Recent Progress in Natural-Product-Inspired Programs Aimed To Address Antibiotic Resistance and Tolerance

Abstract

Bacteria utilize multiple mechanisms that enable them to gain or acquire resistance to antibiotic therapies during the treatment of infections. In addition, bacteria form biofilms which are surface-attached communities of enriched populations containing persister cells encased within a protective extracellular matrix of biomolecules, leading to chronic and recurring antibiotic-tolerant infections. Antibiotic resistance and tolerance are major global problems that require innovative therapeutic strategies to address the challenges associated with pathogenic bacteria. Historically, natural products have played a critical role in bringing new therapies to the clinic to treat life-threatening bacterial infections. This Perspective provides an overview of antibiotic resistance and tolerance and highlights recent advances (chemistry, biology, drug discovery, and development) from various research programs involved in the discovery of new antibacterial agents inspired by a diverse series of natural product antibiotics.

Graphical Abstract

INTRODUCTION

The discovery and routine administration of antibiotics to treat bacterial infections led to their recognition as “wonder drugs” in the mid-20th century.^1–9 To date, antibiotics are still considered one of the most significant interventions in human medicine. The efficacy and selectivity antibiotics exert made many believe that bacterial infections would be a thing of the past. However, instead of the routine eradication of bacterial pathogens in the clinic, we have witnessed an alarming rise in the failure to treat bacterial infections due to the generation of antibiotic-resistant “superbugs”^3,5 and antibiotic-tolerant persister cells and biofilms.^10–12

Clinically approved antibiotics operate via the selective inhibition of critical targets that impede replication and growth or eradicate bacterial cells. It is interesting to note that relatively few bacterial targets have been exploited for antibiotic treatments in the clinic. The mechanisms of action for our arsenal of antibiotics include the inhibition or disruption of (1) cell wall synthesis (e.g., β-lactams,¹³ vancomycin¹⁴), (2) protein synthesis (e.g., tetracyclines,¹⁵ macrolides,¹⁶ aminoglycosides,¹⁷ linezolid¹⁸), (3) RNA poly-merase (e.g., rifamycins^19–21), (4) nucleic acid synthesis (e.g., quinolones^22–25), (5) folate biosynthesis (e.g., sulfonamides, trimethoprim),²⁶ and (6) cell membrane (e.g., polymyxins;²⁶ Figure 1).^4,7

Figure 1.

Antibacterial targets of conventional antibiotics and mechanisms by which bacteria develop resistance to these therapeutic agents.

Upon exposure to antibiotics, bacteria can expediently evade their action using one or more well-defined resistance mechanisms.^{3–6,8,27,28} Bacterial resistance has led to the catastrophic rise of roughly 2 million annual infections, resulting in 23 000 deaths in the United States.²⁹ In Europe, more than 33 000 deaths were attributed to antibiotic-resistant bacterial infections in 2015.³⁰ Universal acquired resistance mechanisms can prevent antibiotics access to the aforementioned intracellular targets. The cell wall of Gram-negative bacteria consists of inner and outer membranes, which upon mutation can act as a permeability barrier to antibiotics leading to reduced penetration into the cells.^31,32 Alternatively, bacteria can actively eliminate, or “pump out”, antibiotics before they bind their corresponding intracellular target via the action of efflux pumps (e.g., efflux of macrolides).^31,32 Mutations and modifications of the target’s binding site are, collectively, another resistance mechanism bacteria utilize to acquire resistance as the resulting changes negatively impact drug–target interactions of antibiotic therapies. Specific examples of mutation and modification to impede antibiotic binding sites include (1) mutation and alteration of penicillin binding proteins to negate the effects of penicillin, (2) structural alterations in DNA gyrase to prevent fluoroquinolone action, and (3) methylation of critical residues within the rRNA of bacterial ribosomes to eliminate the binding of aminoglycosides.^33–35 Additionally, bacteria can enzymatically modify antibiotics to generate inactive agents, and examples include (1) hydrolysis of β-lactam antibiotics by β-lactamase enzymes to cleave and destroy the critical β-lactam warhead (pharmacophore) required for antibacterial activities and (2) modification, or functionalization (e.g., acylation), of hydroxyl and/or amine groups critical for aminoglycoside antibiotics to hydrogen bond (bind) to bacterial ribosomes by aminoglycoside-inactivating enzymes.²⁹ In addition, bacteria can overproduce targets to bypass the effects of antibiotics (e.g., trimethoprim; Figure 1).^29,35

Innate antibiotic tolerance is a distinct microbiological phenomena from acquired resistance and is now recognized as the underlying cause of chronic and recurring bacterial infections.³⁶ Using a communication process known as quorum sensing, bacteria secrete and sense small organic signaling molecules to monitor their population density and coordinate virulent behaviors.^37–41 Bacteria use quorum sensing to coordinate surface attachment and subsequent biofilm formation, where dense communities of bacterial cells are encased within matrixes of polysaccharides and extracellular DNA.^42–46 Once within a biofilm, bacterial cells divide at a much slower rate compared to their planktonic (free-floating) counterparts or they do not replicate at all. The subpopulation of biofilm cells that enter into a dormant state are known as “persister cells” and are characterized as metabolically inactive.^12,47–51 Extracellular polymeric substances (EPSs), or biofilm matrix, provide an additional layer of protection to surface-attached communities, enabling these bacteria to thrive in harsh conditions (e.g., host immune response).^52,53

It is important to note that all classes of FDA-approved conventional antibiotics were initially discovered as bacterial growth inhibitors against actively dividing cultures. Therefore, it should be no surprise that clinically used antibiotics are essentially ineffective against nonreplicating biofilm infections. With the 17 million new cases of biofilm infections in addition to >500 000 deaths each year in the United States that result from these infections, there is a critical need to identify compounds that effectively eradicate biofilms through growth-independent mechanisms (Figure 2).^10,41

Figure 2.

Presentation of the distinctions between acquired antibiotic resistance and innate antibiotic tolerance.

The current need for new and effective antibiotic therapies has motivated several research groups to initiate drug discovery programs inspired by various antibiotic natural products, which may be known or recently discovered. The Lewis group has created a new technology to facilitate isolation of novel antibiotics from previously inaccessible sources, while the Myers lab has developed novel total synthesis platforms for the discovery of new antibiotics. The Hergenrother group elegantly utilized a collection of diverse small molecules to establish guidelines that will allow the targeting of clinically imposing Gram-negative pathogens. The Miller group has pioneered a Trojan horse strategy and continues to make advances in this area by taking advantage of siderophore systems for bacterial import in unique and elegant ways to impart activity against Gram-negative organisms. Recent work by the Wuest lab has focused on a natural product with narrow-spectrum activity against Pseudomonas aeruginosa. Both broad- and narrow-spectrum-based strategies have been employed, but scientists agree that care must be taken to minimize resistance development against these novel agents. With biofilms being inherently tolerant to current antibiotics, the Lewis group and our lab have identified agents capable of eradicating persister cells and established biofilms. Our group has identified a series of synthetically tunable halogenated phenazine small molecules derived from a marine antibiotic that potently eradicates bacterial biofilms through a rapid iron starvation mechanism. Fortunately, recent advances in chemistry, molecular biology, and chemical biology have yielded a bevy of powerful techniques that can be employed to better understand problems associated with pathogenic bacteria and enable the discovery of novel natural-product-inspired therapies. This Perspective will detail each of these innovative and exciting antibiotic discovery strategies (Figure 3).

Figure 3.

Strategies to address the problems associated with antibiotic resistance and tolerance presented in this Perspective.

TEIXOBACTIN, A PROMISING NEW ANTIBIOTIC: iChip DISCOVERY, TOTAL SYNTHESIS, AND NOVEL ANALOGUES

Bacteria compete for finite resources in their environment and often produce antibiotics to compete with each other. Many antimicrobials used in the clinic today have been derived from natural products produced by bacteria. Large numbers of organisms cannot be grown successfully through traditional culturing methods and are thus named “unculturable”.¹ It is believed that approximately 99% of all species in external environments are uncultured.⁵⁴ However, recent development of iChip technology by the Lewis group has enabled the growth of previously uncultured bacteria (Figure 4).^54,55 This new technology has enhanced the potential for many more types of bacteria to be investigated and screened for the identification of new antibiotics.

Figure 4.

iChip (multichannel device) technology developed by Lewis and co-workers to culture previously “unculturable” bacteria, which led to the discovery of the new antibiotic teixobactin 1.

By use of the iChip, extracts from 10 000 bacterial isolates were grown and screened, which led to one isolate (β-proteobacteria, named Eleftheria terrae) that demonstrated antibacterial activity against S. aureus. From the Eleftheria terrae extract, teixobactin 1 was isolated and structurally elucidated using mass spectrometry, NMR, and Marfey’s analysis.⁵⁵ Teixobactin 1 proved to be a unique antibiotic containing a cyclic depsipeptide moiety composed of four amino acid residues, including an enduracididine residue, that are appended to a seven amino acid linear chain bearing a methylphenylalanine. It is also interesting to note that 1 has four d-amino acid residues incorporated into its impressive structure. The biosynthetic gene cluster consists of two nonribosomal peptide synthetase (NRPS) genes named txo1 and txo2 where 11 modules are encoded. The depsipeptide 1 showed excellent antibacterial activities in minimum inhibitory concentration (MIC) assays against significant, Gram-positive pathogens, including S. aureus (MIC = 0.25 μg/mL against methicillin-sensitive and methicillin-resistant strains; 1 also demonstrated antibacterial activities against intermediate vancomycin-resistant S. aureus, VISA), enterococci (MIC = 0.5 μg/mL, vancomycin-resistant enterococci strains, VRE), M. tuberculosis (MIC = 0.125 μg/mL), Clostridium dif f icile (MIC = 0.005 μg/mL), and Streptococcus pneumoniae (MIC ≤ 0.03 μg/mL, penicillin-resistant strain) as reported by Lewis and co-workers.⁵⁴

A common technique for determining the mechanism of action for antibiotics is to generate resistant mutants. Lewis and colleagues were not able to develop resistant mutants of S. aureus or M. tuberculosis, indicating that teixobactin’s target may not be a protein. This hypothesis was supported by the inhibition of peptidoglycan synthesis by 1, while no effect was observed on biosynthesis of DNA, RNA, or proteins. Because true vancomycin resistance in S. aureus is rare and vancomycin is known to bind lipid II, Lewis and co-workers hypothesized that teixobactin could also exert its activity through interaction with this target. When whole cells (S. aureus) were treated with 1, a significant amount of the peptidoglycan precursor undecaprenyl-N-acetylmuramic acid-pentapeptide (UDP-Mur-NAc-pentapeptide) accumulated in a similar fashion to cells treated with vancomycin. Competition enzyme binding assays indicated that 1 does not inhibit MurG, FemX, or PBP2 enzymes directly, while the addition of lipid II prevented 1 from inhibiting growth against S. aureus. These combined results demonstrate that 1 specifically interacts with the peptidoglycan precursor rather than interfering with the activity of the enzymes. Teixobactin retained activity against VRE which has a modified lipid II, suggesting that it binds at a different site than vancomycin.⁵⁴

The therapeutic potential of 1 was also reported by Lewis and co-workers.⁵⁴ Teixobactin retained its antibacterial potency in the presence of serum and demonstrated good chemical and microsomal stability and low toxicity. The pharmacokinetic parameters showed that after a single 20 mg per kg dose (iv injection) of 1 in mice, serum concentrations were maintained above MIC values for 4 h. Animal efficacy studies were then performed with 1 in a mouse septicemia model against MRSA at a dose that leads to 90% death; however, treatment with 1 (iv at 20 mg/kg, 1 h after infection) resulted in the survival of all mice, and a later experiment determined the PD₅₀ (protective dose at which half of the animals survive) of 1 to be 0.2 mg/kg. 1 was also shown to be efficacious in a thigh infection of S. aureus and a lung infection model of S. pneumoniae, causing a 6 log₁₀ reduction of colony forming units (CFU) per mL.⁵⁴

Following the initial report of teixobactin’s discovery, two total syntheses of 1 have been published utilizing solid-phase technology in separate reports by Payne and co-workers⁵⁶ and Li and co-workers.⁵⁷ In both total syntheses of 1, macro-lactamization at the less hindered Thr₈-Ala₉ connection has proved to be the fruitful cyclization approach (see conversion of 5 to 6, Figure 5A, by Li and co-workers) as opposed to a macrolactonization strategy. Following the macrolactamization step, Li and co-workers coupled linear peptide 3 to cyclic peptide 6 using a key ligation step in pyridine/acetic acid before final HPLC conversion to 1. Yuan and co-workers developed a stereoselective and scalable synthetic route to the rare l-allo-enduracididine (L-allo-End)⁵⁸ moiety of 1, which played a critical role in the total synthesis reported by Li and co-workers.

Figure 5.

(A) Total synthesis of the new antibiotic teixobactin 1 and (B) structure–activity relationship (SAR) insights.

In addition to total synthesis campaigns, solid-phase chemistry has enabled initial medicinal chemistry efforts of 1 by multiple groups, including Albericio,⁵⁹ Singh,^60–63 Su and Fang,⁶⁴ Rao,⁶⁵ Li,⁶⁶ and Chan.⁶⁷ Several of these groups have probed the structural requirements of the l-allo-enduracididine moiety at position 10 of 1. If this unique residue is not required for antibacterial activities, then more simplified analogues of 1 can be prepared more readily. Albericio and co-workers were the first to report the replacement of the unusual l-allo-enduracididine residue of 1 by substituting with L-arginine; however, this new analogue lost significant antibacterial activities against S. aureus (MIC = 1.6 μg/mL) and B. subtilis (MIC = 0.40 μg/mL) based on MIC value comparison (1 reported MIC values with ≥8-fold more potent activities when compared to this analogue).⁵⁹ Fang and Su and co-workers synthesized nine teixobactin analogues, which establish that (1) a guanidine or amine group at position 10, (2) the hydroxyl group of serine (position 7), (3) the N–H proton of the terminal phenylalanine (position 1) were all critical for antibacterial activity. During these investigations, Fang and Su also reported that position 4 can tolerate various side chain size and functional group content.⁶⁴

The initial reports of teixobactin analogues gave insightful information into the SAR of 1; however, a “first wave” of synthetic analogues reported by Albericio, Su, Fang, and co-workers^59,64 lost a substantial amount of antibacterial activity and appeared to be less than exciting lead molecules. Rao and co-workers⁶⁵ recently reported the synthesis of 21 new teixobactin analogues, including simplified and equipotent analogue 7 (Figure 5B). This work demonstrated that the N-Me-d-Phe residue at position 1 of teixobactin (1) tolerates additional hydrophobic groups on the phenyl ring and (2) does not require the methyl group on the amine, as the primary amine was well tolerated. These structural modifications combined with a l-lysine substitution of the l-allo-enduracididine residue at position 10 resulted in equipotent analogue 7, which also demonstrated outstanding in vivo efficacy in a S. pneumonia mouse model of infection (100% survival rate of mice infected with bacterial loads that lead to 90% death in controls).

DEFINING RULES TO TARGET GRAM-NEGATIVE BACTERIA AND CONVERTING A DEOXYNYBOMYCIN ANALOGUE TO A BROAD-SPECTRUM ANTIBIOTIC

Gram-negative pathogens are a major clinical problem due to their lack of sensitivity to many commonly used anti-biotics.^68,69 Quinolones were the last new class of antibacterial agents active against Gram-negative bacteria introduced into the clinic in 1968.¹ Despite extensive high-throughput screening efforts against Escherichia coli with large compound libraries of synthetic compounds, no lead compounds were translated into clinical agents.⁷⁰ With two cellular membranes and a lipopolysaccharide-coated outer membrane, Gram-negative bacteria pose significant challenges for compound entry.^68,71–73 Compounds capable of penetrating the outer membrane do so through porins, which are narrow trans-membrane proteins lined with charged amino acids. Once inside Gram-negative bacterial cells, small molecules are susceptible to removal through the function of efflux pumps; therefore, in order for a compound to accumulate within Gram-negative bacteria and engage its intracellular target, compounds are required to enter through porins faster than being removed via efflux.⁶⁸

Hergenrother and co-workers noted that despite initial efforts by others, we have a limited understanding of the physicochemical properties enabling compound accumulation in Gram-negative bacteria.^68,72,73 A structurally diverse set of 100 compounds with unique scaffolds rapidly accessed from available natural products were used to systematically analyze small molecule accumulation in E. coli (whole cells) using a liquid chromatography with tandem mass spectroscopy (LC–MS/MS) method. This approach using whole cells allowed multiple variables affecting compound accumulation to be accounted for, such as multiple porins, efflux pumps, and lipopolysaccharide composition. Conventional antibiotics with known high (i.e., tetracycline) and low (i.e., novobiocin) accumulation properties were evaluated as controls and reported to validate the method.⁶⁸

Natural-product-derived compounds were selected for this study as most clinically used antibacterial drugs are either natural products or derivatives. The compound collection utilized by the Hergenrother lab was synthesized by this team, so these compounds are synthetically accessible with tunable properties to enable structure–activity relationships geared toward accumulation in Gram-negative bacteria. The library of 100 compounds utilized for this work, which was generated utilizing the “complexity-to-diversity” approach,^74–78 included positively charged, negatively charged, and neutral compounds.⁶⁸ The chemical structure, molecular weight, ClogD_7.4, and charge of all compounds were tracked with accumulation results in E. coli.

From the initial accumulation studies, the primary factor that governed accumulation in E. coli was charge, with 12 of 41 positively charged compounds being the most likely to accumulate. Interestingly, 8 of the 12 positively charged compounds were primary amines.⁶⁸ In follow-up SAR investigations, replacement of the primary amine with different functional groups (e.g., carboxylic acid, amide, alcohol) or amines bearing more substitutions significantly reduced accumulation in E. coli. Although the presence of a primary amine was important for accumulation, it alone was insufficient; therefore, chemoinformatics was utilized to understand which factors contribute to amine accumulation. This work revealed that the flexibility, as measured by the number of rotatable bonds, and shape of a compound are important factors that govern accumulation. Compound shape was described by “globularity” to inform the three-dimensionality. Shape was scored on a scale from 0 globularity for a “flat” compound (e.g., benzene) to 1 globularity for a “spherical” compound (e.g., adamantane).

On the basis of the extensive data set generated by Hergenrother and co-workers, the following guiding principles for accumulation in E. coli were developed: (1) compounds are most likely to accumulate if they contain a nonsterically encumbered amine, (2) contain some nonpolar functionality (typically not a problem for organic compounds), (3) contain a rigid structure and (4) have low globularity.⁶⁸ These guidelines are referred to as the “eNTRy rules” (N, ionizable nitrogen, primary amine is best; T, low three-dimensionality, globularity score of ≤0.25; R, relatively rigid, ≤5 rotatable bonds).⁷³ These guidelines were validated retrospectively with antibacterial agents presented by O’Shea and Moser.⁷⁹ In addition, the role of porins was examined by testing the high-accumulation antibiotic controls and test compounds in a strain bearing a diminished number of porins.⁶⁸ As predicted, a decrease in accumulation was observed for all compounds tested in the strain with fewer porins. Molecular modeling was performed with select test compounds and antibacterial agents that transverse the major porin of E. coli, OmpF, revealing a key interaction between the necessary primary, or non-hindered, amine and acidic residues (most often Asp113) that allow high accumulating compounds.

These well-defined guidelines were instructive in converting an analogue of the Gram-positive antibiotic deoxynybomycin (DNM, 9) into the broad-spectrum agent, 6DNM-NH3 (11, Figure 6).⁶⁸ DNM is an inhibitor of DNA gyrase,⁸⁰ however, only active against Gram-positive pathogens, presumably due to minimal accumulation in Gram-negative bacteria.⁶⁸ DNM possesses shape and rigidity features that align with the eNTRy rules; however, no amine group is present. Analogue 6DNM 10 possesses an aliphatic bridge expanded by one methylene unit to give a six-membered ring, compared to DNM’s (9) five-membered ring.

Figure 6.

eNTRy rules defined by Hergenrother and co-workers that enabled the successful conversion of the Gram-positive antibacterial agent 6DNM 10 into 6DNM-NH3 11, a new broad-spectrum antibacterial agent.

As one would expect based on the small structural differences between analogues, DNM 9 and 6DNM 10 possess the same activity profile (S. aureus MIC = 0.06–1 μg/mL; E. coli MIC > 32 μg/mL). However, 6DNM-NH3 11 has incorporated a primary amine appended to the aliphatic region of the six-membered ring, and as predicted based on the eNTRy rules, 11 demonstrates significantly higher levels of accumulation compared to 6DNM 10 (11, accumulation = 1114 nmol per 1012 CFU; 10, accumulation = 298 nmol per 10¹² CFU). In line with eNTRy rule design and observed increases regarding E. coli accumulation, 6DNM-NH3 demonstrates broad-spectrum antibacterial activities (11, S. aureus MIC = 0.03–0.5 μg/mL; E. coli MIC = 0.5–16 μg/mL), including activities against several Gram-negative pathogens (11, A. baumannii MIC = 2–16 μg/mL; K. pneumoniae MIC = 1–8 μg/mL; E. cloacae MIC = 0.5–4 μg/mL; P. aeruginosa MIC = 16 μg/mL).⁶⁸ Steered molecular dynamics simulations indicated that 6DNM-NH3 11 was able to pass through the porin OmpF with the assistance of the key interaction between the primary amine of 11 and Asp113 within the porin channel; however, 6DNM 10 was incapable of this interaction and required distortion in Asp113 and neighboring residues to allow passage. This ground-breaking work will undoubtedly pave the way for the discovery and development of antibiotics against Gram-negative pathogens.⁷²

DEVELOPING TOTAL SYNTHESIS PLATFORMS FOR NEW, FULLY SYNTHETIC ANTIBIOTICS

One of the primary methods to discover and develop new antibacterial agents over the past 60 years has been semisynthesis, which is chemical synthesis using natural products (e.g., antibiotics) as starting points.⁸¹ Early examples of semisynthesis in the antibacterial arena include the conversion of (1) streptomycin to dihydrostreptomycin (via a chemoselective reduction of the aldehyde in streptomycin to a primary alcohol to yield a new, more stable antibacterial agent) and (2) chlorotetracycline to tetracycline (via hydro-genation; tetracycline was later isolated). With these early success stories, medicinal chemists realized the value of semisynthesis for antibacterial discovery and utilized this approach to advance other antibiotic classes (e.g., macrolides, synthetic transformation of erythromycin to azithromycin in four steps); however, this approach has significant limitations as deep-seated modifications are not always accessible via semisynthesis. Myers and co-workers are working to address these limitations by developing modular total synthesis platforms, and they have made significant progress generating novel tetracycline^81–88 and macrolide antibiotics^89–93 bearing structural modifications not possible through current semisynthetic approaches.

In the mid-1990s, the Myers lab began work on fully synthetic tetracycline antibiotics.⁸² This program led to the implementation of a high-yielding, diastereoselective Michael–Claisen annulation between benzoate esters related to 13 and chiral “AB enone” 14, resulting in the constructive C-ring “stitching” of new tetracycline antibiotics.^81–86 The “diversity element” of this route can be clearly seen in structure 15 (Figure 7), which bears four positions poised for structural modification. On the basis of structure–activity relationship knowledge, the diversity element tracks to positions that do not negatively impact bacterial ribosome inhibition by tetracyclines. This incredible chemistry platform has enabled >3000 fully synthetic tetracyclines to date,^81,82,86 several of which have proven to be promising clinical candidates (e.g., 24, 25) including ervacycline 26 (XERAVA), which recently received FDA approval for the treatment of complicated intra-abdominal infections (cIAI).⁹⁴ As presented in Figure 7D, ervacycline 26 has very potent, broad-spectrum activity against several pathogenic bacteria, including methicillin-resistant Staphylococcus aureus.

Figure 7.

(A) Myers’ retrosynthetic analysis of tetracycline. (B) Key Michael–Claisen annulation step for fully synthetic tetracycline antibiotics. (C) Mechanism of Michael–Claisen annulation reaction. (D) Antibacterial profiles of fully synthetic tetracycline antibiotics, including FDA-approved eravacycline 26.

The Myers lab has also recently established a total synthesis platform for the discovery of new macrolide antibiotics.^89–93 Myers and co-workers reported the generation of >300 fully synthetic macrolides (FSMs) that incorporate structural design features that combine structure–activity relationship knowledge (macrolactone, C9 ketone/amine replacement, 14- or 15-membered macrocycle, incorporation of desosamine sugar at C5) with resistance liabilities (replacement of C3-clasinose sugar with ketone to give ketolides; Figure 8A) to inform the synthesis of new macrolides.⁹² Eight simple building blocks (30–33, 38–40, 47; Figure 8B) were utilized to assemble key left-handed fragment 44 (amine) and right-handed fragment 43 (ketone), which were then joined, or converged, through a stereoselective reductive amination reaction followed by sequential macrolactonization and click reactions to yield 48 (a 14-membered azaketolide). The Myers team noted that many past macrolide synthesis efforts have failed at the critical macrolactonization step; therefore, special attention was made to rigidify the substrate via the cyclic carbamate group, which led to an impressive 78% yield for the macrolactonization step (on gram scale). Several clinically effective and promising macrolides have cyclic carbamate moieties fused to the C12/ C13 position of the macrolide, including telithromycin, cethromycin, and solithromycin 50. An additional benefit of the cyclic carbamate moiety is that the aniline residue of solithromycin 50 is known to bind a unique pocket in the bacterial ribosome,⁹² making such structures better inhibitors of this target and more effective antibacterial agents.

Figure 8.

(A) Overview of the structure–activity relationships related to macrolide antibiotics with Myers’ design elements regarding new, fully synthetic macrolides. (B) Overview of the convergent chemical synthesis approach to access fully synthetic macrolide small molecules.

A series of 15-membered azaketolides, including FSM 49, was synthesized using a similar route to the 14-membered azaketolides (Figure 8B).⁹² This new route required modification to only three of the building blocks (34, 35, and 41, the original glycosyl donor reported by Woodward et al.⁹⁵) used to synthesize the 14-membered azaketolides. The key left-handed fragment 44 was also utilized in this route and underwent an analogous (1) reductive amination with right-hand aldehyde 45, (2) macrolactonization (94% yield, 1.9 g scale), (3) click reaction sequence to afford 15-membered azaketolide 49. Myers and co-workers also utilized the key right-hand aldehyde 45 to access a series of 14-membered ketolides, including the clinical candidate solithromycin 50 and approved drug telithromycin (not shown). Intermediate 42, which was accessed through a stereoselective addition of ketone 36 with alkynyl lithium 37, was transformed into a Grignard reagent via the hydromagnesiation of the acetylenic moiety and subsequently reacted with aldehyde 45 to join the key left- and right-hand fragments of the ketolide series. Following a subsequent oxidation/deprotection sequence, the key macrolactionization step was successfully carried out in 66% yield (1.7 g scale) to produce the desired 14-membered macrolactone scaffold. A final reaction sequence involving C2 fluorination and installation of the cyclic carbamate (bearing the important aniline residue) led to the synthesis of solithromycin 50 in 14 steps and 16% yield from 34 and 35.

This total synthesis platform was utilized to generate >300 FSM antibiotic candidates by varying building blocks in addition to modifying readily diversifiable structural features at multiple positions around the macrolactone architecture.⁹² This proved to be a powerful synthetic medicinal chemistry tactic as deep-seated variations at select positions within FSM scaffolds were accessed, which are not possible using semisynthetic approaches. Following chemical synthesis, 305 FSMs were evaluated for antibacterial activities against a diverse panel of Gram-positive and Gram-negative pathogens. Potent antibacterial agents were identified among each of the FSM subclasses (14-membered azaketolides, 15-membered azaketolides, 14-membered ketolides).

Following the initial screen, the most promising FSMs were evaluated against an expanded panel of bacteria with genetically characterized resistance mechanisms to erythromycin (ermB, ribosome-modifying methyltransferase; mefA, efflux; Figure 9) and other antibiotics (MRSA and VRE strains).⁹² Several analogues (e.g., 49, 51, 52; Figure 9) demonstrated significant improvements in activity against this macrolide-resistant panel of pathogenic bacteria. FSM 51 (FSM-100573) is a 14-membered ketolide that demonstrated remarkable activities during these investigations. In fact, 51 outperformed every current clinically used macrolide antibiotic against extremely challenging strains, such as S. pneumoniae with both ermB and mefA genes. FSM 51 also demonstrated significant improvements against Gram-negative pathogens, such as A. baumannii and P. aeruginosa.

Figure 9.

Antibacterial profiles of select FSM analogues identified from microbiological studies. Several analogues demonstrate high antibacterial potencies and overcome specific macrolide resistance mechanisms. Resistant strain details are the following: iErmA, inducible erythromycin ribosome methyltransferase A; MsrA, macrolide streptogramin resistance efflux pump A; ErmB, erythromycin ribosome methyltransferase B; MefA, macrolide efflux protein A.

This incredible synthesis platform has created unprecedented opportunities to explore and develop novel macrolide antibiotics. The synthesis approach to FSMs is highly convergent from multiple building blocks; however, key advanced left-handed and right-handed intermediates were strategically utilized in divergent pathways to access multiple FSM subclasses (e.g., left-handed fragment 44 was used to synthesize 14- and 15-membered azaketolide scaffolds; right-handed fragment 45 was used to synthesize 15-membered azaketolide and 14-membered ketolide scaffolds). Similar to Myers’ tetracycline synthesis platform, one can envision accessing thousands of FSM from the chemistry described here. Myers and co-workers have truly inspired scientists and clinicians with this ground-breaking work, which has great potential to lead to new clinical agents that overcome bacterial resistance in human patients.

PROMYSALIN: A SPECIES-SPECIFIC ANTIBIOTIC THAT TARGETS SUCCINATE DEHYDROGENASE IN Pseudomonas aeruginosa

Broad-spectrum antibiotics target essential pathways in bacteria, including cell wall and protein synthesis, and are capable of treating a wide range of infections.⁹⁶ However, misuse and prolonged exposure to broad-spectrum antibiotics can result in undesirable side effects (e.g., inflammatory diseases, obesity) as commensal population dynamics can be dramatically altered by these agents.^96–98 “Narrow-spectrum” antibiotics (e.g., amoxicillin) target large subsets of bacteria, such as Gram-positive versus Gram-negative bacteria, or anaerobes versus aerobes; however, species-specific agents against significant human pathogens would be extremely useful and could avoid the undesired side effects that result from the misuse of our current antibiotic arsenal.⁹⁶

Over the past several years, Wuest and co-workers have developed an exciting medicinal chemistry and chemical biology program around the species-selective agent promysalin 53 (Figure 10A), which targets the major Gram-negative pathogen Pseudomonas aeruginosa.^96,99–103 The multispecies community found in the root systems of various plants is known as the rhizosphere,^104,105 and in this environment, bacteria are known to utilize chemical warfare strategies for colonization and to defend themselves. Pseudomonads are highly prevalent in the rhizosphere, and these bacteria produce an array of secondary metabolites with biological activities that promote survival, including antibiotics, virulence factors, biosurfactants, siderophores (to obtain iron from their environment).⁹⁹ Promysalin is a secondary metabolite isolated from Pseudomonas putida found in rice plants.¹⁰⁶ This natural product induces swarming and biofilm formation in P. putida; however, it demonstrates unique species-specific antibacterial activity against P. aeruginosa (growth inhibition, IC₅₀ = 0.83 μg/mL; note, MIC values have not been reported for the promysalin work described here).

Figure 10.

(A) Retrosynthesis and (B) synthetic scheme to the Pseudomonas aeruginosa selective natural product promysalin by Wuest and co-workers. (C) Representative promysalin analogues synthesized using diverted total synthesis to probe structure–activity relationships against P. aeruginosa (strain PA14).

Wuest and co-workers reported a convergent and concise total synthesis of (−)-promysalin 53 (Figure 10).⁹⁹ In the original isolation paper by De Mot and co-workers,¹⁰⁶ the three stereogenic centers of promysalin were not defined. Therefore, Wuest’s total synthesis of promysalin involved a unique combination of bioinformatics (reannotation of the biosynthetic gene cluster via AntiSMASH computation) and stereoselective approaches (reagent control, substrate selection; Figure 10B) to yield four diastereomers of promysalin for structural elucidation and biological assessment.

All four diastereomers of the myristic acid fragment 60 (diastereomers 60a–d) of promysalin were synthesized using a stereoselective, five-step synthetic route.⁹⁹ First, Evans’ oxazolidinone 54 underwent a diastereoselective hydroxylation (oxidation) upon treatment with sodium hexamethyldisilazane followed by addition Davis oxaziridine 58, then a TBS-protection afforded 59. This was followed by a cross-metathesis reaction with homoallylic alcohol 55, and then hydrogenation of the resulting olefin (not shown) and final ammonolysis of the oxazolidinone moiety with ammonium hydroxide afforded amide 60. EDC-coupling of the alcohol in 60 with acid (−)-62 and then silyl-group removal with tetrabutylammonium fluoride afforded promysalin 53. All four diastereomers of 60 (60a–d) were synthesized and coupled to (−)-62 to afford all possible diastereomers at C2 and C8 of the myristic acid moiety of promysalin (53a–d).

With the four diastereomers of 53 in hand, NMR comparison and antibacterial assays were utilized in concert to unequivocally identify diastereomer 53a as the natural product. Unfortunately, no optical rotation or authentic sample was available to compare; however, Wuest and co-workers rationalized that the correct enantiomer would exhibit the most potent biological activity.⁹⁹ Promysalin 53a, which had identical spectral data to those reported in De Mot’s isolation paper, proved to have the most potent antibacterial activities against P. aeruginosa strain PA14, which demonstrated an IC₅₀ value of 125 nM (IC₅₀ = 1.8 μM, initial report) and an IC₅₀ value of 1 μM against PAO1. The other diastereomers 53b–d demonstrated >10-to >60-fold less potency against P. aeruginosa strain PA14 than 53a. On the basis of these findings, the absolute stereochemistry of natural promysalin (−)-53 was assigned as (2R,8R,16S).

By use of the total synthesis of promysalin 53 as a platform, a series of 16 related analogues was generated via diverted total synthesis.⁹⁶ This approach allowed several structural features to be probed (i.e., ester linker, proline ring size and functionality, hydroxyl group on the myristic acid fragment) to gain structure–activity relationship information regarding 53. From these investigations, several interesting structural features were identified that led to a complete abolishment of antibacterial activity against PA14, including (1) proline ring size and saturation (analogue 63) and (2) substitution of the ester group with an amide (analogue 64). In contrast, removal of the C2 hydroxyl group (analogue 65) and introduction of a vinyl fluoride on the proline moiety (analogue 66) resulted in promysalin analogues with improved antibacterial activities against PA14.

The SAR insights gained from the diverted synthesis of promysalin analogues enabled the design, synthesis, and utilization of diazirine photoprobe 67, which retained activity, in affinity-based protein profiling (AfBPP) studies.¹⁰¹ Active promysalin photoprobe 67, inactive photoprobe 68 (to identify nonspecific protein binding partners), and promysalin (for competition purposes) were utilized in an elaborate series of experiments (involving irradiation of photoactive probes in cell lysate, subsequent click reaction with biotin azide, pull-down, dimethyl isotope labeling, and LC–MS/MS measurement; Figure 11) that led to the identification of succinate dehydrogenase (SdhC) as the target of promysalin. Further validation that SdhC was the target of 53 was carried out in vitro and via whole genome sequencing of resistant mutants. These findings demonstrate that targeting the tricarboxylic acid (TCA) cycle could be useful in developing narrow, species-selective antibiotic therapies and warrant additional studies regarding SdhC as an antibiotic target.^101–103

Figure 11.

Experimental workflow for pull-down experiments using photoactive diazirines 67 and 68 to identify succinate dehydrogenase (SdhC) as the biological target of species-specific antibiotic promysalin 53.

RECENT ADVANCES OF SIDEROPHORE CONJUGATES FOR “TROJAN HORSE” TARGETED DELIVERY OF ANTIBIOTICS TO PATHOGENIC BACTERIA

Bacteria are able to successfully establish an infection when they experience unlimited growth and proliferation within a host.¹⁰⁷ In order for bacteria to proliferate, adequate supply of key nutrients is required. Iron is a crucial nutrient that bacteria require to thrive and is known to play critical roles in energy metabolism, stabilization of protein structures, and oxygen transport.^107–109 Unfortunately, iron is not membrane soluble and, therefore, is unable to diffuse through bacterial membranes, so acquisition of this key nutrient has its challenges; however, bacteria synthesize and utilize specialized organic molecules known as siderophores to acquire iron from their surroundings.^107,108 Bacteria secrete specific siderophores into their environment where they tightly bind iron(III). The resulting iron(III)-siderophore complex is then recognized by specific cell-surface receptors on bacteria, which shuttles the complex inside the bacterial cell. Once within the reductive cytoplasm of the bacterium, iron(III) is reduced to iron(II), resulting in a loss of affinity by the siderophore and subsequent release of iron(II) for unitization by the bacterial cell.¹⁰⁷

Bacteria utilize highly specific siderophore systems that can be exploited for bacterial targeting and drug delivery. For instance, the Gram-positive pathogen S. aureus utilizes two siderophores, staphyloferrin A and staphyloferrin B (69, Figure 12A).¹¹⁰ Enterobactin (71, Figure 12B) is a siderophore produced by multiple Gram-negative pathogens, including E. coli and S. typhimurium.¹¹¹ P. aeruginosa and other pathogens have effectively developed receptors to recognize and transport iron(III)-siderophore complexes from other bacterial species (xenosiderophores) providing a competitive growth advantage. Although enterobactin is not produced by P. aeruginosa, this siderophore can promote iron uptake in this pathogen.¹¹¹

Figure 12.

(A) Structure and iron(III) complex with the S. aureus siderophore staphyloferrin B. (B) Enterobactin and iron(III) bound complex utilized by multiple Gram-negative pathogens for iron acquisition. (C) Utilization of iron uptake systems as a platform for antibiotic drug delivery by Miller and co-workers (“Trojan horse” conjugate 74 selectively and potently targets P. aeruginosa via iron uptake systems).

These iron-specific uptake systems have provided a promising platform for targeted antibiotic delivery to select pathogens, especially for Gram-negative pathogens that present challenges regarding compound penetration as a result of two cellular membranes. Miller and co-workers have pioneered this area with much success in recent years with several inspiring examples of potent and targeted antibiotic delivery using a “Trojan horse” strategy that links antibiotics to various synthetic siderophores (sideromycins).^111–118 One example is presented in Figure 12C, where Miller and colleagues designed triscatecholate sideromycin 73, inspired by catechol-based siderophore enterobactin 71, which is utilized to target P. aeruginosa.¹¹¹ Sideromycin 73 was designed to have a linker moiety for attaching an antibiotic agent, in this case amoxicillin (ampicillin was also attached, not shown) to afford “Trojan horse” conjugate 74 (note: acetate groups were installed on catechol moieties to prevent pharmacological side effects and serve as prodrugs while circumventing potential methylation via a catechol O-methyl-transferase that would result in the loss of iron binding capabilities of the siderophore). Trojan horse conjugate 74 demonstrated potent antibacterial activities against P. aeruginosa strains in iron-deficient media, which resulted from the hijacking of energy-dependent active bacterial iron uptake systems required for bacterial growth under these conditions.¹¹¹ During these studies, MIC values for 74 ranged from 0.05 to 0.39 μM under iron-deficient media compared to MIC values that ranged from 12.5 to 50 μM in iron-rich media (note: amoxicillin was inactive against these P. aeruginosa strains, MIC > 100 μM).

A more detailed schematic of the Trojan horse strategy to target Gram-negative pathogens can be found in Figure 13A. Recently, Miller’s group has published two outstanding examples of Trojan horse conjugates hijacking iron uptake systems that enable Gram-positive antibiotics (an oxazolidinone¹¹⁷ and daptomycin;¹¹⁶ Figure 13B and Figure 13C) to effectively target Gram-negative pathogens, including Acinetobacter baumannii. These impressive examples of Trojan horse antibiotic delivery also showcase “releasable linkers” versus “nonreleasable linkers” that are an important design component in these systems.

Figure 13.

(A) Illustration of the Trojan horse strategy for targeted antibiotic delivery. (B) Example of Trojan horse strategy with a suicide siderophore–cephalosporin–oxazolidinone 75 (via releasable linker) recently reported by Miller and co-workers.

Oxazolidinones target bacterial ribosomes and are effective at treating Gram-positive infections; however, these agents are inactive against most Gram-negative pathogens as they are unable to penetrate their outer membrane, or if they do permeate the Gram-negative bacteria, they are rapidly effluxed.¹¹⁷ This spectrum of activity along with these resistance mechanisms (lack of penetration, efflux) makes oxazolidinones ideal candidates for Trojan horse conjugation. With that, sideromycin 75 was designed as a dual active agent containing siderophore–cephalosporin–oxazolidinone components, which combine the Trojan horse entry of Gram-negative pathogens with β-lactamase triggered release upon destructive ring-opening of the cephalosporin moiety to liberate an active oxazolidinone (76; Figure 13B).¹¹⁷

In the arsenal of Trojan horse conjugate linkers, it is important to have releasable linker options to avoid conjugate structures that perturb critical drug–target interactions, ultimately preventing antibiotic activities. Early attempts to develop releasable linkers included functional groups that were hydrolyzable (i.e., ester cleavage by esterases¹¹²) or reductively cyclized (i.e., trimethyl lock bioreductive activation,¹¹⁵ not shown) and led to problems with premature antibiotic release before iron-uptake dependent penetration of Gram-negative bacteria.¹¹⁷ The cephalosporin linker in 75 requires the enzymatic activity of intracellular β-lactamases for “suicide” antibiotic release, bypassing premature release issues presented by other releasable linker systems. In initial proof-of-concept experiments, 75 underwent rapid and complete hydrolysis by purified ADC-1 β-lactamases, which are ADC enzymes found in A. baumannii that are capable of hydrolyzing cephalosporin antibiotics.¹¹⁷

The siderophore–cephalosporin–oxazolidinone 75 demonstrated impressive antibacterial activities against multiple Gram-negative pathogens, including A. baumannii (multiple, drug-resistant isolates; MIC = 0.4–6.25 μM), E. coli (MIC < 0.025 μM), and P. aeruginosa (MIC = 0.2–0.4 μM).¹¹⁷ The corresponding oxazolidinone 76, alone or when linked directly to the siderophore moiety (without cephalosporin to release 76), was inactive against these pathogens (MIC from >50 to >500 μM). Interestingly, the siderophore–cephalosporin moiety alone (without oxazolidinone 76; structure not shown) has moderate to good activities against some pathogens but is dramatically enhanced with all three moieties (75) intact against A. baumannii. For instance, against A. baumannii ATCC BAA 1800, the siderophore-cephalosporin components report MIC = 25 μM versus MIC = 0.8 μM for siderophore–cephalosporin–oxazolidinone 75, demonstrating a clear structure–activity relationship for the necessity of all three structural components to display potent antibacterial activities at a level that is clinically relevant.

In separate work, Miller and co-workers coupled a mixed synthetic ligand analogue of A. baumannii siderophore fimsbactin with daptomycin to give conjugate 79.¹¹⁶ Daptomycin is a large, negatively charged lipopeptide anti-biotic that has significant value as a therapeutic agent against Gram-positive pathogens only. Although daptomycin’s mechanism is not fully understood, it has been found to bind and disrupt bacterial cell membranes, resulting in rapid depolarization, concomitant ion efflux and dysregulation of nucleic acid and protein synthesis, and subsequent bactericidal death.¹¹⁹ Prior structure–activity relationship investigations revealed that acylation of the primary amine of the ornithine residue in daptomycin was well-tolerated, and thus the Miller group used this information in the design of daptomycin-siderophore conjugate 79.¹¹⁶ Regarding chemical synthesis, the trichloroethyl chloroformate (Troc) protecting group of 77 was removed using zinc in the presence of glutaric anhydride to afford 78 as a benzyl protected siderophore bearing a (nonreleasable) glutarate linker for subsequent conjugation (Figure 14). A mixed anhydride of 78 was generated in situ before subjecting the siderophore-linker moiety to daptomycin under aqueous reaction conditions for conjugation via amide bond formation. Final hydrogenolytic removal of the benzyl groups provided target daptomycin-siderophore conjugate 79.¹¹⁶

Figure 14.

Chemical synthesis and targeted A. baumannii activities of siderophore–daptomycin conjugate 79.

Siderophore-daptomycin conjugate 79 was shown to bind iron(III) stoichiometrically and demonstrated profound effectiveness against A. baumannii strains (MIC = 0.4–0.8 μM).¹¹⁶ As predicted, daptomycin itself was inactive against A. baumannii strains (MIC > 100 μM) when tested alongside conjugate 79. To demonstrate the importance of iron-binding and uptake regarding conjugate 79, its synthetic precursor bearing bis benzyl group protected catechol moieties (structure not shown) proved to be completely inactive against A. baumannii (MIC > 50 μM). When tested against other Gram-negative pathogens, such as P. aeruginosa, Burkholderia multivorans, and Escherichia coli, conjugate 79 proved to be inactive (MIC > 100 μM), demonstrating the selective targeting of this Trojan horse conjugate toward A. baumannii.

Following in vitro assessment, Trojan horse conjugate 79 was advanced to testing in mice.¹¹⁶ Intravenous administration of 79 at 250 mg/kg in ICR mice resulted in no observed adverse effects. Following favorable toxicity studies in mice, conjugate 79 was evaluated for in vivo efficacy in a sepsis model of infection in female ICR mice using A. baumannii 17961. Mice were infected with A. baumannii via intraperitoneal (ip) administration, and treatments with conjugate 79 (5, 10, 25 mg/kg) were provided intravenously (iv) 30 min and 24.5 h after infection. In this study, all six mice that did not receive treatment with 79 died after 1 day; however, mice treated with 79 at 10 and 25 mg/kg resulted in survival in 4 of 5 mice (at each test concentration) while only 1 of 5 mice survived with 79 at 5 mg/kg, demonstrating a dose-dependent in vivo efficacy against A. baumannii. This is an outstanding demonstration of the clinical applications that the Trojan horse strategy can be utilized to target Gram-negative pathogens in human patients.

Cefiderocol (S-649266; structure not shown) is a catechol-bearing siderophore–cephalosporin conjugate that demonstrates potent antibacterial activities against a broad range of Gram-negative pathogens (e.g., carbepenem-resistant Enterobacteriaceae, P. aeruginosa, A. baumannii).¹²⁰ Cefiderocol has been advanced to clinical studies in human patients, demonstrating the translational potential of the Trojan horse strategy. In a clinical trial for the treatment of complicated urinary tract infection with antibiotic-resistant Gram-negative pathogens, cefiderocol was shown to be noninferior when compared to imipenem–cilastatin treated patients.¹²¹ Currently, cefiderocol is being investigated in clinical trials for hospital-acquired pneumonia and carbapenem-resistant infections. Collectively, this body of work aimed at targeting iron uptake mechanisms in bacteria to deliver antibiotics in a Trojan horse fashion is a promising approach to overcome antibiotic-resistant infections.

ClpP PROTEASE-ACTIVATING AGENT ADEP4 FOR PERSISTER CELL AND BIOFILM ERADICATION

Isolated from Strepomyces hawaiiensis, the acyldepsipeptides (ADEPs) are a class of natural product antibiotics that have been shown to exhibit potent activities against Gram-positive bacteria including methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant Enterococci.^122,123 The antibacterial activities for the natural product ADEP1 (80, Figure 15) and related synthetic analogues were evaluated by Brötz-Oesterhelt and co-workers, wherein ADEP1 reported a 8.76 μM IC₅₀ value against MRSA. In addition, synthetic analogues ADEP2 (81) and ADEP4 (82) also reported therein demonstrated markedly improved activity (500 nM and 64.9 nM, respectively).¹²⁴ Furthermore, ADEP1 and ADEP4 were efficacious in vivo against both S. aureus and E. faecalis infection models. ADEP4 demonstrated the ability to rescue 80% of mice from mortal S. aureus infections following a single dose of 12.5 mg/kg. The group used genomic analyses to identify the target for ADEP analogues as ClpP (caseinolytic protease), a highly conserved, cylindrical bacterial peptidase.¹²⁴

Figure 15.

Structures of the natural product, ADEP1 (80), its synthetic counterparts, and the key intramolecular hydrogen bonding required for activity. Structural modifications have been highlighted with different colors on each synthetic analogue.

The ClpP serine peptidase employs an energy-dependent process implicated in biological functions such as protein quality control, degradation of transient regulatory proteins, and clearance of cellular debris following conditions of bacterial stress.^125,126 Under normal homeostatic conditions, ClpP binds to ATP-dependent cochaperones (e.g., ATPases) which help regulate the unfolding and translocation of proteins into the proteolytic sites of ClpP. Crystal structures of ADEP1 (80) bound to Escherichia coli ClpP showed that ADEP analogues occupy the site of ATPase binding.¹²⁷ The ClpP–ADEP interaction promotes the conversion of the ClpP entrance pore from a closed-to an open-gate form (Figure 16). This allosteric control of the ClpP barrel conformation bypasses the need for an associated cochaperone (an energy dependent process) and permits the unregulated entrance of proteins into the proteolytic active site.¹²⁸ The resultant increase of ClpP activity diverts proteolysis from native physiological proteins to nascent peptides, leading to inhibition of bacterial cell division.

Figure 16.

(A) ADEPs bind to Clp-ATPase binding sites, promoting the entry of unfolded proteins into proteolytic sites, causing indiscriminate degradation of nascent polypeptides, ultimately leading to self-digestion and bacterial cell death. (B) The binding of ADEP analogues engenders the open conformation of the ClpP protease.

In a departure from prior studies (evaluating ADEP-promoted protein degradation against rapidly dividing cells), Lewis and co-workers utilized proteomic analysis to demonstrate that stationary phase MRSA exposed to ADEP4 (82) resulted in a reduced abundance of 24% of cellular proteins compared to the untreated control.¹²⁹ Since stationary phase populations of S. aureus are typically nondividing (and thus notoriously difficult to treat), this finding unveiled a potential therapeutic intervention against dormant persister cells. To evaluate the effect of ADEP4 against dormant bacteria, a population of persister cells was treated with the agent and it was found that ADEP4 was capable of eradicating persister cells to the limit of detection. Conversely, rifampicin reported no effect on persister cell viability.

ADEP4 also reported remarkable killing of stationary phase S. aureus following 2 days of treatment, demonstrating a 4 log reduction of viable bacterial cell counts.¹²⁹ Since ClpP is a nonessential protein for S. aureus, the null clpP (the gene encoding the ClpP protease) mutation frequency is high (∼10⁻⁶), and thus, rebounding populations were observed after day 3 of the experiments. It was surmised that formation of resistant mutants could be suppressed via coadministration of ADEP4 with conventional antibiotics. This co-treatment strategy eradicated persister cells to the limit of detection with no concomitant population rebound. It was found that although treatment of a clpP mutant strain with rifampicin demonstrated similar MIC values as that seen against of the wild-type strain, the mutant strains were less capable of producing persister cells (by an order of 10-to 100-fold). The conclusion was that clpP mutations (e.g., those resulting from ADEP treatment) may render the bacteria less fit and thus more susceptible to other antibiotics.

By use of 96-well plate, non-Calgary Biofilm Device assays, the ability of the ADEP4–rifampicin combination to eradicate bacterial biofilms was also evaluated.¹²⁹ Following a 24-h S. aureus UAMS-1 biofilm establishment, wells were treated with the ADEP4–rifampicin combination (6.49 μM ADEP4, 10 × MIC; 0.49 μM rifampicin, 10 × MIC), resulting in eradicated biofilms (>4 log-fold reduction) following 3 days of treatment. The ADEP4–rifampicin combination was also shown to be highly efficacious in deep-seated in vivo infection models. Following a 24 h infection period, histopathology was used to confirm the adherence of bacterial biofilms to mouse thigh muscle tissue. Treatment of the infection with the ADEP4–rifampicin combination resulted in an eradication of the biofilm-associated infection within 24 h (representing a >4 log-fold reduction). In contrast, rifampicin, vancomycin, or a combination of both could not eradicate the infection although decreased viable cell counts were observed.

The LaFleur group was also able to show that the ADEP–antibiotic combination strategy could be expanded to include additional clinically used antibiotics.¹³⁰ LeFleur and co-workers reported ADEP4 activity against stationary phase multidrug-resistant E. faecalis V583 when treated in combination with ampicillin, ciprofloxacin, daptomycin, oritavancin, or tigecycline. In these cases, the ADEP4–antibiotic combination yielded a 5 log reduction in bacterial cell survival. This was especially interesting as the activity was observed regardless of the mechanistic class of partnered antibiotic. Expression and purification of E. faecium ClpP permitted the use of a fluorometric monitoring assay used to evaluate casein degradation wherein it was reported that ADEP4 activated ClpP with an EC₅₀ of 0.53 μM. A crystal structure was also disclosed, highlighting the binding of ADEP4 to the hydrophobic pockets located between subunits at the apical and distal ends of ClpP. In a murine model of peritoneal septicemia, treatment with ADEP4 (50 mg/kg) and ampicillin (50 mg/kg) yielded a 2 log₁₀ greater reduction in average bacterial burden when compared to antibiotic monotherapy and a 4 log₁₀ reduction of viable bacterial cells relative to the vehicle control.

Encouraged by the in vivo antibacterial efficacy of ADEP analogues, several groups have put forth medicinal chemistry efforts to improve on this therapeutic strategy. In an attempt to reduce entropic binding penalties, Sello and co-workers disclosed an enhancement of ADEP analogues via introduction of conformational restriction.¹³¹ This strategy was realized via replacement of key macrocyclic residues. The result of the rigidified ADEP analogues was a 20-fold improvement of in vitro S. aureus activity from MIC = 649 nM ¹²⁹ for ADEP4 compared to 30.6 nM for ADEP1g (83). Building upon the library of known antibacterial ADEP analogues, the Duerfeldt group described an attempt to remedy a potential metabolic liability of the ADEP series (i.e., the readily hydrolyzed ester within the lactone core).¹³² On the basis of the reported crystal structures of ADEPs bound to ClpP, it seemed evident that replacement of the ester with an amide may have no influence on the interaction of the ADEPs with ClpP. Alas, Duerfeldt and co-workers were surprised to learn that amide-containing ADEP analogues 85 and 86 were markedly less active (by up to 100-fold) than the corresponding ester ADEP analogue 84. It was surmised that although the ester oxygen may not participate in molecular interactions with the target, its replacement with an amide nitrogen disrupts intramolecular hydrogen bonds (87, Figure 15) and thus compromises the compact conformation of the active agent (which presumably assists in membrane permeability). Although no biofilm eradication or in vivo assays were conducted as part of these recent studies, the reported novel agents demonstrate that the ADEP-based therapeutic approaches are fruitful pursuits and initial efforts to advance these agents toward clinical use have been very promising.

HALOGENATED PHENAZINE BIOFILM-ERADICATING AGENTS

Cystic fibrosis (CF) patients are plagued with chronic lung infections, which became a source of chemical inspiration for our lab to identify new biofilm-eradicating agents. Many young CF patients are known to experience initial S. aureus lung infections.¹³³ As these patients age, P. aeruginosa subsequently co-infects the lung and is believed to eradicate S. aureus using a series of phenazine antibiotics, including pyocyanin.^133–137 We believe that the initial S. aureus infections establish surface-attached biofilms on the inside of lung tissues as CF patients have lung infections that span multiple years.

In our initial work, we synthesized a diverse series of 13 phenazine compounds, including five naturally occurring phenazines.¹³⁸ With literature reports of pyocyanin as the main antibiotic of this class responsible for activities against S. aureus, we were sure to include this compound in our initial screening collection. Interestingly, in our initial antibacterial screens against S. aureus and S. epidermidis, we found that 2-bromo-1-hydroxyphenazine, a marine Streptomyces derived phenazine, demonstrated good antibacterial activities with MIC values of 6.25 μM against these pathogens (pyocyanin reported MICs of 50 μM).¹³⁸ In addition, we found that related synthetic analogue 2,4-dibromo-1-hydroxyphenazine (not shown) demonstrated improved antibacterial activities against S. aureus and S. epidermidis with MICs of 1.56 μM. In later work, we found that 2,4-dibromo-1-hydroxyphenazine was able to eradicate methicillin-resistant Staphylococcus aureus biofilms and reported a minimum biofilm eradication concentration, or MBEC, of ∼100 μM (note: when tested alongside 2,4-dibromo-1-hydroxyphenazine, vancomycin reported an MBEC of >2000 μM in Calgary Biofilm Device assays despite an MBEC of 4 μM against planktonic cells in the same experiment, demonstrating a significant level of biofilm tolerance toward vancomycin).¹³⁹

After demonstrating that 2,4-dibromo-1-hydroxyphenazine eradicates MRSA biofilms, we worked to develop modular, convergent synthetic routes to new “halogenated phenazine” (HP; Figure 17A, see structure 95) small molecules.^140–143 Chemical synthesis pathways involving phenylenediamine (88) condensation with quinone 89,^140,141 a Wohl–Aue reaction¹⁴² (aniline 90 condensed with 2-nitroanisole 91), and a Buchwald–Hartwig cross-coupling/reductive cyclization path-way¹⁴³ (fusing aniline 92 with 2-bromo-3-nitroanisole 93) enabled rapid access to a diversity of 1-methoxyphenazines (94) with various substitutions at the 6–9 positions of the HP scaffold (95). A final boron tribromide (BBr₃) demethylation and N-bromosuccinimide (NBS) bromination sequence afforded target HPs 95. These synthetic efforts and subsequent antibacterial studies have led to the identification of several highly potent biofilm-eradicating agents, including HPs 96–98, which demonstrate up to 50-fold more potent biofilm killing activities in Calgary Biofilm Device assays compared to parent 2,4-dibromo-1-hydroxyphenazine (e.g., 98 reported an MBEC of 2.35 μM against MRSA biofilms¹⁴³).

Figure 17.

Chemical synthesis of halogenated phenazines and preliminary SAR findings.

These efforts have enabled a detailed structure–activity relationship understanding regarding halogenated phenazine antibacterial agents (see 99, Figure 17B).^138–143 In general, substitution at the 6- and 7-positions of the HP scaffold is typically well-tolerated and enhances antibacterial activities compared to the corresponding unsubstituted HP; however, the 8-position of the HP scaffold can influence antibacterial potency in a positive (e.g., halogens improve activity) or negative (e.g., methyl reduces activity) fashion. Interestingly, we found that substitution at the 9-position of the HP scaffold completely abolishes all antibacterial activities. We later discovered the reason for a loss in activity regarding 9- substituted HPs after learning that active HP compounds bind metal(II) cations between the hydroxyl oxygen and adjacent nitrogen (forming a five-membered chelate upon metal binding), which plays a critical role in the antibacterial mode of action. Substituents at the 9-position of the HP scaffold, including a relatively small methyl group, impede the critical metal-chelation event for antibacterial activities.¹⁴⁰ Interestingly, general metal-chelating agents (EDTA and TPEN) have been tested alongside HPs and are unable to eradicate bacterial biofilms, highlighting the unique metal-dependent mechanism behind HP small molecules.^141–143

In ongoing efforts to translate HP biofilm-eradicating agents for clinical applications, we are using the SAR knowledge of the HP scaffold to pursue multiple prodrug strategies to functionalize the phenolic hydroxyl group to abolish metal-chelation until selective release within bacterial cells.^142,143 This approach provides a platform for extensive developments, including (1) tuning of physicochemical properties (e.g., installation of PEG group to increase water solubility^142,143) and (2) diverse functional triggers for HP release (e.g., bioreduction of the quinone in 101 is required to liberate the corresponding HP agent¹⁴³). Despite initial success with HPs 100 and 101, continued prodrug efforts are ongoing with the goal of developing an effective clinical agent for biofilm infections.

In separate work, we used HP-14 (96, MBEC = 6.25 μM) as a probe molecule for transcriptomic analysis of treated and untreated MRSA biofilms using RNA-seq technology.¹⁴⁴ This platform allowed us to define new cellular targets and pathways critical to biofilm survival, which is challenging to study as these surface-attached bacterial communities are composed of slow-growing or nonreplicating cells. Treating an established MRSA biofilm with HP-14 at low concentration (0.625 μM; 1/ 10 MBEC) for 20 h enabled the identification of >200 gene transcripts that were either up- or down-regulated (≥2.0-fold change in gene expression) compared to vehicle control.

Using a WoPPER gene cluster analysis tool, we identified 37 gene clusters with alterations in MRSA biofilm gene expression profiles following 20 h treatment with HP-14.¹⁴⁴ Upon exposure of HP-14 to MRSA biofilms, six gene clusters involved in iron acquisition were found to be dramatically up-regulated (“activated”) via WoPPER analysis, including: hts/sfa (staphyloferrin A; siderophore), sir/sbn (staphyloferrin B), isd (iron-regulated surface determinant; heme iron acquisition), MW0695 (hypothetical protein, similar to ferrichrome ABC transporters), and f hu (ferric hydroxamate uptake; two gene clusters). Following validation of these results using RT-qPCR experiments, a time-course assessment of MRSA biofilms treated with HP-14 (0.625 μM; 1/10 MBEC) revealed that four iron acquisition gene clusters (isd, sbn, sfa, MW0695) were activated in 1 h. The rapid activation of multiple iron uptake gene clusters by low concentrations of HP-14 is profound, as bacterial biofilms are notorious for their dormant phenotypes. Following 4 and 8 h of HP-14 exposure at 0.625 μM (1/10 MBEC), MRSA biofilms demonstrated more significant activation levels of these iron uptake genes (e.g., isdB, 22.3-fold activation after 4 h). Interestingly, EDTA and TPEN (5 μM; general metal-chelating agents) were unable to activate iron acquisition genes in MRSA biofilms,¹⁴⁴ which aligns with our previous observations that these metal-chelating agents are unable to eradicate MRSA biofilms (MBEC of >2000 μM) in Calgary Biofilm Device assays.^141–143

On the basis of RNA-seq findings, we believe HP-14 induces rapid iron starvation in MRSA biofilms, which leads to eradication of these surface-attached bacteria.¹⁴⁴ With the lipophilic properties of HP-14 (cLogP = 6.25) combined with potent MRSA biofilm eradication activity (MBEC = 6.25 μM), an iron chelating moiety, and rapid activation of multiple iron uptake gene clusters, we believe that this HP rapidly diffuses into biofilm cells and binds iron(II) following the intracellular release of iron(II) from a siderophore or heme. A proposed scheme of HP-14 inducing iron starvation and MRSA biofilm eradication is presented in Figure 18 and aligns with our WoPPER gene cluster analysis and subsequent RT-qPCR results in time-course studies. In addition, we believe that EDTA and TPEN are not able to efficiently penetrate biofilm cells, which is why they are unable to activate iron uptake genes or eradicate biofilms when tested alongside active HP biofilm-eradicating agents. This work not only demonstrates HP-14’s ability to rapidly induce iron starvation in MRSA biofilms but highlights an impressive level of sensitivity along with rapid response rates of bacterial biofilms to small molecule threats. In addition, these findings suggest that iron starvation of surface-attached bacterial communities could serve as the Achilles heel to persistent and recurring biofilm infections in the clinic.

Figure 18.

Proposed mechanism of MRSA biofilm eradication by HP-14, based on RNA-seq findings.

CONCLUSIONS

Despite increasing concerns regarding antibiotic resistance, antibiotic drug discovery is entering a dynamic and exciting new era. The multidisciplinary approaches of the research programs presented here have, in part, contributed to the reinvigoration of natural-product-based research for antibiotic discovery. Other impactful programs inspired by natural products, yet not included in this Perspectives article, include Genentech’s optimized arylomycin analogues for Gram-negative pathogens¹⁴⁵ and Boger et al.’s vancomycin work.^146–150 The classic tools of synthetic chemistry, microbiology, and chemical biology have been applied in novel ways to answer questions about bacterial pathogenesis and to design innovative therapeutic approaches for combatting infections. Significant challenges lie ahead, but these research efforts along with others have helped pave the way for critical and desperately needed discoveries to address problems presented by antibiotic resistance and tolerance.

ABBREVIATIONS USED

Accum

accumulation

AcOH

acetic acid

ADEP

acyldepsipeptide (antibacterial agent)

AfBPP

affinity-based protein profiling

Asp

aspartic acid

aq

aqueous

BBr₃

boron tribromide

BINAP

2,2′-bis(diphenylphosphino)-1,1′-binaphthyl

Bn

benzyl

Boc

tert-butyloxycarbonyl

C

carbon

ClpP

caseinolytic protease

CF

cystic fibrosis

CFU

colony forming unit (viable bacterial count)

DHF

dihydrofolate

DIPEA

N,N-diisopropylethylamine

DMAP

4-dimethylaminopyridine

DMF

dimethylformamide

DMP

Dess-Martin periodinane

DMPU

N,N′-dimethylpropyleneurea

DNA

deoxyribonucleic acid

DNM

deoxynybomycin

EDC

1-ethyl-3-(3-dimethylaminopropyl)carbodiimide

EDTA

ethylenediaminetetraacetic acid

ETC

electron transport chain

EtOAc

ethyl acetate

FDA

Food and Drug Administration

Fe

iron

HATU

hexafluorophosphate azabenzotriazole tetramethyluranium

Glob

globularity

HF

hydrogen fluoride

HOAt

1-hydroxy-7-azabenzotriazole

HP

halogenated phenazine

Isd

heme-iron complex

KOH

potassium hydroxide

LCMS/MS

liquid chromatography-mass spectrometry

LDA

lithium diisopropylamide

LiOH

lithium hydroxide

Lys

lysine

MBEC

minimum biofilm eradication concentration

mg/kg

milligrams per kilogram

MIC

minimum inhibitory concentration (lowest concentration that demonstrates complete bacterial growth inhibition)

MRSA

methicillin-resistant Staphylococcus aureus

NaBH₄

sodium borohydride

NaHMDS

sodium bis(trimethylsilyl)amide

NaOEt

sodium ethoxide

NBS

N-bromosuccinimide

nmol

nanomole

NMR

nuclear magnetic resonance

NRPS

non-ribosomal peptide synthetase

PABA

p-aminobenzoate

Pd

palladium

Phe

phenylalanine

PhMe

toluene

rRNA

ribosomal ribonucleic acid

RB

rotatable bond (rigidity)

Ru

ruthenium

SAR

structure-activity relationship

sbn

staphyloferrin B

SdhC

succinate dehydrogenase C subunit

SEM

2-(trimethylsilyl)ethoxymethyl

Ser

serine

sfa

staphyloferrin A

TBAF

tetrabutylammonium fluoride

TBDPS

tert-butyldiphenylsilyl

TBS

tert-butyldimethylsilyl

TBSCl

tert-butyldimethylsilyl chloride

TCA

tricarboxylic acid

Tf

triflate

TFA

trifluoroacetic acid

THF

tetrahydrofolate

THF

tetrahydrofuran

Thr

threonine

TMEDA

tetramethylethylenediamine

TMS

trimethylsilyl

TPEN

N,N,N′,N′-tetrakis(2-pyridinylmethyl)-1,2-ethanediamine

Troc

trichloroethyl chloroformate

μg/mL

micrograms per milliliter

μM

micromolar

VRE

vancomycin-resistant enterococci

WGS

whole genome sequencing

Zn

zinc

Biographies

Yasmeen Abouelhassan received a Ph.D. in Medicinal Chemistry with Prof. Robert W. Huigens at the University of Florida in 2018. She received her B.S. in Pharmaceutical Sciences from Cairo University, Egypt, in 2012. Her Ph.D. research focused on the identification of novel biofilm-eradicating agents against Gram-positive bacteria. This research led to the discovery of phenazine antibiotic derivatives as potent methicillin-resistant S. aureus biofilm eradicators operating through a novel iron starvation mechanism.

Aaron T. Garrison received his Ph.D. in Medicinal Chemistry under Professor Robert W. Huigens III at the University of Florida in 2017 where he synthesized potent halogenated phenazine and quinoline small molecules, including diverse prodrug analogues of lead compounds. Thereafter, he accepted a postdoctoral position with Professor Craig W. Lindsley at the Vanderbilt Center for Neuro-science Drug Discovery. His research is presently focused on the design and synthesis of negative allosteric modulators of the muscarinic acetylcholine receptor subtype 5 (M₅) for the treatment of opioid and ethanol dependence. His additional research projects include the development of synthetic methodology to facilitate the discovery of novel therapeutics.

Hongfen Yang received a B.S. in Chemistry from Hubei University in Wuhan, China. She then moved to the University of Chinese Academy of Sciences in Beijing under the guidance of Professor Wu, where she obtained a M.S. in Bioinorganic Chemistry. Currently, Hongfen is a fourth year graduate student in the Medicinal Chemistry Department at the University of Florida under the supervision of Professor Huigens. Her doctoral research involves the design, chemical synthesis, and biological evaluation of new halogenated phenazine biofilm-eradicating agents.

Alejandra Chávez-Riveros obtained her M.S. and Ph.D. in Chemical Sciences from the National Autonomous University of Mexico (Universidad Nacional Autónoma de México, UNAM). Her doctoral work focused on the design and synthesis of steroidal 5α-reductase inhibitors under the mentorship of Professor Eugene Bratoeff. She went to Evestra Inc. for a predoctoral internship, where she synthesized antiprogestins. Upon graduation with a Ph.D. in 2016, she joined Professor Luis D. Miranda’s group at the Institute of Chemistry UNAM for postdoctoral studies and developed the synthesis of novel macrocycles, which demonstrate anti-inflammatory activities. She is currently a Postdoctoral Research Associate in Professor Huigens’ lab at the University of Florida in the Department of Medicinal Chemistry, where she is developing new synthetic methodologies for ring distortion of indole alkaloids.

Gena M. Burch is the Kenneth F. Finger Memorial Fellow in the Infectious Disease and Pharmacokinetics Laboratory at the University of Florida. She received a PharmD in 2016, followed by a M.S. in Medicinal Chemistry in 2018 from the University of Florida under the supervision of Dr. Robert W. Huigens III working on the discovery and development of novel antibacterial and biofilm-eradicating agents. Her current research includes liquid chromatography–tandem mass spectrometry method development for the determination of serum concentrations for antimicrobial drugs. Her clinical research efforts are focused on therapeutic drug monitoring and clinical pharmacokinetics for patients with bacterial and mycobacterial infections.

Robert W. Huigens III received a Ph.D. in Chemistry with Professor Christian Melander at NC State University in 2009. He then moved to the University of Illinois of Urbana–Champaign as an American Cancer Society Postdoctoral Fellow under the guidance of Professor Paul Hergenrother. In 2013, Huigens began his independent career as an Assistant Professor of Medicinal Chemistry at the University of Florida. The Huigens lab has interdisciplinary drug discovery programs inspired by natural products that aim to (1) discover novel antibacterial agents that eradicate bacterial biofilms and (2) rapidly generate architecturally complex and diverse compound libraries from readily available indole alkaloids.