Aa

TDa

b

c

a

ARA

KARLMP

1

mcmToapifrffbi

LSf

0

0h

International Journal of Antimicrobial Agents 41 (2013) 36– 40

Contents lists available at SciVerse ScienceDirect

International Journal of Antimicrobial Agents

j our na l ho me p age: ht tp : / /www.e lsev ier .com/ locate / i jant imicag

nti-anaerobic potential of ranbezolid: insight into its mechanism of actiongainst Bacteroides fragilis

arun Mathura,1,2, Vandana Kaliaa,1,2, Tarani Kanta Barmana,2, Smita Singhala,2, Seema Khana,b,ilip J. Upadhyaya,2, Ashok Rattana,c, V. Samuel Raja,∗,2

Department of Infectious Diseases, New Drug Discovery Research, Ranbaxy Research Laboratories, R&D III, Sector 18, Gurgaon 122 015, IndiaNetaji Subhash Institute of Technology, Dwarka, New Delhi 110 078, IndiaMedanta, The Medicity, Gurgaon 122 001, Haryana, India

r t i c l e i n f o

rticle history:eceived 4 August 2012ccepted 24 September 2012

eywords:

a b s t r a c t

This study reports the anti-anaerobic properties of ranbezolid, a new investigational oxazolidinone. Atime–kill kinetics study against anaerobes showed that ranbezolid was superior to linezolid and killedthe anaerobic pathogens at 4–8 h, except for Bacteroides fragilis where killing was observed at 24 h. In addi-tion, the time–kill kinetics study showed a concentration-dependent bactericidal potential of ranbezolid

naerobesanbezolidinezolidetronidazole

rotein synthesis

against anaerobes. Ranbezolid showed 5.39 log10 reduction and linezolid showed 1.15 log10 reductionin murine disk implant infection with B. fragilis ATCC 25285. Ranbezolid was very potent and showedfast protein synthesis inhibition against B. fragilis, a Gram-negative anaerobe. In addition, non-specificcell wall synthesis inhibition was also observed with ranbezolid. The potent and fast protein synthesisinhibition along with an additional mode of action of cell wall synthesis inhibition could be responsiblefor the cidal effect of ranbezolid against Gram-negative anaerobes.

lsevie

© 2012 E

. Introduction

Anaerobic bacteria are the predominant flora of the nor-al human skin and mucous membranes and are a common

ause of endogenous bacterial infections. Anaerobic infections areore frequent in immunocompromised patients and the elderly.

he increasing resistance to antimicrobial agents among anaer-bic pathogens has been a global problem [1–3]. Treatment ofnaerobic infections is also complicated by slow growth of theseathogens. �-Lactamase production is common in Bacteroides frag-

lis group [4]. Clindamycin resistance is also common in the B.ragilis group as well as in some Clostridium spp. Metronidazoleesistance, common among anaerobic, Gram-positive, non-spore-orming rods, has also been found in the B. fragilis group [1]. The B.

ragilis group is part of the normal bowel flora containing anaero-ic pathogens and is most frequently isolated from intra-abdominal

nfections.

∗ Corresponding author. Present address: Department of Biology, Daiichi Sankyoife Science Research Centre in India, Daiichi Sankyo India Pharma Pvt. Ltd., Villagearhaul, Sector 18, Gurgaon 122 015, India. Tel.: +91 124 284 8513;ax: +91 124 239 7546.

E-mail address: vethakkani.raj.g6@dsin.co.in (V.S. Raj).1 These two authors contributed equally to this paper.2 Present address: Daiichi Sankyo India Pharma Pvt. Ltd., Sector 18, Gurgaon 122

15, India.

924-8579/$ – see front matter © 2012 Elsevier B.V. and the International Society of Chemttp://dx.doi.org/10.1016/j.ijantimicag.2012.09.013

r B.V. and the International Society of Chemotherapy. All rights reserved.

Bacteroides fragilis is the most frequent anaerobic pathogen inman (80% of anaerobic infections) and if left untreated the mortalityrate is ca. 60% [1]. The antimicrobial susceptibility patterns of anaer-obes have become less predictable owing to increasing resistance[2,3,5]. The emergence of highly virulent or multidrug-resistantstrains is challenging current therapy. Furthermore, anaerobicinfections are generally mixed with aerobic pathogens and there-fore these infections often require empirical therapy, with selectionof proper therapy becoming more complicated. To counteract theseproblems, regular resistance surveillance in anaerobes, rationalantibiotic use and evaluation of new treatment alternatives areimportant [2–4]. In this regard, we have evaluated the investi-gational oxazolidinone ranbezolid for its killing potential againstanaerobes.

Oxazolidinones are a new class of synthetic antimicrobial agentsknown for their unique activity against Gram-positive pathogens[6–8]. Linezolid, the first in class, is active against staphylococci,streptococci, enterococci and Gram-positive anaerobes, and itsactivity against Gram-negative anaerobes is moderate [9]. Ranbe-zolid, an investigational oxazolidinone, showed excellent in vitroactivity against a broad range of Gram-positive bacteria [8,10,11].Furthermore, ranbezolid is the first oxazolidinone with similar

activities observed both against Gram-positive and Gram-negativeanaerobes [7,11]. Recently, we showed the mode of action of ranbe-zolid against Staphylococcus aureus and Staphylococcus epidermidisas well as its interaction with the bacterial ribosome [12]. The aim

otherapy. All rights reserved.

l of An

otaf

2

2(

aapdStweS

2

m(AatsBapgteSwtasctei

2

EaCoiet2a

22

w[W

T. Mathur et al. / International Journa

f the present study was to examine the in vitro killing poten-ial of ranbezolid both against Gram-positive and Gram-negativenaerobes as well as to understand its mode of action against B.ragilis.

. Materials and methods

.1. Bacterial strains and minimum inhibitory concentrationsMICs)

The strains used in this study included three Gram-negativenaerobes (B. fragilis ATCC 25285, Bacteroides vulgatus ATCC 29327nd Bacteroides thetaiotaomicron ATCC 29741) and three Gram-ositive anaerobes (Clostridium perfringens ATCC 13124, Clostridiumifficile ATCC 700057 and Peptostreptococcus magnus ATCC 6569).trains were maintained on Wilkins–Chalgren agar (Difco Labora-ories, Sparks, MD). MICs of ranbezolid, linezolid and metronidazoleere determined against selected anaerobic isolates by the ref-

rence agar dilution method according to Clinical and Laboratorytandards Institute (CLSI) guidelines [13].

.2. Time–kill kinetics

The time–kill kinetics study was carried out according to theethod of Yagi and Zurenko [14] using Anaerobe Broth medium

Difco Laboratories) supplemented with 5% lysed horse blood. Thenaerobe Broth medium was pre-warmed to 37 ◦C in reaction tubesnd the drug was added at multiple concentrations ranging fromhe MIC to higher concentrations (1× to 32× MIC). A culture suspen-ion equal to 0.5–1.0 McFarland standard was prepared in Anaeroberoth MIC medium and was added to each reaction tube to give

final inoculum of ca. 1 × 106 CFU/mL. The reaction tubes werelaced at 37 ◦C under anaerobic conditions. One antibiotic-freerowth control was used in each experiment. Exposure of reac-ion tubes to the aerobic environment was kept minimal and thexperiment was performed in an anaerobic chamber (Bactron IV;heldon Manufacturing Inc., Cornelius, OR). Viable counts (CFU/mL)ere performed at 0, 4, 8, 24, 30 and 48 h by plating serial dilu-

ions of suspensions and incubating the plates for 48 h at 37 ◦C inn anaerobic jar under anaerobic conditions using an AnoxomatTM

ystem (Mart Microbiology, Drachten, The Netherlands). Antimi-robials were considered bactericidal at the lowest concentrationhat reduced the original inoculum by ≥3 log10 CFU/mL (99.9%) atach of the time periods, and they were considered bacteriostaticf the inoculum was reduced by 0 to <3 log10 CFU/mL.

.3. Experimental animals and ethical approval

All animal studies were approved by the Institutional Animalthics Committee of Ranbaxy Research Laboratory (Gurgaon, India)ccording to the guidelines of the Committee for the Purpose ofontrol and Supervision of Experiments on Animals (Governmentf India, New Delhi, India). Swiss albino mice of either sex weigh-ng 20 ± 2 g were used for this experiment. Six mice were used inach group. All animals were housed under the same environmen-al conditions with a 12 h:12 h dark:light cycle at a temperature of0–25 ◦C at 40–70% relative humidity and were provided with feednd water ad libitum.

.4. Murine foreign-body infection with Bacteroides fragilis ATCC5285

Murine foreign-body infection with B. fragilis ATCC 25285as established as per the protocols described by Slee et al.

15] with slight modification. Briefly, B. fragilis was grown onilkins–Chalgren agar medium for 48 h and the saline-suspended

timicrobial Agents 41 (2013) 36– 40 37

culture was adjusted to 2 McFarland standard, which was furtherdiluted to 1:10. Sterile paper disks (Becton Dickinson, Sparks, MD)were used for the infection. Disks were inoculated with 20 �L ofB. fragilis having an inoculum of 5 × 106 CFU/disk per mouse. Oneday prior to the experiment, the dorsolateral surface of the micewas shaved off. After 24 h, mice were anaesthetised with isofluraneand the disk impregnated with bacterial suspension was implantedsubcutaneously with the help of a trocar and a cannula. Ranbezolidand linezolid treatment were started 24 h post challenge at a doseof 100 mg/kg body weight twice daily by the oral route and werecontinued for 5 days. At the end of the experiment, animals weresacrificed and disks were removed aseptically and place in 1 mLof broth, vigorously vortexed and plated on agar media to assessthe bacterial counts. In vivo data were analysed using GraphPadPrism 4.1 (GraphPad Software Inc., La Jolla, CA) and a Kaplan–Meiersurvival curve was prepared.

2.5. Macromolecular synthesis inhibition studies

Macromolecular biosynthesis inhibition in B. fragilis was studiedas described by Kalia et al. [12]. The following radiolabelled pre-cursors were used: protein, [14C]isoleucine; DNA, [3H]thymidine;RNA, [3H]uridine; cell wall, [3H]N-acetyl glucosamine; and fattyacid, [14C]acetate. In brief, B. fragilis was grown in AnaerobeBroth medium supplemented with 5% lysed horse blood, andradioactive precursors (1 �Ci/mL for 3H-labelled and 0.1 �Ci/mLfor 14C-labelled compounds) were added during the early loga-rithmic phase (optical density at 600 nm of 0.3). After 5 min, theinhibitors were added (at 2× MIC) and the cells were harvestedat 1, 2, 4 and 6 h. Macromolecules were precipitated with ice-coldtrichloroacetic acid [final concentration of 5% (w/v)] and were fil-tered on glass fibre filters (1.0 �M A/B glass multiwell filter plates;Pall Corporation, Ann Arbor, MI). Plates were dried overnight at37 ◦C and quantification of radioactivity was done after addition ofscintillation fluid. Counting was performed in a Wallac scintillationcounter (Perkin Elmer, Waltham, MA).

3. Results

3.1. In vitro activity of ranbezolid against anaerobes

The MICs of ranbezolid against the anaerobic strains of B. frag-ilis, B. vulgatus, B. thetaiotaomicron, C. perfringens, C. difficile andP. magnus were 0.06, 0.015, 0.06, 0.06, 0.03 and 0.015 �g/mL,respectively, and were significantly lower than those for linezolidand metronidazole (32–128-fold for linezolid and 8–32-fold formetronidazole). The MICs against these strains were 2 �g/mL forlinezolid and in the range of 0.25–0.5 �g/mL for metronidazole(Table 1).

3.2. Time–kill kinetics study

The time–kill kinetics study revealed the bactericidal activ-ity of ranbezolid both against Gram-positive and Gram-negativeanaerobes (Fig. 1). In the time–kill studies, ranbezolid exhib-ited concentration-dependent bactericidal activity (99.9% killing)against C. perfringens, P. magnus and B. vulgatus at drug exposuresof 0.06–0.25 �g/mL (4× MIC) even after 4–8 h of exposure, whereasagainst B. fragilis bactericidal potential was seen at 1 �g/mL (16×MIC) in 24 h. Linezolid also exhibited ≥3.0 log10 CFU/mL reductions

against C. perfringens, P. magnus and B. vulgatus but only at thehigher concentration of 8 �g/mL (4× MIC) and only after 24–48 h ofdrug exposure, whereas bacteriostatic activity was found against B.fragilis at 8× MIC. The other standard drug, metronidazole, showed

38 T. Mathur et al. / International Journal of Antimicrobial Agents 41 (2013) 36– 40

Table 1Minimum inhibitory concentrations (MICs) of antimicrobials against selectedstrains.

Organism MIC (�g/mL)

Ranbezolid Linezolid Metronidazole

Bacteroides fragilisATCC 25285

0.06 2 0.5

Bacteroides vulgatusATCC 29327

0.015 2 0.25

BacteroidesthetaiotaomicronATCC 29741

0.06 2 0.5

Clostridium perfringensATCC 13124

0.06 2 0.5

Clostridium difficileATCC 700057

0.03 2 0.5

br

3

i

Untreated animal Ranbezolid LZD

1

2

3

4

5

6

7

8

9

10

Limit of detection

Lo

g1

0C

FU

/ml

F1�0�

Peptostreptococcusmagnus ATCC 6569

0.015 2 0. 5

actericidal activity at 1 �g/mL (2× MIC) and ≥3.0 log10 CFU/mLeduction was observed within 8 h against all the strains.

.3. In vivo efficacy of ranbezolid against Bacteroides fragilis

Ranbezolid showed excellent in vivo activity against B. frag-lis (Fig. 2). The lower limit of detection was 2 log10 CFU/mL. The

0 10 20 30 40 50

0

2

4

6

8

10

12a

b

Time (hours)

Lo

g1

0C

FU

/ml

0 10 20 30 40 50

0

2

4

6

8

10

Time (hours)

Lo

g1

0C

FU

/ml

ig. 1. Time–kill kinetics against (a) Bacteroides fragilis ATCC 25285 (©, control; �, ran6 �g/mL; �, metronidazole 1 �g/mL), (b) Bacteroides vulgatus ATCC 29327 (©, control;, ranbezolid 0.25 �g/mL; �, linezolid 8 �g/mL; �, metronidazole 0.5 �g/mL), (c) Clostrid.25 �g/mL; �, ranbezolid 0.5 �g/mL; �, linezolid 8 �g/mL; �, metronidazole 1 �g/mL) an, ranbezolid 0.125 �g/mL; �, ranbezolid 0.25 �g/mL; �, linezolid 8 �g/mL; �, metronida

Fig. 2. In vivo activity of ranbezolid (�) and linezolid (LZD) (�) against Bacteroidesfragilis ATCC 25285. Ranbezolid showed 5.39 log10 reduction and linezolid showed1.15 log10 reduction in murine disk implant infection with B. fragilis ATCC 25285.

initial inoculum of B. fragilis was 6.70 log10 CFU/disk per mouse.The mean ± standard deviation bacterial load in untreated controlmice was 8.86 ± 0.19 log10 CFU/mL and ranbezolid-treated miceshowed a bacterial load of 3.47 ± 1.26 log10 CFU/mL. Ranbezolid-

treated mice displayed 5.39 log10 reductions, whereas linezolidshowed 1.15 log10 reductions in the murine foreign-body infectionmodel with B. fragilis ATCC 25285 compared with untreated controlanimals.

c

d

0 10 20 30 40 50

0

2

4

6

8

10

12

Time (hours)

Lo

g1

0C

FU

/ml

0 10 20 30 40 50

0

2

4

6

8

10

Time (hours)

Lo

g1

0C

FU

/ml

bezolid 0.25 �g/mL; �, ranbezolid 1 �g/mL; �, ranbezolid 2 �g/mL; �, linezolid �, ranbezolid 0.03 �g/mL; �, ranbezolid 0.06 �g/mL; �, ranbezolid 0.125 �g/mL;ium perfringens ATCC 13124 (©, control; �, ranbezolid 0.125 �g/mL; �, ranbezolidd (d) Peptostreptococcus magnus ATCC 6569 (©, control; �, ranbezolid 0.06 �g/mL;zole 1 �g/mL).

T. Mathur et al. / International Journal of An

Fig. 3. Protein synthesis inhibition by ranbezolid and linezolid against Bacteroidesfbl

3

islopriwzBGs6ssw

4

asiezGtawf[sno(5

itatsgbsb

[

ragilis ATCC 25285. The drug concentration used for macromolecular synthesis inhi-ition of B. fragilis ATCC 25285 was 0.125 �g/mL for ranbezolid and 4 �g/mL for

inezolid (i.e. 2× the minimum inhibitory concentration).

.4. Inhibition of macromolecular synthesis by ranbezolid

Since ranbezolid showed bactericidal activity and excellentn vivo activity against Gram-negative anaerobes, macromolecularynthesis inhibition in B. fragilis by ranbezolid and compared withinezolid was studied. Ranbezolid was found to be a potent inhibitorf protein synthesis in B. fragilis and showed significantly higherotency (>32 times) (Fig. 3) at concentrations of 0.125 �g/mL foranbezolid and 4 �g/mL for linezolid (i.e. 2× MIC). Protein synthesisnhibition in B. fragilis by ranbezolid was observed at 2 h compared

ith linezolid that inhibited protein synthesis only at 6 h. Ranbe-olid was found to be a fast and potent protein synthesis inhibitor of. fragilis, which could be the reason why it was active against theram-negative anaerobes. Nucleic acid (DNA and RNA) and lipidynthesis were not inhibited by either ranbezolid or linezolid. At

h there was a non-specific or additional inhibition of cell wallynthesis by ranbezolid (ca. 40%; data not shown). Linezolid alsohowed a slight inhibitory effect on cell wall synthesis at 6 h but itas not very significant.

. Discussion

Oxazolidinones are known for their excellent in vitro activitygainst Gram-positive pathogens [6,9,11]. Ranbezolid is a uniquemall molecule in the oxazolidinone class that showed excellentn vitro activity against all groups of anaerobes as reported by Edniet al. [7]. In accordance with Ednie et al., in the current study ranbe-olid also showed in vitro activity both against Gram-positive andram-negative anaerobes (Table 1). In the time–kill kinetics study,

he bactericidal activity of linezolid against the Gram-negativenaerobe B. vulgatus at 24 h at a higher concentration (8 �g/mL)as observed. Similar bactericidal activity of linezolid against a

ew Gram-negative anaerobes was reported by Yagi and Zurenko14]. To our knowledge, ranbezolid is the first oxazolidinone thathowed in vitro and in vivo bactericidal potential against Gram-egative anaerobes. Ford et al. reported that the investigationalxazolidinone U-100766 was active against B. fragilis UC12199MIC = 4 �g/mL) in a murine soft-tissue infection model, with the0% effective dose (ED50) being 46.3 mg/kg [16].

Since the time–kill kinetic potential as well as the in vitro andn vivo efficacy of ranbezolid were better than that of linezolid,he mechanism of action of ranbezolid against the Gram-negativenaerobe B. fragilis was studied. Ranbezolid inhibited protein syn-hesis in 2 h, which is significantly very fast compared with itslow growth rate. Thus, the fast arrest of protein synthesis in slow-

rowing anaerobes could be one of the reasons for effective killingy ranbezolid. Oxazolidinones are protein synthesis inhibitors andtudies on the mechanism of action of ranbezolid showed inhi-ition of protein synthesis, whereas nucleic acid synthesis was

[[

timicrobial Agents 41 (2013) 36– 40 39

not affected [12]. No effects of ranbezolid or linezolid on DNA,RNA and lipid synthesis were observed, which is in agreementwith earlier reports on oxazolidinones [9,12]. In our understand-ing, this is the first oxazolidinone with excellent in vitro activityagainst Gram-negative anaerobes with a defined mode of action.Recently, we showed that ranbezolid is structurally similar to line-zolid, whose piperazine ring makes van der Waals contacts with thesugar residues of A2451, C2452 and U2506 in the 50S ribosome [12].The various conformations of ranbezolid generated through dock-ing studies showed the interaction between the nitrofuran moietywith additional residues of the 50S ribosome at G2505, U2584 andG2583, and this additional interaction could be one of the reasonsfor its potent protein synthesis inhibition [12].

The present study as well as that of Ednie et al. [7] shows thatranbezolid is an excellent inhibitor both against Gram-positiveand Gram-negative anaerobes. Our previous studies showed thatranbezolid is a good inhibitor with a defined mode of actionagainst Staphylococcus spp. [8,11,12]. Considering the broad spec-trum of in vitro activity of ranbezolid both against Gram-negativeanaerobes and Staphylococcus spp., it can be used against mixedinfections of Gram-negative anaerobes with Staphylococcus. Thus,ranbezolid has an additional advantage over linezolid in mixedinfections of anaerobes.

Acknowledgment

The authors thank Dr Pradip Bhatnagar for critical review of thisarticle.

Funding: Financial support for this research from RanbaxyResearch Laboratories (Gurgaon, India) is acknowledged.

Competing interests: None declared.Ethical approval: All animal studies were approved by the Insti-

tutional Animal Ethics Committee of Ranbaxy Research Laboratory(Gurgaon, India) according to the guidelines of the Committee forthe Purpose of Control and Supervision of Experiments on Animals(Government of India, New Delhi, India).

References

[1] Boyanova L, Kolarov R, Mitov I. Antimicrobial resistance and the managementof anaerobic infections. Expert Rev Anti Infect Ther 2007;5:685–701.

[2] Jamal W, Shahin M, Rotimi VO. Surveillance and trends of antimicrobial resis-tance among clinical isolates of anaerobes in Kuwait hospitals from 2002 to2007. Anaerobe 2010;16:1–5.

[3] Roh KH, Kim S, Kim CK, Yum JH, Kim MS, Yong D, et al. Resistance trends ofBacteroides fragilis group over an 8-year period, 1997–2004, in Korea. Korean JLab Med 2009;29:293–8.

[4] Citron DM, Tyrrell KL, Merriam V, Goldstein EJ. In vitro activity ofceftazidime–NXL104 against 396 strains of �-lactamase-producing anaerobes.Antimicrob Agents Chemother 2011;55:3616–20.

[5] Liu CY, Huang YT, Liao CH, Yen LC, Lin HY, Hsueh PR. Increasing trends in antimi-crobial resistance among clinically important anaerobes and Bacteroides fragilisisolates causing nosocomial infections: emerging resistance to carbapenems.Antimicrob Agents Chemother 2008;52:3161–8.

[6] Ednie LM, Jacobs MR, Appelbaum PC. Anti-anaerobic activity of AZD2563, anew oxazolidinone, compared with eight other agents. J Antimicrob Chemother2003;50:101–5.

[7] Ednie LM, Rattan A, Jacobs MR, Appelbaum PC. Antianaerobe activity of RBX7644 (ranbezolid), a new oxazolidinone, compared with those of eight otheragents. Antimicrob Agents Chemother 2003;47:1143–7.

[8] Hoellman DB, Lin G, Ednie LM, Rattan A, Jacobs MR, Appelbaum PC. Anti-pneumococcal and antistaphylococcal activities of ranbezolid (RBX 7644), anew oxazolidinone, compared to those of other agents. Antimicrob AgentsChemother 2003;47:1148–50.

[9] Moellering RC. Linezolid: the first oxazolidinone antimicrobial. Ann Intern Med2003;138:135–42.

10] Das B, Rudra S, Yadav A, Ray A, Rao AV, Srinivas AS, et al. Synthesis andSAR of novel oxazolidinones: discovery of ranbezolid. Bioorg Med Chem Lett

2005;15:4261–7.

11] Rattan A. RBx-7644. Drugs Fut 2003;28:1070–7.12] Kalia V, Miglani R, Purnapatre KP, Mathur T, Singhal S, Khan S, et al. Mode of

action of ranbezolid against staphylococci and structural modeling studies of itsinteraction with ribosomes. Antimicrob Agents Chemother 2009;53:1427–33.

4 l of An

[

[

[

1987;31:1791–7.

0 T. Mathur et al. / International Journa

13] Clinical and Laboratory Standards Institute. Methods for antimicrobial suscep-tibility testing of anaerobic bacteria; approved standard. 8th ed. Document

M11-A8. Wayne, PA: CLSI; 2012.

14] Yagi BH, Zurenko GE. An in vitro time–kill assessment of linezolid and anaerobicbacteria. Anaerobe 2003;9:1–3.

15] Slee AM, Wuonola MA, McRipley RJ, Zajac I, Zawada MJ, Bartholomew PT,et al. Oxazolidinones, a new class of synthetic antibacterial agents: in vitro

[

timicrobial Agents 41 (2013) 36– 40

and in vivo activities of DuP 105 and DuP 721. Antimicrob Agents Chemother

16] Ford CW, Hamel JC, Wilson DM, Moerman JK, Stapert D, Yamcey Jr RJ, et al.In vivo activities of U-100592 and U-100766, novel oxazolidinone antimi-crobial agents, against experimental bacterial infections. Antimicrob AgentsChemother 1996;40:1508–13.