Anthrax lethal factor inhibition

Abstract

The primary virulence factor of Bacillus anthracis is a secreted zinc-dependent metalloprotease toxin known as lethal factor (LF) that is lethal to the host through disruption of signaling pathways, cell destruction, and circulatory shock. Inhibition of this proteolytic-based LF toxemia could be expected to provide therapeutic value in combination with an antibiotic during and immediately after an active anthrax infection. Herein is shown the crystal structure of an intimate complex between a hydroxamate, (2R)-2-[(4-fluoro-3-methylphenyl)sulfonylamino]-N-hydroxy-2-(tetrahydro-2H-pyran-4-yl)acetamide, and LF at the LF-active site. Most importantly, this molecular interaction between the hydroxamate and the LF active site resulted in (i) inhibited LF protease activity in an enzyme assay and protected macrophages against recombinant LF and protective antigen in a cell-based assay, (ii) 100% protection in a lethal mouse toxemia model against recombinant LF and protective antigen, (iii) ≈50% survival advantage to mice given a lethal challenge of B. anthracis Sterne vegetative cells and to rabbits given a lethal challenge of B. anthracis Ames spores and doubled the mean time to death in those that died in both species, and (iv) 100% protection against B. anthracis spore challenge when used in combination therapy with ciprofloxacin in a rabbit “point of no return” model for which ciprofloxacin alone provided 50% protection. These results indicate that a small molecule, hydroxamate LF inhibitor, as revealed herein, can ameliorate the toxemia characteristic of an active B. anthracis infection and could be a vital adjunct to our ability to combat anthrax.

Bacillus anthracis, the etiological agent of anthrax, has been developed as a bioweapon by countries and terrorists largely because of a combination of the spore's durability and the lethal toxemia of the vegetative stage. This Gram-positive bacterium forms spores resistant to adverse environmental conditions and can survive for decades in pastures (1). If ingested or inhaled, even in small numbers, the spores germinate to establish explosive vegetative growth and a resulting toxemia that is usually fatal to the host (2–4). The primary virulence factor is a secreted zinc-dependent metalloprotease toxin known as lethal factor (LF), which is lethal to the host through disruption of signaling pathways, cell destruction, and circulatory shock. The only existing therapeutic intervention for naturally acquired or weaponized anthrax is antibiotic treatment that must be given early after infection and at a time when victims may experience only mild flu-like symptoms (5–9). Delay of treatment, even by hours, substantially reduces survival of infected patients (1, 5). To date, physicians have antibiotic options to eliminate an anthrax infection, but they have no therapeutic options to combat the LF-mediated toxemia and tissue destruction during an ongoing infection or the residual toxemia that persists even after the bacteria have been eliminated by antibiotics.

It is envisaged that, depending on how and when administered, either an LF inhibitor (LFI) could block the proteolytic protection provided by LF in the macrophage and allow that cell to eliminate spores early in infection (which could be used prophylactically if intentional release of anthrax were suspected) or, more probably, an LFI would be used to block late stage effects of LF during an active infection and increase the probability of host survival. This latter aspect would unquestionably be used in adjunct therapy with an antibiotic. Herein, we reveal the crystal structure of a hydroxamate LFI and its intimate interaction with LF and present a sequence of in vitro and in vivo studies, including those with active B. anthracis Ames strain infections, that indicate this interaction has dramatic protective benefits.

Methods

LFI and Recombinant Toxins. The hydroxamate LFI, (2R)-2-[(4-f luoro-3-methylphenyl)sulfonylamino]-N-hydroxy-2-(tetrahydro-2H-pyran-4-yl)acetamide, was used in all studies herein and was synthesized at Merck Research Laboratories (Rahway, NJ). Recombinant LF was purified from Escherichia coli (R. J. Collier, Harvard Medical School, Cambridge, MA) and compared with LF isolated from B. anthracis (S. Leppla, National Institutes of Health, Bethesda, MD). LFI showed identical inhibition versus LF isolated from either source. Recombinant protective antigen (PA) was purified from E. coli (R. J. Collier).

N-Terminally Truncated LF. Forward and reverse PCR primers (5′GGATCCAGGCATGCTGTCAAGATATGAAAAATGGGAAAAG-3′ and 5′-GGATCCTTGCTGCCGCGGGGCACCAGTGAGTTAATAATGAACTTAATCTGA-3′, respectively) were designed to remove a stop codon, add a 3′ thrombin site, and add BamHI restriction sites to the DNA sequence encoding amino acids 264–776 of LF. The PCR product was amplified from pET15b-LF (10) and cloned into pET23+ (Novagen). To add a GST tag to the LF C-terminal coding region, a double-stranded adapter formed by annealing two oligonucleotides (5′-GATCTAAGGATCCGC-3′ and 5′-GGCCGCGGATCCTTA-3′) was inserted between the BamHI and NotI sites of vector pGEX-4T-3 (Amersham Pharmacia), and the resulting vector was linearized with BamHI before the LF BamHI fragment from the pET23+ construct was inserted. This plasmid encodes a GST–LF (264–776) fusion protein with thrombin cleavage sites at the GST–LF junction and the LF C terminus.

X-Ray Crystallography. Crystals of the truncated LF:LFI complex were obtained by the vapor diffusion method in hanging drops with 20–22% polyethylene glycol 8000/100 mM Mg(OAc)₂/100 mM sodium cacodylate, pH 6.8, as precipitant. Crystals were orthorhombic, with unit cell parameters a = 57.3 Å, b = 75.96 Å, and c = 139.0 Å. Data were collected on an ADSC Q210 charge-coupled device detector at beamline 17-ID in the facilities of the Industrial Macromolecular Crystallography Association-Collaborative Access Team at the Advanced Photon Source (Argonne National Laboratory, Argonne, IL) from a crystal that was flash frozen in a liquid nitrogen stream. The cryoprotectant was 25% ethylene glycol in mother liquor. Data were processed with hkl-2000 (11). The resulting data set was 98.9% complete and 7-fold redundant to 2.3 Å, with an average I/σI of 10.8. The structure was solved by molecular replacement with molrep (12) and the coordinates 1J7N.pdb (Protein Data Bank ID code 1J7N). The refinement was conducted by alternating computer-based refinement (13) and manual rebuilding of the model in o (14). The final model had a crystallographic R factor of 19.1% (R_free = 26.9%) and good geometry (rms deviations for bond length and bond angles were 0.011 Å and 1.24°, respectively). The coordinates were deposited in the Protein Data Bank under ID code 1YQY.

LF Protease Enzyme and Macrophage Cytotoxicity Assays. LF protease activity was determined with a fluorogenic peptide slightly modified from earlier work (ref. 10; see also Supporting Materials and Methods, which is published as supporting information on the PNAS web site). Murine J774A.1 macrophage cells (American Type Culture Collection) were used in the cytotoxicity assays. J774A.1 cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% FCS and incubated with LFIs for 1 h at 37°C. LF and PA were then added to make a final concentration of 15 ng/ml and 250 ng/ml, respectively. After 4 h of incubation at 37°C, 3-(4,5-dimethylthiazol-2-yl)-5–3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (Promega) was added, and, after an additional 2-h incubation at 37°C, the OD₄₉₀ was measured.

Mouse Toxemia Model with Recombinant PA and LF. BALB/c mice (weight, 22 g; age, 8 wk) were purchased from Taconic Farms. For the definitive toxemia test, 32 mice were allocated at random to one of four groups and each of the four groups was then allocated at random to treatment with LFI at 0, 1, 10, or 30 mg/kg three times a day (t.i.d.) (–0.25, +1, and + 3 h relative to administration of LF and PA), with saline as the vehicle. At time 0, all mice were coinjected i.v. with 100 μg of recombinant LF and 100 μg of recombinant PA in a 150-μl saline mixture into the mouse tail vein.

Monotherapeutic LFI Protection of Mice Infected with B. anthracis Sterne Strain Vegetative Cells. Jugular vein-cannulated BALB/c mice (weight, 22 g; age, 8 wk) were purchased from Taconic Farms and used in protection trials versus B. anthracis Sterne vegetative cells. A Harvard Apparatus pump-controlled syringe continuously infused a precise rate of LFI or saline through 25-gauge polyethylene tubing to each mouse by means of a counterbalanced lever arm and swivel (Instech Solomon, Plymouth Meeting, PA).

The acapsular B. anthracis Sterne strain was obtained from Hank Heine (United States Army Medical Research Institute of Infectious Diseases, Fort Detrick, MD) and maintained at –70°C. The frozen material was thawed and a loop of material was streaked on a sheep red-blood agar plate and incubated for 18 h at 35°C. Immediately before challenge, the 18-h colonies were suspended in saline, vortex mixed vigorously, and diluted to achieve an inoculum previously shown to give 10⁸ colonyforming units per 0.2 ml. To verify the inoculum, 100 μl from each of the dilution tubes was cultured on sheep red-blood agar plates, and the colonies were counted 24 h later.

Two mouse tests evaluating the monotherapy of LFI against B. anthracis Sterne strain vegetative cells were conducted. Each consisted of one group of 10 BALB/c mice infused with saline at 100 μl/h and a second group of 10 mice infused with LFI in saline at 250 μg·100 μl^–1·h^–1. The mice were challenged i.p. with 10⁸ colony-forming units of B. anthracis Sterne strain 24 h after infusion began, and continuous infusion of LFI or saline was maintained for 9 days.

Monotherapeutic LFI Protection of Rabbits Infected with B. anthracis Ames Spores. Dutch-belted (DB) rabbits (weight, 2 kg; age, 16 wk) were purchased from Covance (Princeton, NJ) and B. anthracis Ames strain spores were obtained from R. Lyons (University of New Mexico, Albuquerque). The initial efficacy test consisted of six DB rabbits dosed s.c. with LFI at 100 mg/kg t.i.d. in saline for 7 days and six rabbits dosed s.c. with saline at the same times. Two hours after the first dose, all rabbits were challenged s.c. with 10⁴ Ames spores and observed for 21 days.

A second study was conducted to confirm and extend the previous monotherapy finding in the DB rabbit. As in the first trial, a s.c. injection of 10⁴ spores of B. anthracis Ames was used to challenge 12 DB rabbits. LFI monotherapy was delivered at 100 mg/kg s.c. t.i.d. in saline for 7 days starting at the time of spore challenge to the first group (n = 4) and for 6 days starting 24 h after challenge to a second group (n = 4). A third group served as a saline-treated control (n = 4).

Ciprofloxacin and LFI Combination Therapy of Rabbits Challenged with B. anthracis Ames Spores. Ciprofloxacin (Cipro) is a first-line drug in anthrax therapy, and a “point of no return” model was developed in DB rabbits to evaluate whether combination of Cipro with LFI would provide additional useful therapy beyond what Cipro would do alone. Probe studies with Cipro monotherapy at twice daily (b.i.d.) 5 mg/kg s.c. administration to rabbit groups (n = 4) challenged with 10⁴ spores of B. anthracis Ames were made stepwise starting at 24 h after challenge and continued through 36, 48, 54, 60, and 66 h after challenge until the point at which 50% of the Cipro-treated rabbits died was determined. Once that determination was made, a test was done in 11 DB rabbits to evaluate whether the LFI and Cipro combination would provide additional protection beyond Cipro alone. All rabbits were challenged s.c. with 10⁴ Ames spores at time 0. At the 50% point of no return, the surviving rabbits were divided into two equal experimental groups of four rabbits each, and the remainder was allocated to a third group given saline only. Of the two experimental groups, the first received Cipro monotherapy at 5 mg/kg s.c. b.i.d. for 2 days starting at the 50% point of no return, and the second group received combination therapy with Cipro by following the same protocol as group 1 on one side of the body and LFI at 100 mg/kg s.c. four times a day for 1 day also starting at the 50% point of no return on the other side of the body.

Animal Welfare and Husbandry. All animal experiments and housing were approved and conducted according to guidelines from Merck's Institutional Animal Care and Use Committee.

Results

LF Protease Enzyme and the Macrophage Toxicity Assays. The LF protease enzyme and the macrophage toxicity assays were used in tandem to identify weak hydroxamate leads from the Merck sample collection and then to guide a medicinal chemistry effort that yielded a potent hydroxamate LFI (Fig. 6A Inset, which is published as supporting information on the PNAS web site). LFI is a time-dependent, reversible inhibitor of LF protease activity with an IC₅₀ value of 60 nM. The kinetic mechanism of inhibition for LFI was determined to be competitive with substrate, with a K_i value of 24 nM (Fig. 7, which is published as supporting information on the PNAS web site). LFI also blocked LF-induced cytotoxicity in mouse macrophage J774A.1 cells with an IC₅₀ of 160 nM (Fig. 6B).

Crystal Structure of LF in Complex with LFI. In parallel with the enzymatic and cell-based assays, the 3D structures of LF complexed with various hydroxamate-based compounds were determined by x-ray crystallography and used to design inhibitors with improved potency and solubility. The N-terminally truncated form of LF, composed of residues 264–776, had identical protease activity compared with full-length LF, contained the entire protease active site with all inhibitor contact points, and yielded much better diffracting crystals than full-length LF. Although the truncated form of LF lacked PA-binding domain I, the overall architecture of domains II–IV was maintained (Fig. 1A). The major differences between the unliganded and inhibited structures were the overall position of domain III and the conformation of the loops spanning residues 427–437, 645–651, and 673–680. Domain III moved in a rigid body motion and appeared to be coupled to the movement of the 673–680 loop, by ≈30° upon inhibitor binding. The movements of loops 427–437 and 645–651 were probably related to different crystal packing of the two forms of the enzyme rather than a direct consequence of inhibitor binding.

Fig. 1.

Crystal structure of LFI bound to LF. (A) Overlay of a C^α trace of full-length apo LF (Protein Data Bank ID code 1J7N, blue) and the truncated (264–776) version (green) used in this study. The domains are numbered in red. LFI and the zinc ion are represented as ball-and-stick models (yellow, carbon; green, sulfur; red, oxygen; blue, nitrogen; cyan, fluorine; and gray, zinc; the same coloring scheme is used in all three models). Despite the absence of domain I in our construct, the overall structure of domains II–IV is conserved. (B) Molecular surface of LF around the inhibitor binding site. The surface is colored according to potential (red, negative; blue, positive). The S1′ site is labeled. (C) Stereoview of LFI bound to LF. Residues within6Åofthe bound ligand are displayed. Water molecules are omitted for clarity. The electron density surrounding the inhibitor is from a 2F_o – F_c difference map contoured at 1.5 s. The orientation is similar to that of A.

LFI bound in the groove at the interface of LF domains III and IV (catalytic domain; Fig. 1 A). The oxygen atoms (–CONHOH) of the hydroxamate chelated the Zn²⁺ ion in a bidentate, planar conformation; the other ligands to Zn²⁺ were His-686, His-680, and Asp-735. The substituted phenyl ring of LFI (corresponding to the P1′ position) was bound in a deep hydrophobic pocket adjacent to the catalytic center (Fig. 1B). From the synthesis and testing of >500 inhibitor analogs, the 4-fluoro-3-methylphenyl group appeared to optimally fill the hydrophobic pocket identified on LF. The tetrahydropyran moiety of LFI was positioned in the large cavity between domains III and IV, but made limited interactions with the protein. Thus, this modification would have little consequence on LF affinity.

Mouse Toxemia Model Using Recombinant PA and LF. LFI was tested for its ability to protect BALB/c mice from a lethal mixture of recombinant PA and LF. Probe studies showed that i.v. coinjection of 100 μg of recombinant LF and 100 μg of recombinant PA in saline into the tail vein of BALB/c mice led to 100% mortality within 48 h and was chosen as the standard toxin challenge. To identify a dosage regime that would maintain trough micromolar blood levels for several hours, a range-finding pharmacokinetic analysis was undertaken. LFI was administered in saline to mice i.p. at 1, 10, or 30 mg/kg t.i.d. (0, +1.25, and + 3.25 h), and sufficient ranges of trough levels of 0.2–0.3, 1.1–2.4, or 1.9–3.1 μM, respectively, and cmax levels of 1.6–2.2, 12.3–13.4, or 20.0–29.9 μM, respectively, were obtained.

Therefore, LFI was examined in the toxemia model by i.p. administration at 0, 1, 10, or 30 mg/kg t.i.d. (–0.25, +1, and + 3 h relative to toxin administration) in saline. After a 48-h evaluation, none of the eight mice survived in the group receiving saline vehicle, whereas one of eight (12.5%), seven of eight (87.5%), and eight of eight (100%) mice survived in the 1, 10, or 30 mg/kg t.i.d. LFI-treated groups, respectively (Table 1). Although seven of eight mice survived in the 10 mg/kg t.i.d. group, clinical signs typical of anthrax infection, such as ruffled hair and huddling, were evident, whereas mice in the 30 mg/kg t.i.d. group appeared normal throughout the trial.

Table 1. LFI protection of mice from recombinant LF- and PA-mediated death.

Compound	Dose, mg/kg t.i.d.	n	Survivors after 48 h	Survival rate, %
LFI	30	8	8	100
LFI	10	8	7	87.5
LFI	1	8	1	12.5
None	0	8	0	0

A mixture of 100 μg of recombinant LF and 100 μg of recombinant PA was injected into the tail vein of BALB/c mice. LFI in saline was administered at various doses i.p. 15 min before and 1 and 3 h after intoxication. Neither LF nor PA was toxic alone.

Monotherapeutic LFI Protection of Mice Infected with B. anthracis Sterne Strain Vegetative Cells. The follow-up to the toxemia model was to examine the ability of LFI to protect mice from an active B. anthracis infection. Probe studies showed that an injection of 10⁸ B. anthracis Sterne vegetative cells into the peritoneal cavity resulted in >90% mortality of BALB/c mice within 2–3 days. The time course for LF production in mouse blood was similar in infections with either Sterne or Ames strains in our laboratory. For both anthrax strains in mice, maximum plasma levels of 10–15 nM LF as determined by LF protease activity were reached ≈12 h before death of the host.

Individual injections of LFI that would maintain micromolar trough levels for multiple days against bacterially generated toxins were eliminated from consideration because of a terminal 0.4-h half-life of the drug in mice. Probe pharmacokinetic studies, however, showed that mice continuously infused with LFI by means of a jugular vein catheter at 250 μg·100 μl^–1·h^–1 in saline maintained LFI plasma levels of 8–10 μM.

In the first evaluation of LFI in this model, 10 jugular-veincannulated BALB/c mice were continuously infused with saline at 100 μl/h, and 10 were similarly infused with LFI at 250 μg·100 μl^–1·h^–1. All 20 mice were challenged with 10⁸ Sterne vegetative cells 24 h after the start of the infusions. Three days after challenge, all 10 mice infused with saline developed clinical signs of anthrax and died. A mean time to death (MTD) of 65 h was calculated for the 10 saline-infused mice, compared with 91 h for the five mice that died in the LFI-infused group. Importantly, five LFI-treated mice survived to the termination of dosing on day 9 (Fig. 2, solid symbols). These five surviving mice underwent an anthrax crisis, albeit delayed and reduced relative to the saline-infused mice, from days 5–7 after challenge, but these mice appeared to resolve the crisis and were normal and healthy at day 9. A repeat study was conducted by using the same conditions with nearly identical results (Fig. 2, open symbols).

Fig. 2.

LFI protection of BALB/c mice infected with B. anthracis Sterne strain vegetative cells. Intravenous continuous infusion of LFI at a rate of 250 mg·100 μl^–1·h^–1 (circles) or saline at 100 μl/h (squares) was started 1 day before (arrow) infection of groups of 10 mice with 10⁸ cells of B. anthracis Sterne strain (day 0). Open and filled symbols represent two independent studies. Inhibitor dosing was discontinued on day 9, and the animals were killed. The proportion of mice surviving when dosed with LFI is significantly greater than those on vehicle alone (P = 0.0019).

Monotherapeutic LFI Protection of Rabbits Infected with B. anthracis Ames Spores. Lethality in rabbit models of anthrax using B. anthracis Ames spores is dominated by effects of toxin; therefore, the rabbit is accepted as the best nonprimate model for human anthrax disease (15, 16). To determine the challenge inoculum, probe studies showed that a s.c. injection of 10⁴ B. anthracis Ames strain spores was >90% lethal to DB rabbits by 72 h, and that inoculum was used in all rabbit studies. To determine dose, pharmacokinetic studies indicated that LFI had a terminal half-life of 2.0 h in the rabbit that necessitated 100 mg/kg s.c. t.i.d. injections to provide trough blood levels of 3–6 μM and cmax levels of 8–12 μM (the latter approximated that of the mouse continuous-infusion experiment described above).

The initial rabbit efficacy test consisted of 12 DB rabbits: six were dosed s.c. with LFI at 100 mg/kg t.i.d. for 7 days and six were dosed s.c. with saline at the same times. Two hours after the first dose, all rabbits were challenged s.c. with 10⁴ Ames spores. From 48 to 72 h, five of six saline-treated rabbits underwent an anthrax crisis and died (Fig. 3). Time to death and the number of deaths in the LFI-treated rabbit group were delayed and reduced: one died at 96 h and one other died at 120 h. The MTD for rabbits dying in the saline group was 65 h, whereas for the two LFI-treated rabbits, it was nearly double (111 h). Surviving rabbits in the LFI-treated group underwent a diminished anthrax crisis from day 4 to day 6, but all rabbits appeared normal by day 7. When dosing ended at day 7, five of the six (83%) saline-treated rabbits were dead compared with only two of the six (33%) LFI-treated rabbits. One additional LFI-treated rabbit died while obtaining a blood sample at the end of the dosing period, but the three remaining rabbits survived and appeared completely normal at the end of the experiment, day 19, when they were killed.

Fig. 3.

LFI protection of rabbits infected with B. anthracis Ames spores. DB rabbits were dosed s.c. with LFI at 100 mg/kg t.i.d. (▴)(n = 6) or saline vehicle (○) (n = 6) starting 2 h before (arrow) infection with 10⁴ B. anthracis Ames spores (day 0). Dosing was discontinued on day 7. Surviving rabbits were monitored to day 21. One LFI-treated rabbit died on day 7 while its blood was drawn. The two survival curves are statistically different (P = 0.013).

To confirm and extend the previous monotherapy finding in the rabbit, a second study was conducted. As in the first trial, a s.c. injection of 10⁴ spores of B. anthracis Ames was used to challenge 12 DB rabbits. LFI monotherapy was delivered at 100 mg/kg s.c. t.i.d. for 7 days starting at the time of spore challenge to the first group (n = 4) and for 6 days starting 24 h after challenge to a second group (n = 4). A third group served as saline-treated control (n = 4). All saline-treated control rabbits died by 80 h after challenge, with an MTD of 72 h (Fig. 4). The first rabbit in the LFI monotherapy group dosed at time of challenge did not die until 112 h after challenge, and the second died at 246 h, providing an MTD of 179 h and more than doubling the survival time relative to the control group. The two other rabbits in this group that received LFI treatment starting at time of challenge survived for 21 days and were healthy and normal.

Fig. 4.

LFI protection of rabbits infected with B. anthracis Ames spores. Twelve DB rabbits were challenged with 10⁴ Ames spores s.c. at time 0. Four rabbits were given saline s.c. t.i.d. starting at time of challenge for 7 days (○). Four rabbits were given LFI at 100 mg/kg s.c. t.i.d. starting at time of challenge for 7 days (▴). Four rabbits were given LFI at 100 mg/kg s.c. t.i.d. starting 24 h after time of challenge for 6 days (□).

In the second arm of this monotherapy trial, the LFI group with 24-h delayed treatment also showed an increase in MTD relative to control (Fig. 4). The first rabbit from this delayed therapy group died at 128 h, and two additional rabbits died at 152 h, giving an MTD of 144 h. Thus, even when LFI treatment was delayed for 24 h, there was still a doubling of MTD relative to vehicle-treated control rabbits. In addition, a single rabbit from this group survived for 21 days and appeared normal at trial end.

Cipro and LFI Combination Therapy of Rabbits Challenged with B. anthracis Ames Spores. A Cipro point of no return model was developed in DB rabbits to evaluate whether combination of Cipro with LFI would provide additional useful therapy beyond what Cipro would do alone. Probe studies with Cipro monotherapy at 5 mg/kg s.c. b.i.d. for 2 days showed it to be completely protective to rabbit groups (n = 4) challenged with 10⁴ spores of B. anthracis Ames when dosing started at 24, 36, 48, or 54 h after challenge. However, one of four rabbits died when Cipro therapy was delayed to 60 h, and two of four died when therapy was delayed to 66 h, and this latter time was defined as the 50% point of no return for Cipro.

In the definitive point of no return test, 11 DB rabbits were challenged s.c. with 10⁴ Ames spores at time 0. At 66 h after challenge, the surviving rabbits were divided into two equal experimental groups of four rabbits each, and the remainder was given saline only. The first experimental groups received Cipro monotherapy at 5 mg/kg s.c. b.i.d. for 2 days starting at 66 h, and the second experimental group received combination therapy with Cipro by following the same protocol as the first group on one side of the body and LFI at 100 mg/kg s.c. four times a day for one day on the other side. The three rabbits that received nothing or saline died by 96 h (Fig. 5), and, when their peritoneal cavities were swabbed and cultured, each was positive for B. anthracis Ames. Cipro protected half of the Cipro monotherapy group and extended the MTD of those dying to 144 h. It is noteworthy that the two rabbits that died in the Cipro monotherapy group were negative for B. anthracis Ames, indicating that the four Cipro treatments given to each rabbit, although sterilizing, did not prevent death. Most importantly, there was not only complete protection in the combination group receiving Cipro and LFI in survival terms, but few clinical signs of anthrax were observed in these rabbits. The six surviving DB rabbits from the two experimental groups were killed on day 11; peritoneal cultures were made, and all were negative for B. anthracis.

Fig 5.

LFI or LFI/Cipro protection in rabbits infected with B. anthracis Ames spores when dosing began at 66 h after challenge. Eleven DB rabbits were challenged with 10⁴ Ames spores s.c. at time 0. Three rabbits were given saline only starting at 66 h after challenge (○). Four rabbits were given Cipro at 5 mg/kg b.i.d. for 48 h starting at 66 h after challenge (▪). Four rabbits were given Cipro as above in combination with LFI at 100 mg/kg s.c. four times a day for 24 h starting 66 h after challenge (▴).

Discussion

The in vivo experiments revealed herein have demonstrated efficacy of the hydroxamate LFI when given as prophylactic monotherapy or in late-stage combination therapy with an antibiotic. Initial LFI monotherapy experiments in mice against lethal cocktails of recombinant LF and PA showed a distinct dose titration whereby all mice receiving saline died and all mice receiving LFI at 30 mg/kg i.p. t.i.d. survived. This finding was proof of concept that inhibition of LF by the hydroxamate LFI had potential to lead to useful therapy against anthrax toxemia; however, to be a practical success, an LFI must deal not with a single pulse of LF but with continuous secretion.

To test LFI against continuous secretion of LF, a murine model was developed by using the B. anthracis Sterne strain. Probe studies showed that an i.p. inoculum of 10⁸ vegetative colony-forming units were consistently lethal to mice in 72 h and produced a bacteremia and circulating LF level equivalent to a 10³ Ames vegetative inoculum that killed in the same time period. The initial murine study against the Sterne strain demonstrated that continuous infusion of LFI by itself delayed MTD in all Sterne-infected mice, suggesting that LFI prophylaxis could provide valuable additional survival time for sterilizing antibiotic therapy to be used. The second unanticipated benefit of LFI monotherapy was that half of the LFI-infused mice survived. It is noteworthy that even the surviving mice underwent a delayed but obvious anthrax crisis from day 5 to day 7 but then resolved the crisis and appeared normal at the end of the study. These data were repeated with almost identical results.

Success against B. anthracis Sterne vegetative cells created an urgency to evaluate LFI directly against B. anthracis Ames spores, and a rabbit model was chosen for such an evaluation. Probe efforts showed that 10⁴ B. anthracis Ames spores delivered s.c. were consistently lethal to rabbits in 72 h. Dose regimens of LFI that produced blood concentration levels similar to those observed in the Sterne model in mice were targeted. The first of these studies was done prophylactically with LFI dosing beginning 2 h before the s.c. Ames spore challenge. LFI treatment produced a doubling of MTD, and half the group survived. These surviving rabbits underwent an anthrax crisis through days 5–7 but resolved the crisis by day 9 with apparent, full recovery. These surviving animals were observed for 21 days before being killed. A follow-up study with LFI dosing beginning at the same time of the Ames spore challenge confirmed these data in rabbits. Moreover, a separate group in the follow-up rabbit study extended the MTD and survival data to include rabbits in which LFI treatment had been delayed by a critical 24 h beyond spore challenge.

The final rabbit salvage study in which antibiotic therapy with Cipro was combined with LFI demonstrated the complementary benefit these two different mechanisms of action offer against anthrax infection. Much probe work with Cipro was done in rabbits to identify the 66-h postchallenge time to be the 50% point of no return. Once that time point was identified and dosing began, the Cipro and LFI combination saved all rabbits at a point at which Cipro alone saved only half. Not only did the Cipro/LFI combination-treated rabbits survive, but few signs of anthrax were observed. It was particularly striking that the two Cipro-treated rabbits that died were found to have sterile blood and peritoneal cultures. These observations illustrated the primary rationale for a toxin inhibitor, because, although Cipro successfully eliminated anthrax infection in the rabbits, the rabbits still died.

Strategically, LF may represent both a prophylactic and therapeutic target because it has been reported to play a vital role at two different points in anthrax infection. Intracellularly, LF has a protective role for the spore after ingestion by macrophages (17). Mutant B. anthracis lacking LF are destroyed in macrophages, indicating that efficient neutralization of LF early in infection could allow host cells to eliminate the initial insult at the locus of infection. Extracellularly, where LF is believed to be the primary virulence factor, it interferes with signal transduction in host defense cells and destroys the integrity of endothelial vessels, which leads to circulatory shock and death of the host. There is little doubt that the LFI in the studies reported here is expressing an effect in the extracellular phase, which is indicated in the mouse studies using vegetative cells, because there was no intracellular phase possible and in the rabbit studies using spores in which treatment was delayed 24 h with monotherapy and 66 h with combination therapy. The role of LFI in the intracellular phase remains to be elucidated, but the fact that surviving mice and rabbits prophylactically dosed with LFI went through anthrax crises and resolved them may hint at an intact immune response and a block of the early influence that LF has in turning off normal immune responses.

The structure-activity relationships observed in the medicinal chemistry program concur with x-ray crystallographic results showing that the binding of LFI to LF is driven by the contacts between the hydroxamic acid moiety and the 4-fluoro-3-methylphenyl substituent of the compound with the active site Zn²⁺ and S1′ pocket of enzyme, respectively. An examination of the substrate cleavage sites in naturally occurring mitogen-activated protein kinase kinases and synthetic peptides from a large combinatorial library (18) indicates that the P1′ position is important for selectivity. The common protein binding sites for inhibitor and substrate suggest that it may be difficult to find LF mutations that weaken inhibitor binding without a concomitant compromise of natural substrate recognition. Thus, the acquisition of anthrax drug resistance by this mechanism, through either deliberate engineering or bacterial evolution, would appear unlikely.

Systemic anthrax, although rare as a natural disease in humans, has recently gained substantial notoriety as an agent of biological warfare and terrorism. Mortality from inhalational anthrax is very high, even with aggressive antimicrobial therapy. In the attacks after September 11th, 2001, mortality of those with inhalational anthrax was 45% (5/11) (19). Specific antimicrobial agents (e.g., Cipro and doxycycline) are very effective against B. anthracis, but, if given too late, they cannot address the primary virulence of anthrax, its systemic toxemia. It is envisaged that complementary combination of an antimicrobial mechanism and a LF-inhibiting mechanism as reported here has the potential to provide additional means of therapeutic intervention and substantially reduce mortality.