Biguanide-Vancomycin Conjugates are Effective Broad-Spectrum Antibiotics against Actively Growing and Biofilm-Associated Gram-Positive and Gram-Negative ESKAPE Pathogens and Mycobacteria

Abstract

Strategies to increase the efficacy and/or expand the spectrum of activity of existing antibiotics provide a potentially fast path to clinically address the growing crisis of antibiotic-resistant infections. Here, we report the synthesis, antibacterial efficacy and mechanistic activity of an unprecedented class of biguanide-antibiotic conjugates. Our lead biguanide-vancomycin conjugate, V-C6-Bg-PhCl (5e), induces highly effective cell killing with up to a 2 orders-of-magnitude improvement over its parent compound, vancomycin (V), against vancomycin-resistant enterococcus. V-C6-Bg-PhCl (5e) also exhibits improved activity against mycobacteria and each of the ESKAPE pathogens, including the Gram-negative organisms. Furthermore, we uncover broad-spectrum killing activity against biofilm-associated Gram-positive and Gram-negative bacteria as well as mycobacteria not observed for clinically used antibiotics such as oritavancin. Mode-of-action studies reveal vancomycin-like cell wall synthesis inhibition with improved efficacy attributed to enhanced engagement at vancomycin binding sites through biguanide association with relevant cell-surface anions for Gram-positive and Gram-negative bacteria. Due to its potency, remarkably broad activity and lack of acute mammalian cell toxicity, V-C6-Bg-PhCl (5e) is a promising candidate for treating antibiotic resistant infections and notoriously difficult-to-treat slowly growing and antibiotic tolerant bacteria associated with chronic and often incurable infections. More generally, this study offers a new strategy (biguanidinylation) to enhance antibiotic activity and facilitate clinical entry.

Graphical Abstract

INTRODUCTION

In 2019, 4.95 million deaths were attributed to antibiotic resistant infections.¹ Without interventions, this global figure is predicted to rise to 10 million by 2050.^2,3 Many of the highest-threat bacterial pathogens are covered under the ESKAPE acronym, which stands for E. faecium, S. aureus, K. pneumoniae, A. baumannii, P. aeruginosa, and Enterobacter species, but this acronym overlooks mycobacteria like Mycobacteria tuberculosis which alone was responsible for 1.3 million deaths globally in 2022.⁴ The CDC places the Enterobacteriaceae family and A. baumannii in the highest threat level category of “urgent” threats to human health, with the rest of the ESKAPE pathogens and M. tuberculosis being defined as “serious” threats.⁵ Cases of non-tuberculosis mycobacteria are also rising at an alarming rate in the U.S. and are associated with lung disease in cystic fibrosis patients and individuals with chronic obstructive pulmonary disease.⁶ Unfortunately, despite the critical need for more drugs against these high-priority pathogens, only 38% of antibiotics currently in development are expected to be active against ESKAPE pathogens.⁷ Modifying existing antibiotics to improve and expand efficacy is one attractive strategy to expedite introduction of drugs to treat high-priority pathogens and minimize the scientific, regulatory, and safety barriers associated with the development of de novo therapeutics. The cell wall inhibitor vancomycin is a promising candidate for this strategy, and vancomycin-conjugates have gained attention for rapid approval for clinical treatment of challenging ESKAPE pathogens.⁸

Vancomycin is a potent glycopeptide antibiotic administered intravenously as a first-line therapy to treat Methicillin-resistant Staphylococcus aureus (MRSA) skin and soft tissue infections, MRSA bacteremia, and C. difficile infections.^9–13 It inhibits cell wall synthesis in bacteria by binding to the D-Ala-D-Ala termini of Lipid II at the cell surface, effectively sequestering key cell wall precursors and inhibiting peptidoglycan assembly.^14–16 However, vancomycin is ineffective against biofilm-associated S. aureus.^17–19 The emergence of resistance in vancomycin intermediate S. aureus (VISA), vancomycin resistant S. aureus (VRSA) and vancomycin resistant enterococci (VRE) also emphasizes the need for new treatment options.²⁰ Vancomycin also inhibits peptidoglycan synthesis in mycobacteria but its higher MIC against mycobacteria is slightly beyond what is considered therapeutically relevant. This comparably lower efficacy than observed in Gram-positive bacteria is generally attributed to restricted penetration of vancomycin through its more complex cell wall which contains arabinogalactan and mycolic acids beyond the peptidoglycan layer. This hypothesis is supported through vancomycin’s improved efficacy against mycobacteria when used in combination with other cell-wall targeting agents.²¹ Due to its high MIC’s in Gram-negative bacteria, vancomycin is simply considered ineffective against these strains. This resistance to vancomycin is attributed to the Gram-negative outer membrane which prevents antibiotic accumulation in the periplasm, therefore the antibiotic cannot inhibit peptidoglycan synthesis.^22–24

Promising vancomycin derivatives that address some of these shortcomings have been introduced, with oritavancin being a prominent example of an FDA-approved semisynthetic derivative with demonstrated in vitro antibacterial activity against biofilm-associated S. aureus.²⁵ Analogues of vancomycin commonly incorporate cationic and lipophilic functionalities to enhance affinity for D-Ala-D-Ala, membrane localization, membrane depolarization, and/or membrane permeabilization, resulting in antibiotics with multiple modes of action.^{19,22,26–29} The cationic functional group is frequently a quaternary ammonium like those found in the Maxamycins of the Boger group²⁷, or peptide-derived ammonium residues like those reported by the Cooper group.³⁰ These studies also incorporated lipophilic moieties like chlorobiphenyl-, n-alkyl-, and 4-phenylbenzoyl- groups to enhance cell surface and membrane interactions. Similarly, our groups have reported that vancomycin modified with one or more arginines or guanidinium groups exhibit profoundly improved efficacy and an expanded spectrum of activity.^19,22,31 Subsequently, others have explored persistently charged guanidinium modifications with notable advances in several cases.^26,27,32,33

The genesis of our interest in guanidinylated antibiotics arose during seminal studies reported in 2000, in which we showed that oligomers of arginine (e.g. D- and L-octa- and nona- arginines), and more generally, polyguanidinylated oligomers (e.g. N-Arg-peptoids)^34,35 readily cross cell membranes and other biological barriers including human skin. These and related guanidinylated transporter systems, some of which are referred to as cell penetrating peptides, have subsequently been used to readily deliver a wide range of cargoes including therapeutic agents, peptides, protein optical probes, metals, RNA, and DNA through cell membranes.^34–37 We proposed that the positively-charged guanidinium groups associate with negatively-charged cell surface phosphates, carboxylates and sulfates through electrostatic and chelative hydrogen bonding interactions^38–40 and that this association converts an otherwise polar polycation into a less polar complex that is driven inward by the cell membrane potential or through endocytotic pathways.^34,35,41 In previous work, we reported that a cell-penetrating guanidinium-rich molecular transporter (octaarginine) conjugate of vancomycin, V-r8, kills vancomycin resistant organisms and eradicates S. aureus biofilms and persister cells.¹⁹ We subsequently showed that a single amino acid addition to vancomycin affording the vancomycin-arginine conjugate V-R, effectively kills Gram-negative pathogens both in vitro and in vivo.^22,42,43 We then uncovered efficacy enhancements of V-R over vancomycin against multiple species of mycobacteria through an improved peptidoglycan binding mechanism via favorable positioning and interaction of the guanidinium group with meso-diaminopimelic acid (mDAP) amino acid residue.⁴⁴

Inspired by these foundational insights into guanidinium-based transporters, here we report an unexplored cationic enhancer of antibiotic activity, the biguanidinium functional group. The biguanidinium and guanidinium functional groups have some properties in common such as similar pKa values (12–14) under physiological conditions and the ability to engage in electrostatic cation-anion and hydrogen bonding interactions with anionic and neutral hydrogen bond accepting substrates, including cell surface anions (phosphates, sulfates, and carboxylates, Figure 1A).^{39,40,45–52} The resulting association of these anions with biguanidinium cations can be stronger than those of guanidinium cations. This was quantified in a synthetic ADP/ATP carrier protein mimic where exchange of a guanidinium for a biguanidinium moiety resulted in a 100-fold increase of the mimic’s affinity for AMP, ADP, and ATP.⁴⁷ Additionally, biguanide-functionalized small molecules have been shown to enter cells through cationic transporter assistance.⁵³

Figure 1.

A) Biguanides can bind anionic functional groups through Coulombic and chelative hydrogen bonding interactions B) Biguanide-vancomycin conjugates are proposed to retain their D-ala-D-ala binding while also engaging surface anions through Coulombic and hydrogen-bonding interactions, synergistic interactions that increase efficacy. C) Previously explored ammonium and guanidinium cations and the biguanidinium cations in this study differ in their pKa values, number and spatial array of hydrogen bond donors, and charge distributions. Note: For clarity, charge delocalization is not shown.

Mono- and poly-biguanidinium agents have been utilized alone as antibiotics and membrane permeabilizing reagents^45,54 and bis-biguanides when co-administered with vancomycin have been shown to sensitize pathogens to antibiotic accumulation through membrane disruption.⁵⁵ Notwithstanding these contributions, covalent chimeras of biguanide-antibiotic dual-function conjugates have not been explored. Here we report the previously unexplored antimicrobial activity of biguanide derivatized vancomycin analogues as representative examples of a design strategy that could be applied more generally to other biguanide-antibiotic chimeras. We describe the synthesis and activity of a novel series of biguanide-vancomycin conjugates with broad-spectrum activity against both actively growing and biofilm-associated ESKAPE pathogens and mycobacteria. Significantly, the lead biguanide-vancomycin conjugate, 5e, shows potent antimicrobial activity against S. aureus, E. faecium, E. coli, A. baumannii, K. pneumoniae, P. aeruginosa, M. smegmatis, and M. abscessus with up to two orders-of-magnitude improvement over the parent compound, vancomycin. Biguanide conjugate 5e additionally eradicates pre-formed biofilm-associated S. aureus, E. coli, and M. smegmatis. Antibacterial studies suggest a multifunctional mode-of-action involving the canonical binding of the vancomycin glycopeptide cleft to Lipid II and uncrosslinked D-Ala-D-Ala peptidoglycan stem termini, with increased cell-association, accumulation, and membrane influence enabled through the unique chemistry of biguanidinium-based conjugation to vancomycin.

RESULTS AND DISCUSSION

Design and Synthesis of Biguanide-Vancomycin Conjugates.

We devised two synthetic routes to boc-aminobiguanides 3a-e (Figure 2). Strategy 1 involves the direct Pinner-like addition of an amine to a cyanoguanidine (2 or 8a) at elevated temperatures in the presence (Figure 2A) or absence (Figure 2B) of solvent.⁵² The second, and preferred, route to boc-aminobiguanides 3a-e involves the biguanidinylation of an amine with an activated pyrazole-biguanide intermediate, 9a-e. The resulting boc-aminobiguanides 3a-e were then converted with acid to the aminobiguanides 4a-e. Subsequently aminobiguanides were conjugated to vancomycin to yield biguanide-antibiotic conjugates 5a-e (V-C6-Bg-R) or N-acetylated to yield control compounds 6a-e (Ace-C6-Bg-R), respectively. (Figure 3).

Figure 2.

Synthesis of Boc-aminobiguanides 3a-e. (A) synthesis of Boc-aminobiguanide 3a from commercially available cyanoguanidine and Boc-protected 1,6-diaminohexane 1. (B) synthesis of lipophilic Boc-aminobiguanides 3b-e from 1, NaN(CN)₂ and ammonium salts, through a fusion procedure. (C) alternative synthesis of Boc-aminobiguanides 3a-e using pyrazole-biguanides 9a-e as biguanidinylation reagents. (*) Compound 8a was acquired commercially. Note: compounds 3a-e do not have any counterions because they are purified in the presence of ammonium hydroxide (See SI). Note: For clarity, a single tautomer of compounds 9a-e is shown.

Figure 3.

Synthesis of biguanide-vancomycin conjugates, 5a-e, and control acetamide-biguanides 6a-e from Boc-aminobiguanides 3a-e. Note: For clarity, a single tautomer of compounds 4a-e, 5a-e, and 6a-e is shown.

A hexamethylene linker, like that found in the bis-biguanide antibiotics alexidine and chlorhexidine, was selected to conjugate vancomycin’s C-terminus to the biguanide moiety by a peptide bond (Figure 1). This strategy required a terminal amine which prompted our selection of the monoprotected diamine 1. Boc-aminobiguanides 3a-e were synthesized by two strategies differing principally by whether an activated pyrazole-biguanide 9a-e is used as an intermediate. Strategy 1 (Figure 2A, Figure 2B), while atom- and step-economical, suffered from poor selectivity and only serviceable yields. Purification of 3a-e from the reaction mixture required silica-gel chromatography to remove unidentified side-products. Strategy 2 (Figure 2C), based on the work of Bardovskyi, et. al,⁵⁶ involves a biguanide transfer reaction utilizing pyrazole-biguanides 9a-e to biguanidinylate amine 1 under mild-conditions which yields 3a-e upon precipitation from diethyl ether. Strategy 2 (Figure 2C) requires an additional synthetic operation to generate the activated intermediates 9a-e but removes a chromatographic purification and increases yields of 3a-e by 1.3- to 2.0- fold over strategy 1.

Cyanoguanidine 8a is commercially available while lipidated cyanoguanidines 8b-d were synthesized from the appropriate amine 7b-d using NaN(CN)₂ and trifluoroacetic acid (TFA) in DMF. The completed reaction was then diluted with 1M NaOH, the organic soluble compounds were extracted with ethyl acetate, and concentrated in vacuo. From the crude concentrate, N-alkylated cyanoguanidines were crystallized from diethyl ether (8b) or THF and hexanes (8c, 8d) and then collected by filtration in 71–91% yield. Cyanoguanidine 8e was synthesized in water from NaN(CN)₂, chloroaniline, and aqueous HCl following a reported procedure.⁵⁷ The product precipitated from the reaction mixture and 8e was collected by filtration in 92% yield.

Pyrazole-biguanide 9a was synthesized in water following the published procedure,⁵⁶ but further development was required to acquire functionalized pyrazole-biguanides 9b-e. N-alkylated pyrazole-biguanides 9b-d were synthesized from 8b-d with stoichiometric pyrazolium chloride in hot DMF. The hot reaction mixtures were added to stirring diethyl ether from which the products precipitated and were collected by filtration in 86–95% yields. The products were further purified by recrystallization from boiling acetonitrile. Acetonitrile proved to be a suitable solvent to synthesize the chlorophenyl analogue 9e in good yields because the addition reaction proceeded slowly or not at all when DMF, water, pyridine, or methanol were used as solvent. Upon heating the reactant solution to reflux at 0.05 M, 9e rapidly crystallized from solution and was collected by filtration in 63% yield (98% BRSM). The starting materials can be recovered from the filtrate and resubjected to yield additional product.

Mono-protected diamine 1 was dissolved in pyridine (1M) with the appropriate biguanidinylation reagent 9a-e and stirred at 60°C for 18 hours. We observed clean conversion to corresponding Boc-aminobiguanide 3a-e. Dropwise addition of the crude reaction into diethyl ether resulted in precipitation of the product which was purified by column chromatography to yield 3a-e in 69–83% isolated yield.

The Boc-aminobiguanides 3a-e were deprotected with TFA in DCM (1:1 v/v) to yield aminobiguanide•TFA salts 4a-e (Figure 3). Each aminobiguanide 4a-e was conjugated to vancomycin through a peptide bond to vancomycin’s C-terminus in the presence of HBTU and DIPEA in a DMSO/DMF cosolvent solution (1:1 v/v). The resulting biguanide-antibiotic conjugates 5a-e (V-C6-Bg-R) were purified by reverse-phase HPLC to yield the corresponding trifluoroacetate salts. The site of conjugation was confirmed with HSQC and HMBC 2D-NMR experiments using 5e as a representative example. Additionally, to investigate antimicrobial activities of the biguanide-moiety itself (without vancomycin), we acetylated the aminobiguanides 4a-e with acetic anhydride to yield the acetamide-biguanides 6a-e (Ace-Bg-R). In summary, biguanide-antibiotic conjugates 5a-e were synthesized in 5 steps in 21–40% overall yield and acetamide-biguanides 6a-e were synthesized in 5 steps in 38–63% yield. Full synthetic details, high-resolution mass analysis, ¹H-NMR, ¹³C-NMR, and 2D-NMR spectra are provided in supporting information.

Evaluation of Biguanide-Vancomycin Conjugates in Antimicrobial Susceptibility Assays with ESKAPE Pathogens and Mycobacteria.

The biguanide-vancomycin conjugates 5a-e and control comparator acetamide-biguanides 6a-e were first evaluated for antimicrobial activity against a panel containing Gram-positive, Gram-negative, and mycobacterial microorganisms (Table 1, Table S1 information). Significantly, the conjugation of a single unfunctionalized biguanide to vancomycin, 5a, resulted in a 2- to 32- fold increase in potency. More generally, all conjugates except 5d were significantly better than vancomycin (V), exhibiting excellent killing activity against multiple organisms that are resistant or insensitive to vancomycin. Vancomycin is efficacious and used clinically to eliminate actively growing S. aureus with an MIC of 1μM against S. aureus 29213. The biguanide-vancomycin conjugates 5a-c and 5e maintain this activity with MICs similar to or slightly lower than vancomycin. E. faecium 51559 is a vanA-type resistance reference strain that produces Lipid II terminating in D-Ala-D-Lac. This modification reduces the binding site affinity for vancomycin and results in greatly diminished vancomycin sensitivity (V MIC of 512 μM). However, conjugates 5a and 5b have 16- to 32- fold lower MICs against this vancomycin-resistant strain (16–32 μM) and more hydrophobic conjugates 5c, 5d, and chlorophenyl conjugate 5e demonstrate potent single-digit MICs from 2–4 μM. Gram-negative organisms, including ESKAPE pathogen strains E. coli, A. baumannii, K. pneumoniae, and P. aeruginosa are inherently insensitive to vancomycin. This resistance primarily stems from the presence of the outer membrane which is absent in Gram-positive organisms and prevents vancomycin from accumulating in the periplasmic space where peptidoglycan synthesis occurs in Gram-negative organisms. Thus, typical MICs for vancomycin against Gram-negative organisms are 128 μM or higher. Here we show that oritavancin is also completely ineffective at killing Gram-negative bacteria and is poorly active against mycobacteria. In striking contrast, the biguanide-vancomycin conjugates 5a-c and 5e all showed significantly lower MICs against Gram-negative organisms, including a striking improvement of 32-fold for 5e against E. coli. Finally, in two strains of mycobacteria, M. smegmatis 700084 and M. abscessus 19977, three biguanide-vancomycin conjugates 5a, 5b, and 5e exhibited more potent antimicrobial activity than vancomycin, including the non-lipidated conjugate 5a which demonstrated 16-fold improvement over the parent compound in M. smegmatis. In M. abscessus 5e was most effective, improving upon vancomycin by 4–8 fold. Biguanide modification of vancomycin, regardless of the lipid, did not enhance efficacy against M. avium. M. avium is notoriously more difficult to treat with some classes of antibiotics than M. abscessus. In our previous study, V-R also was not superior to vancomycin against M. avium.⁴⁴ This may be attributed to non-specific binding in the M. avium cell wall and/or reduced access of the conjugate to vancomycin binding sites at the cell surface.⁴⁴

Table 1.

Minimum inhibitory concentrations (MICs) of biguanide-vancomycin conjugates and comparative compounds. MICs are displayed for vancomycin V, oritivancin, metformin (a control biguanide), biguanide-vancomycin conjugates 5a-e, control biguanide 6e, and an equimolar mixture of V and 6e, against Gram-positive and Gram-negative ESKAPE pathogens and mycobacteria. MICs are bolded when they are single digit μM and more potent than the parent compound, vancomycin.

Minimum Inhibitory Concentration (MIC, μM)
Species	Strain	Vancomycin	Oritavancin	Metformin	V-C6-Bg-H	V-C6-Bg-C4	V-C6-Bg-C10	V-C6-Bg-C16	V-C6-Bg-PhCI	Ace-Bg-PhCI	1:1 Mixture
		V	−	−	5a	5b	5c	5d	5e	6e	V + 6e
Gram-positive
S. aureus	29213	1	1	>512	0.5	0.5	1–2	4	0.25–0.5	>256	1
E. faecium	51559	512	2	>512	32	16	2–4	4	4	>256	>128
Gram-negative
E. coli	25922	128–256	>512	>512	8	8–16	4–8	>128	4–8	>256	128
A. baumannii	19606	64–128	512	>512	16–32	16	4–8	>128	8	>256	128
K. pneumoniae	700603	256–512	>512	>512	32–64	64	32	>128	32	>256	>128
P. aeruginosa	PA14	256–512	>512	>512	64	64	64–128	>128	64–128	>256	>128
Mycobacteria
M. smegmatis	700084	8	32	>512	0.5	1	16	32	4–8	>256	8
M. abscessus	19977	32–64	128	>512	32	8–16	64	>128	8	>256	32–64
M. avium	700898	16	128	>512	16–32	16–32	64	128	16–32	256	8

Taken together, compound 5e represents the most effective and broad spectrum conjugate in this collection with MICs of 4–8 μM, 8 μM, 32 μM and 64–128 μM against E. coli, A. baumannii and M. abscessus, K. pneumoniae, and P. aeruginosa, respectively. Compound 5e is effective against S. aureus and VRE with an MIC of 4 μM against E. faecium 51559. Biguanide-vancomycin conjugate 5e exhibited significantly improved activity against clinically relevant M. abscessus with 4–8 fold greater potency than vancomycin in the MIC assay. In all, 5e is a potent antibiotic with truly broad spectrum activity and the potential to combat many classes of bacterial pathogens.

The excellent activity of the biguanide conjugates required covalent conjugation of vancomycin to the aminobiguanide. For comparison, acetamide-biguanides 6a, 6b, and 6e controls displayed no antimicrobial activity alone, with MICs ≥ 256 μM in all strains (Table 1, Table S1). Similarly, for these three compounds, the unconjugated combination of acetamide-biguanides and vancomycin (1:1 ratio) displayed activity indistinguishable from vancomycin alone (Table S1), indicating the importance of covalent conjugation for the improvements in MICs. For the two aminobiguanides with the longer C10 (6c) and C16 (6d) lipophilic chains, antimicrobial activity was observed from the acetamide-biguanides themselves, presumably attributed to their surfactant-like structure (Table S1).^58,59

In Vitro Evaluation of Mammalian Cell Toxicity.

We evaluated the potential of the biguanide-vancomycin conjugates to exhibit mammalian cell toxicity using an MTS assay with HeLa cells and a hemolysis assay with red blood cells to then focus biochemical and mechanistic activity assays on the most clinically promising compounds that lack acute toxicity (Figure S1). Conjugates 5a and 5e exhibited low levels of cytotoxicity and no hemolysis at concentrations up to 160 μM. Lipidated conjugates 5b and 5d exhibited moderate levels of cytotoxicity at the highest treatment concentrations, but 5c exhibited significant cytotoxicity at and above 40 μM. While conjugate 5b displayed no hemolysis, conjugates 5c and 5d induced unacceptable levels of hemolysis at concentrations as low as 10 μM. The effects of 5c support the notion that its antibacterial activity is likely attributable to general membrane permeabilizing activity associated with medium-length lipophilic molecules.^60,61 The combination of MIC, mammalian cell cytotoxicity, and hemolysis results identified V-C6-BgPhCl (5e) as having activities matching or superior to the other tested agents with no acute toxicity, thus supporting subsequent mode-of-action studies to characterize this lead derivative and examine its potential to eradicate slowly growing and biofilm-associated bacteria that are not eliminated by vancomycin.

Mode-of-Action Characterization of Lead Compound 5e.

The kinetics of vancomycin killing is considered slow as time is required for bacteria to accumulate defective cell walls while the activity of autolysins degrade outer cell wall layers, ultimately eliminating cell viability. We employed a conventional time-resolved viability assay based on the enumeration of viable cells in colony forming units (CFU) per milliliter (mL) following antibiotic treatment to assess whether killing kinetics of 5e were similar to that of vancomycin in each reference strain (Figure 4). In S. aureus, V-C6-Bg-PhCl (5e) displayed similar killing kinetics to vancomycin. In E. coli, where vancomycin is ineffective at clinically relevant concentrations, the MIC evaluation had revealed that 5e was approximately 16-fold more effective than V (Table 1), and here displayed killing kinetics at 16 μM that is comparable to the very high dose of V (256 μM) (Figure 4B). These results are consistent with the hypothesis that 5e demonstrates vancomycin-like inhibition of cell wall synthesis in E. coli, but at a much lower effective dose than vancomycin. In M. smegmatis, 5e evaluated at 16 μM displayed killing kinetics with slightly faster initial killing rate than V (16 μM), but comparable overall, requiring tens of hours to reduce CFU/mL by 4 orders of magnitude to the limit of detection. Each data set for 5e is consistent with a vancomycin-like mode of action whereby inhibition of cell wall synthesis results in a relatively slow killing profile.

Figure 4.

Time-kill kinetics evaluation of V-C6-Bg-PhCl 5e and vancomycin (V) in S. aureus 29213 (A), E. coli 25922 (B), and M. smegmatis 700084 (C). Compounds were added to achieve final 2x MIC concentrations to 1mL aliquots of cells with approximate density of 1 x 10⁶ CFU/mL with subsequent CFU/mL measurements made at indicated time points.

The binding of vancomycin to Lipid II prevents cell-wall synthesis and leads to an accumulation of soluble intracellular peptidoglycan precursors also termed Park’s nucleotide.⁶² To enable molecular detection of peptidoglycan precursors in three reference strains, we performed compound treatments with either 5e or vancomycin followed by cell lysis and LC-MS analysis. As expected, treatment with vancomycin resulted in accumulation of the UDP-MurNAc pentapeptide in S. aureus. Vancomycin also results in accumulation of UDP-MurNAc pentapeptide in E. coli (as expected from treatment with high vancomycin concentrations (256 μM)⁶³) and the related UDP-MurNGlyc pentapeptide in M. smegmatis (Figure 5).⁶⁴ In S. aureus we observed comparable accumulation of UDP-MurNAc pentapeptide following treatment with vancomycin and treatment with 5e (Figure 5A). In E. coli treated with conjugate 5e (16 μM), we observed significantly higher accumulation of the UDP-MurNAc pentapeptide than in E. coli treated with a 16-fold greater concentration of vancomycin (256 μM) (Figure 5B). Lastly, for M. smegmatis we observe slightly greater, but generally comparable accumulation of cell-wall precursor following treatment with 5e compared to cells treated with vancomycin (Figure 5C). The collective results in all three strains underscores that 5e inhibits cell wall synthesis through a vancomycin-like mechanism.

Figure 5.

Molecular detection of peptidoglycan precursor accumulation. HPLC chromatogram display of peptidoglycan precursor accumulation and molecular mass identification by mass spectrometry in (A) S. aureus, (B) E. coli, and (C) M. smegmatis for bacteria treated with 5e and vancomycin. Mass spectra of peptidoglycan precursor pentapeptides are shown for peaks with 4.7 min, 4.7 min, and 3.9 min retention times for S. aureus, E. coli and M. smegmatis, respectively.

We hypothesized the expanded and potent Gram-negative activity observed from our previously reported guanidine-modified vancomycin conjugates and the biguanide conjugates herein could be associated with their enhanced cell-association through engagement with relevant cell-surface anions, resulting in local or broad perturbation of membrane structure and integrity, membrane disruption, and/or enhanced cell permeability.¹⁹ We therefore examined whether enhanced uptake of the fluorescent probe propidium iodide (PI) would be detected in S. aureus cells treated with conjugate 5e as a proxy for enhanced membrane permeabilization. PI does not accumulate in healthy viable cells, but if membrane integrity is perturbed, PI can enter the cytoplasm and bind to DNA, resulting in increased fluorescence. Membrane integrity was evaluated with PI in exponential-phase cells suspended in HEPES-Glucose buffer to maintain metabolic activity.^65,66 Lysostaphin was included as a positive control which disrupts cell membrane integrity through peptidoglycan digestion. Consistent with the enhanced membrane permeabilization hypothesis, treatment of S. aureus with biguanide 5e was accompanied by increased PI fluorescence (Figure 6A), while treatment with vancomycin showed no increase in fluorescence. This observation is consistent with enhanced antimicrobial activity and membrane permeabilization exhibited by our previously reported vancomycin polyguanidino dendrimeric conjugates, wherein the guanidinium-based transporter alone does not have antibacterial activity, but when conjugated to vancomycin the conjugate can facilitate interactions with teichoic acid phosphates and Lipid II pyrophosphate at the cell surface.³¹

Figure 6.

Evaluation of bacterial membrane permeability following treatment with biguanide-vancomycin conjugate, 5e, using fluorescent probe molecules. (A) Propidium Iodide (PI) fluorescence monitored in S. aureus 29213. (B) 1-Nphenylnapthylamine (NPN) fluorescence monitored in E. coli 25922. (C) Ethidium Bromide (EtBr) fluorescence monitored in M. smegmatis 700084.

Next, we examined whether conjugate 5e exhibited activity to support the hypothesis that it influences membrane integrity and membrane permeability in Gram-negative bacteria, such as E. coli. We hypothesized that in E. coli, the vancomycin-conjugated biguanide groups would engage the membrane surface by binding LPS, locally displacing divalent cations, and disrupt local membrane architecture thereby enhancing uptake of vancomycin conjugates, similar to the enhanced activity observed from our vancomycin polyguanidino dendrimeric conjugates against E. coli.³¹ To test this hypothesis we examined whether treatment with 5e would increase incorporation of the probe molecule 1-N-phenylnapthylamine (NPN) into the outer membrane. NPN is a sensitive reporter that exhibits fluorescence upon access to and incorporation into the hydrophobic environment of the phospholipid inner leaflet of the outer membrane. NPN accumulation is typically indicative of LPS destabilization and increased membrane permeability.^67–69 Polymyxin B at 80 μM was included as a control compound known to increase cell permeability. Treatment with 5e at 80 μM (10x MIC) in E. coli 25922 resulted in increased NPN fluorescence, consistent with outer membrane destabilization (Figure 6B), while treatment with vancomycin showed no increase in fluorescence. The observation implies that at the molecular level, local membrane perturbation rather than global membrane damage or rupture is compatible with the slow vancomycin-like killing kinetics observed in Figure 4.

Finally, we evaluated the effects of conjugate 5e on M. smegmatis using ethidium bromide (EtBr). In mycobacteria, EtBr enters cells primarily through porins MspA and MspC and exhibits fluorescence when bound to DNA.⁷⁰ Membrane interacting compounds such as cationic antimicrobial peptides enhance dye uptake and observed fluorescence presumably through porin-independent membrane passage due to perturbed membrane integrity.⁷¹ EtBr accumulation is also enhanced in the presence of efflux pump inhibitors.⁷⁰ The LfrA efflux pump inhibitor verapamil was included as a control compound tested at 200 μg/mL to provide a condition with increased EtBr fluorescence.⁷² Treatment with 5e or vancomycin at 8 μM (1x MIC) in M. smegmatis did not influence EtBr fluorescence (Figure 6C). Thus, for M. smegmatis in this assay, the activity of 5e does not indicate an accompanying influence on membrane integrity or permeabilization. We envision the biguanide functionality provides a similar role to the arginine in V-R when treating mycobacteria, wherein improved anti-mycobacterial activity can be attributed to hydrogen bonding interaction with the mycobacterial peptidoglycan m-DAP residue.⁴⁴ The collective results support the presence of membrane interactions and permeabilization activity associated with lead conjugate 5e in S. aureus and E. coli, but not in M. smegmatis. Importantly, this bacterial membrane-associated activity and antibacterial activity of 5e occur without an accompanying increase in mammalian cell toxicity (Figure S1).

Cellular Localization of V-C6-Bg-PhCl, 5e, using Fluorescent Conjugates.

The models described above for enhanced activity invoke multivalent cell surface interactions that result in enhanced cell-association and uptake of 5e. The readily-scaled synthesis of 5e facilitated generation of a fluorescein-conjugated derivative of 5e (Fl-5e) to directly evaluate cell-association and uptake of Fl-5e in the three reference strains.¹⁹ Indeed, upon treatment of equal concentrations of fluorescein conjugated vancomycin (Fl-V) or FI-5e, S. aureus, E. coli, and M. smegmatis all exhibited significantly higher fluorescence with Fl-5e at identical excitation light intensities and camera exposure time (Figure 7). The difference between Fl-V and Fl-5e cell-association observed by fluorescence microscopy was supported by fluorescence-activated cell sorting to quantify fluorescence on a per cell basis (FACS, Figure 7). In each organism, Fl-5e exhibits stronger cellular association than Fl-V, consistent with the mechanistic hypothesis described above in which Fl-5e exhibits overall stronger binding to Lipid II and/or membrane and teichoic acid phosphates at the cell surface in Gram-positive bacteria, enhanced peptidoglycan recognition in mycobacteria, and enhanced association with the outer membrane to facilitate and enhance transport of Fl-5e into the periplasm to access intracellular targets in Gram-negative bacteria.

Figure 7.

Evaluation of antibiotic cell-association through fluorescein-conjugated vancomycin (Fl-V) and 5e vancomycin conjugate (Fl-5e). (A) S. aureus treated with 5 μM of FI-V and FI-5e were examined by fluorescence microscopy with matched 40-ms exposure times and through FACS. (B) E. coli treated with 5 μM of FI-V and FI-5e were examined by fluorescence microscopy with matched 40-ms exposure times and through FACS. (C) M. smegmatis treated with 30 μM of FI-V and FI-5e were examined by fluorescence microscopy with matched 40-ms exposure times and through FACS. Microscopy images display overlays of each fluorescein-associated fluorescence image and phase contrast image. FACS analysis data bars represent median normalized fluorescence values from three experiments, with data normalized to the highest fluorescence value in each experiment. Error bars represent the range of normalized fluorescence values obtained in three experiments.

Evaluation of Biguanide-Vancomycin Conjugates for Eradication of Bacteria in Established Biofilms.

Having established the efficacy of the biguanide-vancomycin conjugates against actively growing Gram-positive and Gram-negative bacteria and mycobacteria in MIC assays and characterized the influence and molecular activity of 5e in these organisms, we evaluated their efficacy against preformed biofilms of the three representative reference strains. Multiple biguanide-vancomycin conjugates show antibacterial activity against biofilm-associated S. aureus 29213, against which vancomycin has no activity (Table 2, MBEC >512 μM). 5b, 5c, and 5e had the most potent activity here, with minimum biofilm eradication concentrations (MBECs) of 4–8, 16–32 and 8–16 μM. These results are comparable to those for control antibiotics rifampicin (16–32 μM MBEC) and V-r8 (8–16 μM MBEC) and almost as efficacious as oritavancin (2–4 μM MBEC). 5a was less active, with an MBEC of 64–128 μM in this strain, and the C16-functionalized biguanide conjugate 5d exhibited no activity against biofilm-associated S. aureus (MBEC > 128 μM) indicating too large of a hydrophobic moiety can impede efficacy against biofilm-associated S. aureus.

Table 2.

Minimum biofilm eradication concentrations (MBECs) of biguanide-vancomycin conjugates 5a-e against pre-formed biofilms of S. aureus 29213, E. coli 25922, M. smegmatis700084.

Minimum Biofilm Eradication Concentration (MBEC, mM)
Species	Strain	Vancomycin	V-C6-Bg-H	V-C6-Bg-C4	V-C6-Bg-C10	V-C6-Bg-C16	V-C6-Bg-PhCI	Oritavancin	Rifampicin	Polymyxin B	Gentamycin	V-r8	V-R
		V	5a	5b	5c	5d	5e	−	−	−	−	−	−
S. aureus	29213	>512	64–128	4–8	16–32	>128	8–16	2–4	16–32	–	–	8–16	≥128
E. coli	25922	256	32–64	64	64	>128	32–64	>128	–	16–32	8	64–128	64
M. smegmatis	700084	64	2	8	64	>128	32	>256	–	–	–	–	8

We next evaluated the biguanide conjugates for activity against pre-formed biofilms of E. coli 25922 (Table 2). Gram-negative strains like E. coli are inherently insensitive to vancomycin due to their outer membrane and, as expected, vancomycin has a high MBEC value of 256 μM in this assay. All conjugates except 5d demonstrated significant biofilm eradication in this assay with MBECs ranging from 32–64 μM, comparable to the positive control compound polymyxin B (16–32 μM MBEC). Notably, oritavancin exhibits no activity against biofilm-associated E. coli.

Lastly, we evaluated biguanide-vancomycin conjugates against bacteria within pre-formed biofilms of M. smegmatis 700084 (Table 2). MBEC results paralleled those of the MIC assay results. Conjugates with larger lipid substituents had higher (worse) MICs and exhibited higher (worse) MBECs than the smaller 5a and 5b. The unsubstituted biguanide conjugate 5a demonstrated 32-fold greater potency than vancomycin and 4-fold greater potency than V-R in eradicating mycobacterial biofilm-associated bacteria. In this same MBEC evaluation, oritavancin showed no demonstrable activity (MBEC >256 μM).

Of the conjugates evaluated in this study, biguanide-vancomycin conjugate 5e consistently outperformed the parent compound, vancomycin, and displayed remarkable activity across all three biofilm-associated organisms not observed for other clinically used agents. While oritavancin is notably active against biofilm-associated S. aureus, for example, it is ineffective against E. coli and M. smegmatis biofilm-associated pathogens where multiple biguanide-vancomycin conjugates are effective. Biguanide-vancomycin conjugate 5e is uniquely positioned as a broad-spectrum antibacterial effective against actively growing bacteria as well as populations in preformed biofilms across all three bacterial classes.

CONCLUSION

We face a future where drug-resistant microbes and difficult-to-treat biofilms render our currently available antimicrobials ineffective and where routine surgeries and other modes of exposure carry unacceptable risk of a life-threatening infection. The design or discovery of more effective antibiotics is urgently needed to address this issue. Generalizable strategies to increase the potency of existing antibiotics and expand their spectrum of activity provide a fast path to clinical evaluation as needed to address the growing threat of antibiotic-insensitive infections and biofilms. Here, we report the design and synthesis of unprecedented biguanide-antibiotic conjugates and their resulting activity against actively growing ESKAPE pathogens and mycobacteria and the slowly growing or dormant biofilm populations that most often exhibit resistance to currently available antibiotics and are the cause of enormous disease burden and mortality. Our lead biguanide vancomycin conjugate, V-C6-Bg-PhCl (5e), demonstrates excellent cell killing with up to 2 orders-of-magnitude improvement over the parent compound vancomycin against vancomycin-resistant enterococcus. V-C6-Bg-PhCl exhibits truly broad-spectrum activity and is active against representative strains of all ESKAPE pathogens as well as the model mycobacterial organism M. smegmatis and the pathogenic non-tuberculosis mycobacterium M. abscessus. Moreover, V-C6-Bg-PhCl proved to eradicate biofilm-associated bacteria from all three bacterial classes, represented by S. aureus, E. coli, and M. smegmatis. Consistent with multivalency-originating improvements to vancomycin’s activity as opposed to synergy-derived improvements to vancomycin’s activity, conjugation of the biguanide to vancomycin is required for potent antimicrobial activity. Mode-of-action studies included time-kill kinetics assays, molecular detection of peptidoglycan precursor accumulation, membrane permeability assays, and evaluation of antibiotic cell-association through fluorescence microscopy and FACS using fluorescein-conjugated vancomycin and V-C6-Bg-PhCl. Integration of all the results indicated vancomycin-like killing with multifunctional attributes and the ability to engage relevant cell-surface anions and perturb membrane architecture as evidenced by enhanced permeability to fluorescent probe molecules for Gram-positive and Gram-negative bacteria. Thus, V-C6-Bg-PhCl represents a promising new antibacterial agent for difficult-to-treat biofilm-associated infections and urgent threat bacterial pathogens while displaying no acute mammalian cell toxicity. With its expanded spectrum of activity, V-C6-Bg-PhCl could be a potentially invaluable clinical option for the treatment of polymicrobial infections. As exemplified by our seminal studies on guanidinylated antibiotic conjugates and the related advances by our and other labs that followed, we expect that the further investigation of biguanidinylated antibiotics could provide novel leads if not clinical candidates for the treatment of challenging infections.