Vancomycin-Polyguanidino Dendrimer Conjugates Inhibit Growth of Antibiotic-Resistant Gram-Positive and Gram-Negative Bacteria and Eradicate Biofilm-Associated S. aureus

Abstract

The global challenge of antibiotic resistance necessitates the introduction of more effective antibiotics. Here we report a potentially general design strategy, exemplified with vancomycin, that improves and expands the antibiotic performance. Vancomycin is one of the most important antibiotics in use today for the treatment of Gram-positive infections. However, it fails to eradicate difficult-to-treat biofilm populations. Vancomycin is also ineffective in killing Gram-negative bacteria due to its inability to breach the outer membrane. Inspired by our seminal studies on cell penetrating guanidinium-rich transporters (e.g. octaarginine), we recently introduced vancomycin conjugates that effectively eradicate Gram-positive biofilm bacteria, persister cells and vancomycin-resistant enterococci (with V-r8, vancomycin-octaarginine), and Gram-negative pathogens (with V-R, vancomycin-arginine). Having shown previously that the spatial array (linear versus dendrimeric) of multiple guanidinium groups affects cell permeation, we report here for the first time vancomycin conjugates with dendrimerically displayed guanidinium groups that exhibit superior efficacy and breadth, presenting the best activity of V-r8 and V-R in single broad-spectrum compounds active against ESKAPE pathogens. Mode-of-action studies reveal cell-surface activity and enhanced vancomycin-like killing. The vancomycin-polyguanidino dendrimer conjugates exhibit no acute mammalian cell toxicity or hemolytic activity. Our study introduces a new class of broad-spectrum vancomycin derivatives and a general strategy to improve or expand antibiotic performance through combined mode-of-action and function-oriented design studies.

With 700,000 deaths globally per year, drug-resistant bacterial infections have emerged as one of the most urgent threats to public health.^1,2 The overuse of antimicrobial drugs is further fueling an increase in the prevalence of drug-resistant pathogens. Chronic and recurrent bacterial infections pose yet another challenge to effective treatment through the generation of biofilms and persister cells that exhibit tolerance to antimicrobials. New antimicrobials and anti-infective strategies are urgently needed to address these problems.

The World Health Organization has classified a subset of disease-causing bacteria as “high priority” pathogens for drug development that represent the predominant causes of nosocomial infection and death worldwide.^3,4 Collectively termed the ESKAPE pathogens, these include the Enterobacter species, S. aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and E. faecium. Many infections caused by these organisms could be treated with antibacterial agents that target cell-wall synthesis. For example, S. aureus and E. faecium are both Gram-positive organisms that surround themselves with a thick cell wall composed primarily of peptidoglycan and teichoic acids. While E. coli, K. pneumoniae, A. baumannii, and P. aeruginosa are Gram-negative organisms characterized by a dual membrane envelope, they too have a peptidoglycan component positioned in the periplasm between the inner and outer membranes. Peptidoglycan biosynthesis is crucial to bacterial cell wall integrity and cell division in both Gram-positive and Gram-negative bacteria and represents an attractive common target for antibacterial development.

Glycopeptides including vancomycin inhibit peptidoglycan synthesis and have a long history of use in treating Gram-positive infections, while a smaller number of recent efforts have focused on extending the spectrum to Gram-negative organisms.⁵ In Gram-positive bacteria, vancomycin exerts its bactericidal activity by binding and sequestering Lipid II at the cell surface, thereby inhibiting peptidoglycan biosynthesis.^6–9 However, vancomycin is unable to eradicate biofilm-associated S. aureus. The propensity of S. aureus to form biofilms and persister cells is linked to recurrent and chronic infections including infective endocarditis, surgical site and wound infections, and infections of medical devices such as intravenous catheters and artificial joints.^6,10–15 The emergence of vancomycin intermediate-resistance S. aureus (VISA), vancomycin resistant S. aureus (VRSA) and vancomycin resistant enterococci (VRE) also underscores the need for new and more effective therapeutic options.¹⁶ Furthermore, vancomycin is ineffective in killing Gram-negative bacteria at clinically relevant concentrations due to its inability to cross the outer membrane and accumulate to an appreciable level in the periplasm where peptidoglycan assembly occurs.¹⁷ While progress on the design and synthesis of vancomycin derivatives has been made,⁵ more effective, more synthetically accessible, and better tolerated agents are needed.

Clinically used semi-synthetic glycopeptides with improved efficacy against resistant strains include dalbavancin, telavancin and oritavancin, with the latter two uniquely effective against biofilm-associated Gram-positive bacteria.¹⁸ Additional vancomycin conjugates with lipophilic tails and permanent positively charged groups such as trimethylammonium groups have been found to exhibit activity against resistant Gram-positive and Gram-negative bacteria,¹⁹ although killing activity against Gram-negative organisms in this case could be attributed to the lipidated quaternary amine alone. Sulfonium groups also enhance antibacterial activity.²⁰ The Haldar group introduced vancomycin conjugates with groups that bind bacterial cell wall pyrophosphate and display improved killing against Gram-positive pathogens.²¹ Boger and coworkers introduced fully synthetic vancomycin derivatives based on the redesign of the glycopeptide binding pocket, leading to a vancomycin conjugate with dual D-Ala-D-Ala/D-Ala-D-Lac affinity.²² Building on this work, and the chlorobiphenyl (CBP) modification used in oritavancin, they also incorporated a quaternary ammonium cation that further improved activity.^23,24 Other cell surface-interacting and lipophilic derivatives have also shown efficacy in treating vancomycin-resistant Gram-positive pathogens.^25–30 However, the efficacy of many newly introduced conjugates against biofilm-associated bacteria is often not reported. Identification of the best vancomycin conjugates for clinical development should prioritize consideration of biofilm activity and have acceptable tolerability in vivo. Furthermore, the development of vancomycin conjugates with Gram-negative efficacy would open a new therapeutic avenue for the treatment of these infections.

As examples of strategies to enable or enhance drug delivery,³¹ we previously reported the first cell-penetrating arginine- and more generally guanidinium-rich transporters^32,33 that carry drugs across biological barriers including a range of cell membranes and even human skin,³⁴ overcome drug resistance in cancer treatement^35,36 and, more recently, target difficult-to-treat biofilm populations, including vancomycin-insensitive Gram-negative organisms.^17,37 In contrast to existing semisynthetic or lipophilic vancomycin derivatives at that time, we introduced the first vancomycin-transporter conjugate consisting of vancomycin linked by an aminohexanoic acid spacer to a D-octaarginine transporter (r8).³⁷ This conjugate (V-r8) outperformed vancomycin in Gram-positive persister cell and biofilm assays, and demonstrated tighter cellular association and faster bactericidal mode of action against MRSA (99.9% kill in 30 min).³⁷ In vivo, V-r8 eliminated 97% of biofilm-associated MRSA in a murine wound infection model and exhibited no acute toxicity or damage to skin tissues. These results as well as efficacy of V-r8 against VRE are consistent with the hypothesis that V-r8 exhibits multivalent interactions, with the vancomycin portion binding to D-Ala-D-Ala Lipid II stem termini while the r8 subunit synergistically engages proximate anions such as teichoic acids at the cell surface or inside cells. In subsequent studies, we succeeded in transforming vancomycin into a highly active Gram-negative agent by conjugation to a single arginine (V-R) or two arginines (V-RR).¹⁷ We discovered that V-R and related amino acid conjugates are effective against carbapenem-resistant E. coli and A. baumannii. V−R resulted in intracellular accumulation of peptidoglycan precursors, consistent with cell-wall biosynthesis disruption as its mechanism of action, but at much lower concentrations than vancomycin itself. V-R was highly effective in a mouse thigh muscle infection model, reducing E. coli colonization by 4 orders of magnitude at the low dose of 200 mg/kg and greater reductions at higher doses. Vancomycin reduced colonization by only 10-fold at a high dose of 1272 mg/kg.³⁸ Further in vivo studies with V-R demonstrated efficacy against extended-spectrum beta-lactamase-positive and carbapenem-resistant E. coli in a complicated urinary tract infection model, with concentrations of V-R reaching up to 120x MIC in urine 0–8 hr post-administration.³⁹ Thus, our foundational studies identified vancomycin-transporter conjugates that significantly outperform vancomycin, that are better than current clinically used agents in murine biofilm and UTI models, and that exhibit activity consistent with a new or dual mode of action. Following our introduction of V-R, vancomycin and chlorobiphenyl vancomycin conjugates also incorporating single guanidinium groups at the vancomycin C-terminus were reported by Boger and coworkers with efficacy against S. aureus and VRE.⁴⁰ In more recent work, the Boger group reported on new conjugates of their enhanced D-Ala-D-Lac-binding pocket-modified vancomycins that incorporate a C-terminal guanidinium, as in V-R, and additional pocket modifications including an amidine.⁴¹

Unlike V-R and subsequent guanidinium-modified vancomycins, the more highly pursued avenue for improving vancomycin performance has included incorporation of larger cationic lipophilic and/or antimicrobial peptide components, primarily to enhance activity specifically against Gram-positive bacteria. Newer conjugates termed the vancapticins incorporated lysine-rich electrostatic effector peptide sequence (EEPS) and a lipid membrane-insertive element (MIE).⁴² Other work identified a lead conjugate [(Lys⁷(PEG₄-Van)]TP10 with activity against resistant Gram-positive bacteria, intracellular activity in HEK293 cells, and the ability to breach the blood-brain barrier.⁴³ Lipo-cationic vancomycin conjugates paired vancomycin with outer membrane-disrupting adjuvants to potentiate vancomycin activity.⁴⁴ Additionally, a nitric oxide-releasing system associated with vancomycin demonstrated efficacy in the treatment of MRSA abscesses.⁴⁵ Following our report of V-r8, another cell-penetrating peptide with approximately 10 guanidinium groups and ethylene glycol oligomers was found to display attractive in vivo behavior.⁴⁶ The use of triarginine appended to a fatty acid chain also yielded conjugates with improved activity against VRE.⁴⁷ In line with our earlier work on V-r8, subsequent work by the Uhl group examined conjugation of hexa-arginine (R6) to various positions on vancomycin and found improved activity with conjugation to the N-terminus of the glycopeptide.⁴⁸

As an aspirational next-generation goal, we sought to design, prepare and evaluate vancomycin conjugates that would exhibit the remarkable activities of V-r8 against Gram-positive organisms and V-R against Gram-negative organisms in a single agent. To address this goal, we leveraged our seminal discoveries on cell-penetrating guanidinium-rich molecular transporters (Figure 1A) that uncovered superior delivery of probes or drug compounds into cells based on the number and spatial orientation (linear versus dendrimeric) of guanidinium groups attached to a cargo molecule. Highly cationic surfactants such as primary amines⁴⁹ and quaternary ammonium groups^50–52 have since been reported to disrupt bacterial membranes. Our prior work demonstrated that guanidinium groups avidly associate with cell surface anions through electrostatic and double H-bonding, and that a dendrimeric array of guanidinium groups outperforms a linear array.⁵³ Given the superior performance of V-r8 with its linearly arrayed guanidiniums, we hypothesized that a dendrimeric presentation of guanidinium groups could enhance multivalent association with negatively charged groups on the bacterial surface, e.g. lipopolysaccharide in Gram-negative bacteria and pyrophosphate and teichoic acid in Gram-positive bacteria by altering the spatial positioning of guanidinium groups. Dendrimeric and other assemblies incorporating multiple copies of vancomycin have been reported,^54–58 and dendrimers presenting guanidinium groups have themselves shown antibiotic activity.^59–62 Here we report a new class of vancomycin conjugates incorporating dendrimerically displayed guanidinium groups in which both subunits contribute cooperatively to antibiotic activity, arresting cell-wall synthesis and promoting bacterial membrane permeation. We designed two vancomycin dendrimer conjugates (V-triguans), wherein the C-terminus of vancomycin presents three guanidinium groups with different lengths of extension from the dendrimeric branching point. We opted to focus only on a first-generation dendrimeric core for step-economical and raw material cost considerations. The hypothesized cell-surface engagements of a single V-triguan in Figure 1 illustrate how the design of dendrimerically displayed guanidinium groups extending from the vancomycin core could engage the different relevant anions in Gram-negative and Gram-positive organisms. Here, we report on antimicrobial susceptibility, biofilm evaluation, and mechanistic analyses that establish these new dendrimeric V-triguans as broad-spectrum compounds active against both Gram-positive and Gram-negative ESKAPE pathogens and against Gram-positive biofilms.

Figure 1. Polyguanidino dendrimeric vanomcyin conjugates designed to engage relevant cell-surface anions in Gram-positive and Gram-negative bacteria.

(A) Schematic illustrations of various guanidinium-rich molecular transporters including linear transporters such as octaarginine, guanidinium-rich peptoids and backbone spaced transporters, as well as branched transporters such as dendrimeric transporters.³³ G⁺ represents a guanidinium group. (B) Plausible guanidinium interactions with Lipid II pyrophosphate and teichoic acid phosphates add a dual binding functionality beyond the conventional glycopeptide binding to D-Ala-D-Ala of Lipid II and immature peptidoglycan. Wall teichoic acid is shown; lipoteichoic acid is also implicated. (C) Envisioned guanidinium-mediated interactions with LPS displace divalent cations and disrupt local LPS architecture, facilitating V-triguan transport into the periplasm and enabling V-triguan binding to and sequestration of peptidoglycan precursors in a vancomycin-like manner.

RESULTS AND DISCUSSION

Design and Synthesis of V-triguan-6C and V-triguan-2C.

The vancomycin conjugates were designed to explore two linker lengths (6 and 2 carbons) connecting the dendrimerically arrayed guanidinium groups attached orthogonally to the glycopeptide antibiotic D-Ala-D-Ala binding pocket. The dendrimer subunit 1 was synthesized following a procedure by Newkome et al.⁶³ in which three C-C bonds are formed in one step through a triple Michael-addition of nitromethane and three equivalents of t-butyl acrylate. The triester product is then treated with formic acid to yield the intermediate 1 (Figure 2). Conjugation with a mono-protected diamine provided triamides 2a and 2b which, upon reduction of the nitro group and linker introduction, gave the differentially protected tetraamides 3a and 3b. Selective removal of the Boc protecting group followed by guanidinylation and global deprotection afforded the tri-guanidinylated transporters 5a and 5b which upon conjugation to the C-terminus of vancomycin using HBTU as coupling reagent, gave the V-triguan conjugates 6a and 6b. These were purified by reverse-phase HPLC to yield the trifluoroacetate salt (n = 1, 9.0% yield over 7 steps; n = 3, 9.6 % yield over 7 steps based on isolated yields). Full synthetic details, high-resolution mass analyses, ¹H-NMR, ¹³C-NMR spectra and relevant HPLC traces are provided in supporting information.

Figure 2.

Syntheses of V-triguan-2C (6a), V-triguan-6C (6b) ,Triguan-2C (7a) and Triguan-6C (7b). ^aYields determined based on quantitative NMR.

Evaluation of V-triguan-2C (6a) and V-triguan-6C (6b) in antimicrobial susceptibility assays with ESKAPE pathogens.

The vancomycin conjugates, V-triguan-2C (6a) and V-triguan-6C (6b), and the dendrimeric transporters alone (7b) were first evaluated for antibacterial activity against a panel of Gram-negative and Gram-positive organisms. Significantly, both conjugates were superior to vancomycin, exhibiting excellent killing activity against organisms that are resistant or insensitive to vancomycin. Of further significance, each individual conjugate has MIC values against Gram-positive and Gram-negative bacteria that are comparable to or lower than those observed for V-r8 against Gram-positive bacteria (including VRE) and V-R against Gram-negative bacteria (Tables 1 and 2). Gram-negative bacteria are intrinsically resistant to vancomycin because their outer membrane presents a barrier to compound accumulation in the periplasm where cell wall synthesis occurs.¹⁷ However, the V-triguans overcome this challenge and show excellent activity against several Gram-negative organisms (Table 1). Each V-triguan conjugate has an MIC of 8 μM against uropathogenic E. coli strain UTI89, comparable to V-R (MIC of 8–16 μM¹⁷) and superior to V-r8 (MIC of 32 μM³⁷) and vancomycin (128 μM) (Table 1). Furthermore, V-triguan-6C and V-triguan-2C have MIC values of 2 and 4 μM, respectively, against colistin-resistant E. coli. The conjugates were also highly active against carbapenem-resistant E. coli (Table 1). Additionally, we evaluated V-triguan activity against other important Gram-negative ESKAPE pathogens. In particular, V-triguan-6C exhibited excellent improvement of up to 16-fold over vancomycin against A. baumannii, with an MIC of 8 μM. This activity is superior to V-R and V-r8, which have MICs of 16–32 and 64 μM against A. baumannii, respectively. V-triguan-6C also exhibited enhanced killing activity over vancomycin against extended-spectrum beta-lactamase-positive Klebsiella, with an MIC of 32 μM (Table 1). Finally, V-triguan-6C and V-triguan-2C each exhibited an approximately 10-fold and 32-fold improved efficacy over vancomycin against P. aeruginosa, with an MIC of 32–64 μM against PA14 and 16 μM against PA01, respectively (Table 1). Across all the Gram-negative strains tested, we observed an 8-fold or greater improvement in the MIC for the vancomycin dendrimer conjugates compared to vancomycin. This enhanced activity of V-triguan-6C and V-triguan-2C in Gram-negatives is consistent with the hypothesis that the guanidinium-presenting dendrimeric transporter allows the conjugates to cross the outer membrane such that they can reach and bind peptidoglycan precursors in the periplasm, as reported for V-R.¹⁷

Table 1.

Minimum inhibitory concentrations (MICs) of vancomycin-dendrimer conjugates and dendrimer scaffolds against Gram-negative pathogens.

	MIC
Strain	Vancomycin (V)		V-triguan-6C (6b)	V-triguan-2C (6a)	V +128 μM Triguan-6C (7b)	Triguan-6C (7b)	V-r8	V-R
	(μg/mL)	(μM)	(μM)	(μM)	(μM)	(μM)	(μM)	(μM)
E. coli 25922	185	128	8–16	8	64–128	>128	16–32	8–16
E. coli UTI89	185	128	8	8	64	>128	32	8–16
E. coli BAA-2469 (carbapenem resistant)	93–185	64–128	4–8	4–32	64	>128	16	8–16
E. coli NCTC 13846 (colistin resistant)	93–185	64–128	2	4	64	>128	32	8
A. baumannii 19606	93–185	64–128	8	32	64	>128	64	16–32
K.pneumoniae 700603	371–742	256–512	32	128	128–256	>128	>128	64
P. aeruginosa PA14	742	512	32–64	64	512	>128	8	64
P. aeruginosa PA01	742	512	64	16	512	>128	8	32

Table 2.

Minimum inhibitory concentrations (MICs) of vancomycin-dendrimer conjugates and dendrimer scaffolds against Gram-positive pathogens.

	MIC
Strain	Vancomycin (V)		V-triguan-6C (6b)	V-triguan-2C (6a)	V +128 μM Triguan-6C (7b)	Triguan-6C (7b)	V-r8	V-R
	(μg/mL)	(μM)	(μM)	(μM)	(μM)	(μM)	(μM)	(μM)
Vancomycin-resistant
E. faecium ATCC 51559 (VRE)	>742	≥512	8–16	16–32	512	>128	1–2	32
E. faecalis ATCC 51575 (VRE)	185	128	16	32	128	>128	32a	128–256
Vancomycin-susceptible
S. aureus 29213	0.73–1.4	0.5–1	0.5–1	0.5	0.5–1	>128	0.5–1	0.5–1
S. aureus USA400 MW2 (MRSA)	0.73–1.4	0.5–1	0.5–1	1	1	>128	0.63–1	1

^aPreviously determined MIC.³⁷

Covalent conjugation between the transporter and vancomycin was essential for activity against Gram-negative pathogens. In these strains, an acetylated version of the 6-carbon-linker synthetic dendrimeric scaffold alone (Triguan-6C, 7b, SI) had no activity at the highest concentrations tested (MICs > 128 μM). Furthermore, the unconjugated combination of vancomycin with 128 μM Triguan-6C (7b, sub-MIC) demonstrated no significant improvement in MIC over vancomycin alone. This activity differentiates our dendrimeric scaffold from polymyxin B derivatives like polymyxin B nonapeptide and NAB741, as well as other membrane-active sensitizing adjuvants, which themselves sensitize Gram-negative bacteria and reduce the MIC for antibiotics like vancomycin when tested in combination.^64–67

For Gram-positive organisms, vancomycin is already highly active against exponentially growing S. aureus and the V-triguans maintain this activity. In contrast, vancomycin is not effective in killing vancomycin-resistant enterococci (VRE) at reasonable concentrations, primarily due to their cell-wall reprogramming with production of D-Ala-D-Lac-terminating cell-wall precursors in place of D-Ala-D-Ala. The MIC of vancomycin is 512 μM against vancomycin-resistant E. faecium 51559 (with VanA-type resistance) and 128 μM against vancomycin-resistant E. faecalis 51575 (with VanB-type resistance).^68–70 Remarkably, V-triguan-6C and V-triguan-2C have MIC values of 8–16 μM and 16–32 μM, respectively, against E. faecium 51599, and MIC values of 16 μM and 32 μM against E. faecalis 51575 (Table 2).⁶⁶ The activity of V-triguans against VRE could be attributed to the branched array of guanidiniums engaging proximate anions such as teichoic acids and Lipid II-associated pyrophosphate at the cell surface, compensating for the reduced binding affinity of the glycopeptide binding cleft to the D-Ala-D-Lac peptidoglycan stem termini. Additionally, or concurrently, the conjugate could exhibit enhanced penetration into the cytoplasm, where it could engage cell-wall precursors. If the V-triguans operate in this way with vancomycin-like activity, but achievable with lower concentration, we hypothesized that we would be able to detect the accumulation of cell-wall precursors. Consistent with vancomycin-like activity, the accumulation of cell-wall precursors was detected in E. faecium 51559 and E. faecalis 51575 (Figure S2).

Covalent conjugation of vancomycin and the dendrimeric transporter was also essential for Gram-positive activity. The transporter unit (7b) alone did not exhibit antibacterial activity (MIC ≥ 128 μM, Table 2). A mixture of vancomycin and sub-MIC concentrations of the dendrimeric transporter also displayed activity indistinguishable from that of vancomycin alone in Gram-positive strains. This differentiates the dendrimeric transporters from octaarginine, which displayed some antibacterial activity alone (MIC of 20 μM in MRSA strain USA400, also termed MW2).³⁷ The lack of activity of the Triguan-6C and Triguan-2C transporters also differentiates these materials from other work with much denser arrays of polyamines⁷¹ or guanidinium groups, in which the guanidinium-containing dendrimers themselves are antimicrobial.^{49,50,72–74} Thus, we observed efficacies of dendrimeric guanidinium-presenting V conjugates that are superior or comparable to V-R against Gram-negative bacteria and demonstrate efficacy against Gram-positive VRE. Comparing V-triguan-2C and V-triguan-6C, V-triguan-6C was more effective than V-triguan-2C against A. baumannii and K. pneumoniae and thus presents the best antibacterial activity of V-R and V-r8 in a single agent.

Further evaluation against S. aureus: biofilm activity, killing kinetics and membrane permeability.

Having established the efficacy of the V-triguan conjugates against planktonic Gram-positive and Gram-negative bacteria in MIC assays, we evaluated their efficacy against biofilms. Many antibiotics, including vancomycin, are active only against planktonic bacteria and are unable to successfully eradicate cells in biofilms. Indeed, the formation of bacterial biofilms poses a tremendous challenge to treatment and is linked to difficult-to-treat persistent and chronic infections.^{11,13,75–77} V-triguan-6C successfully eradicated bacteria within pre-formed S. aureus biofilms with a minimal biofilm eradication concentration (MBEC) of 16–32 μM (Table 3). This activity is comparable to MBEC values for V-r8 (8–16 μM) and rifampicin (10–39 μM). As expected, vancomycin was unable to eradicate S. aureus biofilms even at concentrations up to 512 μM. We also evaluated activity against pre-formed Gram-negative biofilms of E. coli. Here, we did not observe significant anti-biofilm activity (MBEC > 128 μM). Thus, V-triguan-6C encompasses the favorable characteristics of V-r8 in efficacy against VRE and pre-formed S. aureus biofilms and V-R in killing actively growing Gram-negative pathogens. The ability of V-triguan-6C to eradicate S. aureus biofilms is consistent with the idea that V-triguan-6C is better able to penetrate the biofilm matrix to reach target cells, and/or that the conjugate has improved activity against slow-growing or dormant cells within the biofilm. We next sought to probe other mechanistic aspects of the activities of the V-triguan conjugates against S. aureus.

Table 3.

Minimal biofilm eradication concentrations (MBECs) against preformed biofilms of S. aureus 29213.

MBEC (μM)
Compound	S. aureus 29213
Vancomycin	>512
V-triguan-6C	16–32
V-triguan-2C	≥64
V-r8	8–16
V-R	≥128
Rifampicin	10–39

In order to dissect the new activity unique to the dendrimeric vancomycin conjugates, we employed a time resolved viability assay, based on enumeration of viable cells in colony forming units per mL, CFU/mL, following antibiotic treatment. V-triguan-6C displayed killing kinetics indistinguishable from vancomycin against S. aureus 29213, and slower than that observed for V-r8 (Figure 3A). V-triguan-2C displayed similar time-kill kinetics to V-triguan-6C (Figure 3A). The additional proposed modes of action for the V-triguans could result in increased conjugate permeation and cell membrane permeability as seen previously for V-r8. To gain additional insight into the mode of action of the V-triguans in S. aureus, we used the fluorescent probe propidium iodide (PI) to examine possible enhancement of membrane permeability during V-triguan treatment. PI is unable to penetrate and accumulate in healthy viable cells. However, if membrane structure or integrity is perturbed, PI enters the cytoplasm and binds to DNA, resulting in increased fluorescence. Treatment of S. aureus with V-triguans in HEPES-Glucose buffer to maintain metabolic activity^78,79 was accompanied by increased PI fluorescence. Lysostaphin was included as a positive control which abolishes cell envelope integrity through peptidoglycan digestion (Figure 3B). The influence of V-triguan treatment on PI fluorescence is dependent on the suspension media, similar to activity observed for V-r8. No increase in PI permeability was observed when cells were suspended in PBS, which lacks glucose and has a higher osmolarity than the HEPES-Glucose buffer (Figure S1). The increased membrane permeability is consistent with a mechanism of action where the dendrimeric transporter facilitates transport of the conjugate into the cytoplasm, and passage of the bulky V-triguan conjugate permeabilizes the membrane to molecules such as PI. Alternatively or additionally, the topologically separated guanidinium groups may participate in multivalent interactions at the cell surface with conventional binding of vancomycin to D-Ala-D-Ala Lipid II stem termini while the positively charged guanidinium groups synergistically engage proximate anions such as the pyrophosphate of Lipid II and/or teichoic acid at the cell surface. This dual binding could also lead to increases in cell permeability as the V-conjugates tug at components of the cell surface as cells try to grow and compromise membrane integrity. The unconjugated mixture of V + Triguan-6C/Triguan-2C (7a/7b) and Triguan-6C/Triguan-2C alone did not have any cell permeabilizing effect, supporting the conclusion that conjugation is crucial for activity either by enabling simultaneous dual binding that disrupts permeability, or via transport of the bulky conjugate into the cell.

Figure 3.

V-triguan-6C exhibits vancomycin-like killing kinetics in S. aureus and enhances probe dye uptake. (A) Killing kinetics of vancomycin (V), V-triguans, and V-r8. (B) PI fluorescence evaluation of membrane permeability in HEPES-Glucose buffer for stationary phase S. aureus 29213.

V-triguans exhibits vancomycin-like activity in E. coli at low concentration.

In Table 1, we observed an 8-fold or greater improvement in MIC for the V-triguans compared to vancomycin in various Gram-negative organisms. Furthermore, V-triguan activity was superior to V-R among several pathogens. In Gram-negative bacteria, the outer membrane is composed of primarily lipopolysaccharides (LPS) in the outer leaflet and phospholipids in the inner leaflet. The outer membrane serves as a barrier to transport of large compounds such as glycopeptides. We hypothesized that the V-triguans exhibit improved activity over vancomycin because the guanidinium groups on the conjugate enable transport across the outer membrane via enhanced interactions at the membrane surface through binding interactions with LPS anions, concomitant displacement of divalent cations, and perturbation or dynamic remodeling of the membrane structure.⁸⁰ Furthermore, following uptake into the periplasm, we hypothesized that the V-triguan killing activity would be through vancomycin-like inhibition of peptidoglycan synthesis, as observed for V-R.¹⁷ Similar to experiments described above for S. aureus, we first examined the time-kill kinetics of the V-triguans in E. coli 25922. The kinetics of cell killing of the V-triguans were similar to that of V-R at a concentration of 16 μM, which is 1–2x MIC for each conjugate (Figure 4A). As expected, vancomycin is ineffective against Gram-negative bacteria at low concentrations, and a 16 μM vancomycin (V) treatment condition mirrored the untreated growth control (Figure 4A). At the much higher concentration of 256 μM required for bacterial killing, vancomycin time-kill kinetics are identical to those for V-R and the V-triguans.¹⁷ Time-kill kinetics analysis supports the mechanism in which the V-triguans exhibit vancomycin-like inhibition of cell-wall synthesis in Gram-negative bacteria as reported previously for V-R, but at a lower concentration than that required for vancomycin.

Figure 4.

Mechanistic activity of V-triguans against E. coli. (A) Time-kill kinetics analysis of vancomycin (V), V-triguans, and V-R in E. coli 25922. (B) Outer membrane permeability assessed with NPN fluorescence in HEPES-Glucose for exponential phase E. coli 25922.

Given our molecular hypothesis that vancomycin-conjugated guanidinium groups engage the membrane surface through binding LPS and affecting local membrane architecture to enhance uptake of vancomycin conjugates, we examined whether treatment with the V-triguans would increase intercalation 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 and is typically indicative of LPS destabilization and increased membrane permeability.^67,81,82 In contrast to V-R, V-triguan-6C treatment at 32 μM (4x MIC) in E. coli 25922 resulted in NPN fluorescence similar to that of the positive control, polymyxin B at 32 μM (Figure 4B). V-triguan-2C also exhibited significant NPN fluorescence, although with a peak fluorescence value lower than that for V-triguan-6C treatment. Remarkably, treatment with the isolated Triguan-6C transporter also exhibited significant NPN fluorescence and we note that Triguan-6C on its own exhibited no antibacterial activity. Indeed, Triguan-6C concentrations up to 128 μM showed no antibacterial activity (Table 1). This result reveals that Triguan-6C can disrupt LPS architecture and dynamics. Such an influence without causing killing is also observed for other membrane-interacting agents and permeabilizers such as polymyxin B nonapeptide.^83,84 The combination of vancomycin + Triguan-6C was similar to that of Triguan-6C alone. In contrast, Triguan-2C treatment (as well as vancomycin + Triguan-2C treatment) did not yield remarkable NPN fluorescence (Figure 4B). These collective results reveal membrane disturbance and permeabilization activity associated with the vancomycin polyguanadino dendrimers (V-Triguan-6C and V-Triguan-2C). The dramatic distinction between the longer-branched Triguan-6C influence on NPN fluorescence and the lack of fluorescence attributed to Triguan-2C is consistent with a model in which the longer linker length of Triguan-6C allows for greater membrane surface area engagement than that for Triguan-2C, leading to greater local disruption of divalent cation arrangements and enhanced NPN uptake in this sensitive assay. Overall, NPN assay results illuminate the molecular influence of the polyguanidino dendrimeric transporters in which the V-triguans interact with LPS and lead to detectable local membrane disturbance. This activity is not observed for V-R and is not necessarily required for V-triguan killing activity, but is a feature of their influence on E. coli. The uptake of NPN upon V-triguan treatment is consistent with antibiotics that exhibit self-promoted uptake. The V-triguans as well as other cationic antibiotics, have the potential to competitively displace divalent cations such as Mg²⁺ that bridge neighboring LPS phosphate groups,⁸⁵ facilitating antibiotic association with the cell surface and transport through the outer membrane. An increase in exogenously available divalent cations through medium supplementation typically reduces potency of compounds that exhibit this type of self-promoted uptake. In addition, LPS-defective mutants of E. coli are significantly more susceptible to vancomycin.⁸⁶ Consistent with this view, we found that growth medium supplementation with 5 mM Mg²⁺ increased the MIC of V-triguan-6C and V-triguan-2C from 8 to 64 μM in E. coli (Table S2).¹⁷ The vancomycin MIC also increased from 128 to 512 μM, as vancomycin is positively charged near neutral pH. Chloramphenicol was included as a control compound that enters cells via porins and does not exhibit an altered MIC with Mg²⁺ addition.

Finally, we additionally examined whether treatment with the V-triguans would result in intracellular accumulation of peptidoglycan precursors, as observed for V-R¹⁷ and as expected for treatment with high concentrations of vancomycin.⁸⁷ HPLC-MS analysis revealed that V-triguan-6C and V-triguan-2C resulted in greater accumulation of peptidoglycan precursor UDP-MurNAc pentapeptide than that seen for a matched treatment of 16 μM vancomycin in E. coli 25922 (Figure 5). An increase in the accumulation of peptidoglycan precursors was achieved for vancomycin by increasing the vancomycin concentration from 16 μM to the MIC value of 128 μM, as expected (Figure S3).⁸⁷ The collective data and direct comparisons of V-R, the V-triguans, isolated transporters, and combinations against E. coli provide data that support a self-promoted uptake mechanism and underscore the ability to influence antibiotic interactions with bacterial surfaces through strategic molecular design.

Figure 5.

Accumulation of peptidoglycan precursors in E. coli. (A) HPLC chromatograms showing peptidoglycan precursor accumulation for treated cells. Data are representative of at least 3 biological replicates. The intensities of the chromatograms were normalized by the internal standard, terfenadine, with an elution time of 6.0 min. The average percentage increase in UDP-MurNAc pentapeptide accumulation upon antibiotic treatment is shown in Figure S4. (B) Mass spectrum of peptidoglycan precursor UDP-MurNAc pentapeptide detected at the 4.12 min retention time.

V-triguan-6C and V-triguan-2C demonstrate no adverse toxicity in hemolysis and mammalian cell culture assays.

Towards the future consideration of V-triguan-6C in animal models for advancement in pre-clinical investigation, compounds were evaluated in red blood cell hemolysis assays and mammalian cell toxicity assays. No or minimal hemolysis was observed for all conjugates, with values under 2% for each antibacterial conjugate (Table S1). No adverse toxicity was observed up to 160 μM in mammalian cell MTS assays using HeLa cells grown in DMEM medium (Figure S5). Thus, V-triguan-6C presents a promising new antibacterial candidate with little to no toxicity as evaluated in mammalian cell culture and red blood cells.

CONCLUSION

We face a future where our current arsenal of antibiotics could be rendered ineffective at eliminating drug-resistant microbes and difficult-to-treat biofilm populations. The design or discovery of new or improved antibiotics is critically needed to address the current and expected further rise in infections and their broad impact on global human health. Prompted by this view, we have initiated studies that would afford a fast path to the clinic by improving and/or expanding the activity of current generation, often monofunctional, antibiotics through introduction of multifunctional characteristics. We previously reported that V-r8 exhibits superior activity relative to vancomycin against Gram-positive bacteria, including VRE and biofilm-associated S. aureus.³⁷ Furthermore, we transformed vancomycin into highly active Gram-negative antibacterial agents through single and double arginine conjugation to vancomycin, with lead candidate V-R showing exemplary in vivo activity in mouse infection models and activity against drug-resistant E. coli, K. pneumoniae and A. baumannii.^17,38,39 Here, we describe the design and evaluation of the first members of a new class of vancomycin derivatives that represent the best of V-r8 and V-R in a single agent. V-triguan-6C uniquely incorporates both the antibiotic vancomycin and dendrimeric array of guanidinium groups as a transporter subunit. Supportive of the importance of this novel spatial array of guanidinium groups, V-triguan-6C displays order of magnitude improvement in efficacy compared with vancomycin alone against both Gram-positive and Gram-negative bacteria and eradicates biofilm-associated MRSA. Mechanistic studies, including time-kill kinetic assays and membrane permeability evaluations, support a model that combines the self-promoted uptake activities we ascribed separately to V-r8 and V-R. V-triguan-6C and V-triguan-2C exhibit no acute mammalian cell toxicity or hemolytic activity. The vancomycin-polyguanidino dendrimer conjugates represent a promising class of broad-spectrum vancomycin derivatives and more generally illustrate how transporter modifications, in this case the distinct dendrimeric spatial array of guanidinium groups, could provide dual or polyfunctional antibiotics, establishing a fast path to clinical candidates as needed to address the urgent threat of rising antimicrobial resistance.

METHODS

Preparation of stock solutions

Stock solutions of vancomycin and vancomycin conjugates V-triguan-6C and V-triguan-2C were prepared in sterile Milli Q water. To account for differences in molecular weight based on the number of counterions associated with the prepared vancomycin conjugate samples, absorbance readings at 280 nm were used to calculate the stock solution concentrations. Absorbance readings were taken on a Themo Scientific NanoDrop One and the extinction coefficient for vancomycin published by Sigma Aldrich (5940 M⁻¹ cm⁻¹) was used to calculate concentration according to the Beer-Lambert Law. Solutions of unconjugated triguan-6C were prepared by weight in Milli Q water. All solutions were stable to be stored at 4 °C.

MIC assays

MICs were determined using broth microdilution in accordance with CLSI methods.⁸⁸ One day prior to each MIC experiment, a single colony of the appropriate bacterial strain was added to 4 mL nutrient rich medium (TSB for S. aureus strains and E. coli ATCC 25922, LB for other Gram-negative strains) and growth overnight at 37°C with 200 rpm shaking to create a stationary phase bacterial suspension. For E. faecium ATCC 51559 (VRE) and E. faecalis ATCC 51575, each overnight culture was prepared in BHI media with 4 μM vancomycin and cells were pelleted, washed in PBS and resuspended in fresh media prior to preparing the inoculum. The overnight bacterial suspension was diluted to 1 × 10⁸ CFU/mL in PBS (OD₆₀₀ of 0.1–0.4, depending on strain), and this 10⁸ CFU/mL suspension was diluted 1:100 in Mueller Hinton broth (MHB, Difco 257530) just prior to inoculating the 96- well polypropylene treatment plate (Costar 3879). A 50 μL portion of inoculum was added to a treatment plate containing 2-fold serial dilutions of compound in MHB (50 μL of treatment/well) to lend a final total volume of 100 μL/well and a final inoculum density of ~5 × 10⁵ CFU/ml. The completed assay plate was sealed with Parafilm, placed in a lidded plastic tray lined with moistened paper towels, and incubated at 37 °C for 18–24 hours. The MIC was read as the lowest treatment concentration where no bacterial growth occurred, as determined by OD₆₀₀ measurements on a microplate reader. Based on the 2-fold broth dilution method, a 2-fold variation in MIC determination is typically accepted as the standard error of the MIC measurement. Results with quality control strains, e.g. E. coli 25922 and S. aureus 29213 used here, have been selected for their consistency across laboratories and exhibit ≥ 95% consistency within +/− 1-fold dilution.⁸⁹ All MIC measurements were performed in replicate as indicated and sometimes resulted in the same value for every one. In cases where a single number is shown in the table, there was no range. When a range was observed, it is indicated.

MBEC assays

To cultivate biofilms of S. aureus ATCC 29213, 150 μL of inoculum (1 × 10⁷ CFU/mL, prepared in MHB) was added to each inner well of a 96-well base plate (Nunc 269787), and a plate lid containing 96 pegs (Nunc 445497 Immuno TSP lid) was added to the base. Outer wells were filled with PBS and were not used in screening to avoid evaporation and growth inconsistencies. The entire apparatus was sealed with paper tape and small holes were created in the tape to allow for air flow. The taped plate was placed in a sealed plastic bag lined with moistened paper towels for humidity control and grown for 48 hr at 37 °C with 150 rpm. shaking. Prior to treatment, biofilms were rinsed in a 96-well plate containing 200 μL PBS/well for 1 min and then the peg lid was transferred to a treatment plate containing 2-fold serial dilutions of compound in MHB (final treatment volume: 200 μL/well). The plate was sealed and placed in the plastic bag as above and incubated for 24 hr at 37 °C with 150 rpm shaking. After 24 hr of treatment, the peg lid was removed, rinsed twice in 200 μL PBS buffer for 1 min each, then transferred to a recovery plate containing 0.0016% resazurin dye in MHB. The plate was sealed again and placed in the plastic bag as above and incubated for an additional 24 hr at 37 °C with 150 rpm shaking. After 24 hr of growth recovery time, the MBEC was read as the lowest treatment concentration where no bacterial growth occurred, as determined visually according to the color change of the resazurin dye and confirmed by fluorescence readings on a microplate reader (resazurin excitation: 530 nm; emission: 590 nm).

For E. coli ATCC 25922 biofilms, 100 μL of inoculum (1 × 10⁷ CFU/mL, prepared in M63 minimal media with 0.08% glucose) was added to the inner wells of a sterile PVC plate. To help with growth consistency of the biofilms and evaporation, 150 μL of PBS was added to the outer row and column of each plate. The lidded plate was placed in a plastic tray and the tray was sealed with paper tape containing holes for air flow. The tray was incubated statically at 37 °C for 24 hr to establish biofilms in the wells of the plate. Upon 24 hr incubation, the growth media was carefully removed from the inner wells by pipette and 160 μL of PBS was added to rinse away any planktonic cells. The rinse was carefully removed by pipette and repeated two more times for a total of three rinses. After rinsing, 150 μL of treatment media containing 2-fold serial dilutions of compound in M63 + 0.08% glucose was added to the plate and the plate was incubated statically in the plastic tray for an additional 24 hr at 37 °C. After 24 hr of treatment, the media was removed by pipette and the wells were rinsed three times with 160 μL PBS as above. Fresh recovery media (150 μL M63 with 0.8% glucose) was added and the plate was incubated statically in the plastic tray for an additional 24 hr at 37 °C. The next day, the media was transferred to a clear, flat-bottom 96-well plate and the MBEC was determined as the lowest treatment concentration where no bacterial growth occurred, as determined by OD₆₀₀ measurements on a microplate reader.

Time-kill experiments

A stationary-phase culture of bacteria was diluted to ~5 × 10⁵ CFU/mL in MHB (OD 0.2–0.3 is approx. 1 ×10⁸ CFU/mL), and 1 mL of this suspension was mixed with 1 mL of MHB containing twice the desired final concentrations of compounds in a polypropylene overnight culture tube (Fisher 352059). Tubes were incubated at 37 °C with 200 rpm shaking. At determined time points, aliquots were removed, serially diluted in PBS, and plated on TSA to enumerate CFU/mL (detection limit: 2.3 log CFU/mL).

Propidium iodide assay

The propidium iodide (PI) assay was adapted from existing protocols in the literature.^17,37 Overnight cultures of S. aureus ATCC 29213 were grown in TSB, pelleted by centrifugation (5000 g for 1 min), washed three times in PBS, and subsequently resuspended in either PBS or HEPES−glucose buffer (5 mM HEPES, 5 mM glucose, pH 7.2) to an OD₆₀₀ of 0.3. Cultures were treated with 10 μM PI solution (from a 5 mM PI stock in DMSO), and 130 μL of bacterial suspension was added to a 96-well, black-walled, clear-bottom plate (Costar 3603). The plate was incubated 5 min at 37 °C with shaking in a fluorescent microplate reader (SpectraMax M5, Molecular Devices), and the fluorescence signal was measured (excitation 535, emission 617). Twenty microliters of compounds at desired concentrations was added (prepared in either HEPES−glucose or PBS to match bacterial buffer), and changes in fluorescence post-treatment were monitored. Lysostaphin was used as a positive control, where concentrations of lysostaphin tested were based on MICs determined in the absence of bovine serum albumin (BSA).

N-Phenylnaphthalen-1-amine (NPN) assay

Outer membrane permeabilization assays with the probe NPN were performed based on literature procedures.¹⁷ Stationary-phase cultures of E. coli ATCC 25922 were diluted 1:100 in LB and grown to OD₆₀₀ of 0.6 a at 37 °C with 200 rpm shaking. The bacteria were subsequently pelleted by centrifugation at 10,000 rpm for 5 min at room temperature and washed in HEPES/Glucose buffer (5 mM HEPES, 5 mM Glucose, pH 7.2). The bacteria were pelleted again and resuspended in buffer to achieve a bacterial density corresponding to an OD₆₀₀ of 0.625. A 5 mM solution of NPN (Sigma-Aldrich) was freshly prepared in acetone for each assay and added to the bacterial culture to achieve a concentration of 6.25 μM. 80 μL of the bacterial culture with NPN was added to a black-walled, clear bottom treatment plate (Costar 3603) and the plate was incubated for 5 min at 37°C with shaking in a fluorescent microplate reader (SpectraMax M5, Molecular Devices) while monitoring the fluorescence signal (l_excitation = 350 nm; l_emission = 420 nm). 20 μL of compounds at desired concentrations (prepared in HEPES-Glucose buffer) was then added to plate wells, yielding a final cell density corresponding to OD₆₀₀ = 0.5 and [NPN] = 5 μM, and the fluorescence intensity was monitored.

HPLC-MS

Bacterial samples were prepared for HPLC-MS analysis by diluting a stationary-phase cultures 1:500 and growing to an OD₆₀₀ of 0.6 at 37°C with 200 rpm shaking. E. coli ATCC 25922 were grown in MHB and VRE were grown in BHI medium. Each OD₆₀₀ 0.6 culture was treated with 130 μg/mL of chloramphenicol for 15 min at 37°C with 200 rpm shaking and then subsequently aliquoted (25 mL per aliquot) for treatment with compounds, where an untreated sample was included as a control. Upon 1 hr treatment, CFU/mL were enumerated on LB agar or BHI agar to determine cell density of samples, and the aliquots were centrifuged at 4°C for 20 min at 3000 rpm. The cell pellets were washed with 1 mL HEPES/Glucose buffer (5 mM HEPES, 5 mM Glucose, pH 7.2) and centrifuged at 8000 rpm for 5 min at room temperature. The supernatant was removed, and the bacteria were resuspended in 500 μL HEPES Buffer and heated at 100°C for 15 min. The samples were centrifuged at 13,000 rpm for 5 min, and the supernatant was removed and lyophilized. The lyophilized lysate was resuspended in MilliQ water to achieve a concentration of 8 mg/mL and filtered through a 0.2 μm filter. Terfenadine was used an internal standard and added to achieve a concentration of 0.3 mg/mL. Each sample was analyzed by reverse-phase HPLC (RP-HPLC) using a 15 min method on a C18 column (Zorbax Poroshell 120 SB-C18, 2.1×50mm 2.7u, PN 689775-902 with Poroshell 120 SB-C18 2.1×5mm 2.7u guard column, PN 821725–912) with a flow rate of 0.2 mL/min. The primary starting solvent was 0.1% formic acid in water, followed by a 2–95% gradient over 8 min of 0.1% formic acid in acetonitrile. The acetonitrile solvent was then reduced back to 2% and ramped up to 95% and ramped down a second time over the subsequent 7 min. Fractions corresponding to peaks of interest were collected and identified using a Waters Single Quadrupole mass spectrometer with an electrospray ionization source in negative ion mode. HPLC chromatogram intensities were normalized by the internal standard, terfenadine, with an elution time of 6.0 min.

Hemolysis assay

Hemolytic activity of the vancomycin conjugates was assessed according to a published protocol.³⁷ Rabbit red blood cells were purchased from Innovative Research Inc. (IRBRBC25ML-33785) and resuspended in PBS to yield a 1% volume/volume suspension. A 100 μL aliquot of 1% erythrocyte solution was added to the wells of a V-bottomed 96 well microtiter plate (Costar 3894). Treatments (100 μL) were added to yield the desired final compound concentrations. Blank PBS was used as a negative control and 1% Triton X-100 as a positive control. The plate was incubated statically for 1 h at 37 °C and subsequently centrifuged for 5 min at 1,500 g at room temperature. A 100 μL portion of the resulting supernatant was transferred to a flat-bottomed microtiter plate and analyzed on a microplate reader via absorbance measurements at 450 nm. Percent hemolysis was determined by dividing background-corrected absorbance measurements by background-corrected measurements for 1% Triton X-100.

Mammalian cell cytotoxicity assay

Cellular cytotoxicity of V-triguans was assessed using an MTS viability assay. Briefly, HeLa cells were plated at 5,000 cells/well in serum-containing DMEM in a 96-well plate. HeLa cells were then allowed to adhere for 24 h at 37 °C in a humidified, 5% CO₂ atmosphere. Cells were washed with 200 μL PBS buffer two times prior to treatment. In a separate 96-well plate, V-triguans were serially diluted in serum-free DMEM over a range of 160 μM to 5 μM to a final volume of 100 μL. The prepared antibiotic treatments were transferred to the plate with cells, and incubated at 37 °C for 2 h, at which point the media was removed. The cells were rinsed with 200 μL PBS two times, then 100 μL serum-containing DMEM was added to each well and the cells were incubated for an additional 18 h. 20 μL of CellTiter 96^® AQ_ueous One Solution Reagent was added to each well. After incubating for 2 hours, the absorbance at 490 nm was measured. Percent viability was determined by dividing the absorbance obtained for treated cells by that for untreated cells.