In Vivo Targeting of Escherichia coli with Vancomycin-Arginine

The ability of vancomycin-arginine (V-r) to extend the spectrum of activity of glycopeptides to Gram-negative bacteria was investigated. Its MIC toward Escherichia coli, including β-lactamase expressing Ambler classes A, B, and D, was 8 to 16 μg/ml.

MICs were determined in alignment with CLSI guidelines as previously described for V-R and cationic antimicrobial peptides (14,16). The MIC range of V-r against 29 different E. coli strains was 8 to 16 mg/ml (MIC 90 , 16mg/ml), including those with multiple resistance mechanisms ( Table 2). The MIC of V-r against the efflux pump mutant strain JW0451-2 was 8 mg/ml, suggesting that V-r is unlikely to be a substrate for efflux in this pathogen. Notably, the MIC of V-r was also 8 mg/ml against two out of five of the Acinetobacter baumannii strains tested. In comparison, the MICs of vancomycin were significantly higher, at 64 to 256 mg/ml, against all E. coli and A. baumannii strains tested. Importantly, the antimicrobial potency of V-r towards a number of Gram-positive bacteria remained intact ( Table 2). In frequency-of-resistance (FoR) assays at 8 times the MIC of V-r (128 mg/ml), E. coli ATCC 25922 demonstrated an extremely low FoR, at ,2.32 Â 10 210 , which is similar to or lower than those with standard-of-care therapies, such as ciprofloxacin (17,18). Time-kill assays were performed against uropathogenic E. coli strains, including the sequence type 131 (ST131) NCTC 13341 isolate. V-r, but not vancomycin, demonstrated rapid bactericidal activity to limits of detection (i.e., 100 CFU/ml) within 1 or 4 h of exposure, and this was maintained up to 24 h (Fig. 2).
Plasma pharmacokinetics (PK) of V-r after subcutaneous (s.c.) administration (20 and 121 mg/kg) was determined in naive male CD-1 mice (n = 3/group) using liquid chromatography-tandem mass spectrometry for analysis with a lower limit of quantitation of 5 ng/ml (Table 3). V-r displayed first-order elimination, similar to vancomycin, after s.c. administration (19,20). Prior to efficacy studies, a single s.c. administration of V-r  was shown to be well tolerated in male CD-1 mice (n = 3) at the highest dose tested (800 mg/kg). Using a screening-based strategy, preliminary proof-of-concept studies with V-r employed an abbreviated 9-h thigh muscle infection model in male CD-1 mice rendered neutropenic (21). To that end, an E. coli ATCC 25922 isolate was inoculated at 9.7 Â 10 4 CFU into both thigh muscles per mouse (n = 5 per experimental group). V-r was administered s.c. every 2 h (110 to 880 mg/kg total dose) starting 1 h postinfection. At 9 h, thigh homogenates were prepared, and CFU were enumerated after culture on CLED (cystine-, lactose-, and electrolyte-deficient) agar. Compared to pretreatment and  vehicle burdens of 5.1 6 0.2 and 7.1 6 0.1 log 10 CFU/g tissue, respectively, V-r exhibited a dose-dependent reduction in bacterial burden of 1.2 to 3.4 log 10 compared with vehicle (Kruskal-Wallis one-way analysis of variance using StatsDirect Statistical Analysis Software) ( Table 4). V-r doses at 440 and 880 mg/kg afforded 1.0-and 1.3-log 10 reductions below stasis, respectively, with an extrapolated static dose of 215 mg/kg. As anticipated, vancomycin failed to significantly impact E. coli burden at a dose equivalent to the highest dose of V-r. In a 24-h thigh muscle infection model, E. coli UTI89 was inoculated at 7.8 Â 10 4 CFU into one thigh muscle per mouse (n = 5 to 8 per group) and treated with V-r (total dose, 200 to 1,400 mg) using an every-6-h dosing regimen from 1 h postinfection. All doses of .200 mg/kg significantly reduced burden below stasis by up to 2.7 log 10 CFU/g. These bactericidal effects of V-r were statistically superior to those of ciprofloxacin, which induced a 1.4 log 10 reduction from stasis ( Fig. 3 and Table 5). Overall, V-r caused an ;4 to 7.5 log 10 reduction in bacterial burden, compared with vehicle control, over the entire dose range. The MIC data confirm previous findings that the coupling of arginine with vancomycin bestows significant antimicrobial activity of the V-r conjugate against E. coli infection while remaining effective against methicillin-resistant Staphylococcus aureus (MRSA) (14). Such in vitro findings were effectively translated into thigh muscle infection models, where a total 24-h dose of 250 mg/kg V-r reduced E. coli burden to pretreatment (stasis) levels. Since area under the curve over 24 h in the steady state divided by the MIC (AUC/MIC ratio) is the primary PK/pharmacodynamic predictor of vancomycin (5), this static dose corresponds to a total AUC/MIC of 47.3. Based on a free (f) fraction of 35%, as determined in plasma protein binding studies (Table 1), the fAUC/MIC of V-r was 16.5. As an approximation of exposure using allometric scaling (22), this would be equivalent to a human dose of ;20 mg/kg, with a dose of 28 mg/kg  required to elicit an additional 1-log 10 kill. Such allometric doses of V-r are in line with the daily and loading doses of vancomycin in humans (5). The positive efficacy data support the notion that the cationic feature of arginine within V-r allows for breaching of the stubborn outer membrane of E. coli isolates and possibly other Gram-negative bacteria (14). The sequelae of events leading to V-rmediated E. coli eradication likely involve (i) improved cell surface association with negatively charged groups, (ii) effective translocation across the outer membrane leading to enhanced drug uptake, and (iii) disruption of peptidoglycan synthesis within the periplasmic space (6,14). To our knowledge, the current findings describe the first report of a marked abrogation of E. coli burden in vivo with a minimally modified vancomycin-cationic transporter conjugate. Previously, it was reported that vancomycin-QC14, a strongly lipophilic/cationic molecule, reduced thigh muscle infection of a carbapenem-resistant A. baumannii strain (23). Because V-r was highly effective in time-kill assays against E. coli NCTC 13441, a pandemic uropathogenic clone (24), a logical next step would be to evaluate the conjugate in a model of urinary tract infection (UTI). Based on the high renal elimination of vancomycin in humans (25) in a nonmetabolized form (26), it is reasonable to hypothesize that V-r may drive a highly targeted therapeutic intervention to combat E. coli-associated UTIs.
These data further underscore a precedent for creating a novel Gram-negative active agent by transforming a commonly used and selective Gram-positive antibiotic by introducing certain cationic features through a simple and scalable synthesis protocol (14). Such an approach, in consort with effective in silico predictions (27,28), might expedite antibiotic development and increase the overall probability of success of

ACKNOWLEDGMENTS
We are grateful to Patricia A. Bradford (Antimicrobial Specialists LLC, USA) for critically reviewing the manuscript. We also thank D. Corbett and J. Gould (Evotec, UK), D. Turner (Cyprotex, UK), Y. Huang, M. Gassen, and J. Li (WuXi AppTec, China) and their respective teams for support with synthesis, in vitro, and animal studies described herein.
All studies described in this report were financed solely by SuperTrans Medical Ltd. (Israel).
Stanford University has filed patent applications on this and related technology, supported in part by the National Institutes of Health grants R01GM117278 (L.C.) and NIH-CA031845 and NSF CHE-1566423 (P.A.W.), which has been licensed by SuperTrans Medical for the treatment of bacterial infectious diseases. P.A.W. and L.C. serve as consultants to the company.