DNA, while ciprofloxacin increases the rate of cleavage on both strands.

The bioactivity of MccB17 is contingent on successful import into the cell. Spontaneous resistance mutations occurred in genes encoding the outer membrane protein ompF and the inner membrane peptide transporter sbmA.²⁸⁴ These data suggest that OmpF and SbmA are involved in the import of MccB17. SbmA has been implicated in the translocation of other antibacterial peptides including bleomycin, microcin J25 (a lasso peptide, see section 8.1), the mammalian peptide bactenecin-7 (Bac7), peptide nucleic acids (PNAs), and synthetic peptide phosphoramidate morpholino oligomers.^285–290 Strains deficient in SbmA are non-susceptible to MccB17.²⁸⁴

SAR studies of MccB17 demonstrated the importance of the two 4,2 thiazole-oxazole bis-heterocycles (Figure 23). The first bis-heterocycle (the A site) may serve as a biosynthetic checkpoint, as the C15S variant resulted in lower titers of MccB17.²⁹¹ The second bis-heterocycle (the B site) was crucial for bioactivity.²⁹¹ DNA gyrase supercoiling kinetic experiments conducted in vitro confirmed the importance of these bis-heterocycles.²⁹² Replacement of the B site heterocycles (with glycine) decreased the affinity of the MccB17-DNA gyrase interaction.²⁹³ Small heterocyclic derivatives of MccB17 mimicking the A and B sites, did not induce DNA cleavage by GyrB.²⁹² To further isolate the region responsible for bioactivity, MccB17 N27K was digested to produce two fragments consisting of Val1–Ser26 and Gly28–Ile43. Each fragment retained about ~20 to 50% of gyrase inhibition as compared to full length MccB17. The C-terminus of MccB17 was shown to be important for DNA cleavage, as removal of the C-terminal Ile43 reduced in vitro activity, while deletion of Ser41–Ile43 abrogated all activity. Conversely, removal of eight N-terminal core residues yielded an MccB17 fragment that retained wild-type DNA cleavage activity.²⁹²

A longer fragment of MccB17 encompassing residues Gly20–Ile43 displayed greater inhibitory activity and more significantly stabilized the gyrase-DNA cleavage complex in vitro (as determined by measuring the amount of single-stranded DNA).²⁹² However, this C-terminal fragment did not inhibit E. coli growth, implicating the N-terminus of MccB17 as important for membrane translocation. Independent studies elaborated on other residues required for cellular import. Removal of two unmodified C-terminal residues increased the MIC from 0.06 μM to 150 μM.²⁹⁴ The most critical residue was Ile43 as MccB17 variants ΔIle43, I43T, I43K, and I43E resulted in MIC increases of 10–500-fold. In contrast, the I43L variant retained wild-type bioactivity. In vivo studies of SbmA-overexpressing E. coli demonstrated that replacement of I43 with a stop codon (ΔI43), I43K, and I43E reduced transport across the membrane. In vitro gyrase inhibition assays also showed that ΔI43, I43T, and I43K possessed reduced inhibitory activity (MIC = 0.6–6 μM), while the I43E variant was inactive (MIC = 30 μM). Additional studies, such as those utilizing DNA gyrase point variants, are needed to elaborate the molecular details of the MccB17-DNA gyrase interaction.

A class of structurally related LAPs has demonstrated activity against the bacterial ribosome. Klebsazolicin (50, Figure 24) was discovered through genome mining from the opportunistic human pathogen Klebsiella pneumonia subsp. ozaenae. Heterologously produced klebsazolicin exhibited modest growth suppressive activity against various Proteobacteria including E. coli, K. pneumoniae, and Yersinia pseudotuberculosis with MICs in the 16–65 μM range.²⁹⁵ Klebsazolicin sensitivity was dependent on OmpF and SbmA, suggesting a role in cellular import, an uptake mechanism similar to that of MccB17 described previously. An E. coli fluorescent reporter assay indicated that klebsazolicin targeted the ribosome in vivo and inhibited protein translation in vitro with an IC₅₀ of 340 nM. Mutational analysis identified the N-terminal 14-residue fragment as the minimal bioactive core of klebsazolicin (Figure 24). Genetic sequencing of spontaneous resistance mutants identified two mutations in the 23S rRNA gene, U2609G and U2609A, which reside at an exposed site at the peptide exit tunnel.²⁹⁵

Figure 24.

Chemical structure of klebsazolicin (50). Ser1 to Ala14 comprises the minimal bioactive core of klebsazolicin (blue). The klebsazolicin precursor peptide is shown above with the core peptide shown in bold. The numbering scheme for (50) is reflected in the labels above the core peptide.

Further evidence for the MOA of klebsazolicin was provided by the co-crystal structure of the molecule bound to the Thermus thermophilus 70S ribosome (Figure 25).²⁹⁵ Klebsazolicin occludes the peptide exit tunnel and binds in a similar manner to the streptogramin antibiotics. Translation terminates as the nascent di- or tri- peptide cannot be accommodated. Klebsazolicin generates extensive van der Waals contacts with U2609, which also is a site of spontaneous mutation after selection of klebsazolicin-resistant strains. Direct contacts were also observed for A2058, A2059, U2506, and U2585, which were previously demonstrated to be protected from chemical modification by dimethyl sulfate, 1-cyclohexyl-3-(2-morpholinoethyl)-carbodiimide, and methyl-p-toluenesulfonate in the presence of klebsazolicin. Furthermore, Thz7, Thz10 and the N-terminal amidine ring form key stacking interactions with nucleobases in the 23S rRNA (C2586, A2062, and C2610, respectively). The C-terminal portion of the peptide (Ser15–Gly23) occupies the exit tunnel.

Figure 25.

(A) Positioning of klebsazolicin (Kleb, dark purple) bound to the complex of the ribosome with tRNA (PDB ID: 5W4K). tRNA is bound in the A-site by A-tRNA (A: aminoacyl); P-site tRNA by P-tRNA (P: peptidyl); and E-site by E-tRNA (E: exit). (B) Overview of klebsazolicin blocking the exit tunnel. (C) Interactions of the 23S rRNA (carbon atoms in yellow) with klebsazolicin (carbon atoms in purple). Stacking interactions are shown as blue regions between regions of interest.

Rhizobium sp. Pop5, a symbiont of the common bean plant Phaseolus vulgaris, produces a related LAP phazolicin (51, Figure 26). Phazolicin inhibits the growth of members of the genera Rhizobium, Sinorhizobium, and Azorhizobium sp. (MIC = 1–4 μM).²⁹⁶ However, no activity was observed against other plant-associated bacteria from Proteobacteria, Firmicutes, and Actinobacteria (e.g. Erwina amylovora, Pseudomonas, Bacillus, and Arthobacter sp.). Growth inhibition of E. coli was only observed at very high concentrations of 5–10 mM. Phazolicin insensitivity was overcome using an E. coli strain with a deletion in tolC, which encodes a major outer membrane multidrug efflux porin, suggesting that phazolicin is subject to TolC-dependent efflux. As with klebsazolicin, in vivo fluorescent reporter systems were used to show that the ribosome is the target, which was then confirmed using in vitro translation assays. The cryoEM structure of phazolicin bound to the E. coli ribosome complex demonstrated that phazolicin binds within the ribosomal exit tunnel in a “curled” conformation (Figure 27). Numerous intramolecular interactions stabilize the exit tunnel placement and π-stacking interactions occur between Thz3, Oxz15, and Oxz18 of phazolicin and A751, C2611 and U2609 of the ribosome. Although both phazolicin and klebsazolicin occlude the ribosome exit tunnel, the set of peptide/ribosome interactions is different.

Figure 26.

Chemical structure of phazolicin (51). Residues in blue are involved in binding to nucleotides in the exit tunnel. The phazolicin precursor peptide is shown above with the core peptide shown in bold. The numbering scheme for (51) is reflected in the labels above the core peptide.

Figure 27.

Close-up of interactions between E. coli 23S rRNA (carbon atoms in pink) and phazolicin (carbon atoms in blue) (PDB ID: 6U48). Interactions and residues are shown. Stacking interactions are shown as yellow regions between residues of interest. The view in panel B has been rotated by 90°.

To understand the narrow spectrum of activity, the structure of E. coli ribosome-bound phazolicin was superimposed onto the structure of the T. thermophilus 70S ribosome. Phazolicin binds between uL4 and uL22 loops of the E. coli 50S ribosome (Figure 28). However, in the T. thermophilus ribosome, this site is occupied by uL4-His69 and uL22-Arg90, which would not permit phazolicin to bind.²⁹⁶ The E. coli ribosome possesses a smaller residue in the uL4 loop (Gly64), which permits binding although it has a positively charged and large residue in the uL22 loop (Lys90). In complementation studies, the uL4-G68H variant rendered Sinorhizobium melliloti resistant to phazolicin, while complementation of S. melliloti with uL22-K90R did not confer resistance. These results were rationalized by the observation that position 90 of the uL22 has more space in the phazolicin-bound structure, hence substituted side chains are not as conformationally restricted as in substitution at His69 of the uL4 loop.

Figure 28.

Position of phazolicin (blue carbon residues) in relation to E. coli 50S ribosome uL4 (light green), uL22 (gold) loops (PDB ID: 6U48), and E. coli 23S rRNA (carbon atoms in pink) Key residues on both proteinaceous loops are highlighted.

In addition to serving as antibiotics, other known LAPs function as signaling molecules. Goadsporin (52, Figure 29) was originally isolated in a screening effort to identify compounds that induced morphological changes and secondary metabolite production in Streptomyces lividans.²⁹⁷ S. lividans cultures treated with fermentation broths of the goadsporin producer Streptomyces sp. TP-A0584 induced pigmentation, aerial hyphae formation, and sporulation. Goadsporin possessed antibiotic activity against S. lividans, S. coelicolor and S. scabies with MICs of 4.0, 2.0 and 0.12 μM, respectively, but lacked antibiotic activity against all non-streptomycetes tested. Goadsporin also lacked cytotoxic, apoptotic, and anti-protein kinase C activities. In a disk diffusion assay using 42 randomly selected streptomycetes, goadsporin was growth suppressive in 32 of the strains and induced pigmentation in 20 strains. Goadsporin also induced the production of an anti-B. subtilis antibiotic in strain TP-A0593. Structural studies demonstrated that goadsporin is a 19 residue, N-terminally acetylated LAP containing four oxazoles, two thiazoles, and two alkenes (Figure 29).^298,299

Figure 29.

Chemical structure of goadsporin (52). Dehydroamino acids are shown in blue; azoles that can be substituted without loss of activity are shown in orange; Gly10 is shown in green. The goadsporin precursor peptide is shown above with the core peptide shown in bold. The numbering scheme for (52) is reflected in the labels above the core peptide.

Introduction of the self-immunity gene godI into S. lividans endows the strain with goadsporin immunity.³⁰⁰ Sequence analysis demonstrated that GodI is ~50% identical to the E. coli Ffh signal recognition particle protein.^300–302 Ffh binds the signal peptide of membrane proteins to facilitate membrane integration.³⁰³ These findings suggest that Ffh inhibition by goadsporin is responsible for the morphological and metabolic changes observed in susceptible bacteria.^301,302 Genomic analyses of previously reported goadsporin-susceptible strains for the absence of GodI/Ffh has yet to be carried out. Such studies may inform on how changes in the primary sequence of Ffh across susceptible and non-susceptible strains affects the response to goadsporin.

SAR studies of goadsporin utilizing gene deletions have partially elucidated important functional groups for bioactivity. Deletion of godF and godG resulted in production of a goadsporin variant that lacked both alkenes (derived from Ser4 and Ser14), which lacked antibiotic activity.³⁰⁴ The G10A variant did not produce a zone of inhibition against S. lividans, but retained activity against S. scabies.³⁰⁰ Individual switching of methyloxazoles and oxazoles (T5S and S15T variants) did not diminish bioactivity, suggesting the importance of the heterocycle itself rather than methylation status of the ring. Interestingly, addition of a single Lys residue at the C-terminus of goadsporin eliminated activity. More detailed studies are needed to shed light on the MOA of this LAP.

3.2. Pyritides (including thiopeptides)

Pyritides are highly modified peptides defined by the presence of a six-membered, nitrogenous heterocycle that can exist in one of several oxidation states (most often pyridine). The pyritide class includes thiopeptides, which contain thiazoles and dehydroamino acids in addition to the central nitrogenous heterocycle, although many other secondary modifications have also been documented.^305,306 Thiopeptides can be classified into one of five series (a through e), which are distinguished based on the oxidation state of the 6-membered, nitrogenous heterocycle (Figure 30).²⁹ Many thiopeptides show impressive activity against Firmicutes by inhibiting ribosomal protein synthesis, though other bioactivities have been reported. Broadly speaking, the MOA of thiopeptides can be divided into one of four main categories: 1) binding of ribosomal RNA (rRNA) L11-binding domain (L11BD), 2) binding to Elongation Factor Thermo unstable (EF-Tu), 3) TipA_L transcription factor binding, and 4) FOXM1 transcription factor inhibition. We will devote most of our discussion to these four categories; other bioactivities will only be briefly mentioned.

Figure 30.

(A) Classification of thiopeptides based on the oxidation state of the central heterocyclic ring. (B) A generalized scheme of prokaryotic translation. Thiopeptides known to inhibit any steps in the process are noted. E, exit site; P, peptidyl site; A, aminoacyl site; EF-G, elongation factor G.

3.2.1. Ribosome inhibitors

Thiostrepton (53, Figure 31), the prototypical and best-known thiopeptide, contains a 26-member primary macrocycle (series b) and was originally isolated in the 1950s from Streptomyces azureus ATCC 14921.³⁰⁷ Thiostrepton demonstrated potent activity against Firmicutes, such as S. aureus (MIC = 0.06 μM), B. subtilis (MIC = 0.06 μM), and hemolytic streptococci (MIC = 6 nM). Growth of Proteobacteria (such as E. coli, Salmonella, and Shigella) and fungi were not affected by treatment with thiostrepton at concentrations up to 12 μM.³⁰⁸ Early studies established that thiostrepton inhibits bacterial protein synthesis.^309,310

Figure 31.

Structures of thiostrepton (53) and micrococcin P1 (54) and P2 (55). Residues that project into the cavity between uL11 and the 23S rRNA are shown in blue. The C-terminal dehydroalanine tail of thiostrepton that is required for activity is highlighted in red. Conjugating chemical species to the penultimate Dha16 (magenta), as opposed to Dha17 results in an analog devoid of activity.

Thiostrepton inhibits protein synthesis by binding to the 50S ribosomal subunit and preventing both mRNA-tRNA translocation by the GTPase Elongation Factor G (EF-G), and EF-Tu-catalyzed amino-acylated tRNA delivery (Figure 31).^310–314 The exact site of thiostrepton binding on the 50S subunit was identified based on the self-resistance mechanism of the producer S. azureus ATCC 14921, through identification of spontaneous resistance mutations, and through biophysical studies.^{309,312,315–317} Self-immunity in S. azureus is conferred by an RNA methyltransferase that modifies A1067 of the 23S ribosomal RNA component of the 50S subunit.³¹⁸ This thiostrepton resistance mechanism was also observed in several other streptomycetes, as well as two producers of the structurally related thiopeptides siomycin and sporangiomycin.³¹⁹ Resistance was also observed in variants of B. subtilis and B. megaterium that either possessed point mutations in uL11 or lacked this protein altogether.^320,321 uL11, along with (uL7/uL12)₄ and uL10, form the GTPase-associated center in the 50S ribosomal subunit that serves as a binding site for GTPases on the ribosome.³²² uL11 was also shown to be necessary for thiostrepton inhibition of the E. coli translational machinery in vitro.^{320,323–326} These data demonstrate that thiostrepton interacts with both the protein and RNA subunits of the ribosome.

The findings that uL11 bound the 23S rRNA along the same sequence that is targeted for methylation or mutation in resistant organisms provided a unifying rationale for the observed resistance determinants.^327–329 Subsequently, thiostrepton was demonstrated to directly bind at this rRNA sequence, suggesting that binding disrupted the native uL11–23S rRNA interaction.³³⁰ Thiostrepton binding was proposed to occur through allosteric cooperativity, as thiostrepton inhibited EF-G-dependent and -independent translocation and had no effect on the P site binding of tRNAs. An allosteric model was further supported by the observation that micrococcin P1 (54, Figure 31), a related 26-membered macrocyclic thiopeptide (series d), bound the same region of the 23S rRNA but stimulated EF-G GTPase activity, which is the opposite of the response observed for thiostrepton.^330,331 Structure-function studies of the 23S rRNA subunit demonstrated significant conformational changes upon uL11 binding, which stabilized select conformations and allowed for the cooperative binding of thiostrepton.^331–334

Structural studies of the uL11–23S rRNA interface revealed a cleft that can accommodate a thiopeptide (Figure 32).^335–338 These studies showed thiostrepton binding prevents any conformational transitions of uL11.^{321,339–342} Furthermore, thiostrepton was shown to induce tightening of the uL11–23S rRNA junction. Additional interactions between the uL11 N-terminal domain (NTD) and thiostrepton contribute to further stabilization of a conformation that does not allow the interactions with elongation and release factors necessary for translation.^{339,343–347} Disruption of these conformational changes has been shown to inhibit complexation of EF-G with amino-acylated tRNA during elongation.³⁴⁸ Studies on thiostrepton-resistant T. thermophilus uL11 variants demonstrated that amino acid substitutions increased the flexibility of uL11, potentially overcoming the conformational stabilizing effects of thiostrepton.³⁴⁹ These findings were further supported by NMR studies of Thermotoga maritima uL11 variants, demonstrating a compact, stable uL11-thiostrepton-23S rRNA structure. Substitution at conserved Pro sites increased uL11 flexibility and permitted translation to proceed in the presence of thiostrepton.³⁵⁰

Figure 32.

(A) Isolated view of interactions of thiostrepton with uL11 (green) and helices 43 and 44 of 23S rRNA (pink) from the structure of thiostrepton bound to the 50S ribosomal subunit from D. radiodurans (PDB ID: 3CF5). Residues in the primary macrocycle of thiostrepton that bind between uL11 and the 23S rRNA are shown in cyan. CTD, C-terminal domain; NTD, N-terminal domain. (B) Close-up of the interactions between thiostrepton, uL11 and the 23S rRNA. Key residues on uL11 and the 23S rRNA are shown in yellow. Thz6 is stacked between Pro21 and Pro25 of uL11. The quinaldic acid moiety (QA, carbon atoms in olive) engages in stacking interactions with A1078 of 23S rRNA and Thz1 is stacked against A1106. These positions correspond to A1067 and A1095 respectively in E. coli 23S rRNA.

Partial elucidation of the disparate activities of thiopeptides on the GTPase-associated center in the 50S subunit showed that thiostrepton increases the dissociation rate of GTP, thereby slowing GTP hydrolysis, while micrococcin P1 increases the dissociation of GDP+P_i from EF-G, subsequently increasing the rate of GTP hydrolysis.³⁴³ These observations are reconciled by structural studies on micrococcin P1 bound to the 50S ribosome of Deinococcus radiodurans.³⁵¹ Micrococcin P1 does not share the thiostrepton mode of binding and instead stabilizes interactions between the uL11 and uL7 ribosomal proteins.³⁵¹ It was hypothesized that in this conformation, uL7 is strategically poised to interact with EF-G, thus promoting GTP hydrolysis. As only GTP bound EF-G can associate with the ribosome, translocation, and protein synthesis ceases. These effects were rationalized by the presence or absence of the quinaldic acid (QA) heterocycle that directly interacts with the D. radiodurans 23S rRNA at A1078, which corresponds to A1067 in E. coli 23S rRNA (Figure 32).^351,352 In addition to targeting elongation steps, these thiopeptides also possess inhibitory activity toward translation initiation by destabilizing the interactions between the ribosome, tRNA, and initiation factor 2.^353–358

The discovery that extrachromosomal organellar DNA in Plasmodium sp. plastids encode rRNAs inspired efforts to target Plasmodium falciparum organellar protein synthesis using thiostrepton.^359,360 The plasmodial 23S rRNA contains adenine at the position equivalent to E. coli A1067. Consequently, thiostrepton binds to plastid rRNA to terminate organellar, but not nuclear, protein translation and arrests the trophozoite to schizont transition of the parasite life cycle.^361–363 Using the Malstat viability assay, thiostrepton displayed modest antimalarial activity (IC₅₀ = 8.9 μM).³⁶⁴ Thiostrepton induced conformational changes in plastid rRNA fragments analogous to those that occur in bacterial rRNA, despite only 60% sequence identity between P. falciparum and E. coli rRNA.³⁶⁵ A single nucleotide substitution at the position equivalent to A1067 in the plastid rRNA prevented thiostrepton binding.³⁶⁵ Similar anti-plasmodial activity was observed for micrococcin P1, and is presumed to occur through a thiostrepton-like mechanism.³⁶⁶ Growth suppression towards other plastid organelle-containing parasites, specifically Babesia sp., has been reported for thiostrepton.³⁶⁷ Further exploration of the anti-parasitic activity of thiostrepton identified the P. falciparum 20S proteasome as a target.³⁶⁴ Treatment of trophozoites with the thiopeptide resulted in accumulation of ubiquitinated proteins, similar to the phenotype observed when treated with known proteasome inhibitor MG132, suggesting that thiostrepton impaired proteasome activity. Thiostrepton was found to inhibit both the chymotrypsin and caspase-like activities of human 20S proteasomes (IC₅₀ = 5.2 μM and 3.8 μM, respectively).³⁶⁸ The dual-targeting ability of thiostrepton reduces the risk of developing resistance, and thus makes it an attractive anti-malarial drug.

3.2.2. Inhibitors of Elongation Factor Tu (EF-Tu)

The discovery of the d series thiopeptide GE2270A (56, 29-membered ring; Figure 33) in Planobispora rosea ATCC 53733 further expanded the notion of thiopeptides as inhibitors of the translational apparatus.^369,370 GE2270A inhibits the growth of many Firmicutes, including Staphylococcus, Streptococcus, Enterococcus, and Clostridium (MIC = 23–190 nM), but shows no activity against Proteobacteria.³⁶⁹ Early studies determined that GE2270A inhibits protein synthesis³⁶⁹ by specifically targeting the GTP-bound form of EF-Tu (K_d = 1 nM) and preventing the aminoacyl-tRNA^aa (aa-tRNA^aa) interaction.^371,372 Other thiopeptides known to bind EF-Tu include GE37468,³⁷³ thiomuracin A (57, Figure 33),³⁷⁴, and amythiamicin A (58, Figure 33),³⁷⁵ each of which contains a 29-membered macrocycle.

Figure 33.

Chemical structure of GE2270A (56), thiomuracin A (57), and amythiamycin (58).

The MOA of GE2270A was revealed by structural elucidation of E. coli (Ec) EF-Tu-GDP and T. thermophilis (Tt) EF-Tu-GNP (GNP is 5’-guanylyl imidodiphosphate, a non-hydrolyzable GTP analog) bound to GE2270A.^376,377 The GE2270A-Ec EF-Tu-GDP structure showed that GE2270A binds primarily to domain 2 of EF-Tu, and is locked into a pocket by a salt bridge formed by Arg223 (Arg234 in Tt EF-Tu) and Glu259 (Glu271 in Tt EF-Tu) from Ec EF-Tu (Figure 34). The interactions between GE2270A and EF-Tu are dominated by van der Waals contacts rather than hydrogen-bonding interactions. Comparison of the hydrogen-bonding residues in GE2270A with the peptide sequences of GE37468 and thiomuracin A suggests that the following interactions may be found across other 29-membered ring thiopeptides and EF-Tu (E. coli numbering used here): Phe261_EF-Tu and Tyr7; Asp216_EF-Tu and Thr228_EF-Tu with the b-hydroxyl group on Phe8 of thiomuracin.³⁷⁸ As these residues are not conserved in amythiamicin A (58) (Figure 33), the precepts of its interactions with EF-Tu have yet to be determined. Studies on the native resistance mechanism of the GE2270A producer P. rosea ATCC 53733 showed that this organism contains an alternate EF-Tu (EF-Tu1) which was resistant to GE2270A and other classes of EF-Tu-targeting antibiotics.³⁷⁹ EF-Tu1 has six amino acid substitutions in the GE2270A binding pocket as compared to Ec EF-Tu, and these substitutions limit the number of van der Waals contacts with GE2270A (Figure 34).

Figure 34.

(A) Crystal structure of Ec EF-Tu (PDB ID: 1D8T) bound to GE2270A (orange carbon atoms) and GDP. Ec EF-Tu residues involved in hydrogen bonding are shown in yellow. Salt bridge that locks GE2270A into the binding pocket is shown in red dashed lines. The six amino acids that differ between P. rosea EF-Tu1 and Ec EF-Tu are shown in pink. (B) Overlay of the position of GE2270A (spheres) bound to Tt EF-Tu (PDB ID 2C77) superimposed onto the structure of T. aquaticus EF-Tu bound to GTP (PDB ID: 1EFT). GE2270A intrudes into the region between domains 1 and 2 of “on” state EF-Tu. (C) Overlay of the position of GE2270A (spheres) as bound to Tt EF-Tu superimposed onto the structure of T. aquaticus EF-Tu GNP (light green) bound to yeast Phe-tRNA (yellow, PDB ID: ITTT). GE2270A clashes with the 3’ end of the tRNA molecule (red arrow). Tt EF-Tu in panels (B) and (C) are otherwise hidden from view.

EF-Tu adopts a compact structure when bound to GTP (“on state”) that is competent for binding to aa-tRNA^aa. Upon GTP hydrolysis, the nucleotide-binding domain of EF-Tu (domain 1) undergoes a conformational change (“off state”) resulting in aa-tRNA^aa dissociation (Figure 35).^380,381 Superimposition of the structure of Tt EF-Tu-GDP bound to GE2270A with the structure of Thermus aquaticus-EF-Tu-GTP reveals that bound GE2270A would sterically clash with domain 1, suggesting that GE2270A prevents the conformational switching of EF-Tu that allows binding of charged tRNA (Figure 35).³⁷¹ These observations are supported by the superimposition of the GE2270A-Tt EF-Tu-GDP structure with that of T. aquaticus EF-Tu bound to Phe-tRNA,³⁸² which showed that GE2270A makes contacts with residues that would otherwise bind the aminoacylated 3’ end of tRNA (Figure 34).^376,377

Figure 35.

EF-Tu in “on” (A) and “off” (B) states. EF-Tu adopts a compact state when bound to GTP or non-hydrolysable analogs such as GNP (PDB ID: 1EFT). EF-Tu adopts a relaxed state when bound to GDP (PDB ID: 1EFC).

3.2.3. Induction of the TipA promoter

In addition to stalling protein synthesis, thiopeptides are also known to affect prokaryotic pan-thiopeptide resistance pathways. This activity was first observed when thiostrepton (at concentrations as low as 0.3 μM) was shown to induce protein expression in S. lividans.³⁸³ TipA_L and TipA_S were among the most highly upregulated proteins during this response. TipA_L belongs to the MerR family of stress response regulators. The N-terminus of TipA_L is a winged-helix-turn-helix DNA-binding domain, while the C-terminal domain binds thiostrepton.³⁸⁴ An alternative start codon within the tipA gene encodes for TipA_S, a discrete thiostrepton-binding domain. The TipA_L-thiostrepton complex induces expression from the TipA promoter (p_tipA). Binding of this complex, in turn, increases expression of TipA_L and TipA_S, with the latter expressed at a ~20-fold excess over TipA_L. It is hypothesized that increased levels of TipA_S sequester thiostrepton, serving as a resistance mechanism against thiostrepton and related antibiotics.³⁸⁴ Though TipA homologs are present across streptomycetes and other species (e.g. B. subtilis, Caulobacter crescentus, and Klebsiella oxytoca), the occurrence of non-thiopeptide producers that utilize this resistance mechanism has not yet been evaluated. Further studies along these avenues are warranted and may reveal that this resistance mechanism is employed against other classes of antibiotics.

Many other thiopeptides including thioxamycin (60), thiotipin (61), promothiocins A (62), nosiheptide (63) (Figures 36–38) are known to induce p_tipA,^{383,385–387} which prompted efforts to discern the nature of interactions between thiopeptides and TipA_L/A_S. Detailed studies on TipA_S demonstrated the formation of a lanthionine linkage between Cys214 in TipA_S and a dehydroalanine residue in the C-terminal tail of thiostrepton.^388,389 Thiopeptides including promothiocins A & B, nosiheptide, and berninamycins A & B (68, 69, Figure 40) were also able to form covalent linkages in vitro with TipA_S,³⁸⁹ suggesting that this interaction is independent of macrocycle size. Tail-deleted variants of thiostrepton and promothiocin A were unable to form TipA_S-antibiotic complexes, underscoring the necessity of the tail to form this covalent linkage.

Figure 36.

(A) Chemical structure of cyclothiazomycin (59), a thiopeptide that does not elicit the TipA response (i.e. lacks dehydroamino acids in the tail region), and (B) structures of those that do elicit the TipA response, thioxamycin (60), and thiotipin (61). The stereochemistry at Ala8 of thioxamycin is not defined.

Figure 38.

(A) Close up of TipA_S bound to promothiocin A (carbon atoms shown in green, PDB ID: 2MBZ). (B) Close up of TipA_S bound to nosiheptide (carbon atoms shown in teal, PDB ID: 2MC0). Key interacting residues on TipA_s are shown in yellow. π-π stacking interactions are shown as opaque yellow regions. “Pyr” denotes the pyridine ring in panels (A) and (B). Each thiopeptide and TipA_S interact mainly through hydrophobic packing. Limited hydrogen bonding contacts are made between the protein and the ligand. (C) Chemical structure of nosiheptide (63). (D) Generalized motif in various thiopeptides that is used for recognition by TipA_S.

Figure 40.

Chemical structures of thiopeptides that do not act as proteasome inhibitors. Thiocillin I (67), berninamycins A (68) and B (69), and YM-266183 (70).

Studies on the requirements for TipA_L-dependent p_tipA induction showed that induction levels increased upon treatment with compounds with greater numbers of dehydroamino acids in the tail region.³⁸⁹ Thiopeptides that lacked a dehydroamino acid containing tail, such as GE2270A (56, Figure 33), amythiamicin A (58, Figure 33), and cyclothiazomycin (59, Figure 36) were unable to activate expression from p_tipA. The importance of a C-terminal carboxamide is also underscored by the observation that higher dosages of thiopeptides with a C-terminal carboxylate (e.g., thioxamycin and thiotipin, Figure 36) were required for p_tipA induction, suggesting that a charge at the C-terminal residue may disfavor interactions between the thiopeptide and TipA_L. As tail-deleted variants of GE2270A and promothiocin A could still activate expression from p_tipA at low levels, a secondary structural motif or a non-covalent mechanism may play a role.³⁸⁹

Thiopeptide recognition by TipA_S/A_L was structurally interrogated through NMR spectroscopy of TipA_S-thiopeptide complexes.^390,391 The structure of apo-TipA_S revealed an antiparallel helical arrangement, which forms a V-shaped cleft with Cys214 located in a solvent-exposed region (Figure 37). The N-terminal region of apo-TipA_S, encompassing helices α6 through α8, was disordered. Chemical shift mapping of promothiocin A-treated TipA_S revealed stabilization of the N-terminal region, and suggested a hydrophobic cavity in which the thiopeptide was bound.³⁹⁰ These observations were validated by the co-crystal structures of TipA_S bound to promothiocin A and nosiheptide where half of the “A ring” of both compounds are accommodated by helices a10–13, while the disordered N-terminus of TipA_S is stabilized by binding the other half of the “A ring” (Figure 37).³⁹¹ Extensive hydrophobic contacts are observed, with few if any specific interactions between the protein and thiopeptide ligand (Figure 38). Comparison of the structures reveals a common motif required for TipA_S binding that includes the pyridine ring, Thz2, Oxz5, and Oxz9 (promothiocin A numbering; Figure 38). This motif explains the broad substrate tolerance of TipA_S/A_L, as it is found in micrococcin P1 (54), thioxamycin (60), thiotipin (61), nosiheptide (63), siomycin A (64), and many others.

Figure 37.

Promothiocin binding to TipA_S. (A) Chemical structure of promothiocin A (62). (B) Structure of apo TipA_S (PDB ID: 1N79). (C) Structure of TipA_S bound to promothiocin A (green, PDB ID: 2MBZ). Helices α6-α8 are disordered in the apo structure and become ordered upon thiopeptide binding.

3.2.4. Anticancer Properties of Thiopeptides

Selected series a-e thiopeptides display low micromolar IC₅₀ values against gastric and hepatocellular carcinomas.³⁹² Investigations into anticancer MOAs have largely focused on the quinaldic acid containing compounds thiostrepton (53) and siomycin A (64) (Figures 31 and 39). Each of these thiopeptides targets the oncogenic transcription factor Forkhead box M1 (FOXM1), which activates the expression of proteins involved in cell cycle progression, angiogenesis, and metastasis.^393,394 FOXM1 has received considerable interest as a chemotherapeutic target as it is overexpressed in breast tumors,³⁹⁵ hepatocellular carcinomas,³⁹⁶ non-small cell lung carcinomas,³⁹⁷ and other cancers.

Figure 39.

Structures of siomycin A (64), thiostrepton A hydrolyzed or methanolyzed at the ester in the B ring (65), and bisdehydroalanine fragment with anticancer activity (66).

Two independent studies revealed the anticancer potential of thiopeptides, with thiostrepton shown to inhibit FOXM1 mRNA and protein expression in a breast cancer cell line³⁹⁸ and siomycin A reducing FOXM1 mRNA levels in C3-Luc cells (embryonic mouse cells transformed with the human papillomavirus 16 genome, with luminescence reporter).³⁹⁹ Both thiostrepton and siomycin A exhibit cytotoxic activity in cancer cell lines (IC₅₀ ~0.5–5.0 μM), but not in non-transformed MRC-5 cells (human lung fibroblasts). Multiple studies have correlated a decrease in FOXM1 expression to an increase in apoptosis, thus a pro-apoptotic action of thiopeptides has become the prevailing theory for their anticancer activity.^18,398–401

The transcriptional activity of FOXM1 can be inhibited by direct inactivation or by preventing interaction with promoters. FOXM1 is activated via Thr596 phosphorylation⁴⁰² and immunoblotting studies show that siomycin A-treated C3-Luc cells display lower levels of phosphorylated FOXM1. These data support a direct inactivation model wherein FOXM1 cannot be activated to promote transcription of its targets.³⁹⁹ Other biophysical studies support a promoter blockade model as FOXM1 binding to promoters decreases with increasing concentrations of thiostrepton.³⁹⁸ ITC experiments verified that thiostrepton bound to both oncogenic splice variants of FOXM1, but preferred FOXM1c.⁴⁰³ Thiostrepton inhibits the binding of FOXM1c to its DNA target with a K_I of 11 μM (as compared to a K_I of 31 μM for FOXM1b and its DNA target). In the same study, chromatin immunoprecipitation experiments demonstrated that thiostrepton-treatment of MCF-7 breast cancer cells reduced FOXM1 binding to genes implicated in cellular proliferation, i.e. Myc, Cdc25B, and cyclin B1.

Though specific interactions between thiostrepton and FOXM1 have yet to be determined, regions of the thiopeptide that are critical for antitumor activity have been identified. Truncation of the Dha terminus (Dha16 and Dha17) resulted in an analog that is unable to inhibit cell growth.⁴⁰³ A variant with Dha16 as the terminal residue displayed reduced activity as compared to the native compound (GI₅₀ = 22 μM vs. 3.7 μM), whereas addition of a thiol-linked biotin tag selectively to Dha16 and not Dha17, resulted in a 10-fold reduction in activity as compared to thiostrepton. This observation suggests that the presence of at least one Dha is required for bioactivity. The importance of these residues should also hold for siomycin A, as the core peptides only differ at positions 1 and 2 (Ile1 vs. Ala1, and Val2 vs. Ser2). Lastly, SAR studies of the antitumor properties of thiostrepton suggest that the Dha-containing fragment 66 (Figure 39) is critical for bioactivity. This fragment exhibits an LC₅₀ of 0.6 – 1.2 μM against various cell lines.⁴⁰⁴ However, the molecular target of this fragment has not been determined.

The cytotoxic activity of thiostrepton and siomycin A has also been attributed to proteasome inhibition^18,405 and inhibition of mitochondrial translation.⁴⁰⁶ Treatment with siomycin A or thiostrepton resulted in stabilization of several marker proteins including p21, Mcl-1, p53, and hdm2, indicative of a non-functional proteasome.^18,405 Other thiopeptides including micrococcin P1/P2 (Figure 31), thiocillin I (67), berninamycins (68, 69), YM-266183 (70), and thiostrepton that was hydrolyzed at the ester in the B-ring (65) (Figures 39–40) were unable to stabilize levels of these marker proteins nor were they able to inhibit FOXM1 (up to 10 μM).⁴⁰⁵ However, higher dosages of berninamycin (20 μM) reduced FOXM1 transcript and protein levels in prostate-derived epithelial cells.⁴⁰⁷ Apart from a strong dose-dependent effect, these results suggest that thiopeptides that only contain a primary macrocycle do not exert these bioactivities. This hypothesis is also supported by data on the methanolysis product of thiostrepton that converted the B ring to the methyl ester 65 (Figure 39).⁴⁰⁵ Taken together, these observations highlight that the presence of both primary and secondary macrocycles in addition to Dha16 and Dha17 are important for FOXM1-mediated bioactivity.

3.2.5. Additional bioactivities

In addition to the bioactivities discussed in sections 3.2.1-3.2.4, other therapeutic targets of thiopeptides have been explored. Cyclothiazomycin is an inhibitor of bacteriophage RNA polymerase⁴⁰⁸ and human renin (IC₅₀ = 1.7 μM).⁴⁰⁹ Thiostrepton reactivates latent HIV-I cells in CD4+ T cells, which provides a means to target latent HIV infection.^410,411 Thiostrepton also has been reported to rescue misfolded a-sarcoglycan (a major cause of limb-girdle muscular dystrophy type 2D),⁴¹² inhibit psoriasis-like inflammation induced by TLR7–9,⁴¹³ and repress the sonic hedgehog signaling pathway in a triple-negative breast cancer model.⁴¹⁴ Lastly, the antibacterial effects of nosiheptide⁴¹⁵ (formerly multhiomycin) and the nocathiacins extends to clinically-relevant drug resistant strains including vancomycin-resistant S. aureus (VRSA), vancomycin-resistant Enterococci (VRE), and MRSA.^416,417

3.3. Bottromycins

The bottromycins (71-74) are a group of macrocyclic RiPPs that possess growth-suppressive activity against several multidrug-resistant bacteria.³³ The chemical structure of the bottromycins has undergone several revisions since its original isolation but its identity has been unequivocally established by total synthesis (Figure 41).⁴¹⁸ Bottromycins contain a class-defining macrolactamidine and a suite of other modifications, including a decarboxylated C-terminal thiazole as well as several C- and O-methylations on the side chains (Figure 41). Studies on the MOA of the bottromycins were carried out primarily by studying bottromycin A2 (71), which was among the first two isolated members of this class and the subject of the total synthetic effort.^418,419 The antibacterial activity of bottromycin has been demonstrated for Mycoplasma, MRSA, VRE, and the rice plant pathogen Xanthomonas oryzae pv. oryzae with MIC values in the low μM range.^418,420,421

Figure 41.

Chemical structure of bottromycin A2 (71), and its congeners B2 (72), C2 (73), and D (74). The methyl group on C_β of Phe6 (blue background) is critical for activity. The methyl ester (orange) is important because the corresponding carboxylic acid is inactive; however, replacement of the methyl ester with amides, ethyl/propyl esters, and propyl/isopropyl thioesters resulted in compounds with improved activity.

The bottromycins were discovered from soil-dwelling Streptomyces bottropensis and identified by the antibacterial activity present in spent media.^419,422 Subsequently, isolation of bottromycin from Streptomyces No. 3668-L2 identified major and minor congeners (A2 and B2 respectively, Figure 41) that demonstrated similar antibacterial activities towards Firmicutes.⁴²³ Despite promising in vitro activity, bottromycins are not efficacious under physiological conditions owing to the hydrolytic instability of the methyl ester.

Though bottromycins are known protein synthesis inhibitors, their target was difficult to resolve. Isotopic labeling studies showed that bottromycin A2 inhibits protein synthesis in vitro and in vivo.⁴²⁰ Inhibition was dependent on base composition as significant translational stalling was observed for poly-C but not poly-U or poly-A. These early studies also showed that bottromycin did not affect tRNA charging, and based on these data, the mechanism for inhibition was proposed to occur through direct interaction of bottromycin with ribosome-bound mRNA in a nucleotide specific manner.

A series of subsequent studies pointed towards inhibition of translocation as a mechanism of action. In vitro translation experiments using poly-U mRNA demonstrated that bottromycin A2 inhibited the formation of triphenylalanine more than it did the formation of diphenylalanine, suggestive that translocation may be affected.⁴²⁴ To further study the MOA of bottromycin A2, interference of the puromycin reaction was employed. Puromycin is an aminonucleoside antibiotic that resembles aminoacylated Tyr but contains a non-hydrolyzable amide bond between the amino acid and the ribose sugar.^425,426 Puromycin can enter the ribosomal A-site and be incorporated into a growing peptide; however, a second molecule of puromycin cannot be attached to the growing peptidyl-puromycin chain as the first molecule contains an inert bond. This results in the dissociation of the ribosome subunits, and chain termination.

Bottromycin A2 partially inhibited the puromycin reaction in the presence of EF-G and GTP, which are both needed for peptidyl-tRNA translocation (see Figure 30 for an overview of ribosome catalysis). In these studies, translocation of A-site loaded ³H-labeled initiator N-formylmethionine-tRNA (fMet-tRNA) and polylysyl-tRNA in the presence of EF-G, GTP, and puromycin was examined. More radioactivity was retained when bottromycin was added in the presence of GTP and EF-G than in the puromycin reaction alone, suggesting that bottromycin prevents translocation, and subsequent chain termination by puromycin.⁴²⁷ Similar experiments utilizing ³H-tRNA and ¹⁴C-polylysine revealed that bottromycin does not affect tRNA release from the P site, but is able to reduce the level of puromycin-dependent chain termination in the presence of GTP and EF-G. Based on these studies, bottromycin A2 was proposed to interfere with translocation and movement of mRNA on the ribosome,⁴²⁸ which was subsequently supported by mapping of the bottromycin binding site to the 50S ribosomal subunit.⁴²⁹

Subsequent studies cast doubt on inhibition of translocation as the mode of action of bottromycin A2. Experiments using E. coli ribosomes revealed that bottromycin A2 inhibited N-acetyl-¹⁴C-Phe-puromycin formation and also inhibited polyphenylalanine synthesis in the presence and absence of EF-G and GTP to similar degrees.^424,430 Thus, it was concluded that bottromycin A2 inhibits peptidyltransferase activity. Additional studies complicate understanding of the mode of action, as bottromycin did not inhibit the puromycin reaction with native E. coli polysomes,⁴³¹ and also failed to inhibit puromycin-dependent peptide release in protoplasts from B. megaterium.⁴³²

These conflicting results regarding the MOA of bottromycin A2 were resolved in a series of three papers. Bottromycin A2 was shown to inhibit the puromycin reaction, but increasing concentrations of puromycin reduced the levels of bottromycin-dependent inhibition suggesting that puromycin and bottromycin A2 occupy the same site on the ribosome.⁴³³ Next, bottromycin A2 was demonstrated to release peptidyl- and aminoacylated-tRNA from the ribosome A site,^434,435 but not induce radiolabeled (oligo)-Phe-tRNA from the P site. These data suggest a model where bottromycin inhibits translation by releasing bound aminoacyl tRNAs from the A site. Translation studies utilizing E. coli polysomes reaffirmed that bottromycin A2 releases A site bound aminoacyl-tRNAs. This MOA also resolved conflicting studies regarding the inhibitory effect against polysomes with nascent polypeptides.⁴³¹ Large peptidyl-tRNA binds more tightly to the A site than aminoacyl-tRNA, thus impeding bottromycin-dependent release.⁴³⁵ Thus, the conclusion of ~20 years of research established that bottromycin acts by binding the ribosomal A site to prevent recognition of charged tRNAs. An avenue for future research is to gain structural insights into this interaction for bottromycin A2 or its congeners.

The total syntheses of bottromycin A2 and its analogs have revealed which structural features are important for antibiotic activity. Bottromycin A2 and B2 differ by a methyl group at Pro2, with the latter compound four times less potent as an inhibitor against methicillin resistant S. aureus and vancomycin resistant Enterococci strains (MIC = 1.2 μM vs. 4.9 μM).⁴³⁶ Analogs of bottromycin A2 and B2 lacking the methyl group on C_β of Phe6 (Figure 41) demonstrated no antibiotic activity (MIC > 32.5 μM). Likewise, removal of the Thz-β-Ala-OMe moiety from bottromycin A2 was detrimental to activity. However, a derivative in which the Thz-β-Ala-OMe was replaced with benzylamine showed effectiveness similar to bottromycin against MRSA and VRE strains (MIC = 1.3–2.7 μM). An analog that only possesses the β-Ala-OMe group (no Thz) had slightly poorer activity compared to the native compound or the benzylamine analog, suggesting that the terminal Thz can be altered. The methyl ester is also critical for bioactivity as the carboxylic acid of bottromycin A2 displayed MICs > 98 μM against MRSA and VRE.⁴³⁷ The hydrolytic lability of the Asp7 methyl ester to yield the inactive acid product led to synthetic efforts to generate a more stable compound in vivo. Substitution of the methyl-ester with various amides yielded bottromycin A2 analogs with improved in vivo efficacy against a mouse model of S. aureus infection compared to the parent compound; however, in vitro efficacy suffered.⁴³⁸ Replacement of the methyl ester with a propyl/isopropyl thioester or with an ethyl/propyl ester resulted in bottromycin A2 analogs with comparable, if not better, activity against S. aureus strains, MRSA, and VRE.⁴³⁷

3.4. Thioamitides

Thioamitides are RiPPs defined by the presence of a backbone thioamide(s) in place of amide(s), although other secondary modifications are typically also present. The first thioamitides to be discovered were the thioviridamides (75), and related compounds have been reported more recently (Figure 42).⁴³⁹ Generally, these compounds are cytotoxic against various cancer cell lines, but do not demonstrate significant antibacterial activity.^20,439–442 The structural diversity of the thioamitides is reflected in the MOAs, which can impact primary and secondary metabolism.

Figure 42.

Chemical structures of thioviridamide (75), pre-thioviridamide (76), thioalbamide (77), and thiostreptamide s4 (78).

Thioviridamide (Tvr) from Streptomyces olivoviridis was first identified in a screen for antitumor compounds.⁴³⁹ Although thioviridamide is an acetone adduct of the true natural product prethioviridamide (pTvr, 76) (Figure 42), both compounds have comparable cytotoxic activities.^439,441 Both Tvr and pTvr selectively induce apoptosis in cells transformed with the adenovirus type 12 (Ad12) or the adenoviral E1A apoptosis-suppressing oncogene with IC₅₀ values in the low nM range. pTvr also inhibits proliferation of human cervical carcinoma cells (IC₅₀ = 0.36 μM),⁴⁴³ but similar studies have not been extended to Tvr.

The cytotoxicity of pTvr was attributed to activation of the integrated stress response (ISR).⁴⁴³ pTvr induced significant cellular morphological changes, dissipation of mitochondrial membrane potential, a time-dependent reduction in O₂ consumption as well as global translation inhibition. Screening of shRNA libraries revealed that pTvr induced the GCN2-ATF4 metabolic stress response pathway, which is central to the ISR and apoptotic induction.^443,444 Photoaffinity labeling studies identified the target of pTvr as the ATP5B subunit of the mitochondrial F₁ ATP synthase.⁴⁴³ Inhibition of F₁F_o-ATPase activity was responsible for the upregulation of the GCN2-ATF4 pathway. The selectivity for E1A-expressing cells was due to increased upregulation of ATF4 upon treatment with pTvr as compared to cells that did not express E1A.

Thioalbamide (Tab, 77) and thiostrepamide s4 (78, Figure 42) are related to pTvr. Tab is similar to pTvr in that it also induces apoptosis in human cancer cell lines.^20,445 Mechanistic studies on Tab demonstrated that treated cells were arrested at the G1 stage, and confirmed that both intrinsic and extrinsic apoptotic pathways were activated. Tab-induced apoptosis was attributed to increased production of reactive oxygen species (ROS). Perplexingly, superoxide dismutase isoform 2 was overexpressed in cells treated with Tab, suggesting that the mitochondria are responding to increased levels of ROS, but are failing to neutralize excess ROS.⁴⁴⁵ Additionally, exposure to Tab in vitro led to a >60% reduction of glycolysis and oxidative respiration in breast cancer cell lines. These findings support a general mechanism of metabolic disruption.

Mechanistic studies on both Tab and pTvr suggest that they perturb oxidative respiration, though a comprehensive understanding of the mechanism is lacking. Similarly, the molecular-level details regarding the pTvr:F1 ATP synthase β-subunit interaction is unknown. Derivatization studies identified pTvr analogs with improved cytotoxicity,⁴⁴⁶ and further studies on understanding of interactions between pTvr (and analogs) with ATP synthase are warranted.

4. Cyanobactins

Cyanobactins are RiPPs that are produced by cyanobacteria in both marine and freshwater environments.^447–450 The majority of isolated cyanobactins are macrocyclic and contain (methyl)(thi/ox)azoline(s) installed by YcaO enzymes (section 3). However, linear cyanobactins⁴⁵¹ and macrocyclic cyanobactins lacking azoles⁴⁵² have also been characterized. Cyanobactins are often decorated with isoprenoid groups.⁴⁸ Members of this class possess diverse bioactivities including antimalarial,⁴⁵³ anti-HIV,⁴⁵⁴ antifungal,^16,455 and antitumor^{16,17,19,456,457} activity, though few MOA studies have been performed. The bioactivity of cyanobactins has been reviewed,^19,458 and thus we only cover here bioactivities with known or preliminary MOAs.

Antineoplastic activity has been documented for several cyanobactins.^{16,17,19,456,457} Of these, ulithiacyclamide (79, Figure 43) is among the most potent, possessing an in vitro IC₅₀ of 0.05 – 0.47 μM against L-1210 murine leukemia cells.^459,460 Mechanistic work revealed that ulithiacyclamide inhibits protein and RNA synthesis, but detailed study of the inhibition mechanism has yet to be performed. Preliminary MOA studies have also been performed on cycloxazoline (80, Figure 43), which is cytotoxic against simian virus 40 transformed fetal lung (MRC5-CV1) and human bladder carcinoma (T24) cells at 0.66 μM.⁴⁶¹ Cycloxazoline increased the proportion of cells arrested at the G2/M phase as well as the population of polyploid cells in a dose-dependent manner.⁴⁶² This observation suggests the target of cycloxazoline is involved in cytokinesis. Trunkamide A (86) is a low micromolar growth inhibitor of cancer cell lines (IC₅₀ = 0.8 – 1.5 μM) and has been pursued in clinical trials as an anticancer agent.⁴⁶³ Owing to the structural similarity to telomestatin (83),⁴⁶⁴ future studies could evaluate macrocyclic cyanobactins as telomerase inhibitors.

Figure 43.

Chemical structures of the cyanobactins ulithiacyclamide (79), cycloxazoline (80), dendroamide A (81), agardhipeptin A (82), telomestatin (83), (R,R,R) QZ59 (84), (S,S,S) QZ59 (85), trunkamide (86), haliclonamide A (87), haliclonamide B (88), sphaerocyclamide (89), kawaguchipeptin A (90), and kawaguchipeptin B (91).

Certain cyanobactins possess the ability to reverse multidrug resistance (MDR) phenotypes.^465–467 MDR can arise from overexpression of ABC efflux pumps, e.g. P-glycoprotein (Pgp), which prevent the intracellular accumulation of chemotherapeutic compounds. Pgp accommodates a myriad of substrates due to its ability to repack its transmembrane helices upon substrate binding.^468,469 Dendroamide A (81, Figure 43) chemosensitizes Pgp-overexpressing breast carcinomas (MCF-7/ADR) cells to daunomycin and actinomycin D.⁴⁶⁶ Cotreatment of MCF-7/ADR cells with either aforementioned chemotherapeutic and dendroamide A increased the percentage of cells killed versus treatment with chemotherapeutic alone. Direct interaction of dendroamide A with Pgp was confirmed using photoaffinity-labeling studies. Dendroamide A was able to outcompete the MDR modulator ³[H]-azidopine for binding to Pgp, suggesting the two compounds occupy a similar binding site.^466,470 Pgp was proposed to contain at least three binding sites.^471,472 The H-site binds Hoechst 3342, quercetin, and colchicine,^471,473,474 the R-site binds anthracyclins such as daunomycin and doxorubicin,^471,475 whereas the P-site binds prazosin and progesterone.⁴⁷² Co-crystal structures and modeling studies of murine Pgp with stereoisomers of an analog of dendroamide A containing 1,3-selenazoles termed QZ59 (Figure 43) revealed that (S,S,S)-QZ59 (85) occupies both the H- and R-sites, while (R,R,R)-QZ59 (84) occupies a distinct site from the H- and R-site.^476–478 Patellamides B and C (93, 94, Figure 44) were reported to chemosensitize human leukemic lymphoblasts to vinblastine, while patellamide D (95, Figure 44) chemosensitizes the same cell line to colchicine.^465,467 Whether the patellamides interact with Pgp or with another efflux pump has yet to be determined.

Figure 44.

Chemical structure of the patellamide A (92), and patellamides B-E (93–96).

The resemblance of cyanobactins to porphyrins and cyclic crown ethers have led to the hypothesis that cyanobactins may coordinate metals and function as siderophores or chalkophores.^479,480 Examples of metal-coordinating cyanobactins include haliclonamide A (87) (Fe³⁺-binding),⁴⁸¹ and patellamides A, B, C, and E (Figure 44, Cu²⁺-binding).^482,483 Furthermore, the cyanobacterial symbiotic Prochloron sp. has been shown to uptake Cu²⁺-complexed patellamide,⁴⁸⁴ suggesting that these organisms produce and reabsorb these molecules. However, the physiological relevance of metal binding by cyanobactins has not been investigated in detail.

Although cyanobactins have been associated with various bioactivities, the exact MOAs remain unknown. For example, agardhipeptin A (82, Figure 43) produced by Plankothrix sp. displays modest plasmin inhibitory activity.⁴⁸⁵ Similarly, sphaerocyclamide (89), a prenylated cyanobactin produced by Sphaerospemopsis sp. LEGE 00249 (Figure 43), showed only antimicrobial activity against the fouling bacterium Halomonas aquamarina CECT 5000.⁴⁸⁶ The kawaguchipeptins (90, 91, Figure 43) inhibit the growth of S. aureus (MIC ~0.7 μM) through a yet to be determined mechanism.⁴⁸⁷ Given the promising bioactivities observed, further characterization of the MOAs of these compounds is needed. Lastly, additional studies are necessary on cytotoxic cyanobactins such as microcyclamide,⁴⁸⁸ the lissoclinamides^456,489 and ascidiacyclamide,⁴⁹⁰ to identity the cellular targets of these compounds.

5. Post-translationally modified microcins

Microcins are a class of ribosomally produced antimicrobial peptides that are produced predominantly by Enterobacteriaceae and demonstrate activity against a narrow range of related bacteria. The microcins are divided into two classes; class I microcins are post-translationally modified (i.e., RiPPs) and generally <5 kDa in size.^37,491 Examples of a subset of class I microcins are discussed in other sections of this review, such as MccB17 (section 3.1 on LAPs) and MccJ25 (section 8.1 on lasso peptides). This section will review the known the MOA of all other class I microcins. Class II microcins are larger, ranging from 5–10 kDa in size.^37,491 Class II microcins are further demarcated into subclasses IIa, which do not contain post-translational modifications, and IIb, which are post-translationally modified through conjugation with a siderophore.^37,491 These class IIb siderophore-peptides will be also discussed in this section. Several prior reviews focus on the unmodified subclass IIa microcins that are not RiPPs.^28,492,493

5.1. Microcin C7

Microcin C7 (McC) is a RiPP produced by E. coli and other Proteobacteria, which is active against strains of E. coli, Klebsiella, Salmonella, and Shigella.⁴⁹⁴ The MccA precursor peptide is encoded by one of the shortest known genes (21 nucleotides) and this length allows for facile iterative peptide synthesis via ribosome recycling without mRNA dissociation.⁴⁹⁵ Unprocessed McC contains a formylated N-terminus and an adenosine monophosphate (AMP) linked to a C-terminal Asp through a P-N bond (Figure 45, 97). Additional modification includes the attachment of an aminopropyl moiety on the phosphate group of AMP (Figure 45, 97).^496–498 The final bioactive compound, termed here processed McC is formed after removal of the peptide by intracellular proteases in the susceptible cell (Figure 45, 98).

Figure 45.

Chemical structures of unprocessed McC (97) and processed McC (98). Essential residues required for recognition by YejA are highlighted in blue.

Early MOA studies suggested that unprocessed McC was an inhibitor of protein translation.⁴⁹⁹ However, this hypothesis was disproven, as neither 10 μM of unprocessed McC nor the MetArgThrGlyAsnAlaAsp heptapeptide inhibited in vitro translation.⁵⁰⁰ Treatment of unprocessed McC with crude cell extracts yields processed McC, a translation inhibitor that blocks the aminoacylation of aspartyl tRNA.⁵⁰⁰ Structural elucidation revealed that processed McC is an inert analog of the aspartyl-adenylate intermediate generated by aspartyl tRNA synthetase (AspRS). Inhibition of AspRS by processed McC disrupts tRNA^Asp charging, and eventually leads to cell death via the inhibition of protein synthesis. Processed McC is a nanomolar competitive inhibitor of prokaryotic AspRS.^496,500 Additionally, processed McC inhibited in vitro translation in a wheat germ extract system, suggesting that the compound is effective against plant AspRS; however, import of McC into plant cells was not tested.⁵⁰⁰

Active McC is only matured by cell extracts, suggesting that unprocessed McC must be imported into the cytosol of target organisms.⁵⁰⁰ This observation was confirmed as the E. coli YejABEF transport system was shown to actively transport unprocessed McC by recognition of the peptide portion of the conjugate (Figure 46).⁵⁰¹ The N-terminal formyl methionine improves uptake by YejABEF, and conjugates shorter than the fMetArgThrGlyAsnAlaAsp heptapeptide are not imported.⁵⁰² Further studies identified that the R2H, R2Y, A6F, and A6M variants of McC produced smaller zones of growth inhibition against a susceptible E. coli strain, but retained in vitro inhibitory activity, suggesting that Arg2 and Ala6 are required for import.⁵⁰³ Once inside the cytoplasm, unprocessed McC is matured into its bioactive form, wherein endogenous peptide deformylase removes the N-terminal formyl group from Met1,⁵⁰⁴ which is then removed by methionine aminopeptidase.⁵⁰⁵ The remaining polypeptide is digested up to the C-terminal Asp by the endogenous oligopeptidases PepA, PepB, or PepN.⁵⁰⁵ The processed mature aspartyl-adenylate analog can then bind to AspRS. The presence of the aminopropyl group improves inhibition of AspRS in an E. coli B test strain by 10-fold as compared to McC lacking the post-translational modification.⁴⁹⁶ This finding suggests that more favorable interactions are made between the aminopropylated compound and AspRS than the unmodified compound. Lastly, studies utilizing sulfamoyl analogs of McC demonstrate that replacement of Asp7 with other amino acids resulted in analogs that targeted other aaRSs.^506,507 Substitutions to hydrophobic amino acids resulted in less E. coli growth inhibition, suggestive of poor YejABEF uptake or slower metabolism inside the target cell.⁵⁰⁷

Figure 46.

Schematic of McC processing. Unprocessed McC (97) is synthesized in the cytoplasm of the producer. The adenylation reaction converts the C-terminal Asn to an Asp amide. Met1-Ala6 are shown as blue circles, while the decorated Asp amide is shown as an orange circle with a star. After synthesis and export (left portion), unprocessed McC is imported into a susceptible cell via outer membrane transporters OmpC and OmpF, and the inner membrane transporter YejABEF (right portion). Unprocessed McC (97) is deformylated in the cytoplasm of the target cell, allowing for downstream proteolysis. Cellular proteases PepA, PepB, and PepN then remove the hexapeptide (blue circles) revealing the bioactive compound. Processed McC (98) inhibits the aminoacylation reaction carried out by AspRS.

Peptide-cytidinylates are a related class of protein synthesis inhibitors (Figure 47, 99). The first peptide-cytidinylate isolated was from B. velezensis (formerly, B. amyloliquefaciens) and was shown to possess a carboxymethyl modification on the cytidine.⁵⁰⁸ The unprocessed peptide-cytidinylate (pro-McC^Bam) was not bioactive against B. subtilis 168 or several strains of E. coli. Reported reasons for the lack of bioactivity were poor recognition by transporters, poor intracellular transport and processing, or lack of an appropriate cellular target. Subsequently, a second peptide-cytidinylate was isolated from Y. pseudotuberculosis IP32953 (termed McC^YPs), which contains both an aminopropyl and a carboxymethyl moiety (Figure 47, 99).⁵⁰⁹ Like processed McC^Bam, processed McC^YPs showed no bioactivity against an E. coli test strain. Intriguingly, bioactivity was demonstrated after proteolytic processing by TldD/E proteases in the producer strain, which reduces the 42-residue precursor peptide to 11 residues attached to the antibiotic warhead (MIC = 3 μM). The 11-residue conjugate (Figure 47, 100) can be imported by E. coli YejABEF, and the compound undergoes further processing by non-specific proteases in the target cell to generate McC^YPs which inhibits AspRS. Hence, it is probable that the lack of bioactivity observed for pro-McC^Bam could be attributable to either a failure of import of the conjugate or a lack of proteolytic processing in the host strain.

Figure 47.

Peptide-cytidinylate moiety found on McC^Yps/Bam showing the cytidine and carboxymethyl moieties in blue (99). pro-McC^Yps (100) produced after TldD/E cleavage, the 11 amino acids that remain after cleavage are shown in orange. Neither the site of carboxymethylation nor the stereochemistry at phosphorus of these compounds have been established.

4.2. Siderophore-peptide conjugates

Siderophore-conjugated microcins, such as microcin E492 (MccE492) produced by Klebsiella pneumoniae RYC492, consist of linear peptides to which a linearized catecholate siderophore is attached via a bridging b-D-glucose.^510–512 A trimer of N-(2,3-dihydroxybenzoyl)-L-serine (DHBS) anchored at the C-terminus of these peptide-sugar conjugate mimics the siderophores enterobactin and salmochelin, both of which are used in ferric iron acquisition.⁵¹³ Discussion of the MOA of these peptides is complicated by the fact that early studies utilized the unmodified form of these peptides (i.e. the peptide lacking the DHBS and bridging glucose).^514–516 Thus, we will use the “m” notation to denote the fully modified peptide (e.g., “MccE492m” refers to the form that is glycosylated and siderophore-conjugated), while “MccE492” is not modified or partially modified. Examples of siderophore-peptides include MccE492m (101), MccH47m, MccMm (Figure 48), and the uncharacterized MccI47m. These conjugates are hybrids of NRPS and RiPP pathways, similar to the amidinotides (section 11), as the siderophores are likely derived from an NRPS.

Figure 48.

Peptide sequences of characterized siderophore peptides. Residues colored in blue are predicted to be membrane spanning. The glycosylated trimer of N-(2,3 dihydroxybenzoyl)-L-serine (DHBS) is attached to the C-terminal serine carboxylate on MccE492m (101), and the bridging glucose is shown in orange. MccH47m and MccMm are thought to contain the same post-translational modification as MccE492m.

Of these peptides, MccE492 and MccE492m are the most extensively studied. Prior to 2004, the addition of a glycosylated siderophore was not a known modification in MccE492. These earlier studies utilized a recombinant over-producing E. coli strain that produced a compound with a mass consistent with that of Mcc492 (lacking the glycosidic linkage and the siderophore). Despite the lack of modification, Mcc492 was growth-suppressive against various E. coli (MIC = 0.02–0.14 μM), Salmonella ser. Enteritidis (MIC = 0.60 μM) and Salmonella ser. Typhimurium (MIC = 1.2 μM) strains.⁵¹⁶ However, Mcc492 did not inhibit the growth of Klebsiella, Staphylococcus, or Vibrio at 10 μM. Though MccE492 was known to depolarize the E. coli membrane,^514,516 clues related to means of cellular uptake were provided by evidence that enterobactin antagonized the activity of MccE492.⁵¹⁷ This suggested that uptake of MccE492 occurred through the catecholate siderophore receptors Fiu, Cir, and FepA. These receptors bind to the siderophore enterobactin and its hydrolysis products (i.e. DHBS).

Optimization of the growth medium of the same recombinant overproducing E. coli strain revealed that MccE492 could be isolated with a 831 Da adduct at the C-terminus, which was confirmed to be a glycosylated trimer of DHBS (Figure 48).^510,511 This product (namely MccE492m) was detected in the culture medium of the natural producer, K. pneumoniae RYC492.⁵¹⁰ Heterologous production studies and enzymatic studies demonstrate that unmodified or partially modified forms of MccE492 (containing a glycosylated monomer/dimer of DHBS) are the result of incomplete processing.^511,512,518 The antibacterial activity of fully modified MccE492m was 4–8 times greater than that of unmodified peptide against the same panel of Proteobacteria.⁵¹⁰

Additional studies confirmed the role of the FepA, Cir and Fiu as the outer membrane receptors that import the peptide conjugate in a “Trojan horse” manner into target cells (Figure 49). The FepA, Cir, and Fiu receptors serve redundant function as MccE492m import is not compromised if any one or two of these receptors are deleted. The fepA⁻, cir⁻, or fiu⁻ single deletion mutants, and the double mutants cir⁻/fiu⁻ and cir⁻/fepA⁻ remained susceptible to MccE492m. However, the fepA⁻/fiu⁻ double mutant and cir⁻/fiu⁻/fepA⁻ triple mutant were 8- fold or 200-fold more resistant to MccE492m, respectively.⁵¹⁰ Unexpectedly, MccE492 could also be imported by the FepA, Fiu, and Cir outer membrane receptors, albeit the MICs are two to four times higher for MccE492 as compared to MccE492m for various receptor deletion strains.^516,519 These findings underscore that MccE492 and MccE492m are recognized by the same receptors, although activity is notably enhanced in the presence of the DHBS trimer. However, another report found that secreted MccE492 lacking the C-terminal Ser and by consequence the siderophore, was unable to produce zones of inhibition on a lawn of susceptible E. coli.⁵²⁰

Figure 49.

Cartoon schematic of the MOAs for MccE492m and MccH47m. The FepA, Fiu, and Cir outer membrane receptors are responsible for recognizing the siderophore-peptide conjugate. The action of TonB is required to translocate the siderophore-peptide into the periplasm. MccE492(m) targets ManYZ in the occluded state to impede mannose import and utilizes the ManYZ as a scaffold for subsequent membrane insertion. Oligomerization of embedded MccE492 leads to pore formation, leakage of solutes (labeled M⁺) out of the cell, and eventual cell death. MccH47m targets ATP synthase and affects proton translocation.

Screening of E. coli K12 resistant mutants to MccE492 suggested that the cytoplasmic membrane components TonB, ExbB, and SemA could be involved in the mode of action.⁵²¹ TonB/ExbB/SemA are inner membrane proteins that transduce the proton motive force of the cytoplasmic membrane to drive active transport by high-affinity outer membrane transporters, such as FepA, Cir, and Fiu.⁵²² A TonB-deficient strain of E. coli was resistant to MccE492⁵¹⁶ and MccE492m⁵¹⁰ (MICs >10 μM, as compared to 0.02 μM for the wild-type strain). Furthermore, depolarization of the cytoplasmic membrane was not detected in the TonB-deficient strain treated with MccE492.⁵¹⁶ The TonB-deficient strain of E. coli was 500 times more resistant to MccE492 than ExbB-deficient E. coli.⁵¹⁶ These results support a model in which MccE492 or MccE492m binds to the catechol receptors, and require energy transduction by TonB/ExbB/SemA for translocation across the outer membrane.

Transposon mutagenesis revealed that the mannose permease complex (man-PTS) is required for the bactericidal activity of MccE492/MccE492m. Mutations in the manY and manZ genes, which encode for the inner membrane components of the man-PTS, confer resistance to both MccE492 and MccE492m.⁵²⁰ Furthermore, MccE492 co-purifies with ManZ, suggesting that the two may form a physiologically relevant complex.^520,523 Thus, it was hypothesized that ManYZ may serve as a docking protein for MccE492/MccE492m for insertion into the membrane (Figure 49).⁵²⁰

The mechanism of action of MccE492 or MccE492m has been resolved with the cryoEM structure of MccE492 bound to ManYZ.⁵²⁴ The structure of apo-ManYZ revealed an outward facing transporter conformation,⁵²⁵ which upon binding to substrate will shift to an inward facing orientation. This sequence of steps is referred to as the “elevator mechanism”.⁵²⁶ The structure of MccE492 bound to ManYZ reveals the transporter in an occluded state wherein residues Leu61 – Thr77 of MccE492 are embedded in a pocket that prevent the core domain from rotating, thus impeding the ability of the transporter to engage in mannose import. The structure also reveals that the C-terminal Gly63 – Ser71 of MccE492 forms a hairpin structure that anchors ManYZ, while Asp4 – His51 forms an amphipathic α-helical domain that can insert into the inner membrane,⁵²⁴ cause pore formation, and disrupt the proton motive force, leading to cell death.

Additional bioactivities have been reported for MccE492, including cytotoxic activity against some human cell lines.⁵²⁷ MccE492 induces apoptosis in HeLa cells by disrupting mitochondrial function. It was proposed that once inside the cell, MccE492 inserts into the mitochondrial membrane leading to depolarization, release of cytochrome c, and activation of the intrinsic apoptosis pathway. Additional studies are needed to identify the cell surface receptor for MccE492. Lastly, MccE492 was reported to form amyloids-like fibrils which demonstrate reduced bactericidal activity.⁵²⁸ The post-translationally modified MccE492m demonstrates decreased amyloid formation.⁵²⁹ As soluble MccE492m is toxic to cells, these amyloids have been proposed to serve a functional role by serving as a reservoir of active microcins, which can be released in response to small environmental or intracellular changes.^528,530,529

Studies on the MOA of MccH47 (Figure 48) are similarly complicated by difficulties in obtaining pure siderophore-modified peptide (MccH47m). A mixture of MccH47 and MccH47m inhibited growth of E. coli, Shigella, and Salmonella at concentrations < 2 μM, but was inactive against K. pneumoniae, K. oxytoca, Acinetobacter baumannii, Pseudomonas aeruginosa, and S. aureus up to ~100 μM.⁵³¹ Like MccE492m, MccH47m relies on FepA, Fiu, Cir and additionally, the IroN receptors for cellular import.^532,533 Further work on MccH427m demonstrated that TonB- and ATP synthase-deficient strains of E. coli were resistant, suggesting that ATP synthase may be the biological target.⁵³⁴ ATP synthase consists of a membrane-bound component (F₀) and a cytoplasmic component (F₁). Mutations that conferred resistance to MccH47m were localized to the a, b, or c subunits of F_0, which corresponds to the proton channel of ATP synthase. These results suggested that MccH47m affects proton translocation and it was proposed that MccH47m dissipates the proton motive force by dysregulating proton transport (Figure 59).⁵³²

Figure 59.

Sequences and connectivity of lasso peptides referred to in this review. Green lines represent the isopeptide bond, while orange lines represent disulfide linkages.

Lastly, MccM (Figure 48) was shown to be most potent against strains of E. coli, while possessing some activity against Salmonella.⁵³⁵ The bactericidal effect of this peptide is also dependent on the FepA, Fiu, and Cir outer membrane receptors, and TonB.⁵¹⁸ The molecular target of this RiPP is unknown.

5.3. Pantocin

Pantocin A (102) can be classified as a class I microcin as it is a post-translationally modified peptide produced by Pantoea agglomerans and shows biological activity only against related Enterobacteriaceae. Pantocin A, which contains a fused bicyclic core structure formed from two glutamic acid residues to yield an active tripeptide product, was the first identified member of this class and has served as a family prototype (Figure 50).¹⁰ The best studied target of pantocin A is the fire blight pathogen Erwinia amylovora, but the compound is also known to target other plant pathogenic Erwinia species (such as E. rhapontici and Dickeya dadantii).^26,536 Cellular uptake is proposed to occur via unidentified tripeptide transporters as evidenced by suppression of activity in the presence of Ala-Gly-Gly tripeptides during cross feeding experiments.^26,537 This hypothesis is supported by the requirement of a nitrogen-poor environment for bioactivity, wherein the activity of these transporters is essential. Pantocin A activity can be reversed by L-histidine and L-histidinol, and the compound has been shown to inhibit L-histidinol phosphate aminotransferase, a histidine biosynthetic enzyme. This interaction has not been structurally characterized. Other examples of such histidine-reversible antibiotics include herbicolin O⁵³⁸ and MccEh252,⁵³⁹ which likely share the same bioactive warhead as pantocin A, and are all active against plant pathogens. Moreover, the identity of the tripeptide transporters involved in import of these molecules has not yet been identified.

Figure 50.

(A) Structure of pantocin A (102). (B) Reaction catalyzed by L-histidinol phosphate aminotransferase.

6. Methanobactin

Methanobactins (mtbs) are involved in Cu(I) chelation and import by methanotrophic bacteria. These chalkophores were originally identified because of accumulation of an unidentified copper binding ligand in media containing methanotrophs with copper uptake mutations.^540–543 In cultures of wild type cells, methanobactin was shown to accumulate extracellularly and copper binding initiates cellular uptake of the mtb-Cu⁺ complex. A crystal structure of the mtb-Cu⁺ complex (PDB ID: 2XJH) as well as NMR spectroscopy indicated that mtb interacted with the copper ion through two oxazolone-thioamide groups (103, Figure 51).^544,545 Additional characterization demonstrated that mtb binds Cu(I). However, copper loading has been shown to occur with the addition of Cu(I) or Cu(II), and it has been hypothesized that reduction of Cu(II) may occur after loading onto mtb.⁵⁴⁶ Kinetic and spectral data were used to support this hypothesis and to develop a proposed model for mtb binding and reduction of Cu(II) to Cu(I) (Figure 51C). Methanobactins have a high degree of variability in the precursor sequences,⁵⁴⁷ but the mature structures maintain a similar ligand constellation, which is thought to be crucial for copper binding.⁵⁴⁸

Figure 51.

Structure of and proposed model for methanobactin Cu(II) binding. (A) Chemical structure of methanobactin from Methylosinus trichosporium bound to Cu⁺ (103). (B) Crystal structure of methanobactin binding copper (PDB ID: 2XJH). (C) Proposed model for the process of copper binding to methanobactin in the structure shown in panel A.⁵⁴⁶

Because of its toxicity, copper transport is tightly regulated. Methanotrophic bacteria have high copper requirements because one of the methane-converting enzymes is copper dependent. It has been proposed that methanobactin provides an acquisition mechanism, particularly from insoluble copper sources.⁵⁴⁹ Copper uptake and transport experiments concluded that while free copper ions undergo passive uptake through porins, the mtb-Cu complex and apo-mtb undergo active transport. Apo-mtb can act as an inhibitor of mtb-Cu uptake indicating that the same transport system is likely used for both forms.⁵⁵⁰ This transport system was later identified to be a TonB-dependent transporter, MbnT, ensuring import of mtb, and a periplasmic binding protein, MbnE, transporting periplasmic mbt-Cu.^550–552

In addition to its involvement in copper acquisition, it has been proposed that mtb also plays a larger role for copper use within the cell. Two enzymes in methanotrophs are used in the process of methane oxidation, particulate methane monooxygenase (pMMO, a copper-dependent enzyme encoded at the pmo locus) and soluble methane monooxygenase (sMMO, an iron-dependent enzyme encoded at the mmo locus).^553,554 The mtb-Cu complex has been shown to upregulate pmo gene expression. Initial purification attempts of pMMO resulted in the copurification of mtb leading to the conclusion that mtb could be acting as a copper-chelating cofactor.⁵⁴³ A series of structural and spectral characterizations of pMMO demonstrated that the enzyme did not co-crystallize with mtb but instead had three mononuclear copper centers suggesting that mtb does not act as a cofactor but may play a role in Cu delivery to the enzyme.⁵⁵⁵ EPR studies were performed that indicate mtb-Cu increases electron flow to the copper centers of pMMO although the mechanism remains to be fully elucidated.⁵⁵⁶

In addition to copper transport and roles in enzyme activity, mtb and another protein, the transcription factor MmoD, have been proposed to be involved in regulation of the pMMO and sMMO operons. A “copper switch” that upregulates pMMO and down regulates sMMO in the presence of copper^557,558 was hypothesize to directly involve mtb. However, this hypothesis has been questioned because bacteria lacking the mtb biosynthetic genes still exhibit a copper switch.²⁴

7. Signaling peptides

Quorum sensing (QS) is a phenomenon in which small molecules are utilized to modulate gene expression in response to changes in cell density. In Firmicutes, peptide hormones are used to facilitate the QS response, and examples include unmodified peptides and post-translationally modified peptides. Unmodified QS peptides are ribosomally synthesized as propeptides that are cleaved to elaborate their bioactive form.^559,560 As these peptides do not undergo further modifications, they will not be covered in this review.

7.1. Autoinducing Peptides (AIPs) and gelatinase biosynthesis-activating pheromone (GBAP)

Peptides that are post-translationally modified to become signaling hormones belong to one of two classes: staphylococci utilize the macrocyclic thiolactone-containing AIPs (section 7.1.1), while enterococci utilize the macrocyclic lactone-containing GBAP (section 7.1.2). The QS response associated with these two peptides results in the up-regulation and expression of virulence factors, leading to disease states and the ability of the pathogen to evade host immune responses. An understanding of how these compounds interact with their cognate receptors could lead to the development of antagonists that subvert pathogenicity.^561,562

7.1.1. AIPs from S. aureus

S. aureus is a human pathogen that causes a wide range of clinical pathologies including endocarditis and osteomyelitis.⁵⁶³ Among the many S. aureus virulence regulatory systems, a major one is encoded by the accessory gene regulatory locus (agrBDCA) that elaborates the peptide signal in a QS system (Figure 52).⁵⁵⁹ In this locus, agrD encodes the precursor peptide, and AgrB is an integral membrane endopeptidase that macrocyclizes the core peptide to form a thiolactone and cleaves off the C-terminal region of the precursor peptide. The peptide is presumably transferred to the extracellular face of the membrane by AgrB. The peptide is released from the membrane after the leader peptide is removed by the signal peptidase SpsB.⁵⁶⁴ Extracellular accumulation of the AIP leads to signal transduction through the Agr two-component regulatory system (TCS). Binding of the AIP to the AgrC histidine receptor kinase results in autophosphorylation, and subsequent phosphorylation on the transcriptional regulator AgrA (Figure 52). Phosphorylated AgrA can either bind to the P2 promoter, which up-regulates the agrBDCA operon, or to the P3 promoter,⁵⁶⁵ which up-regulates transcription of virulence factors including hemolysins,⁵⁶⁶ enterotoxins,⁵⁶⁷ and proteases.⁵⁶⁸ A complete list of genes and their protein functions regulated by the agr locus has been reviewed.^567,569

Figure 52.

Schematic representation of the (A) agr and (B) fsr signaling pathways. The precursor peptide is synthesized with a leader region (orange). AgrB or FsrB is thought to install the thiolactone or lactone linkage (green) on the core peptide. The C-terminal portion of the AIP precursor peptide is removed and cyclized by AgrB. The AIP is secreted into the extracellular environment by an unknown mechanism. SpsB facilitates removal of the leader peptide. The GBAP leader peptide is thought to be removed by FsrB. Secreted proteases GelE (gelatinase, brown) and SprE (serine protease, blue) are factors associated with pathogenicity.

Allelic variations in the agrBDCA locus have given rise to at least four agr specificity groups (denoted by the roman numerals I, II, III, or IV; Figure 53) amongst different S. aureus strains.^570–572 Each specificity group forms a cognate AIP signal:AgrC receptor pair. The autoinducing cascade is elicited only when a cognate signal:AgrC pair is formed. AIPs from other specificity groups generally inhibit agr activity, as do AIPs from other Staphylococcus sp.^570,573 For certain toxin-mediated diseases, there is a strong correlation between the agr specificity group and the disease phenotype.⁵⁷⁴

Figure 53.

Chemical structures of the AIPs: AIP-I (104), AIP-II (105), AIP-III (106), and AIP-IV (107). Critical hydrophobic residues, which are part of the “triangular knob” motif, are shown in blue (see Figure 54). Substitution of the residues shown in orange greatly perturbs the three-dimensional structure of the AIPs.

SAR studies have focused on delineating the roles of the thiolactone and other residues of the AIPs with respect to agonist/antagonist effects on transcriptional activity. Comparison of synthetic, linear free acid and a linear thioester analog with cyclized AIP-II demonstrated that neither elicited or inhibited the agr response,⁵⁷⁵ highlighting that the macrocycle is functionally indispensable. The necessity of the thiolactone linkage has been probed using lactone or lactam containing analogs of AIP-II (105), each of which failed to activate AgrC-II. Similar behavior was also observed for the AIP-I lactam analog, wherein half-maximal activation was only achieved at 24 μM (~1000-fold higher than the EC₅₀ of AIP-I).⁵⁷⁶ However, the thiolactone linkage is not critical for cross group inhibition, as both lactone and lactam variants of AIP-II were low nanomolar inhibitors of group I agr systems.⁵⁷⁵

Two key features govern the ability of an AIP to bind its receptor: the presence of a hydrophobic patch within the macrocycle and the identity of the residue following the ring-forming cysteine. Alanine scanning of residues within the thiolactone revealed that hydrophobic residues are critical for receptor activation and cross-inhibition. The F6A and I7A variants of AIP-I resulted in an increase in the EC₅₀ of ~500-fold relative to wild-type AIP-I.⁵⁷⁶ AIP-II-L8A and -F9A were unable to activate AgrC-II or inhibit AgrC-I.⁵⁷⁵ AIP-III-F5A, -L6A, and -L7A also failed to illicit cross-group inhibition against AgrC-I, -II, and -IV (IC₅₀ > 200 nM).⁵⁷⁷ The hydrophobic residues in AIPs can tolerate substitution to other bulky non-polar amino acids without loss of activity, as the activity profiles for the AIP-I-M8I or AIP-IV-Y5F did not differ from wild-type.⁵⁷⁸ These observations are reconcilable from the NMR structures of AIP-I (104), AIP-III (106), and AIP-IV (107), which identified a triangular knob formed by the hydrophobic patch. This knob may serve as a point of steric engagement with the receptor (Figure 54).⁵⁷⁹ AIP-II is unstructured in solution⁵⁷⁹ and contains only two hydrophobic residues, as opposed to the three possessed by the other AIPs. Presumably, AIP-II may engage its cognate receptor in a different manner than AIP-I, III, or IV.

Figure 54.

NMR structures of native AIPs. (A) AIP-I (carbon atoms in purple), (B) AIP-II (carbon atoms in light green), (C) AIP-III (carbon atoms in cyan), and (D) AIP-IV (carbon atoms in orange). Hydrophobic residues inside the macrocycle are labeled. Orientations of the triangular knobs are shown in dark blue.

SAR studies have also demonstrated that the residue following the ring-forming Cys is critical for activity. AIPs I, III, and IV possess Asp at this position. AIP-I-D5A and AIP-III-D4A inhibited AgrC across all four specificity groups.^577,578 However, as AIP-I-D5N could still activate AgrC-I, (EC₅₀ = 90 nM vs. 28 nM for native AIP-I), suggesting that receptor activation does not require ionic interactions. The NMR structure of AIP-III-D4A revealed that the Ala substitution collapses the triangular knob giving a more globular conformation (Figure 55).⁵⁷⁹ As a result, AIP-III-D4A presumably cannot make the necessary activation contacts. On the other hand, the NMR structures of AIP-I-D5A and -D5N retain the hydrophobic knob, but reveal a different orientation for the exocyclic tail than the wild-type peptide.⁵⁸⁰ These data suggest that in the context of AIP-I, and the highly similar AIP-IV, the exocyclic Asp is required for positioning the tail in a activation-inducing conformation. Structures of AIPs bound to their cognate receptors would directly answer what AIP conformations are necessary for receptor activation.

Figure 55.

NMR structures of (A) AIP-III, (B) AIP-III D4A, depicting the mutated residue in gold, (C) AIP-I, and (D) AIP-I D5A, depicting the mutated residue in gold. The triangular knob is lost in the AIP-III D4A variant and yields a more globular fold (orange line). The triangular knob is maintained in the AIP-I D5A variant, but the position of the exocyclic tail is different (green vs. blue line are pointing in opposite directions). These changes are likely to impact receptor binding.

The size of the macrocycle is also critical for activity of the AIPs. Contraction (via deletions) or additions (via introduction of L-b-homo amino acids) in the size of the macrocycle proved to be detrimental to agonist activity, without affecting antagonist activity.⁵⁸¹ Taken together with previous observations, these data suggest that receptor agonism is intolerant to changes in AIP structure, while receptor antagonism is generally forgiving. The exocyclic tail also contributes to receptor activation. For instance, tail-less AIP-I (tAIP-I) is a poor activator of its cognate receptor (EC₅₀ = 5.8 μM vs. 28 nM for AIP-I), while tail-less analogs of AIP-II (tAIP-II) and AIP-IV (tAIP-IV) were unable to activate their cognate receptors.⁵⁷⁸ Removal of the tail region results in weak inhibition against the cognate AgrC receptor, e.g. tAIP-I (IC₅₀ = 4.5 μM), tAIP-II (IC₅₀ = 230 nM), tAIP-III (IC₅₀ > 200 nM), and tAIP-IV (IC₅₀ = 4.1 μM).^577,578 As the inhibitory concentrations of the tail-less variants against non-cognate receptors do not significantly differ, it can be surmised that the exocyclic tail is critical for agonist activity.

Little is known about what residues of AgrC mediate AIP binding. AgrC is a transmembrane protein that possesses receptor and signaling domains. To evaluate AIP recognition, chimeric AgrC receptor domains containing the N-terminal half from one class and the C-terminal half from another class were constructed. These studies revealed that the C-terminal region (residues 90–205) of the AgrC receptor is responsible for cognate AIP recognition.⁵⁸² Further investigations of orthologous receptors AgrC-I and AgrC-IV, which only differ by 27 amino acids, revealed that swapping AgrC-I residues 104, 107, and 116 to their equivalents in AgrC-IV switched the activation profile upon presentation with AIP-I and -IV. These substitutions correspond to residues in the predicted second extracellular loop of AgrC receptors,⁵⁸³ which may directly engage the AIP. The first extracellular loop may also play a role in recognition, as swapping residues in this region of AgrC-IV for equivalent residues from AgrC-I, resulted in a 100-fold increase in EC₅₀ when presented with AIP-IV.⁵⁸⁴ AgrC-II and AgrC-III are more sequence divergent than AgrC-I and IV, and further investigation is required to identify if their cognate ligands bind to the same extracellular loops. It should be noted that recognition may not be restricted to a single locus on the protein, and that conformational changes upon peptide binding may also occur. Structural data of ArgC or homologous receptors bound to the cognate signal would help provide insights into the determinants of signal specificity.

7.1.2. Fsr system from Enterococcus faecalis

E. faecalis is an opportunistic pathogen that can cause urinary tract infections, endocarditis, and prosthetic joint infections, which can be life threatening if left untreated.^585,586 Analogous to the agr system in S. aureus, E. faecalis possess the fsr system that contributes to virulence (Figure 52B). An integral membrane protease/transporter installs a lactone (rather than the thiolactone found in the agr system) onto the FsrD precursor peptide to form the mature peptide hormone (GBAP, 108). Binding of extracellular GBAP to the histidine kinase receptor FsrC activates a two component regulatory system in which the phosphorylated FsrA response regulator activates the expression of genes encoding for gelatinase (GelE) and the serine protease SprE (Figure 52B).^587–589 The fsr QS system has been shown to govern the pathogenicity of E. faecalis by regulating the formation of biofilms.^590,591 GelE-positive E. faecalis strains have been reported to induce disease phenotypes in mice and rabbit models.^592,593 In contrast, GelE deletion strains have been associated with increased survival in nematode models.⁵⁹⁴ Moreover, the expression of GelE is positively correlated with biofilm formation.^595,596 For more details we refer to a recent review on virulence factors and E. faecalis pathogenicity.⁵⁹⁷

Two SAR studies have been conducted on GBAP (Figure 56) in which the role of the lactone, the residues in the macrocycle, and the necessity of the tail for agonist activity were tested.^598,599 Replacement of the lactone linkage with a lactam maintained similar levels of agonist activity but replacement by a thiolactone or a disulfide was not tolerated.⁵⁹⁸ The tail region was found to be dispensable for agonist activity, as its complete removal, single alanine substitutions (Q1A and N2A), or substitutions with the corresponding D-amino acid resulted in similar EC₅₀ values as compared to GBAP. Interestingly, the D-Phe7 variant demonstrated only a four-fold increase in EC₅₀ over GBAP, suggesting that the FsrC binding pocket can accommodate the epimer. Alanine substitutions of residues in the macrocycle severely reduced agonist activity, as the I6A, F7A, and G8A variants failed to activate the QS response at 10 μM, suggesting that these residues are the most critical for interactions with the receptor. The NMR structure of wild type GBAP reveals that Ile6-Phe7-Gly8 forms a triangular knob similar to that reported in the structure of AIP-III (Figure 56).^579,598 Hence, it is likely that the GBAP:FsrC engagement is governed by similar interactions.

Figure 56.

(A) Chemical structure of GBAP (108). Critical hydrophobic residues are shown in blue. (B) NMR structure of GBAP highlighting the triangular knob formed by Ile6-Phe7-Gly8.

Although specific residues of FsrC that engage GBAP have yet to be identified, the FsrC receptor-binding domain is predicted to consist of six transmembrane helices and three extracellular loops (akin to AgrC). Synchrotron-radiation circular dichroism spectroscopy on FsrC and FsrC bound to GBAP was used to study the interactions between these receptor-signal pairs.⁶⁰⁰ While no changes to the secondary structure occurred upon GBAP binding, changes to the tertiary structure of FsrC were inferred from spectral shifts in the near UV region (corresponding to signals from Tyr and Trp). Notably, several Tyr and Trp residues reside in the first or second extracellular loops of FsrC, suggesting movement of these residues upon signal binding. Further studies are necessary to further demarcate the regions and residues in FsrC required for GBAP binding.

7.2. ComX

B. subtilis also utilizes peptide pheromones for intra-species signaling. B. subtilis possesses multiple master regulatory proteins including Spo0A, DegU, and ComA. These regulators coordinate the ability of the organisms to differentiate into different subpopulations, as well as to enter states of natural competence.⁶⁰¹ Both responses are important for proper niche establishment and survival. One of the most studied quorum-sensing pathways in B. subtilis comprises the ComQXPA system, in which ComX encodes for a ribosomally synthesized and post-translationally modified peptide signal.

The first ComX isolated was from B. subtilis 168, and was termed ComX168 (Figure 57, 109).⁶⁰² Analysis of the ComX168 open reading frame revealed that the gene encodes for a 55-residue peptide, while the secreted product is a prenylated 10-residue peptide. A farnesyltransferase, ComQ, attaches a C15 lipid (farnesene) to the precursor peptide. Alanine scanning of the last nine amino acids of the core peptide revealed that the unmodified W53A variant was nonfunctional as a signal.⁶⁰³ The NMR structure of a related geranylated pheromone ComX_RO-E-2 from B. subtilis strain RO-E-2 (Figure 57, 110) revealed that the mature compound features a tricyclic tryptophan modification geranylated at the C3 position.⁶⁰⁴ SAR studies on ComX_RO-E-2 revealed that the minimal peptide that retained activity was the geranylated tripeptide [Phe-Trp-Glu] ComX _RO-E-2.⁶⁰⁵ The length of the isoprenyl chain (i.e. C10 vs C15) may dictate specificity groups within this class of molecules.⁶⁰⁶ Currently, there are no studies on the effect of isoprene substitution on bioactivity of ComX peptides.

Figure 57.

Chemical structures of ComX168 (109) and ComX_RO-E-2 (110). The minimal structural features required for ComX_RO-E- activity are highlighted in blue. The inset shows a schematic of ComX pheromone production and activity. ComX peptides activate phosphorylation of ComA mediated by the ComP receptor. Phosphorylated ComA activates expression of surfactin biosynthesis, which is necessary for cellular differentiation in response to environmental changes and production of secreted proteases that are used for cells to escape biofilms.

The interactions that dictate specificity between ComX and its transmembrane receptor are not known, but the biological effects of the signaling cascade have been determined. ComX is sensed by the ComP-ComA two-component regulatory system. Upon binding of ComX to its second extracellular loop, the histidine kinase ComP undergoes autophosphorylation.^607,608 The phosphate group is then transferred to the response regulator, ComA, and phosphorylated ComA activates the transcription of genes found in the sfrA operon.^609–611 One of the proteins encoded in this operon is ComS,⁶¹² which prevents the degradation of transcription factor ComA⁶¹³ permitting the expression of genes related to competence, surfactin biosynthesis, and biofilm formation (Figure 57, inset).^614,615 Surfactin is secreted as a QS signal for neighboring cells to upregulate genes required for cellular differentiation, but is not directly involved in regulating competence.⁶¹⁶ Phosphorylated ComA also activates the transcription of the DegQ transcriptional regulator that enhances the rate of DegU phosphorylation. High concentrations of phosphorylated DegU leads to the production of secreted proteases, which may allow individual cells to escape the biofilm.^617,618 Lastly, as with other classes of autoinducing peptide receptors (section 7.1), polymorphisms in the comQ, comX, and N-terminal portion of comP reveal that B. subtilis strains can be classified into one of four pherotypes,⁶¹⁹ wherein members of the same group can interact effectively with each other, but not between groups.

8. Peptide atropisomers

Atropisomerism is a type of conformational chirality that results from hindered rotation about a single bond, when the energetic barrier for rotation is high enough to allow for the isolation of individual conformers. Non-canonical atropisomers have identical connectivities and configurations at all stereogenic centers but cannot be converted by rotation along a single bond. For example, threaded lasso peptides may be considered as non-canonical atropisomers of the unthreaded counterpart.⁶²⁰

8.1. Lasso peptides

The lasso peptides are a class of RiPP with a class-defining macrolactam formed between the N-terminus and an internal Asp or Glu residue (an isopeptide bond). This macrolactam ring encircles the peptide forming a lariat knot topology. The C-terminal tail is locked within the ring and held in place either sterically by the presence of a large residue (often referred to as “plugs”) or by one or more disulfide bonds. This threaded, rotaxane-like conformation is known colloquially as the “lasso fold”, as opposed to an unthreaded, “branched-cyclic” conformation (Figure 58). Formally, lasso peptides may be considered as members of the class I microcins (see Section 5 on microcins), and microcin J25 is one of the first and among the best studied lasso peptides. Lasso peptides are divided into four subclasses based on the presence and connectivity of disulfide bonds (Figure 58). The sequence diversity of the lasso peptides and their attendant MOAs are broad, including antibacterial activity, receptor antagonism, and antiviral activity; these features make the lasso peptides a promising class of RiPPs for further discovery and development.^621,622

Figure 58.

(A) Schematic structure and classification of lasso peptides. Important regions of a generic lasso peptide are shown. (B) Class I lasso peptides have two disulfide bonds (orange) that covalently join ring-to-loop and ring-to-tail. A large majority of lasso peptides belong to class II, which lack disulfide linkages; instead, the loop is sterically thought to be locked into place by the side chains of “plug” residues (orange circles) in the tail that are positioned below the ring and sometimes also above the ring. Class III lasso peptides are joined by ring-to-loop disulfides while class IV display a tail-to-tail disulfide linkage. The numbering of the residues in the schematic structure of class I shows that the lasso peptide is right-handed, which is the natural isomer for all currently known lasso peptides.

Prior to the genomics revolution, virtually all natural products were discovered via bioassay-guided approaches. Lasso peptides discovered during this time (pre-2000) are primarily active against cell-surface receptors. The first, anantin (Figure 59), is an antagonist of the atrial natriuretic factor receptor, with a reported K_d of 600 nM and an in vitro IC₅₀ of 1.0 μM.⁶²³ Currently, little is known about the molecular level details of the peptide-receptor interaction. Subsequent discovery and evaluation of RES-701–1 (Figures 59, 60) demonstrated selective antagonism of the endothelin type B receptor (IC₅₀ = 10 nM) over endothelin type A receptors.⁶²⁴ Related lasso peptide congeners RES-701–2, −3, and −4 were also selective antagonists of the endothelin type B receptor.⁶²⁵ This activity was dependent on a threaded conformation as chemically synthesized, branched-cyclic RES-701 lacked the selectivity/activity of naturally isolated threaded RES-701–1.^626,627

Figure 60.

(A) Structure of endothelin type B receptor bound to the human hormone endothelin-1 peptide (PDB ID: 5GLH). The transmembrane (TM) domain is shown in green, while the intracellular domain is shown in orange. (B) Close up of endothelin-1 bound to the endothelin type B receptor. Portions of the linear tail of endothelin-1 (Ile19-Trp21) project into the cavity of the receptor. RES-701–1 and its congeners may bind in a similar way. Peptide sequences of the hormone endothelin 1 and the lasso peptide RES-701–1 are shown (the isopeptide linkage is shown in green and disulfide links in orange).

SAR studies demonstrated that the C-terminal Trp of RES-701–1 is directly involved in the selectivity for the endothelin type B receptor. Methyl esterification of the C-terminus yielded a dual selective analog that demonstrated affinity for both endothelin type B and type A receptors. Deletion of the C-terminal carboxylate to yield tryptamine resulted in a 5-fold loss in IC₅₀ against the endothelin B receptor (IC₅₀ = 10 nM vs. 49 nM).⁶²⁸ Further studies demonstrated that substitution of the C-terminal Trp to Phe, Tyr, or removal of Trp had only a modest effect on the binding activity against endothelin B.⁶²⁹ Substitution with Gly or Ala substantially weakened the in vitro IC₅₀ against endothelin B by 12- and 50-fold, respectively. Substitution at the C-terminus with β−2-naphthyl-alanine modestly enhanced the IC₅₀ from 10 nM to 6.3 nM. These SAR studies suggest that larger/aromatic sidechains at the C-terminus are involved in receptor binding, and replacement with smaller residues destabilized the interaction of RES-701–1 with the endothelin B receptor. The bioactivity of RES-701–1 was also demonstrated in isolated tissue samples.^630,631 There are no structures of a complex between RES-701–1 and the endothelin B receptor, which may provide a framework for further optimization of this compound.

The lasso peptide BI-32169 (Figure 59) inhibits glucagon-dependent cAMP production by the human glucagon receptor with an IC₅₀ of 440 nM.^632,633 The activity of BI-32169 was enhanced by methyl esterification of the C-terminus. The first, and to date only, successful total chemical synthesis of a lasso peptide (BI-32169) in a threaded conformation was achieved using a cryptand-imidazolium support.⁶³⁴ The threaded conformation of synthetic BI-32169 was confirmed using a number of analytical and biophysical methods, and using the natural product as a standard. Moreover, synthetic BI-32169 afforded opportunities for SAR studies. Specifically, anti-glucagon receptor antagonism was enhanced from 520 nM to 160 nM when a BI-32169 variant consisting of all D-amino acids was used against BHK-21 cells expressing the human glucagon receptor. To date, no detailed SAR studies have shed light on the glucagon receptor:BI-32169 interaction.

Several characterized lasso peptides exhibit protease inhibitory activity. Propeptin (Figure 59) displays low micromolar activity against human, bovine, and flavobacterial prolyl endopeptidases (IC₅₀ ~0.35 μM), and antibacterial activity against Mycobacterium phlei.^635,636 Propeptin-2, which lacks the two most C-terminal residues of propeptin (Ser-Pro), displays two-fold decreased activity against flavobacterial prolyl endopeptidase and loses antibacterial activity against M. phlei.⁶³⁷

The lasso peptide lassomycin (Figure 59) is growth-suppressive against multiple clinical and drug-resistant strains of M. tuberculosis (MICs = 0.83–1.65 μM) and Mycobacterium sp.⁶³⁸ Spontaneous lassomycin-resistant bacteria possessed mutations to the molecular chaperone casinolytic protease C (ClpC1), which unfolds misfolded or aggregated proteins in an ATP-dependent manner and translocates these proteins to ClpP1/P2 proteases for protein recycling.^638–640 In the presence of lassomycin, purified mycobacterial ClpC1 exhibited increased ATPase activity but a decrease in the ability to degrade the studied substrate casein. These data suggested that lassomycin inhibited translocation of a bound protein substrate to the proteolytic active site. Structural modeling studies indicated lassomycin likely interacts with the N-terminal half of ClpC1 (Figure 61). Resistance mutations against lassomycin were localized to the N-terminal region of ClpC1. The NMR structure of lassomycin was initially and incorrectly assigned as a branched-cyclic conformer (PDB ID: 2MAI), leading to questions on whether the threaded lasso conformation was necessary for bioactivity. Chemically synthesized lassomycin was isobaric to the naturally occurring compound, but exhibited markedly different NMR chemical shifts and lacked any anti-M. tuberculosis bioactivity up to 55 μM.⁶⁴¹ These results proved that the synthetic, branched-cyclic form of natural lassomycin was incorrect and the threaded conformation is now presumed to be essential for bioactivity.

Figure 61.

Crystal structure of ClpC1 (PDB ID: 3WDB). The N-terminal portion of the protein (circled and labeled) is proposed to interact with lassomycin.

Siamycin I and II (Figure 59)⁶⁴² were originally identified in a syncytia inhibition (SI) screen for inhibitors of viral fusion; both compounds showed inhibitory activity against HIV and HSV.⁶⁴³ In the SI assay, a cell line expressing the HIV envelope proteins gp160 (or the cleavage products gp120 and gp41) is co-cultured with HeLa cells expressing the CD4 receptor.⁶⁴⁴ Cell fusion generates polynuclear cells (syncytia), the formation of which is prevented by viral fusion inhibitors. Anti-HIV activity (ID₅₀) as measured in a cell viability assay for siamycin I and II were determined to be 3.2 and 4.1 μM, respectively; while anti-HSV-1 ID₅₀ for siamycin I was 22 μM and siamycin II was 12 μM.⁶⁴³ Furthermore, siamycin I possesses antiviral activity against laboratory strains of HIV-I (ED₅₀ = 0.05–4.5 μM) as well as clinical isolates that are resistant to HIV protease and reverse transcriptase inhibitors (ED₅₀ = 0.89–5.7 μM).⁶⁴⁵

Investigations of siamycin I-resistant HIV strains revealed multiple point mutations in both gp160 and gp41. Utilizing the SI assay, it was demonstrated that preincubation with siamycin I, followed by removal of the compound restored syncytia formation, suggestive of non-covalent inhibition. A third anti-HIV lasso peptide, siamycin III (originally RP 71955) was discovered using a fluorometric assay to screen for inhibitors of HIV-1 protease.^646,647 Solution structure determination of siamycin III revealed a structure similar to that of siamycin I/II.⁶⁴⁷ Detailed mechanistic analyses on the anti-HIV activities of these peptides have not yet been reported.

In addition to inhibition of viral fusion, siamycin I showed antibacterial activity. S. aureus ATCC 29213-resistant mutants against siamycin I exhibited cross-resistance to both vancomycin and nisin (see section 2), which are both antibiotics that target the cell wall.⁶⁴⁸ Treatment of B. subtilis with siamycin I induced the liaI (lipid II-interacting antibiotic) cell wall stress response, which is generally indicative of a cell wall target (sections 2.2 and 15.4).⁶⁴⁹ Follow-up studies revealed that fluorescently labeled siamycin I localizes to the division septum of S. aureus, and in doing so, delocalizes penicillin-binding protein 2 (PBP2) from its usual site of action.⁶⁴⁸ In vitro studies determined that siamycin I inhibits the transglycosylation reaction catalyzed by PBP2, as detected by reduced undecaprenyl pyrophosphate release from lipid II.

Siamycin I at sub-lethal concentrations attenuates the fsr quorum-sensing system of E. faecalis. This quorum-sensing system is autoinduced by the peptide GBAP (section 7.1.2) and leads to the expression of two virulence factors: the gelatinase GelE and a serine protease SprE. Siamycin I inhibits GelE expression by decreasing GBAP production, which was attributed to disruption of the fsr two-component signal transduction proteins FsrC-FsrA (Figure 52B).⁶⁵⁰ Subsequent work showed siamycin I inhibited the autophosphorylation activity of FsrC,⁶⁵¹ and spectroscopic studies supported a direct interaction between siamycin I and the FsrC kinase at a site distinct from the binding site of the native GBAP ligand.⁶⁵² Siamycin I also possessed in vitro inhibitory activity against several ATP-dependent enzymes, including membrane sensor kinases, protein kinases, and ATPases of both prokaryotes and eukaryotes.⁶⁵¹ Despite these in vitro activities, siamycin I lacks anti-Proteobacterial activity in vivo, suggesting that this lasso peptide cannot cross the bacterial outer membrane.^643,651 The related class I lasso peptide sviceucin also demonstrated anti-fsr activity.⁶⁵³ Sviceucin has an MIC of 10 μM against E. faecalis, but treatment with sublethal dosages, below 1 μM, reduced GelE production by 50–70%.⁶⁵³ It has yet to be determined if sviceucin also targets FsrC.

Streptomonomicin (Figure 59), produced by Streptomonospora alba, is a cell wall-targeting lasso peptide that showed low μM MICs against multiple Bacillus sp., but little to no activity against fungi or Proteobacteria.⁶⁵⁴ Sequencing of spontaneously resistant B. anthracis colonies revealed five mutations to the walR gene and a single mutation in a regulatory region directly upstream. WalR is a response regulator that functions along with the histidine kinase WalK to form a two-component signal transduction system that controls the expression of proteins involved in cell wall metabolism and division.^655,656 Of the isolated WalR variants, three contained amino acid substitutions in the DNA-binding domain, whereas the other two had substitutions in the phosphate group receiver domain.⁶⁵⁴ These B. anthracis mutants demonstrated a chaining phenotype, indicative of a septal cleavage defect, where daughter cells do not separate after division.⁶⁵⁴ The WalRK system regulates the expression of LytE, an autolysin involved in septal cleavage in Bacillus sp,^654,657 and decreased expression of LytE would be consistent with the elongated phenotype. Downregulation of lytE was demonstrated for streptomonomicin-resistant mutants, leading to the hypothesis that streptomonomicin targets cell wall biosynthesis through the WalR protein.⁶⁵⁴ Despite these studies of resistant mutants, the details of the MOA of streptomonomicin remains unknown. Further study is needed to confirm or refute WalR as the target of streptomonomicin.

Several lasso peptides target proteobacterial RNA polymerase (RNAP). The best-studied of these is the class II lasso peptide microcin J25 (MccJ25), which demonstrates in vivo and in vitro activities against clinically relevant strains of E. coli, Salmonella sp., and Shigella sp.^658–660 Studies of spontaneously resistant E. coli mutants identified a variant possessing a T391I substitution in RpoC, the β’ subunit of E. coli RNAP.⁶⁶¹ RNAP as the target for MccJ25 was supported by complementation studies and in vitro and in vivo transcription assays. MccJ25 did not demonstrate in vitro activity against non-proteobacterial RNAP, including homologs from T. aquaticus, B. subtilis, and yeast.⁶⁶² The site of MccJ25 binding was suggested to be the E. coli RNAP secondary channel, which was supported by identification of resistant mutations that mapped within and near this channel (Figure 62). Biophysical characterization demonstrated that MccJ25 binds to and occludes the secondary channel, preventing entry of nucleotides to the polymerase active site and locking RNAP into an inactive conformation.^663,664 Saturation mutagenesis of the secondary channel demonstrated that substitutions at multiple sites within this channel endowed resistance to MccJ25.⁶⁶⁴ The co-crystal structure of MccJ25 bound to E. coli RNAP revealed that MccJ25 is bound primarily through hydrophobic contacts in the region between the bridge helix, activation loop, and F loop.⁶⁶⁵ MccJ25-Tyr9 is essential for RNAP inhibition as substitutions to any other amino acid abolished activity.⁶⁶⁶ The C-terminal tail is necessary for RNAP interaction and inhibition as confirmed by structural studies.⁶⁶⁵ Amidation of the C-terminal Gly of MccJ25 inhibited RNAP interaction in vitro but did not abolish peptide uptake.⁶⁶⁷

Figure 62.

(A) Surface model of E. coli RNAP β’ subunit bound to MccJ25 (PDB ID: 6N60). (B) Close up of bound MccJ25 (carbon atoms in green). MccJ25 interacts with the E. coli RNAP β’ subunit primarily through hydrophobic interactions. The activation loop (purple), F loop (red), and bridge helix (tan) and active site Mg²⁺ are also shown. (C) Location of resistance-conferring variants are shown as spheres in a wire model of E. coli RNAP β’ subunit. Spheres corresponding to residues in the activation loop (purple), F loop (red), or bridge helix (tan) are colored as in panel B. Sites of other variants are shown in blue.

As MccJ25 shows in vivo activity against Proteobacteria, the compound must cross both the inner and outer membranes to bind to the intracellular RNAP target. Transposon mutagenesis endowing MccJ25 resistance identified insertions in the fhuA gene, which encodes a multifunctional outer membrane protein largely involved in siderophore transport.⁶⁶⁸ A model of MccJ25 internalization was established with the identification of additional resistant E. coli strains that possessed mutations in tonB, exbB/D, and sbmA genes.²⁸⁶ These genes encode components of the TonB membrane transport pathway. Binding of TonB, ExbB, ExbD, and SbmA to an unidentified proton channel induces a conformational change in FhuA, causing it to open and presumably import MccJ25.

S. ser. Typhimurium contains an FhuA homolog but is otherwise resistant to MccJ25. However, S. ser. Typhimurium can be sensitized to MccJ25 when provided with E. coli fhuA. This result provides a framework for understanding how sequence differences in FhuA may account for MccJ25 import.^669,670 Subsequently, mutational studies on E. coli FhuA suggested that MccJ25 bound the exterior loops of the protein, as the removal of these loops abolished MccJ25 bioactivity.⁶⁷¹ MccJ25 binds to E. coli FhuA with a K_d of 1.2 μM and stoichiometry of 2:1 MccJ25 to FhuA.⁶⁷² The co-crystal structure of E. coli FhuA-bound MccJ25 provided definitive evidence of a direct interaction of MccJ25 with the exterior cavity of FhuA, which is similar to the binding mode of the native ferrichrome ligand (Figure 63).⁶⁷³ However, ferrichrome interacts with FhuA through extensive polar contacts while MccJ25 makes limited contacts with FhuA via the MccJ25-His5 imidazole and the MccJ25-Ala3 backbone carbonyl. MccJ25-His5 engages in a critical hydrogen bonding contact with the backbone amide of FhuA-Phe115.

Figure 63.

Co-crystal structures of FhuA-MccJ25 and FhuA-ferrichrome. (A) The co-crystal structure of MccJ25 (green spheres) bound to FhuA (light blue ribbons) demonstrates that the lasso peptide is bound to the exterior cavity of FhuA (PDB ID: 4CU4). Ferrichrome (brown spheres) bound to FhuA (PDB ID: 1BY5) occupies a similar region. (B) Close-up of the ferrichrome binding pocket. FhuA residues engaging in hydrogen-bonding interactions are shown in yellow (black numbering), while carbon atoms of ferrichrome are shown in brown. (C) Close-up of the MccJ25 binding pocket. Residues engaging in hydrogen-bonding interactions are shown in yellow (black numbering for FhuA, green borders for MccJ25; carbon atoms of MccJ25 shown in green).

SAR studies of MccJ25 established the role of the threaded conformation for bioactivity. Thermolysin cleavage of the β-hairpin loop region of MccJ25 (between Phe10 and Val11) generated an interlocked dual-chain species (t-MccJ25), which increased the MICs, as measured by growth inhibition in liquid culture, to 80–150 nM against S. enteritidis (MccJ25, <2 nM) and 300–600 nM against a TolC-deficient strain of E. coli (MccJ25, 2–5 nM).^674,675 In vitro transcription studies demonstrated that both MccJ25 and t-MccJ25 inhibited E. coli RNAP and that losses in bioactivity resulted from impaired FhuA-dependent uptake.^676,677 Modification of MccJ25-His5 by carbethoxylation also reduced activity, likely by interfering with peptide uptake.⁶⁷⁸ Additional studies on the role of MccJ25-His5 suggested that this residue was important for internalization by the SbmA transporter.⁶⁷⁹ Bypass of FhuA by osmotic shock did not restore antibacterial activity to the MccJ25-H5A or -H5R variants, supporting the importance of the inner membrane transporter SbmA for MccJ25 to gain cytosolic access. Recently, the cryo-EM structure of SbmA bound to a monoclonal antibody has been determined to 3.2 Å resolution, albeit in the absence of bound ligand such as MccJ25, klebsazolicin or MccB17 (see section 3.1 on LAPs).⁶⁸⁰ The structure shows a previously unknown fold in an outward-open conformation exposing a large cavity. This transport cavity is of sufficient size (~23 Å) to accommodate MccJ25 (or a LAP), and targeted mutation of residues lining the transport cavity decreased translocation of MccJ25. Currently, no resistance-conferring mutations for sbmA have been reported, which would shed additional light on the interaction of MccJ25 and the SbmA inner membrane transporter.

Microcin Y(MccY) is a recently discovered lasso peptide produced by S. enterica serovar Birkenhead which possesses 52% sequence identity (13 of 25 residues) with MccJ25 (Figure 59).⁶⁸¹ Despite this high relatedness, the antibacterial spectrum of MccY differs from MccJ25. MccY inhibits the growth of Salmonella infantis while MccJ25 is inactive (MIC = 0.04 μM vs. >200 μM).⁶⁸¹ Unlike MccJ25, MccY does not inhibit the growth of E. coli (tested upwards of 200 μM).⁶⁸¹ E. coli transformed with FhuA from S. enterica serovar Typhimurium became susceptible to MccY (MIC = 12.5 μM), suggesting that MccY and MccJ25 share an outer membrane uptake mechanism, though the specific intracellular target of MccY is unknown. Furthermore, while MccJ25 is translocated across the inner membrane by SbmA, it is currently unknown if MccY also utilizes this protein, as spontaneously resistance mutants have not been studied. Additional studies are necessary to determine if receptor (FhuA/SbmA) or putative target (RNAP) amino acid sequences are the determinate factor for MccY/MccJ bioactivity spectra.⁶⁸¹

Secondary MOAs for MccJ25 have been proposed, including dissipation of the cytoplasmic membrane potential, superoxide production, and antimitochondrial activity. In S. enterica serovar Newport, MccJ25 was shown to inhibit respiration, as measured by oxygen consumption in the presence of 10 mM D-glucose.⁶⁸² Assays using the tetrazolium dye MTT demonstrated a marked decrease in the activities of reduced nicotinamide adenine dinucleotide (NADH) and succinate dehydrogenases in isolated membrane fractions of MccJ25-treated S. ser. Newport, though these observations were not observed in assays against E. coli. Effects on oxygen consumption were observed in S. ser. Typhimurium, S. ser. Derby, and S. ser. Enteritidis strains transformed with the E. coli FhuA protein.^670,677 Studies using bacteriomimetic vesicles showed that MccJ25 inserted into and increased the fluidity and permeability of these membranes in a FhuA-independent manner.^683,684 2,4-Dinitrophenol-dependent dissipation of the proton motive force, which FhuA uses for active transport of MccJ25, did not drastically impair the antibacterial activity of this lasso peptide, suggesting that internalization by FhuA is not the only means of import.⁶⁸⁵

Reactive oxygen species (ROS) production has been observed in E. coli overexpressing FhuA treated with MccJ25.⁶⁸⁶ Utilizing a C-terminally amidated analog of MccJ25, which does not interact with RNAP (discussed above), E. coli FhuA-overexpression strains were sensitized to MccJ25 through the production of ROS.⁶⁸⁷ E. coli strains expressing FhuA with a deletion of either cytochrome bd-I or bo₃ remained resistant to MccJ25, implicating these two enzymes as potential targets of MccJ25. MccJ25 was shown to weakly inhibit purified cytochrome bd-I, slowing the rate of electron transfer by the enzyme, and resulting in a net increase of superoxide production.⁶⁸⁸ Follow up studies confirmed the role of MccJ25-Tyr9 in superoxide production, and electron paramagnetic resonance studies demonstrated the formation of a tyrosyl radical and its participation in the redox reaction cycle.^689,690

Antimitochondrial activities of MccJ25 have been observed in rat heart, wherein MccJ25 inserted into the membrane, dissipated the proton motive force, reduced intracellular ATP, and inhibited complex III of the cytochrome c reductase respiratory chain.⁶⁹¹ Through ROS production, MccJ25 was shown to activate the mitochondrial permeability transition pore, leading to mitochondrial swelling and cytochrome c release.⁶⁹² Furthermore, MccJ25 was shown to induce carbonylation of mitochondrial proteins and oxidation of mitochondrial lipids, both of which would induce the release of cytochrome c.⁶⁹³ MccJ25-dependent effects on the mitochondrial permeability transition pore and membrane were only observed in the presence of 20 μM MccJ25, suggesting this may not be a physiologically relevant MOA.^692,693 Despite this, interference with oxidative respiratory cycles demonstrates the mechanistic plasticity of lasso peptides, which could be improved through engineering efforts.

Though MccJ25 has been extensively investigated in terms of its bioactivity, other lasso peptides including capistruin,⁶⁹⁴ acinetodin,⁶⁹⁵ klebsidin,⁶⁹⁵ citrocin,⁶⁹⁶ and ubonodin (Figure 59) are also known to inhibit E. coli RNAP.⁶⁹⁷ Capistruin, a class II lasso peptide, displays growth-suppressive activity against certain Proteobacteria (e.g., Burkholderia sp. and Pseudomonas sp.).⁶⁹⁸ Capistruin inhibited transcription activity of E. coli in vitro and MccJ25-resistant RNAP was also resistant to capistruin.⁶⁹⁴ Structural work on capistruin bound to RNAP demonstrated MccJ25-like binding within the secondary channel (Figure 64).⁶⁶⁵ However, capistruin is only weakly active (MIC = 25 μM) against E. coli, suggesting that capistruin is not able to cross the E. coli outer membrane as effectively as MccJ25.⁶⁹⁸ In support of this hypothesis, structural studies of E. coli FhuA demonstrated that capistruin could not be modeled into the exterior cavity without steric clashes.⁶⁷³

Figure 64.

Capistruin binding to RNAP. Capistruin (blue carbon atoms) occupies a similar binding site as MccJ25 (green carbon atoms, see Figure 62) but is not positioned as close to the RNAP active site. The closest residue to the catalytically requisite Mg²⁺ ion is Arg15 (12.8 Å away), as compared to His5 in MccJ25 (5.9 Å).

Capistruin is active against Burkholderia suggesting that Burkholderia FhuA can bind the lasso peptide.⁶⁷³ This proposal is speculative as currently no data suggest that capistruin is imported by a TonB-dependent siderophore receptor. Like capistruin, acinetodin and klebsidin are active against E. coli RNAP, and MccJ25-resistant variants exhibited cross-resistance to both lasso peptides.⁶⁹⁵ Additionally, while these compounds were inactive against E. coli, klebsidin was active against K. pneumonia. E. coli could be sensitized to klebsidin through expression of the K. pneumonia FhuA protein, which possesses 60% amino acid sequence identity to E. coli FhuA. These findings support a mechanism where these lasso peptides are imported by a FhuA/TonB-dependent receptor and rationalize the relatively narrow ranges of bioactivity.

The lasso peptides citrocin produced by Citrobacter pasteurii type strain CIP 55.13 and ubonodin produced by Burkholderia ubonesis MSMB2207 (Figure 59) display RNAP inhibitory activity in vitro, although the physiological relevance of this activity has not yet been established.⁶⁹⁶ At 1 μM, citrocin reduced transcription levels of E. coli RNAP to 15%. E. coli mutants resistant to citrocin contained alterations in sbmA but not in fhuA, indicating that citrocin is imported into susceptible cells through a different transporter than MccJ25, which engages FhuA. Notably, citrocin has more potent E. coli RNAP inhibitory activity despite possessing a higher MIC against E. coli BW25113, which expresses TolC (citrocin, 16–125 μM; MccJ25, 1 μM as determined in the same study by spot-on-lawn assays). This finding is likely due to differences in ability for citrocin to cross the outer membrane of Proteobacteria. The activity of ubonodin was initially tested against E. coli RNAP, revealing only 60% transcription inhibition at a 100 μM dose.^696,697 The high concentration required for inhibitory activity suggests that E. coli RNAP may not be a physiologically relevant target of ubonodin. Moreover, ubonodin is unable to produce zones of inhibition on E. coli lawns but inhibits growth of several Burkholderia sp., particularly Burkholderia cepacia (MIC = 4 μM).⁶⁹⁷ Information on the viability of Burkholderia RNAP as the target of ubonodin is not currently available.

Lariatin A and B (Figure 59) were isolated from Rhodococcus sp. K01-B0171, and were initially observed to be growth-suppressive against M. smegmatis and M. tuberculosis.⁶⁹⁹ SAR studies of lariatin A have been performed, identifying residues necessary for bioactivity.^700,701 Tyr6 could be replaced with other aromatic residues (Trp, Phe) without impacting the size of the zone of growth inhibition. Alanine substitutions at Gly11 and Asn14 resulted in the complete loss of activity. Lastly, Lys17 could be replaced with Arg, but not Ala, without any loss of activity, suggesting that a positively charged residue is critical for activity.⁷⁰¹ The molecular target(s) of lariatin A/B have not been reported.

Humidimycin (Figure 59) is an antifungal lasso peptide that was identified by high-throughput screening for potentiators of the antifungal drug caspofungin against pathogenic Aspergillus fumigatus and Candida albicans.⁷⁰² Humidimycin, when introduced with caspofungin, prevented phosphorylation/activation of SakA, the primary controller kinase of the high osmolarity glycerol pathway. Deletion mutants of SakA exhibited higher sensitivity to humidimycin, suggesting that this pathway is the target of the lasso peptide, though no details have been reported. These findings are of interest given the previously observed kinase inhibitory activity observed in the closely related lasso peptide siamycin I.⁶⁵¹

8.2. Atroptides

Tryptorubin A (111), the only reported atroptide, was isolated from the fungus-derived bacterium Streptomyces sp. CLI2509 (Figure 65).⁷⁰³ Structural studies demonstrated that tryptorubin A contained a Trp-Trp crosslink and an unusual Tyr-Trp linkage. These two features give rise to a highly strained macrocycle. Initial bioinformatics predictions suggested that a hybrid type III PKS/NRPS encoded the hexapeptide, but subsequent analysis suggested that tryptorubin A is the product of a cytochrome P450 modification of a genetically encoded precursor peptide. P450 enzymes are primarily associated with hydroxylation reactions, but they are known to install C-C and C-N bonds,^704,705 but experimental evidence of the biosynthetic mechanism has yet to be established. Structural elucidation of tryptorubin A revealed that it could exist as one of two non-canonical atropisomers that cannot be interconverted by rotation along a single bond;⁶²⁰ chemical synthesis was used to access both atropisomers. There are no reports associated with the bioactivity of tryptorubin A.

Figure 65.

Chemical structure of tryptorubin A (111). The macrocycle formed by Trp2, Tyr3, and Trp5 is shown in blue.

9. Graspetides

The graspetides are a class of RiPPs defined by the presence of one or more lactone/lactam rings formed between the ω-carboxy groups of Glu/Asp and the hydroxy groups of Ser/Thr (macrolactone) or the ε-amino group of Lys (macrolactam) (Figure 66).⁷⁰⁶ The formation of the ester or amide linkages are catalyzed by enzymes of the ATP grasp ligase superfamily, hence the classification of these products as graspetides.¹⁰ Initially, graspetides were largely identified in various genera of freshwater Cyanobacteria, including the bloom-forming Microcystis and Planktothrix. More recent genome mining efforts have identified graspetides in phyla outside of Cyanobacteria, including numerous terrestrial species.^707–709 In general, the graspetides are known serine protease inhibitors (Table 2).

Figure 66.

Alignment of the microviridin core sequences. Residues that are involved in macrolactone (orange and purple background) and macrolactam (green background) formation are shown. The identity of residue at position 5 (blue background) largely dictates target protease specificity. A similar figure is reported in ref (³⁴) but has errors at positions 12 and 13 in the last two entries.

Table 2. IC₅₀ values of characterized graspetides.

IC₅₀ (μM) values of characterized graspetides, unless otherwise indicated. N.D. indicates not determined.

Graspetide	P1 Residue	Trypsin	Elastase	Chymotrypsin	Ref.
Microviridin A	F	>58	>58	>58	710
Microviridin B	L	34	0.03	1.5	711
Microviridin C	L	18	0.05	2.8	711
Microviridin D	M	>55	0.4	0.7	712
Microviridin E	Y	>60	0.4	0.7	712
Microviridin F	Y	>59	3.5	>59	712
Microviridin G	L	>55	0.01	0.77	713
Microviridin H	L	>55	0.017	1.57	713
Microviridin I	L	N.D.	0.19	N.D.	714
Microviridin J	R	0.02–0.09	>6	1.7	715
Microviridin K	M	>100	N.D.	N.D.	716
Microviridin L	F	58	>58	42	717
Microviridin M	L	>34	2.9	>34	718
Microviridin N3	L	>30	2.5	>30	718
Microviridin SD1634	R	8.2	N.D.	15.7	719
Microviridin 1777	L	>10	0.16	0.10	720
Plesiocina	L	>100	0.016	0.0075	721
Marinostatinb	M	>50	N.D.	N.D.	722

^aReported as K_i; Produced by heterologous expression in E. coli.

^bReported as K_i; Performed using chemically synthesized compound. For IC50 values against a wider panel of proteases, see a recent review.³⁴

The majority of characterized graspetides belong to the microviridin family. Group I microviridins (e.g., microviridins A, B, G, I, J, K, L) contain two macrolactones (Thr-Asp and Ser-Glu) and one macrolactam (Lys-Asp/Glu). Groups II and III microviridins (e.g., microviridins C, D, E, H and microviridin SD1652, respectively) retain the macrolactam linkage displayed by group I but only possess one macrolactone (group II: Thr-Asp; group III: Ser-Glu). Group IV microviridins also retain the macrolactam linkage of groups I-III but do not possess any other crosslinks (e.g. microviridin F, 122, Figure 69).³⁴ The tricyclic compound microviridin A (112, Figure 67) was the first characterized graspetide, and was identified during a screen for human tyrosinase inhibitors (IC₅₀ = 330 μM).⁷¹⁰ Mammalian tyrosinase plays a key role in the development of melanin.⁷²³ Excessive melanin production (hyperpigmentation) has been linked to Parkinson’s disease and some skin cancers, and tyrosinase has been validated as a therapeutic target for treating hyperpigmentation.⁷²⁴ The discovery of microviridin A as an inhibitor of tyrosinase suggested the possibility of utilizing the tricyclic caged scaffold for therapeutic purposes. However, to date, there are no additional details on the MOA and no structural data of tyrosinase bound to microviridin A has been reported.

Figure 69.

Chemical structures of microviridin F (122), marinostatin (123), and plesiocin (124). The hairpin motifs in plesiocin are outlined in green.

Figure 67.

Structures of group I microviridins A (112), B (113), G (114), J (115), K (116), and L (117). Group I microviridins contain two macrolactones (shown in blue) and one macrolactam (shown in orange) which are installed by ATP grasp ligases. Residues involved in the microviridin L mutagenesis study are labeled.

Microviridins B (113, Figure 67) and C (118, Figure 68) were identified through screening efforts to identify inhibitors of the serine protease elastase.⁷¹¹ Elastase inhibitors are of therapeutic interest for the treatment of pulmonary emphysema and cystic fibrosis.⁷²⁵ Similar screening efforts identified microviridins D-H (Figure 67–69), which show inhibitory activity towards other serine proteases.^712,713 The initial studies of microviridin G/H lead to the speculation that the identity of the amino acid at position 5 of the core peptide might dictate protease specificity. Lastly, the bicyclic marinostatin (121, Figure 69) was identified during a screen for subtilisin inhibitors with inhibitory activity against several serine, cysteine, and metalloproteases.^726,727

Figure 68.

Structures of the group II microviridins C (118), D (119), E (120), and H (121). Group II microviridins contain two macrolactones (shown in blue) and one macrolactam (shown in orange) which are installed by ATP grasp ligases.

The polycyclic structures of the graspetides suggest that they may bind to the active sites of target enzymes as conformationally restrained structural mimics of substrates. Support for this hypothesis emerged from studies of Microcystis UWOCC MRC, which induced a lethal, molting defect when fed to the cyanobacterial predator Daphnia sp., a species of planktonic crustacean.⁷²⁸ The cyanobacterial metabolite that induced toxicity in daphnids was identified as microviridin J, an inhibitor of trypsin-like proteases (115, Figure 67).⁷¹⁵ Subsequently, purified microviridin J was shown to induce the same molting defect in Daphnia as cells of Microcystis UWOCC MRC, demonstrating daphnid protease inhibition as a likely mechanism of toxicity.⁷²⁹

Atomic details of the MOA were established through structural elucidation of microviridin J bound to bovine trypsin.⁷³⁰ In the structure, Thr4, Arg5, and Lys6 in microviridin J adopted a binding conformation similar to that of bona fide trypsin substrates, and this conformation is rigidified by the macrolactone (Figure 70). It was proposed that the rigid, compact structure of the microviridin prevented cleavage of the target amide bond despite the binding of Arg5 of microviridin J in the S1 specificity pocket of the enzyme.

Figure 70.

Co-crystal structure of bovine trypsin (teal) bound to microviridin J (PDB ID:4KTU). Residues Thr4, Arg5, and Leu6 of microviridin J (purple carbon atoms) mimic the substrate of trypsin. The Ser195 nucleophile is proximal to the scissile bond (black arrow), but peptide cleavage is prevented by the rigid microviridin J structure.

Biochemical studies of the related microviridin L (117, Figure 67) supported the hypothesis that the identity of the residues at position 5 of the graspetide core peptide contributed to protease specificity. Naturally occurring microviridin L is a selective inhibitor of subtilisin (IC₅₀ = 5.8 μM) and contains Phe5 at the P1 site.⁷³¹ Random mutational analysis identified a microviridin L G2A variant that was produced at high levels with few unprocessed side-products, and this backbone was used for further SAR studies. Microviridin L variant G2A/F5R demonstrated a marked inhibition of trypsin (IC₅₀ = 2.5 μM). Likewise, variant G2A/F5I inhibited elastase (IC₅₀ = 5.2 μM).⁷³⁰ Activity of additional microviridin L variants is summarized in Table 3. The specificity of these variants could be further tuned to accommodate substrate preference for a particular protease target. For example, plasmin is a serine protease with a preference for basic residues at the P1 site but biases Lys over Arg at this position.⁷³² The variant G2A/F5K demonstrated inhibitory activity against plasmin (IC₅₀ = 2.0 μM), while the G2A/F5R variant was a far less effective plasmin inhibitor (IC₅₀ = 47 μM).

Table 3. Protease inhibition profile of microviridin L variants.

The microviridin core sequence was altered at position 5 (the P1 site) and also contained a G2A substitution that facilitated high-level production of the tricyclic product.

	Inhibitory activity (IC50, μM)
Microviridin L variant	Elastase	Chymotrypsin	Trypsin	Subtilisin	Thrombin	Plasmin
G2A/F5L	0.65	14	>65	0.78	>65	>65
G2A/F5I	5.2	>33	>65	>33	>65	>65
G2A/F5V	0.16	66	>66	>0.4	>66	>66
G2A/F5K	>65	>65	2.5	0.32	>65	2
G2A/F5R	64	>64	1.8	0.64	>64	47

Microviridin K (116, Figure 67) contains Met5 at the specificity-determining P1 site and is an inhibitor of the serine protease subtilisin (IC₅₀ = 5.2 μM).⁷¹⁶ Notably, the M5R variant strongly inhibited trypsin (IC₅₀ = 0.41 μM), consistent with the preference of this protease for a basic residue at the P1 site. The cage-like tricyclic structure was shown to be important but not critical for inhibitory activity as a bicyclic variant of microviridin K showed a modest decrease in inhibitory activity against subtilisin (IC₅₀ = 8.3 μM), and the bicyclic M5R microviridin K variant showed a near five-fold loss of potency against trypsin (IC₅₀ = 1.8 μM). These studies highlight the importance of both the specificity determining residue, as well as the tricyclic ring structure for protease inhibition.

Structural studies of the subtilisin inhibitor marinostatin (123, Figure 69) by solution NMR spectroscopy demonstrated a tricyclic structure like the microviridins. The structure also revealed a Thr3-Met4-Arg5-Tyr6 motif that is proposed to structurally mimic the P2-P2’ residues of a protease substrate (Figure 71) and bind to the active site of subtilisin as a non-cleavable inhibitor.⁷³³ Replacement of the native ester linkages with disulfide bonds generated an analog that mimicked the structure of a P2-P2’ substrate binding motif. The backbone amides of residues Thr3-Met4-Arg5 exhibited similar rates of hydrogen-deuterium exchange for marinostatin and the disulfide linked variant.⁷³³ Though the disulfide variant exhibits similar activity against subtilisin as the native compound (3.4 pM vs. 1.5 pM), it possesses differences in chemical shifts (|δ| > 0.2 ppm) from native marinostatin, suggestive of differences around the Met4-Arg5-Tyr6 reactive center.⁷³³ These findings suggest that conformational restraints arising from intramolecular linkages involving specific residues play a significant role in stabilizing a non-cleavable conformation of marinostatin and other graspetides.

Figure 71.

Four peptidic protease inhibitors. (A) Marinostatin, (B) microviridin J, (C) SSI, and (D) OMTKY3. All four inhibitors adopt similar structural conformations of residues that mimic the P2-P2’ sites in substrates.

The sequence of marinostatin that is believed to mimic the P2-P2’ residues of a protease substrate is similar to that found in members of the Streptomyces subtilisin inhibitor protein (SSI) family and to the Kazal protein domains which function as non-classical serine protease inhibitors (Figure 71).⁷³⁴ Based on structural alignment, the average dihedral angles of the residues in the sequence Thr3-Met4-Arg5-Tyr6 of marinostatin were close to those observed for the structure of SSI bound to subtilisin BPN,⁷³⁵ and to the Kazal protein turkey ovomucoid third domain (OMTKY3) bound to chymotrypsin.⁷³⁶ In the co-crystal structures with each of these proteinaceous inhibitors, the scissile peptide bond (P1-P1’) is intact but the carbonyl carbon of the P1 residue is distorted towards a tetrahedral geometry.⁷³⁷ The region encompassing the P2-P2’ residues in both SSI and OMTKY3 is flanked by a C-terminal β-sheet and an N-terminal disulfide bond, which stabilizes the protease substrate region (Figure 71).^735,736 It has been suggested that the compounds may undergo proteolysis but are not released from the active site.⁷³⁸ Similarly, the macrolactone in marinostatin and the microviridins provide conformational restraints accounting for their activity as protease inhibitors (Figure 71). The importance of the crosslinks is reflected in experiments demonstrating that tricyclic microviridins generally show better inhibition than variants containing fewer rings.^707,716,730 Additional experimental and computational studies are needed to provide a conclusive mechanistic rationale as to how these graspetides act as protease inhibitors.

Plesiocin (124, Figure 69) is the first characterized bicyclic graspetide that contains four hairpin loops, each of which are formed by residues participating in macrolactone linkages (Figure 69).⁷²¹ Plesiocin inhibits the serine proteases elastase and chymotrypsin. This preference is rationalized as both proteases prefer hydrophobic residues at the P1 position, which corresponds to Leu in the Leu-Ala-Ile-Gly motif found in each hairpin (Figure 66). The number of hairpin repeats has been demonstrated to directly affect inhibitory activity. In vitro produced plesiocin containing four hairpin motifs possessed K_i values of 16 nM against elastase and 7.5 nM against chymotrypsin. In contrast, inhibition assays against an analog with only the first of four hairpin motifs displayed weaker inhibitory activity (K_i = 380 nM against elastase and 63 nM against chymotrypsin). The P1 Leu within each hairpin motif was critical for protease inhibition. Substitution of all four Leu residues to Ala increased the K_i against chymotrypsin to 3.3 μM.⁷³⁹ Variants that retained Leu in a single hairpin, whilst the others were replaced with Ala, also demonstrated lowered chymotrypsin inhibitory activity (K_i = 7.2–230 nM). This result demonstrates that each individual hairpin can act independently as a protease inhibitor. It remains unknown if plesiocin is cleaved into smaller fragments.

In vitro produced plesiocin demonstrated near-complete inhibition of chymotrypsin at a plesiocin:chymotrypsin ratio of ~0.16, where a ratio of ~1.0 would indicate a 1:1 stoichiometry.⁷³⁹ A 1:1 inhibitor-to-chymotrypsin ratio was demonstrated for a single bicyclic hairpin motif, consistent with super-stoichiometric inhibition by the four-hairpin plesiocin.⁷³⁹ Protease inhibition plasticity similar to that of the microviridins was also shown for plesiocin, which shows better inhibition against elastase/chymotrypsin over trypsin due to the hydrophobicity of the bicyclic motif. The introduction of a basic residue to the scissile site of the hairpin can bias inhibition towards trypsin over chymotrypsin.^721,739

10. Circular bacteriocins

Circular bacteriocins contain a circular peptide backbone created by a macrolactam linkage between the N- and C-terminus.²⁵ These compounds range from 58–70 amino acids and can be divided into two subclasses based on their physicochemical properties.²¹ The majority of circular bacteriocins fall under subclass I, which possess a high isoelectric point (pI > 9). Subclass II circular bacteriocins tend to feature lower isoelectric points (pI < 7).²⁵ Examples of structurally characterized circular bacteriocins include enterocin AS-48 (hereafter AS-48), carnocyclin A, and acidocin B.^740–742 Bacterial growth inhibition studies have been carried out with other circular bacteriocins such as enterocin NKR-5–3B, lactocyclicin Q, and leucocyclicin Q.^743–745

The growth-suppression spectrum of known circular bacteriocins is primarily directed against Firmicutes (Table 4). Notable genera inhibited include Lactobacillus, Lactococcus, and Enterococcus. The range of activity also encompasses food spoilage genera within Firmicutes, such as Clostridium, Staphylococcus, Bacillus, and Listeria.⁷⁴⁶ These compounds have no or weak bioactivity against Proteobacteria, including Klebsiella, Acinetobacter, Pseudomonas, or Enterobacter. High concentrations of AS-48 (>6.6 μM), lactocyclicin Q (~17–34 μM), or leucocyclicin Q (34–38 μM) were required to observe growth inhibition of E. coli.^743,744,747

Table 4. Bacterial growth-suppression activity of circular bacteriocins.

+ indicates that a zone of inhibition was observed (activity), − indicates that no activity was observed, w (“weak”) indicates that a zone of inhibition was observed only at higher concentrations (91 μM); N.D., not determined.

Bacteriocin	Class	Lactobacillus	Lactococcus	Enterococcus	Staphylococcus	Bacillus	Listeria	Ref
AS-48	I	N.D.	N.D.	+	+	+	N.D.	747
Garvicin ML	I	+	+	+	N.D.	N.D.	+	759
Carnocyclin A	I	+	+	+	+	N.D.	+	761
Circularin A	I	+	N.D.	+	N.D.	N.D.	N.D.	762
Uberolysin A	I	N.D.	+	+	+	N.D.	N.D.	763
Lactocyclicin Q	I	+	+	+	w	+	+	743
Amylocyclicin	I	N.D.	N.D.	N.D.	N.D.	+		764
Enterocin NKR-5–3B	I	+	+	+	N.D.	+	+	745
Pumilarin	I	N.D.	+		−	+	N.D.	765
Gassericin A	II	N.D.	N.D.	N.D.	+	+	+	766
Acidocin B	II	+	N.D.	N.D.	−	−	−	767
Butyrivibriocin AR10	II	+	N.D.	N.D.	N.D.	+	N.D.	768
Plantaricyclin A	II	+	+	−	−	−	−	769

Early work on AS-48 demonstrated that the toxin forms pores of ~7 Å in diameter in artificial lipid bilayers, thus allowing cations, anions, and small molecules to pass through in a voltage-independent manner.⁷⁴⁷ Carnocyclin A demonstrated the same ability to form pores in lipid bilayers, but these channels were only permeable to small anions in a voltage-dependent fashion.^741,748 Differences in ion selectivity amongst these peptides suggest that the circular bacteriocins may not share a universal MOA.

AS-48 was the first circular bacteriocin to be structurally characterized,⁷⁴⁹ revealing the same fold as NK-lysin,^750,751 a non-RiPP cationic antibacterial and tumor lysing peptide.^751,752 Both peptides adopt a helical bundle conformation, and thus were predicted to exhibit the same MOA. As NK-lysin and AS-48 both lack membrane-spanning helices, pore formation was hypothesized to result from a phenomenon known as molecular electroporation. When a peptide with sufficient charge density binds to a membrane, the peptide local electric field can induce destabilization of the membrane, resulting in pore formation. This mechanism does not necessarily imply that the peptide inserts into the membrane.

Further characterization of AS-48 provided further insight into its MOA. Crystallographic studies of AS-48 bound to decyl-β-D-maltoside highlighted the amphipathic nature of the molecule.¹⁸ These structures reveal that AS-48 exists as one of two possible dimers, one associated through its hydrophobic face, or one associated by its hydrophilic face (Figure 72). The aliphatic tail of the detergent is bound to the hydrophobic core of AS-48, suggesting that AS-48 would similarly associate with membrane lipids. AS-48 has an amino acid distribution and surface properties often found in membrane inserted proteins.⁷⁴⁰ The highly cationic α4 helix (AS-84_49–69) alone does not possess bactericidal activity, indicating that its charge density alone is not sufficient to induce pore formation.⁷⁵⁴ However, the bactericidal effect of AS-48 against Listeria monocytogenes CECT 4032 decreased when cultures were co-treated with AS-84_49–69 indicating that the cationic helix competes for binding at the membrane. These findings suggest the antibacterial action of AS-48 requires all its aliphatic helices to be present. Finally, coarse-grained simulations reveal that AS-48 fully inserts into the membrane. In 9 out of 15 simulations, a toroidal pore was observed (Figure 73).⁷⁵⁵ In this model, the membrane adopts a highly curved arrangement (due to the association of helices a4–5 with hydrophilic phospholipid head groups) that leads to membrane destabilization, and eventual cell death.^755,756

Figure 72.

(A) Crystal structure of dimeric AS-48 obtained in Tris-HCl pH 7.5 (left, PDB ID: 1O83). Hydrophobic helices are shown in orange, while hydrophilic helices are shown in blue. Crystal structure of dimeric AS-48 in the presence of the detergent decyl-β-D-maltoside (right, PDB ID: 1084). (B) Hydrophobic surface models of AS-48 in aqueous solution (left) and in the presence of detergent (right). Hydrophobic residues are shown in orange; hydrophilic residues are shown in blue. (C) Structure of carnocyclin A (PDB ID: 2KJF). Hydrophobic helices are shown in silver, while hydrophilic helices are shown in green. (D) NMR solution structure of acidocin B obtained in deuterated sodium dodecyl sulfate and D₂O (PDB ID: 2MWR). Amphipathic helices are shown in salmon, while the hydrophobic helix is shown in cyan.

Figure 73.

Cartoon representation of the toroidal pore arrangement of AS-48 (shown as blue cylinders) in the lipid bilayer. Three copies of AS-48 are hypothetically required for pore formation.⁷⁵⁵

The NMR structure of the anion-selective peptide carnocyclin A shows an electropositive helix, which may direct the peptide to the membrane (Figure 72).¹³ However, currently available data do not explain why carnocyclin A is anion selective and why it exerts voltage-dependent activity. The NMR structure of acidocin B,⁷⁴² a class II circular bacteriocin, has also been elucidated (Figure 72). Unlike AS-48 and carnocyclin A, acidocin B possesses one hydrophobic helix and three weakly amphipathic helices. The largely hydrophobic helix of acidocin B may facilitate membrane interaction. MOA studies have yet to be reported on the anionic class II bacteriocins. One testable hypothesis is that the MOA may involve lipid extraction and membrane solubilization, similar to the negatively charged saposin-like proteins.^757,758

Though pore formation and membrane disruption is thought to be a primary MOA for the circular bacteriocins, garvicin ML (GarML)⁷⁵⁹ exerts its bactericidal activity by targeting the maltose ABC transporter.⁷⁶⁰ Resistant mutants to GarML possessed a large chromosomal level deletion that included the genes for the maltose ABC transporter (malEFG).⁷⁶⁰ A direct correlation between increased expression levels of the maltose ABC transporter and the increased sensitivity of cells to GarML was shown.⁷⁶⁰ The transporter is proposed to become constitutively open upon GarML binding, leading to the efflux of intracellular solutes, followed by cell death.⁷⁶⁰ It is worth noting that other RiPPs also target sugar transport systems, including MccE492/MccE492m (section 4.2) the glycocins (section 13).

11. Amidinotides

Amidinotides are characterized by the presence of a phenyl-2-guanidinoacetic acid-containing molecule linked to a ribosomally synthesized peptide. The pheganomycins (PGMs, 125–129) were first isolated in the late 1980s from Streptomyces cirratus (Figure 74).⁷⁷⁰ The amidinotides are hybrid NRPS/PKS-RiPP products owing to the presence of the nonproteinogenic (S)-2-amino-2-(3,5-dihydroxy-phenyl)acetic acid moiety.^771,772 An ATP-grasp ligase links a precursor peptide with this non-proteinogenic amino acid to yield the final product. Although the related compounds resorcinomycins A and B⁷⁷³ also possess the phenyl-2-guanidinoacetic acid moiety, they are not true RiPPs, as they contain a single Gly residue instead of a ribosomally synthesized peptide. Known amidinotides display a narrow spectrum of activity, primarily inhibiting the growth of Mycobacterium sp.

Figure 74.

Chemical structures of the pheganomycins (125–128) and deoxypheganomycin D (129). A substituent at position 4 of the aromatic ring appears important for bioactivity of these compounds (blue). The terminal aspartate (orange) of deoxypheganomycin D may contribute to its bioactivity. The proteinogenic portion of both compounds (boxed) is labeled with the original residues above the structures prior to post-translational modification.

Early mechanistic studies performed on a chemically derivatized pheganomycin with increased stability termed deoxypheganomycin D (129, Figure 74) suggested that the molecule acts on the mycobacterial cell wall. Mycobacterium smegmatis ATCC 607 treated with 0.27 μM of deoxypheganomycin D retained only 13% of ¹⁴C-glycerol label in its purified cell wall extracts as compared to an untreated sample.⁷⁷⁰ Deoxypheganomocyin D was proposed to affect cell wall synthesis and lead to increased permeability. Notably, treatment of mycobacteria with the compound increased both influx and efflux of radiolabeled amino acids and nucleotides, consistent with increased permeability. Deoxypheganomycin D does not inhibit protein or DNA/RNA synthesis. No additional MOA studies have been reported on any amidinotide.

SAR studies revealed that the peptide portion of the PGMs alone did not possess any inhibitory effects.⁷⁷⁴ Additionally, substitution of the 4-hydroxymethyl group with a hydrogen in PGM abolished bioactivity. However, deoxypheganomycin D lacking the 4-hydroxy group but containing a C-terminal Asp inhibited the growth of M. smegmatis.⁷⁷⁰ Additional MOA studies on amidinotides are warranted.

12. Crocagins

Crocagins A and B (130, 131, Figure 75) from the soil-dwelling Myxococcales Chondromyces crocatus Cm c5 were first identified in a screen for compounds that disrupt the interaction between CsrA and its small noncoding inhibitory RNA (see below).⁷⁷⁵ They possess a highly substituted tetrahydropyrimidone moiety that is derived from the tripeptide Ile-Tyr-Trp (by the formation of two C-N bonds) and a carbamate moiety.⁷⁷⁶ The compounds are structurally alike but crocagin B (131) contains a 2,3-dehydro-Tyr residue (Figure 75).

Figure 75.

Chemical structure of crocagin A (130) and crocagin B (131). Crocagin B is dehydrogenated between Cα and Cβ of Tyr (C=C bond shown in blue).

Pathogens can alter their physiology and virulence properties in response to changing environmental conditions. Regulation of gene expression is predominantly controlled at transcription initiation.^777,778 However, certain bacteria can regulate gene expression at the post-transcriptional level by controlling translation initiation efficiency, transcript elongation efficiency, and/or rates of RNA decay.⁷⁷⁹ Bacterial post-transcriptional regulation is mediated by small RNAs and RNA-binding proteins (RBPs) that engage the nascent RNA transcript. Upon binding to transcripts, these effectors can induce changes to the secondary structure of RNA. Examples include facilitating or preventing access to the Shine-Dalgarno sequence,^780,781 Rho-utilization sites,⁷⁸² and preventing mRNA degradation.⁷⁸³ CsrA/RsmA (carbon storage regulator A and repressor of secondary metabolites A, respectively) are RBPs that can either activate or inhibit translation from a transcript.^{780,784–786} CsrA/RsmA systems have been extensively reviewed.⁷⁸⁷ Crocagin A inhibits the interaction of CsrA from Y. pseudotuberculosis YPIII with its inhibitory RNA ligand at a modest IC₅₀ of 35.8 μM, and was shown to directly bind to CsrA to elicit these effects (K_d = 12.9 μM). Crocagin A was also screened for antibacterial and cytotoxic effects, but no such activities were observed.⁷⁷⁶ Bioactivity for crocagin B has not been reported.

13. Glycocins

Currently characterized glycosylated bacteriocins (glycocins) contain at least one mono-or disaccharide moiety installed on Cys or Ser of peptides containing two disulfide bonds.^9,23,788 Whereas all known glycocins are glycoactive (i.e. require the sugar for bioactivity), bioinformatic studies suggest currently uncharacterized members will not always contain disulfides and thus the glycosylation serves as the class-defining post-translational modification.⁷⁸⁹ The glycocin class has been divided into four subclasses based on sequence homology and named for the original glycocin identified from the respective subclass: sublancin (132), glycocin F (133), enterocin 96, and enterocin F4–9 (134) (Figure 76).²³ The exact structure of enterocin 96 has not been established. Several glycocins possess growth-suppressive activity against Firmicutes and in rare cases Proteobacteria. The mechanism(s) through which glycocins impart this bioactivity remains unknown, but several studies have made progress towards understanding the MOA.²³ Despite having similar three-dimensional structures, characterized glycocins have either bacteriostatic or bactericidal activities and display considerable SAR differences. Taken together, these observations suggest the potential for multiple MOAs within the glycocin class.⁷⁹⁰

Figure 76.

Glycocin subclasses and NMR structures. (A) Chemical structures of sublancin (132), glycocin F (133), and enterocin F4–9 (134). (B) NMR three-dimensional structures of sublancin (PDB ID: 2MIJ) and (C) glycocin F (PDB ID: 2KUY).

Sublancin 168 (132, Figure 76) is the archetypical glycocin with growth-suppressive activity against multi-drug resistant human pathogens including MRSA. The compound is produced by B. subtilis 168, and while initially reported to be a lantibiotic,⁷⁹¹ a structural revision corrected the earlier report.⁷⁹² The first MOA insights came from studies showing that mutations in two different regions of the B. subtilis genome confer resistance to sublancin. The first region encodes a large mechanosensitive channel, MscL, and the second is an operon (ptsGHI) encoding the phosphoenolpyruvate:glucose phosphotransferase system (PTS) involved in sugar uptake.^793,794 To gain further insight, variants of sublancin were prepared via heterologous expression and utilized for SAR studies.^795,796 These investigations revealed that in addition to the glycosylation post-translational modification, Asn31 and Arg33 are important for activity. A positively charged residue at position 33 is likely important for activity, as replacement of Arg33 with Lys resulted in similar activity as native sublancin. Although Arg33 is important for sublancin activity, charged residues are not conserved at this position across all glycocins suggesting niche-specialization.⁷⁹⁰ Indeed, several glycocins have a narrow spectrum of antibacterial activity, and are only active against organisms within the same genus as the producing organism.⁷⁹⁰

In addition to the amino acid sequence, the sugar moiety is essential for antibacterial activity.^795,796 The glucose on sublancin has been substituted with several other sugars, including galactose, N-acetylglucosamine, mannose, and xylose with minimal influence on the MICs.^796,797 Notably, whereas mannose and N-acetylglucosamine are imported by the PTS, galactose and xylose are not. When testing bioactivity under minimal media conditions, addition of free glucose suppressed the bioactivity of sublancin. Other tested sugars had no effect on the activity of wt sublancin nor analogs carrying different sugars. Therefore, it appears that glucose competes with sublancin and its analogs for the glucose PTS, and that changing the sugar on sublancin does not direct the compound to different sugar import systems. Consistent with this conclusion, resistance mutations in ptsGHI imparted resistance against sublancin analogs carrying different sugars.⁷⁹⁶ The glucose PTS phosphorylates the C6 hydroxy group of glucose during import raising the question whether the glucose on sublancin must be phosphorylated for activity. However, the high antibacterial activity of a sublancin analog carrying xylose argues against phosphorylation at the C6 hydroxy group because xylose does not possess such a group (Figure 77).⁷⁹⁷

Figure 77.

(A) Schematic representation of the glucose PTS used to import glucose into B. subtilis. A relay mechanism transfers a phosphate from phosphoenolpyruvate (PEP) via several protein carriers to the hydroxyl at C6 of the incoming glucose. A separate phosphorylation mechanism is involved in catabolite repression. (B) Replacement of the glucose on sublancin with xylose (which lacks C6) did not compromise antimicrobial activity.⁷⁹⁷ R = sublancin aglycon.

To gain further insight into the sublancin MOA, macromolecular synthesis assays were performed. Radiolabeled precursors were utilized to measure incorporation into the cell wall, DNA, RNA, and proteins.⁷⁹⁶ Sublancin affected DNA, RNA, and protein biosynthesis, suggestive of a process upstream of macromolecule biosynthesis that is interrupted by sublancin. This observation is consistent with the hypothesis that the PTS is involved in the bioactivity of sublancin through a yet unknown MOA. However, several findings argue against a simple blockage of glucose import. First, sublancin is bactericidal, not bacteriostatic. Second, sublancin is bioactive even when bacteria are grown on a non-PTS carbon source. Finally, mutations that inactivate the glucose PTS provide resistance, but overexpression of ptsG, ptsH, or ptsI renders cells hypersensitive.⁷⁹⁷ Collectively, these observations suggest sublancin acts through a mechanism that requires an active glucose PTS, and that possibly induces a lethal gain of function in the glucose PTS. For other natural products that target the sugar PTS (e.g. garvicin ML, section 10 and MccE492, section 4.2), such gain of function has been shown to involve pore formation or constitutive channel opening, but such a mechanism was not observed for sublancin.⁷⁹⁷

Another study focused on the effect of sublancin on the mouse host immune system.⁷⁹⁸ Sublancin induced macrophage migration, upregulates production of select anti-inflammatory chemokines, downregulated pro-inflammatory cytokines, and modulated the gut microbiota populations compared to control samples.⁷⁹⁸ All of these effects could be mechanisms by which sublancin may be effective in treatment of bacterial infections in a host.

Glycocin F (133, Figure 76) is produced by Lactobacillus plantarum and exhibits bacteriostatic activity. A model for glycocin F-receptor interactions was proposed after SAR studies with analogs prepared by solid-phase synthesis (Figure 78).⁷⁹⁹ These investigations identified the disulfide bond nearest the interhelical loop, the size of the interhelical loop, and the O-linked N-acetyl-D-glucosamine (GlcNAc) as essential for activity.^799,800 The S-linked GlcNAc and C-terminal tail contribute to the activity, but were not essential. Similar to sublancin, even when the identity of the C-terminal sugar was changed, bacteriostasis was abolished with the addition of free GlcNAc, but not other sugars, suggesting that changing the identity of the C-terminal sugar did not alter the PTS system being targeted.^788,799 Similar SAR studies were also performed for ASM1, a glycocin F-type family member with ~91% sequence identity to glycocin F, yielding the same conclusions on the importance of these structural features for bioactivity.⁸⁰¹ The results led to a model in which the C-terminal tail is utilized to localize glycocin F to the correct cell surface receptor. This initial interaction is proposed to facilitate binding of the O-linked GlcNAc and the interhelical loop to a currently unknown primary target to initiate bacteriostasis.⁷⁹⁹

Figure 78.

Proposed model for glycocin F-receptor interactions. The O-linked GlcNAc in the loop between the helices is black while the C-terminal S-linked GlcNAc is white. (A) Both sugars bind the GlcNAc-specific PTS membrane spanning EIIC domain (purple) followed by (B) the loop GlcNAc binding its primary target (blue) that has yet to be identified.⁷⁹⁹ Disulfides are shown in orange.

Enterocin 96 and enterocin F4–9 (134, Figure 76) produced by E. faecalis also exert bacteriostatic activity. SAR studies on enterocin 96 indicated that unlike other glycocins, the disulfides were dispensable. Furthermore, the di-glucosylated form was more active than the mono-glucosylated form, and replacement with galactose at the terminal position eliminated activity.⁸⁰² For enterocin F4–9, the sugar moieties, proposed to be GlcNAc, and disulfides were essential for activity.⁸⁰³ These studies suggest that the details of the MOAs of different glycocins may differ in subtle but important ways.

A glycocin produced by a thermophilic bacterium, Aeribacillus pallidus (also referred to as Geobacillus sp. 8), was independently identified by two research groups and termed pallidocin and geocillicin, herein referred to as pallidocin (135, Figure 79).^790,804 Similar to all other glycocins, pallidocin was glycoactive, and bioactivity could be reversed by the addition of glucose. However, no reduction in bioactivity upon glucose addition was observed for bacillicin BAG2O and bacillicin CER074 even though these glycocins are glucosylated.⁷⁹⁰ This finding further suggests that glycocins display unique MOAs.

Figure 79.

Schematic structure of pallidocin (135).

14. Sulfatides

Xanthomonas oryzae pv. Oryzae (Xoo) is a plant pathogen that causes bacterial blight disease in rice.^805,806 Early efforts to understand the pathogenicity of Xoo were focused on identifying strains of rice that were resistant to bacterial blight.^807,808 One such strain, IRBB21 possesses a gene encoding a histidine kinase receptor, XA21, which confers pathogen immunity to transgenic rice.^808,809 Subsequently, the genes required for Xoo virulence have been identified, and were termed raxXSTAB (required for activation of XA21).^810,811 RaxX, the product of the rax gene cluster, is a RiPP featuring a sulfated tyrosine (136, Figure 80). RaxX activates the XA21-mediated immune response, which suppresses infection in rice.⁸¹²

Figure 80.

Chemical structure of RaxX21 (136). Post-translational modiications are colored in blue. The C-terminus of the peptide contains residues that are important for recognition by XA21 (orange). Amino acid sequences of Arabidopsis PSY1, RaxX21 and Rax13 are shown (top), highlighting the position of the tyrosine residue in each (green). Additional chemical modifications are found on Arabidopsis PSY1 (Pro-4-hydroxylation and O-glycosylation).

Knowledge of the XA21-mediated immune response is incomplete but is briefly summarized below. XA21 is a cell-surface receptor that provides the first line of defense in plants against bacterial pathogens.⁸¹³ XA21 consists of an extracellular ligand binding domain, a transmembrane domain, a juxtamembrane domain, and an intracellular Ser/Thr kinase domain.^808,814,815 RaxX activates the XA21-mediated immune response by inducing autophosphorylation of XA21 and/or phosphorylation of downstream proteins, many of which are unidentified.^816,817 Although the order of the signaling cascade is not known, several XA21-binding (XB) proteins and regulators have been identified.^{809,818–822} For example, both XB3 and XB25 have been shown to be required for XA21-mediated immunity.^818–822 The former protein is hypothesized to induce the MAPK cascade, which leads to transcriptional activation of defense-related genes.⁸²³ Several studies have investigated the details of the XA21-mediated immune response and interactome.^823–825

SAR studies of RaxX have identified residues required for XA21 activation. The RaxX precursor peptide contains 60 amino acids and undergoes sulfation on Tyr41. A putative leader peptide cleavage site has been identified, but has not been validated in vitro.⁸²⁶ The unmodified precursor peptide is not bioactive. Though the sulfated 60-mer precursor peptide activates XA21 at an EC₅₀ of 100 nM, a shorter construct termed RaxX21 (precursor peptide residues 35–55) is a more potent activator (EC₅₀ = 20 nM). An even shorter construct, encompassing precursor residues 32–49, activates XA21 at an EC₅₀ of > 10 μM, suggesting that key binding residues may reside near the C-terminus of the peptide. The binding dissociation constant (K_d) of RaxX21 to the extracellular domain of XA21 is 16 nM, whereas non-sulfated RaxX21 binds with a K_d of 205 nM.⁸²⁶ The specific interactions between the receptor and the peptide have yet to be investigated.

RaxX possesses additional bioactivity. RaxX21 resembles a sulfotyrosine-containing plant peptide hormone, PSY1, which induces growth in Arabidopsis (Figure 80).^827,828 Treatment of an Arabidopsis mutant that lacks the ability to produce sulfated and bioactive PSY1 with low nanomolar concentrations of either RaxX21 or PSY1 increased root growth by at most two-fold. The minimal peptide required to promote root growth encompasses precursor peptide residues 40–52 (termed RaxX13). Unlike RaxX21, PSY1 does not induce an immune response in XA21 expressing rice, indicating that RaxX21 is recognized by both hormone and XA21 receptors, whereas PSY1 only targets the former.⁸²⁸ The growth promoting property of RaxX is rationalized by the fact that Xoo is a biotrophic pathogen that requires a living host to obtain nutrients for its survival. Thus, by mimicking of the growth stimulating activity of PSY1, RaxX reprograms the host environment to benefit the pathogen. It is likely that RaxX evolved specifically to mimic PSY1.⁸²⁸ Consequently, rice may have evolved the XA21 receptor, which specifically invokes the immune response in the presence of RaxX, but not in the presence of PSY1 and related peptides.⁸²⁸

15. RiPPs with class-defining features installed by radical SAM enzymes

The radical SAM (rSAM) enzymes form a large and diverse superfamily with more than a million predicted members.^829,830 These enzymes employ a [4Fe-4S] center to reductively cleave S-adenosylmethionine, generating L-Met as a byproduct and a 5’-deoxyadenosyl (5’-dA) radical. In most known examples, the 5’-dA radical abstracts a hydrogen atom from the substrate, which then undergoes further chemistry to yield many different outcomes for individual members of the superfamily.⁸³¹

15.1. Polytheonamides

Originally discovered in the early 1990s, the polytheonamides were not immediately recognized as RiPPs owing to their extensive modifications, which include N-acylation, β-hydroxylation, N-methylation, C-methylation, and a large number of epimerizations.^832–834 Polytheonamides are potent cytotoxins against cultured murine cells (IC₅₀ = ~13 pM). Biochemical and analytical chemical studies have largely focused on polytheonamide B (137, Figure 80), which is structurally homologous to the other polytheonamides, and will be the focus for the remainder of this discussion.

Polytheonamide B was proposed to act as a cationic membrane channel.⁸³⁵ Detailed solution NMR spectroscopy revealed that polytheonamide B adopted a b-helical structure 45 Å in length with a central pore diameter of ~4 Å.⁸³⁶ Upon insertion into membranes, polytheonamide B affords selective permeation of monovalent cations. The β-helical structure is stabilized by hydrogen bonding interactions between five side chain N-methylated D-Asn and one D-Asn, which all reside on the same face of the helix (Figure 81). The N-terminal half of polytheonamide B is largely nonpolar while the C-terminus is relatively polar leading to the hypothesis that insertion of polytheonamide B into the membrane is facilitated by its N-terminal portion. As the structure was acquired in a solvent (1:1 methanol/chloroform) that mimics the dielectric constant in a cell membrane, the reported conformation is believed to reflect the native conformation of polytheonamide B in its active form.

Figure 81.

Structure of polytheonamide B (137, pTB). The N-terminal 5,5-dimethyl-2-oxohexanoate is indicated in pink. L-amino acids are shown in blue, while D-amino acids are shown in orange. SAR studies utilized derivatives of pTB (137a-i) that were modified at the N-terminal 5, 5-dimethyl-2-oxohexanoate group (R₁), residue 44 (R₂), β-substituents of residues 2, 22, 29, and 37 (X, and Y), and residue 47 (Z).

In synthetic planar membranes, polytheonamide B treatment led to permeabilization of K⁺ and Cs⁺ ions in a periodic manner. Opening and closing events were monitored by voltammetry on the side of the membrane that contained polytheonamide B.⁸³⁷ The probability of an opening event was lower at −200 mV than at +200 mV, indicating a voltage-dependent gating mechanism. These findings suggest an asymmetric polytheonamide B structure orients in the membrane, as ion permeability was unidirectional i.e., permeabilization of K⁺ and Cs⁺ ions only occurred in one direction across the membrane. Transmembrane current amplitude was directly proportional to polytheonamide B concentration, indicating that this compound forms a unimolecular ion channel instead of a channel composed of multiple polytheonamide B subunits. The current recordings demonstrated ion selectivity for monovalent cations and impermeability to divalent cations, and that polytheonamide B is not released from the membrane once it is incorporated.⁸³⁷

The selective orientation of polytheonamide B is attributed to its asymmetric polarity, as the N-terminus is more hydrophobic than the C-terminus (Figure 82). Addition of polytheonamide B to eukaryotic membrane mimetic liposomes (10:1 phosphatidylcholine:cholesterol) encapsulating carboxyfluorescein as a pH indicator (internal pH = 6.5; bulk solution pH = 5.5) dissipated the pH gradient leading to liposomal fluorescence.⁸³⁸ However, polytheonamide B did not disrupt the membrane sufficiently to allow transmembrane passage of carboxyfluorescein, which with a molecular radius of ~5 Å in solution, is larger than the polytheonamide B pore diameter of ~4 Å.

Figure 82.

(A) Solution NMR structure of polytheonamide B (PDB ID: 2RQO). The hydrogen-bonding network (dashed lines) occurring between D-Asn residues is presumed to stabilize a pore-forming conformation. (B) Polytheonamide B exhibits a relatively polar C-terminus (left) and a more nonpolar N-terminus (right). Hydrophobic residues are colored from orange (hydrophobic) to blue (hydrophilic).

Synthetic studies supported this mechanism and localized the membrane disruption activity to the C-terminus and membrane insertion/orientation to the hydrophobic N-terminus of the peptide.^838–841 Smaller fragments exhibited lower H⁺/Na⁺ exchange across liposomal membranes than full length polytheonamide B with exchange increasing as fragments were added from the C-towards the N-terminus (Figure 83).⁸³⁸ In planar membrane channel current assays, the B+C+D fragment demonstrated a similar open/close gating mechanism to native polytheonamide B, though with a lower voltage dependency.

Figure 83.

Bioactivity of polytheonamide B fragments. The C-terminal fragment D exhibited the highest membrane perturbation measured by carboxyfluorescein (CF) release from egg yolk phosphatidylcholine and cholesterol (10:1) liposomes but lacked ion transport ability measured by H⁺/Na⁺ exchange. Conversely, N-terminal fragments lacked membrane perturbation activity. Additions of N-terminal fragments to fragment D yielded a compound that lost membrane disruption activity but gained ion transport activity. Fragment IC₅₀ values were determined in vitro against murine p388 leukemia cells. Ion exchange and CF release values were estimated from the maximal percentage of each variable released over the time course of the experiment. Percentage exchange monitors loss of the starting material from inside the liposome, i.e. 5% exchange means 5% left the liposome interior. Figure adapted from reference.⁸³⁸

These data support a mechanism where the C-terminus functions as a membrane disruption domain, while the N-terminus forms the ion channel. These N-terminal fragments add sufficient length to span the entire lipid bilayer.⁸³⁸ Conversion of the N-terminal acylamide group (5,5-dimethyl-2-oxohexanamide) to more polar groups like amino, acetylamide, or trimethyl ammonium increased IC₅₀ values by 240-, 480-, or 2,500-fold respectively (137a, 137b, 137c, Figure 81 inset).⁸³⁹ Conversely, introducing a hydrophobic octanamide (137d) functionality only decreased toxicity by 5-fold while a palmitamide (137e) decreased IC₅₀ by a factor of three. Current measurements in planar bilayers demonstrated ion channel activity for all variants except the trimethyl ammonium analog, which was proposed to engage in a channel-blocking mechanism caused by the quaternary cation.

Derivatization studies have also been performed to elucidate SAR of polytheonamide B. These studies were performed using a synthetically accessible analog of polytheonamide B, which retained all intramolecular hydrogen bonding interactions and chiral centers, but the β-substituents of residues 2, 22, 29, and 37 were removed (Figure 81), and residues 44 and 47 were exchanged for propargyl glycine, and D-Thr respectively. A dansyl group was installed at position 44 via Huisgen cycloaddition to serve as a fluoregenic marker. These six modifications resulted in an analog that possessed an IC₅₀ of 12 nM against p388 mouse leukemia cells, approximately 100-fold less potent than polytheonamide B (Figure 81, analog 137f).⁸⁴² Despite the bulky fluorophore substituent, the dansylated analog could insert into liposomal membranes, which permitted its use in further SAR studies. Improper insertion in the membrane, e.g. perpendicular to the membrane, would have decreased bioactivity of 137f. In vitro analyses of liposomal H⁺/Na⁺ exchange also demonstrated this mimic can efficiently permeabilize phospholipid bilayers to small cations with a gating mechanism like that of polytheonamide B.

Additional SAR studies were performed using dansylated polytheonamide B (Figure 80, 137f). Truncation studies demonstrated that removal of 12 N-terminal residues (Δ_1–12) resulted in an IC₅₀ of 3.7 nM against murine p388 mouse leukemia cells, compared to 12 nM for the full length dansylated polytheonamide B.⁸⁴³ The Δ_1–11 and Δ_1–13 constructs possessed IC₅₀ values of >420 and 81 nM, respectively. Ion transport assays performed using liposomes demonstrated that Δ_1–11 is unable to permeabilize membranes to H⁺/Na⁺. Utilizing pH-indicator dyes, Δ_1–11 was also unable to induce a pH shift in murine P388 leukemia cells, further supporting an inability to induce ion transport across membranes. This finding was consistent with truncation studies on native polytheonamide B, where N-terminal truncations reduced membrane disruption and cation transport ability (Figure 83).⁸³⁸ Despite an inability to efficiently transport cations, the Δ_1–11 dansylated polytheonamide B analog suggests a novel MOA.

Modification of the N-terminus of full length dansylated polytheonamide B further increased potency. Introduction of a palmitamide on the N-terminus improved the IC₅₀ from 12 to 0.84 nM (Figure 81, 137g).⁸⁴¹ This change was attributed to the increased logP of this analog, which was determined to be 4.9, compared to 4.5 for polytheonamide B. Introduction of a more polar N-terminal functional group like tetramethylammonium (Figure 81, 137h) or a free amine (Figure 81, 137i) increased the IC₅₀ to 83 and 25 nM, respectively, likely due to the decreased logP of these compounds (2.6 for the trimethylammonium, and 3.2 for the free amine derivatives). These studies highlight that the relative hydrophobicity of the N-terminus is also critical to the bioactivity of polytheonamide B.

The published NMR structure has served as a starting point to perform molecular dynamics simulations. The network of hydrogen bonds between Asn and methyl-Asn side chains served to stabilize the channel structure, allowing polytheonamide B to be “trapped” in the membrane and to retain channel activities over long residence times, contributing to its potency.^836,837,844 Cation transport through the channel is proposed to occur through coordination by backbone carbonyls and is additionally facilitated by the movement of water molecules through the channel.^845,846 Water molecules residing within the polytheonamide B channel at resting state function as a water wire, which transport H⁺ during membrane depolarization.⁸⁴⁷ This ion transport mechanism was demonstrated to be reversible depending on the voltage potential, as proton transport is at equilibrium at ~0 mV, confirming the ability of polytheonamide B to depolarize cell membranes regardless of potential direction.

In addition to membrane depolarization, cell localization studies demonstrated the ability of polytheonamide B to neutralize lysosomes through a micropinocytosis pathway.^848,849 Chemically-synthesized polytheonamide B was functionalized with an N-terminal BODIPY group, facilitating studies in live cells. The dye bis-(1,3-dibutylbarbituric acid) trimethine oxonol [DiBAC₄(3)] was used to study changes in membrane potential. Treatment of DiBAC₄(3)-exposed cells with 5 nM polytheonamide B led to complete membrane depolarization in MCF-7 cells within 1 h.⁸⁴⁸ Imaging studies with BODIPY-functionalized polytheonamide B demonstrated that this compound did not localize to the plasma membrane but instead to smaller organelles.⁸⁴⁸

Colocalization studies with LysoTracker (which accumulates in acidic lysosomes) demonstrated BODIPY-polytheonamide B was located within lysosomal membranes.⁸⁴⁸ The lysosomal pH-indicator fluorescein-tetramethylrhodamine-tagged dextran (FRD) demonstrated a pH increase from 4 to 7 in lysosomes after treatment with polytheonamide B, presumably as a consequence of proton transport. Internalization of polytheonamide B was shown to be reduced by 5-(N-ethyl-N-isopropyl)-amiloride, which is an inhibitor of micropinocytosis, suggesting polytheonamide B is actively internalized by a native cellular pathway. Flow cytometry studies with polytheonamide B demonstrated induction of the apoptotic pathway due to membrane depolarization and dissipation of the lysosomal pH gradient. This secondary effect was observed at concentrations near the measured IC₅₀ values, supporting the hypothesis that this secondary MOA is potentially physiologically relevant.

15.2. Darobactin

Darobactin (138) is a first-in-class, heptapeptide RiPP that features a C-O linkage between the C7 indole of Trp1 and the β-carbon of Trp3, and a C-C bond between the C6 indole of Trp3 and the β-carbon of Lys5 (Figure 84). The C-O and C-C bonds are remarkably installed by the same rSAM enzyme.⁸⁵⁰ Darobactin is selectively active against several clinically relevant Proteobacteria such as P. aeruginosa and K. pneumoniae (MICs = 2–16.5 μM). Conversely, the compound was found to possess little activity against Firmicutes. The first clues regarding the darobactin MOA were uncovered when resistant mutants from E. coli MG1655 were selected and found to map to the bamA gene. Given that Firmicutes lack BamA, darobactin is a rare example of a Proteobacteria-selective antibiotic.

Figure 84.

(A) Structure of darobactin (138). Strands β1 (blue) and β16 (orange) surround the lateral pore of BamA. (B) The open state of the lateral pore shown in ribbons. (C) The open state of the lateral pore highlights the hydrogen bonding contacts between β1 and β16. (D) The closed state of the lateral pore. (E) Location of resistance-conferring mutations in BamA, noting proximity to the lateral pore. (F) Close-up of β-strand mimicry interactions between darobactin (pink carbon atoms) and β1 of BamA (blue carbon atoms).

BamA is an essential β-barrel outer membrane protein (OMP) that functions as a chaperone that assists other OMPs to properly fold and insert into the OM of Proteobacteria. A critical feature of BamA is the lateral pore, which is an opening between the N- and C-terminal β-strands.⁸⁵¹ Structural analyses have captured the lateral pore in an “open” and “closed” state (Figure 84).^852–855 The open state allows for proper folding of the nascent OMP polypeptide, and subsequent insertion into the OM.⁸⁵⁶ Sequencing of darobactin-resistant mutants revealed three strains that possessed two to three substitutions located in or near the lateral pore.⁸⁵⁰ Isothermal titration calorimetry was used to demonstrate darobactin binding to BamA (K_d = 1.2 μM), and NMR analysis revealed that BamA was stabilized in the closed state upon binding darobactin.^850,857,858

The structure of the BamA-darobactin complex reveals that the antibiotic mimics a β-strand that binds to the open conformation of BamA along β1 (Asn422-Phe428, Figure 84).⁸⁵⁹ Consequently, binding of darobactin occludes the lateral pore such that BamA is unable to fold or insert proteins into the OM. The unmodified core peptide of the darobactin precursor peptide does not possess any inhibitory effects, presumably as it cannot serve as a β-strand mimic.⁸⁵⁰ Alanine scanning analyses and binding experiments revealed that the most critical residue of BamA in strand β1 for darobactin binding is Asn427, which makes direct side chain contacts with Asn2 in darobactin.⁸⁵⁹ Single alanine substitutions of Asn422 through Phe426 did not significantly affect binding to darobactin.

Molecular dynamic simulations suggest how three distinct BamA substitutions (G429V, E435K, and Q445P) confer resistance to darobactin.⁸⁵⁹ The BamA-G429V replacement appears to disrupt the interactions between darobactin Asn2 and β16 of BamA. The BamA E435K variant introduces unfavorable allosteric effects with nearby negatively charged residues on BamA, leading to the destabilization of β1-β2 of BamA, and a 32-fold increase in the MIC over wild-type. Lastly, the Q445P substitution induces a conformational change in β2, which in turn leads to the destabilization of the β1-darobactin interaction.

Heterologous expression has been independently pursued by two groups to produce darobactin and derivatives thereof.^860,861 Pseudomonas heterorhabitis VMG was predicted to produce darobactin B, which differs only by two substitutions (S4T and S6R). The MICs for darobactin B against various Proteobacteria were approximately the same as darobactin A: including P. aeruginosa (5.7–15.2 μM), K. pneumoniae (1.9–3.8 μM), A. baumanii (7.6–121 μM), and E. coli (0.95–3.8 μM). Analogs with a K5R or a N2S substitution were 8–16 fold less potent than darobactin A.⁸⁶⁰ Another study produced a new-to-nature F7W analog with 2–4 fold greater inhibitory activity against clinically relevant Proteobacteria.⁸⁶¹ These studies demonstrated that the darobactin scaffold is amendable to engineering.

15.3. Sactipeptides

Sactipeptides are RiPPs with a class-defining, intramolecular sulfur-to-alpha carbon thioether (sactionine) crosslink installed by a rSAM enzyme. Most structurally characterized sactipeptides display a hairpin conformation arising from the cysteine donor residues being more N-terminal than the sactionine acceptor residues (Figure 85). Subtilosin A (139, Figure 85) and sporulation killing factor (SKF, 144, Figure 87) are further head-to-tail cyclized. All characterized sactipeptides possess antibacterial activities. The majority display growth-suppressive activity against closely related bacteria, typically other Firmicutes, such as the human pathogens C. difficile, L. monocytogenes, and Staphylococcus sp., or the actinobacterium Gardnerella vaginalis.^{306,862–872} MOA studies on sactipeptides strongly support the plasma membrane as the principle target.^{863,868,873–875}

Figure 85.

(A) Chemical structure of subtilosin A (139). (B) Hydrophobicity surface map of subtilosin A (PDB ID: 1PXQ). Hydrophobic regions are shown in orange, while hydrophilic regions are shown in blue. The hydrophilic patch composed of Asn1, Lys2 and Trp34 is highlighted by the red curved line. The right panel shows the same surface rotated 180° about the vertical axis.

Figure 87.

Chemical structures of sporulation killing factor (144), RumC (145), and streptosactin (146). NMR solution structure of RumC is shown in the inset (PDB ID: 6T33). Stereocenters marked with * do not have an assigned absolute stereochemistry.

Subtilosin A (Figure 85) was first isolated from B. subtilis 168, and was found to inhibit the growth of various Bacilli strains and other Gram-positive bacteria such as L. monocytogenes ATCC 19115 and E. faecium (formerly S. faecium IFO 3181).^864,876 Though subtilosin A is reported to produce zones of inhibition against Gram-negative bacteria including K. pneumoniae and Porphyromonas gingivalis, the susceptibility of Gram-negative strains is dependent on the absence of a polysaccharide capsule.⁸⁷⁶ Though subtilosin A is not hemolytic, a T6I variant demonstrated hemolytic activity and proved to be more potent against a number of bacteria tested (Table 5).⁸⁷⁷

Table 5.

MICs/MLCs of subtilosin A and variant T6I.

Indicator	Subtilosin A MIC/MLC (μM)	Subtilosin A T6I MIC/MLC (μM)
Bacillus subtilis	250	100
Bacillus anthracis	16	2.56
Bacillus cereus	40	6.4
Bacillus thuringiensis	40	6.4
Enterococcus faecalis	100	16
Staphylococcus carnosus	100	16
Listeria monocytogenes	40	6.4
Streptococcus pyogenes	100	6.4
Rabbit blood agar*	>250	16

^*Reported as a minimum lytic concentration (MLC). Hemolysis was assessed using rabbit blood agar, as opposed to purified erythrocytes. Table adapted from (⁸⁷⁷).

Subtilosin A permeabilizes membranes to small molecules and ions. Treatment of G. vaginalis with subtilosin A at 0.6 μM induced ATP release, causing extracellular ATP content to increase to 40% of the total (intracellular + extracellular) ATP concentration.⁸⁶⁸ Utilizing the fluorophore 3,3’-dipropylthia-dicarbocyanine and the K⁺/H⁺ ionophore exchanger nigericin, subtilosin A was found to not affect the transmembrane potential (ΔΨ). Though subtilosin A did not deplete ΔΨ, it was shown to completely deplete the transmembrane pH gradient (ΔpH).⁸⁶⁸ The fact that subtilosin A only dissipated the ΔpH component of the proton motive force suggests the formation of transient pores in the bacterial membrane. Additional studies with L. monocytogenes demonstrated minimal effect on the pH gradient and ATP efflux, further suggesting that subtilosin A activity depends on the composition of the cell membrane.⁸⁶⁹

Fluorescence spectroscopy studies confirm that subtilosin A targets the membrane.^873,878 The intrinsic fluorescence of Trp34 in subtilosin A shifted to shorter wavelengths in the presence of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) and 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylglycerol (POPG) small unilamellar vesicles (SUVs), indicating the Trp residue is buried inside the phospholipid bilayer in each case.⁸⁷³ In a separate experiment, carboxyfluorescein-containing SUVs were exposed to subtilosin A and dye leakage was observed in a concentration-dependent manner. No dye leakage was observed at concentrations lower than 16.4 μM, which is the lowest concentration at which subtilosin A exists as a monomer. It was hypothesized that subtilosin A has detergent-like behavior, in which a critical concentration is required for aggregation and membrane permeabilization. Subtilosin A exhibits selectivity for membrane composition as no dye leakage was observed for SUVs with compositions mimicking mammalian membranes.

NMR studies of subtilosin A in lipid bilayers further refine the model of membrane association. NMR studies of ¹⁵N-labeled subtilosin A in aligned POPC lipid bilayers demonstrated that the compound exists in two conformations, one perpendicular to the membrane surface and a smaller population aligned nearly parallel to the membrane surface.⁸⁷³ Utilizing ³¹P NMR spectroscopy to study lipid conformation, upfield shifts of ~2–4 ppm of POPC, POPG, and 1,2-dimyristoyl-sn-glycero-3-phosphatidylcholine (DMPC) bilayers in the presence of subtilosin A indicate an altered lipid head group orientation. This conformational change was supported in studies of quadrupolar splitting of the α- and β-methylenes of choline in DMPC membranes, indicating a subtilosin A-dependent conformational change as well as a modest increase in lipid acyl chain conformational exchange. The NMR solution structure of subtilosin A reveals a hydrophilic patch consisting of Asn1, Lys2 and Trp34 (Figure 85),⁸⁷⁹ which is posited to associate with phospholipid head groups.⁸⁷³ Therefore it was proposed that subtilosin A does not fully insert and span the plasma membranes, but rather binds at the bilayer surface creating local permeabilization defects. However, as the concentrations necessary to induce membrane permeability were higher than the MIC of subtilosin A (Table 5), it was hypothesized that bactericidal activity may depend on interaction with a surface receptor.^873,877 Further study will be required to determine if there is a cell-surface receptor for subtilosin A.

Subtilosin A also demonstrates spermicidal activity and inhibits late-stage replication of HSV-1.^880–882 Subtilosin A has in vitro activity against human spermatozoa with an IC₅₀ of ~19 μM, and treatment of ectocervical cell cultures with 40 μM subtilosin A reduced cell viability by a modest 5–10% over 24 h.⁸⁸¹ Subtilosin A inhibited the forward progression of spermatozoa and induced tail coiling in a dose dependent manner, though no detailed mechanism for this observation has been proposed. Concerning HSV-1, subtilosin inhibited viral replication in vitro by ~92% at concentrations as low as 1.0 μM.⁸⁸² Pretreatment of viral particles with 17.3 μM subtilosin A reduced the viral titer by 99%. Inactivation of HSV-1 was shown to occur not in the early stages of viral replication but instead the compound likely inhibited viral assembly or release.

Additional MOA studies of subtilosin A indicated inhibition of quorum sensing-dependent biofilm formation.⁸⁸³ At sub-lethal concentrations, subtilosin A inhibited biofilm formation of G. vaginalis (229 nM, >90% reduction), L. monocytogenes (4.4 μM, 80% reduction) and E. coli (4.4 μM, 60% reduction). Subtilosin A was also implicated as a potential inhibitor of quorum sensing in Chromobacterium violaceum. C. violaceum produces the pigment violacein only when neighboring cells secrete N-acyl homoserine lactone (AHL) quorum sensing molecules.⁸⁸⁴ Subtilosin A at concentrations of 2–37 μM reduced C. violaceum violacein production by >70%, suggesting that subtilosin A inhibits AHL production or secretion.⁸⁸³ Similarly, autoinducer-2 (AI-2) is an AHL found in a wide variety of bacteria.⁸⁸⁵ Production of AI-2 in G. vaginalis was inhibited by subtilosin A at 882 nM and 1.2 μM with no influence on bacterial growth. However, inhibition of AI-2 production was not observed for L. monocytogenes at 82 nM–1.3 μM subtilosin A, but cell viability declined to ~50% when treated with the highest concentration (1.3 μM).⁸⁸³ To summarize, in addition to membrane disruption, the anti-biofilm activity of subtilosin A suggests a second molecular target, providing opportunities for further MOA studies. A recently described sactipeptide hyicin was isolated from Staphylococcus hyicus 4244 and possesses 68% and 85% sequence identity and similarity to subtilosin A, respectively. This peptide demonstrated bactericidal activity against S. aureus and Staphylococcus saprophyticus, and was able to inhibit biofilm formation and disrupt preformed biofilms.⁸⁷⁰ Currently, the structure of hyicin is unknown, and more detailed MOA studies have not been performed, but given the sequence similarity to subtilosin A, it is reasonable to assume these two sactipeptides share an MOA.

MOA studies have also been undertaken for thuricin CD, thuricin Z/huazacin, and thurincin H (140, Figure 86). B. thuringiensis DPC 6431 produces thuricin CD, a two-component antibiotic composed of the separate sactipeptides thuricin α (141) and β (142), both of which are necessary for full activity.^866,886 Thuricin CD irreversibly depolarizes the membrane and induces pore-formation in Bacillus firmus at 125 nM.⁸⁷⁵ Membrane polarization was studied using the dye 3,3’-diethyloxacarbocyanine iodide, which loses fluorescence and serves as a probe for bacterial membrane potential. Flow cytometry experiments revealed that both thuricin α and thuricin β (individually at 3 μM) induced depolarization and reduced cell size in isolation. At higher concentrations (9 μM), thuricin CD rapidly elicited depolarization to a greater degree than its individual components. Light microscopy showed B. firmus contained low-density regions of the plasma membrane after thuricin CD treatment. The individual components of thuricin CD seem to act synergistically to disrupt the bacterial membrane, though the potential role of a native cell surface receptor has not been ruled out. Furthermore, the membrane-bound thuricin CD structure is unknown, and higher-order structures beyond heterodimerization may be involved in membrane disruption.

Figure 86.

The structures of the sactipeptides thuricin H (140), thuricin CD, composed of two components thuricin α and thuricin β (141, 142), and huazicin/thuricin Z (143). Thuricin Z/huazacin stereocenters marked with * do not have assigned stereochemistry.

Similarly, huazacin³⁰⁶ (143, Figure 86; synonymous with thuricin Z, produced by B. thuringiensis serovar Huazhongensis) interacts directly with the bacterial cell membrane and induces ion permeability of artificial bacterial membranes.⁸⁶³ Confocal fluorescence microscopy demonstrated B. cereus uptake of propidium iodide upon huazacin treatment, indicating membrane permeabilization. Transmission electron microscopy of treated B. cereus showed an intact cell envelope with partial lysis of the inner bilayer. Scanning electron micrographs additionally demonstrated a collapsed bacterial cell surface and membrane perforations. Huazacin possesses a relatively narrow spectrum of activity, with MICs of 2–8 μM against B. cereus sp. and no notable activity towards other bacteria. The narrow spectrum activity is likely a consequence of species-specific differences in cell envelope composition. Bacteriomimetic large unilamellar vesicles (LUVs) (9:1 1,2-dipalmitoyl-sn-glycero-3-phosphatidyl glycerol to 1,2-dipalmitoyl-sn-glycero-3phosphatidylethanolamine) were permeabilized to K⁺/H⁺ upon huazacin treatment. Additionally, fungi-like LUVs (4:1 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine to ergosterol) did not demonstrate permeabilization in a huazacin-dependent manner.

Thurincin H (140, Figure 86), produced by B. thuringiensis SF36, also possesses potent activity against Bacillus and Listeria sp.⁸⁸⁷ with reported MICs against B. cereus strains 0.5–1 nM.⁸⁸⁸ Preliminary studies of thurincin H against B. cereus demonstrated a reduction in cell viability of ~3-log CFU/mL at 16 × MIC, but no decrease in optical density even at 256 × MIC, suggesting that thurincin H lacks lytic activity. B. cereus treated with 256 × MIC of thurincin H demonstrated a collapsed rod-shaped morphology. Moreover, surviving B. cereus did not display membrane punctures, as was the case when cells were treated with nisin (section 2.1). In assays utilizing propidium iodide, treatment with thurincin H did not permeabilize the membrane.⁸⁷⁴

Several other bioactive sactipeptides have been reported, though with no associated MOA. The gene cluster encoding sporulation killing factor (SKF, 144, Figure 87) was originally identified in B. subtilis as a target of the Spo0A transcriptional regulator.^889,890 The Spo0A protein is activated by phosphorylation as the terminal regulator of the Stage 0 sporulation phospho-relay pathway, which is responsible for initiating sporulation in B. subtilis. During times of nutrient limitation, Spo0A activates transcription of SKF, which induces lysis of other B. subtilis cells, presumably to release nutrients for surviving cells. SKF was subsequently determined to be a secreted compound containing sactionine linkages.^862,891 The molecular target of SKF remains unknown.

Two recently discovered sactipeptides further expand their structural diversity.^892–895 Ruminococcin C (RumC, 145, Figure 87) was identified as an anti-clostridial antibiotic in a gnotobiotic mouse model. When colonized with Ruminococcus gnavus, sterile mice generated feces that inhibited the growth of Clostridium perfringens.^892,896 RumC was purified from the fecal material, and shown to have antibacterial activity against C. perfringens, B. subtilis, and L. monocytogenes.⁸⁹² The solution NMR structure of RumC demonstrated a bi-hairpin structure (Figure 87), which is unique among sactipeptides, as these RiPPs generally contain only a single hairpin motif.⁸⁹⁴ Bacterial cytological profiling demonstrated that RumC does not exhibit pore-forming activity, but did induce phenotypes similar to metronidazole, which is known to inhibit nucleic acid synthesis.⁸⁹⁷ RumC has also demonstrated antibiofilm activity and can even disrupt biofilms at high (>2 × MIC) concentrations.⁸⁹⁸ Furthermore, RumC exhibits antifungal activity against Heterobasidion annosum, a property that is unique among known sactipeptides.⁸⁹⁸

The recently characterized streptosactin (146, Figure 87) from Streptococcus sp. exhibited narrow spectrum bactericidal activity against Streptococcus thermophilus LMD-9 and LMG 18311 at 1 μM.⁸⁹⁵ Streptosactin also displayed bactericidal activity against the producing strain S. thermophilus JIM 8232 at 1 μM. In all cases of bactericidal activity, streptosactin was dosed in the pre-logarithmic growth phase against strains possessing identical streptosactin biosynthetic gene clusters relative to the S. thermophilus producing strain leading to the hypothesis that this compound acts as a fratricidal agent. Previous studies also demonstrated that expression of the streptosactin gene cluster correlates with the development of competence in S. thermophilus.⁸⁹⁹ Currently, the fratricidal mechanism of streptosactin is proposed as a strategy for competent S. thermophilus cells to incorporate genomic DNA from non-competent neighboring cells.

15.4. Epipeptides

Epipeptides are RiPPs with D-amino acids installed by a rSAM enzyme. B. subtilis produces the epipeptide EpeX* (147) (formerly YddF*, where * indicates the epimerized peptide; Figure 88) that has potent bactericidal effects against the producing strain.^900,901 Such intra-species competition is believed to aid the survival of a sub-population at the expense of other members of the same population. Coordinated cell death induces lysis of susceptible cells, which release nutrients and transferrable genetic traits that can be utilized by the toxin-producing cells.^902–904 Although the precise function of EpeX* is not currently known, we focus on the bioactivity of EpeX*, and its possible role in cannibalism and fratricide - two distinct outcomes of coordinated cell death.

Figure 88.

Structure of EpeX* (147). Val4 and Ile12 are epimerized to the D-configuration (blue).

The LiaFRS system is a three-component regulatory system that senses cell envelope stress in B. subtilis (Figure 89). These proteins are encoded by the liaIH-liaGFSR locus,¹⁴¹ wherein LiaS is a membrane-bound histidine kinase, LiaR is a DNA-binding response regulator, and LiaF a negative regulator of LiaRS-mediated signal transduction.⁹⁰⁵ Upon cell wall perturbations caused by antibiotics like bacitracin, nisin (section 2.1), nukacin ISK-1 (section 2.2), cacaoidin (section 2.5), siamycin I (section 8.1), and vancomycin, the LiaFRS cell-envelope stress response is triggered.^141,900,906 LiaR binds to its primary target, the liaI promoter, and induces expression of the liaIH genes,^141,907 which encode for a transmembrane protein (LiaI) and a homolog of phage-shock protein A (LiaH). These two proteins are hypothesized to mediate cell membrane stress,^907,908 although the molecular-level details are unknown. The EpeX* encoding operon epeXEPAB (formerly yddFGHIJ) was identified in a search to uncover additional regulators of LiaFRS signaling. Cells deficient in the EpeAB ABC transporter showed increased expression from the P_liaI promoter.⁹⁰⁰ This observation indicates that the cell envelope stress response was triggered by the inability to export a self-produced toxic compound. Thus, the EpeAB transporter likely exports EpeX*.⁹⁰⁰

Figure 89.

(A) Schematic of the lia cell stress response signaling pathway. (B) Schematic of the lia signaling pathway in response to EpeX*. EpeP is a transmembrane protease that processes and exports EpeX*.⁹⁰⁹ EpeAB is an ABC transporter involved in providing self-immunity against the antibacterial compound.

Subsequent characterization of EpeX* has revealed it triggers the LiaRS response in stationary B. subtilis cells when exogenously supplied.⁹¹⁰ LiaIH mediates resistance against EpeX*, as a ΔliaIH strain was 3-fold more susceptible to EpeX* than wild type cells. EpeX* inhibits growth of B. subtilis (MIC < 0.94 μM), whereas synthetic analogs containing either all L-amino acids, or only one of the two modifications (D-Val4 or D-Ile12), did not inhibit growth of B. subtilis.⁹⁰¹ Investigation into the MOA of EpeX* revealed depolarization of the cell membrane, leading to membrane permeation as monitored by fluorescence dyes.⁹¹⁰ Furthermore, EpeX* rigidifies the membrane, which may account for its delayed polarization kinetics.

One proposal for EpeX* is that it functions as a cannibalism toxin, a strategy in which a toxin-producing sub-population benefits from nutrients released by the lysis of susceptible cells.^891,902,903 Two characterized cannibalism toxins in B. subtilis are sporulation delaying protein (SDP) and sporulation killing factor (SKF, section 15.3).^{891,911–913} Expression of both SDP and SKF are controlled by Spo0A, the master transcriptional regulator of sporulation.^889,890 Expression of EpeX* is correlated with the expression of SDP, suggesting that the two peptides could act synergistically.⁹⁰¹ In support, it was noted that the SDP and EpeX* producing operons are controlled by the same transcriptional repressor.⁹¹⁴ There is no evidence that the epe (formerly yyd) operon is regulated by Spo0A.⁹¹⁰

An alternative hypothesis is that EpeX* functions as a fratricidal toxin. Fratricide is a process that initiates along with natural competence, wherein non-competent sub-populations are exposed to antimicrobials for which they lack immunity. Any DNA resultant from cell lysis can be naturally incorporated into the competent population.^915,916 This process is suggested to enhance genetic diversity via exchange of extracellular DNA.⁹¹⁷ Deletion of ComA (section 7.2), the transcriptional regulator responsible for inducing natural competence in B. subtilis, indirectly leads to decreased levels of the epeXEPAB mRNA levels.^611,918 Further characterization of EpeX* is needed to validate whether EpeX* indeed functions as a fratricidal toxin.

15.5. Mycofactocin

Mycofactocin was bioinformatically predicted when co-occurrence was observed between genes encoding a specific set of redox-active enzymes and the mycofactocin biosynthetic gene cluster in mycobacteria.⁹¹⁹ These findings led to the hypothesis that mycofactocin could be a new redox cofactor for these enzymes.⁹²⁰ In vitro studies indicated that the putative oxidoreductases had non-exchangeable NAD cofactors and had low, single turnover activity, but that activity could be increased through the introduction of redox mediators.^919,921 Furthermore, disruption of the mycofactocin biosynthetic genes inhibited the growth of the producing strains on minimal media as first shown for M. tuberculosis^919,922 and M. smegmatis.^923,924 Additionally, mycofactocin-related genes were shown to be strongly upregulated in the presence of ethanol.⁹²⁵ Collectively, these studies were consistent with a possible redox cofactor role of the product of the mycofactocin biosynthetic gene cluster.

Because efforts to isolate mycofactocin from producing organisms were unsuccessful, the first direct evidence of mycofactocin acting as a cofactor was not obtained until 2019.⁹²⁴ Cell lysates from wild-type M. smegmatis were utilized to illustrate that mycofactocin can reduce dichlorophenolidonphenol (DCPIP), while mycofactocin-deficient lysates could not. A null variant was also generated for methanol dehydrogenase (Δmno), one of the enzymes that was thought to require mycofactocin as a cofactor, and the Δmno lysate was also unable to reduce DCPIP. However, DCPIP reduction was observed when lysates of the two mutants were mixed, suggesting mycofactocin was supplied by the Δmno strain and the Mno protein was provided by the mycofactocin deficient strain.⁹²⁴

Since these discoveries, the structure of pre-mycofactocin (148) has been elucidated by reconstitution of its biosynthetic enzymes (Figure 90).^925,926 The final structure of mycofactocin has yet to be confirmed and is believed to be further modified, potentially via glycosylation.⁹²⁷ Notably, pre-mycofactocin and glycosylated pre-mycofactocin were accepted as substrates for enzymatic catalysis by carveol dehydrogenase.^925,926

Figure 90.

Structure of pre-mycofactocin (148).

Pyrroloquinoline pyrrole is another RiPP cofactor that has been extensively studied. We refer the reader to previous reviews for its function.^928–932

16. Concluding remarks and outlook

Studies of the mechanisms of action of almost all major classes of RiPPs predate their identification as post-translationally modified peptides. Such studies have been foundational in enabling research advances in RiPP identification, characterization, and biosynthetic enzymology. No doubt spurred by the discovery of classes of RiPPs with coveted bioactivities, the identification and dissection of targets for these compounds continue to stimulate further advances including engineering and structure-activity relationship studies. The synteny of genes encoding all necessary precursors and biosynthetic proteins for RiPP production allows for a comparatively straightforward pipeline for prediction, production, and engineering.

While analytical, biochemical, and biosynthetic studies of RiPPs have largely been fueled by the genomic revolution and advances in instrumentation, efforts on identification of MOAs have not advanced as quickly. The challenge in this area is the fact that, despite the simplistic biosynthetic logic, the diversity of molecular scaffolds that can be elaborated by RiPP pathways is expansive. Thus, the bioactivities associated with even closely related structural classes of RiPPs cannot be easily predicted. Conversely, the slow rates at which MOAs are identified provide numerous opportunities for further investigations not only of newly discovered RiPPs but also for known compounds for which such studies are either limited in scope or have yet to be carried out. This review compiles what is currently known about the MOA of RiPP natural products and highlights the many gaps in our current understanding and thus the opportunities for further studies. Considering the staggering pace at which new RiPP scaffolds continue to be discovered, a large number of compounds are available for targeted MOA studies. Given that many known classes of RiPPs can be genetically configured for affinity purification from a heterologous source, it is entirely practical for the community to establish a common source collection to enable such future efforts.

Another area with much potential for future exploration is the various microbiomes, such as in the human gut and oral cavity or the plant rhizosphere. RiPPs are encoded in the genomes of a remarkably wide spectrum of different phyla and applications of either the producing organisms or the compounds as growth promotors or probiotics are exciting prospects. However, such applications will require much more investigation of aspects of RiPP research that have received much less attention. These include assessment of toxicity and immunogenicity, pharmacokinetics, and gastrointestinal stability.^38,933–937

One caveat with agnostic approaches towards MOA studies of RiPPs is that a number of these natural products may not have easily assayable activities or have pleiotropic activities. This aside, advancements in the technological state of instrumentation and computing have set the framework necessary for further in-depth studies. In-depth studies of each RiPP class could also result in more quantitative analysis of MOA and allow for useful comparisons across different classes of RiPPs that share the same target. Such comparisons are currently lacking and opportunities to fill this knowledge gap are ample. Given what is already known about the MOAs that are recapitulated here, additional studies will likely continue to provide unexpected insights for translating these compounds into modalities that are useful for society.

Biographies

Emily K. Desormeaux

Emily was born near Houston, Texas in 1994. She received her B.S. in Chemistry from Texas A&M University in 2017. The following year, she began her graduate career at the University of Illinois Urbana-Champaign working in the lab of Prof. Wilfred van der Donk. Her doctoral work is focused on the mechanistic enzymology of lanthipeptide synthetases and the identification of lanthipeptide gene clusters.

Adam J. DiCaprio

Adam was born in 1990 in Poughkeepsie, New York and received a B.S. in Biology from Ursinus College in 2012. After 3 years as an NMR research associate at Process NMR Associates, he joined the group of Prof. Doug Mitchell and obtained his Ph.D. in Chemistry from the University of Illinois at Urbana-Champaign in 2021. His thesis work centered on applications of NMR spectroscopy to elucidate protein-protein interfaces of biosynthetic importance as well as natural product structure determination. He recently joined Merck & Co. as a Senior Scientist within the NMR Structure Elucidation Group.

Chayanid Ongpipattanakul

Chayanid (Ing) was born in Bangkok, Thailand in 1994. She earned a B.S. in Biochemistry from the University of California, Los-Angeles in 2016. In 2021, she received her Ph.D. in Biochemistry from the University of Illinois at Urbana-Champaign having conducted research in Prof. Satish Nair’s laboratory. Her thesis work was focused on the structural and biochemical characterization of natural product biosynthetic enzymes. She is currently a postdoctoral researcher in Prof. Charles Craik’s group at the University of California, San Francisco.

Wilfred A. van der Donk

Wilfred was born in Culemborg, the Netherlands. He obtained his B.S. and M.S. at Leiden University under the direction of Jan Reedijk and Willem Driessen. In 1989 he moved to Rice University where he completed his Ph.D. in organic chemistry in 1994 with Kevin Burgess. After postdoctoral studies with JoAnne Stubbe at MIT, he started his independent career in 1997 at the University of Illinois at Urbana-Champaign, where he currently holds the Richard E. Heckert chair in the Department of Chemistry. Since 2008, he is an Investigator of the Howard Hughes Medical Institute. He is affiliated with the Departments of Biochemistry and Bioengineering and is a faculty member of the Carle R. Woese Institute for Genomic Biology. Research in his laboratory uses chemistry, enzymology and molecular biology to better understand enzyme catalysis and to use that knowledge for synthetic biology.

Douglas A. Mitchell

Doug was born near Pittsburgh, Pennsylvania in 1980. He received a B.S. in Chemistry from Carnegie Mellon University in 2002. After a short internship in medicinal chemistry at Merck Research Laboratories, he obtained his Ph.D. from the University of California, Berkeley in 2006 while working with Prof. Michael Marletta. For postdoctoral studies, he worked with Prof. Jack Dixon at the University of California, San Diego. Prof. Mitchell joined the Department of Chemistry at the University of Illinois at Urbana-Champaign in 2009 and currently holds the John and Margaret Witt Professorship. Mitchell is affiliated with the Department of Microbiology and is a faculty member of the Carle R. Woese Institute for Genomic Biology. The Mitchell lab employs a wide variety of approaches to identify and characterize novel natural products. The lab also has strong interests in bioinformatics, biosynthetic enzymology, and new reaction discovery.

Satish K. Nair

Satish was born in Bihar but comes from Kerala, India. His family emigrated to the U.S. during the bicentennial (1976), and he was educated in the New York City public school system. He received his B.S. in Chemistry (with Honors) from Brown University in 1989 and carried out his doctoral work with David Christianson at the University of Pennsylvania where he received his Ph.D. in Chemistry in 1994. He was a Leukemia Society of America post-doctoral fellow in the laboratory of Stephen K. Burley at Rockefeller University. Satish joined the Department of Biochemistry at the University of Illinois at Urbana-Champaign in 2001. He currently serves at the Head of Biochemistry and holds the Gregorio Weber Chair. He is also the Director for the Center of Biophysics and Co-Director of the Macromolecular CryoEM and MicroED facility. The Nair lab is interested in structure-function studies of enzymes involved in the biosynthesis of primary and secondary metabolites.