Antidotes to anthrax lethal factor intoxication. Part 1: Discovery of potent lethal factor inhibitors with in vivo efficacy

Abstract

Sub-nanomolar small molecule inhibitors of anthrax lethal factor have been identified using SAR and Merck L915 (4) as a model compound. One of these compounds (16) provided 100% protection in a rat lethal toxin model of anthrax disease.

Three forms of the disease anthrax caused by Bacillus anthracis are characterized by the route of exposure. Infection of an open wound leads to cutaneous anthrax, and ingestion of contaminated food causes gastrointestinal anthrax. While each of these forms can be fatal, they are less severe compared to inhalation anthrax, where mortality rates can be very high (>85%).¹ Once inhaled, germination of spores leads to release of bacteria in a vegetative state and eventual propagation leading to septicemia. To aid in infection of the host, the bacteria secrete three proteins; edema factor (EF) and lethal factor (LF), each of which combine with protective antigen (PA) to form two binary toxins that act as virulence factors possessing potent immunosuppressive activity.² EF is a Ca⁺²/calmodulin-dependent adenylate cyclase which appears to impair immune function. LF is a Zn²⁺-dependent metalloproteinase which disrupts MAPK pathway signaling and leads to suppression of pro-inflammatory gene expression. PA aids in the translocation of EF and LF into the target cells.³ If left unchecked, the resulting toxemia and sepsis lead to vascular collapse, shock, and the death of the host within a few days. While EF clearly contributes to the lethality of anthrax, LF has been shown to be the primary causative agent leading to death of the host due to toxemia.⁴ Given the effectiveness of B. anthracis as a weapon of bioterrorism,⁵ the major role LF plays in the pathogenesis of anthrax, and validation of LF as a target for small molecule drug intervention,⁶ we began our search for an antidote to LF intoxication. Presented below are the results from our early discovery phase of a project leading to the identification of novel small molecule anthrax LF inhibitors (LFIs).

A number of small molecule inhibitors of anthrax LF are known, with those shown in Figure 1 being representative of various structural classes.⁷ Of these, the sulfonamide-based series represented by Merck L915 (4) is the best characterized⁶ and provided a good starting point for discovery of new lead series.

Figure 1.

Representative known small molecule inhibitors of anthrax lethal factor

Using 4 as a design model, we began our investigation with the goal of identifying novel X-Y linking groups (Figure 2) using a FRET based assay to guide the SAR.⁸ Replacing the NH-group of the sulfonamide with a methylene group (X = CH₂, Y = SO₂) provided sulfone analogs with equivalent potency while the corresponding amide (X=NH, Y = CO) was inactive. Other two atom linking groups, such as benzylamines or ethers (X=NH, O, Y = CH₂) afforded lower (>10×) potency analogs. Based on the H-bond implicated in the co-crystal structure between the sulfonyl group pro-R oxygen of L915 to backbone amide protons (K656 and G657) of LF,⁶ we reasoned that a hydroxyl or ether functional group may provide for a similar interaction. Indeed, use of these two atom links (X = CH₂, Y = CH(OH), CH(OMe)) afforded active analogs, with the 4-methylether series displaying significantly better potency compared to the alcohol. We also investigated the corresponding one atom linking series (X = NH, O, S, and CH₂) in the absence of a Y atom. We were pleased to discover that all of these compounds were potent inhibitors of anthrax LF, with the aniline and phenylether series (X=NH, O) being specific (>300×) for the target metalloprotease versus several MMPs (data not shown). These results led to the further examination of four possible core structures as novel LFI lead series (Figure 3).

Figure 2.

Design model based upon sulfonamide hydroxamic acids

Figure 3.

One atom and two atom linking group series selected for study.

Our initial goals were to identify the best R¹-groups and substitution pattern for the phenyl ring of these core structures, as well as the preferred stereochemistry at the C2 and C4 positions in each linking group series. In the one atom series, the aniline derivatives were selected for further study due to their ease of synthesis (Scheme 1). Our initial work explored the effect of changing the size, polarity, and position of the R¹-group on inhibitor potency using racemic compounds and the preparation of various mono-, di-, and tri-substituted aniline derivatives (Figure 3; X = NH, R² = n-Bu). The resulting SAR led to the identification of the 3,5-dimethyl-4-fluoroaniline analog as the most potent inhibitor possessing sub-micromolar inhibitory activity.⁹ In the two atom linking series, a similar approach of varying the R¹-group in the racemic 4-methoxy (γ-ether) analog series led to the identification of the 3-methyl-4-fluorophenyl and 4-fluorophenyl analogs as having the best potency (Figure 3, R² = H). This result was consistent with the two atom sulfonamide linking group found in Merck L915 (4). Determining the preferred stereochemistry in each series required the development of synthetic schemes capable of providing optically pure analogs in the aniline¹⁰ (Scheme 1) and the 4-methoxy-4-phenylbutanoic acid (γ-ether) series¹¹ (Scheme 2).

Scheme 1.

Reagents and conditions: (a) 2 eq of 3,5-dimethyl-4-fluorophenylboronic acid, 1.1 eq of Cu(OAc)₂, 2 eq of Et₃N, MS 4Å, DCM, open air, rt; (b) TFA/DCM (1:1), rt; (c) 1 eq of R-ZCl (Z=CO, SO₂), 5 eq of Et₃N, DCM, rt; or RCHO, 2 eq of Et₃N, 1.4 eq of NaBH(OAc)₃, DCE, rt; (d) KCN (5 mol %), THF/MeOH/50% NH₂OH-H₂O (2:2:1), rt.

Scheme 2.

Reagents and conditions: (a) LiHMDS, HMPA, R-CH=CHCH₂I, −70°C to rt; (b) KOH, H₂O, dioxane, rt; (c) MeI, NaH, THF, rt; (d) conc. H₂SO₄, MeOH, rt; e) O₃, Ph₃P, DCM, −70 °C to rt; (f) diphenylamine, NaBH(OAc)₃, AcOH, DCE, rt; (g) NH₂-OH, KCN, H₂O, MeOH, THF, rt.

In the case of the aniline series, a stereospecific synthesis was developed following the work by Lam and co-workers¹² which begins with optically pure amino acid derivatives. Using this synthetic route with the available (R) and (S) amino acids, a clear preference (>10 to 100×) for the (2R) stereochemistry versus (2S) at C2 in the aniline series was observed.

The presence of two stereogenic centers in the γ-ether series led to the development of the synthetic pathway shown in Scheme 2. Readily available and optically pure 5-aryl-2-furanones¹³ provided access to all four diastereomers. Stereoselective alkylation¹⁴ of furanone 9 using allylic halides led to an easily separable mixture of the desired major isomer 10 and the minor syn-isomer. Of the four possible diastereomers, the greatest inhibition of LF activity was observed with the (2S,4R)-diastereomers (11), while the (2S,4S)-isomers were found to be >100-fold less potent. The remaining two diastereomeric series displayed intermediate inhibitory activity.

Having identified the preferred R¹-groups and stereochemistry for the aniline and γ-ether series, we began to explore variations to the C2-side chain in an effort to further increase inhibitor potency. In the aniline series, the primary amine derived from 6 (Scheme 1) provided a convenient branch point for analog synthesis, while in the γ-ether series intermediate 11 proved to be good starting point for exploring diversity (Scheme 2, R = H). Tables 1 and 2 provide a representative sample of our findings in each series.

Table 1.

LF inhibitory data for selected aniline derivatives

Compd	Z	R	LF (FRET)
			Ki (nM)a
13	-	-H	8.2
14	CO	3-Me-4-F-Ph	7.3
15	SO2	3-Me-4-F-Ph	3.1
16	CH2	3-Me-4-F-Ph	0.05
17	C(O)NH	3-Me-4-F-Ph	9.3
18	CH2	2-F-Ph	0.69
19	CH2	3-F-Ph	0.85
20	CH2	4-F-Ph	0.24
21	CH2	4-Cl-Ph	1.1
22	CH2	4-Me-Ph	0.41
23	CH2	4-OMe-Ph	0.44
24	CH2	4-CF3-Ph	0.47
25	CH2	4-NO2-Ph	0.30
26	CH2	4-t-Bu-Ph	1.4
27	(CH2)2	4-F-Ph	2.1
28	(CH2)3	4-F-Ph	1.3

^aValues are means of three experiments; standard deviation is <10%

Table 2.

LF inhibitory data for selected γ-ether derivatives

Compound	R	LF (FRET)
		Ki (nM)a
29	-N(Me)2	96
30	Pyrrolydin-1-yl	122
31	Piperidin-1-yl	230
32	Morpholin-4-yl	350
33	(R,S)-3-hydroxymethyl-piperidin-1-yl	99
34	(R,S)-3-(4-F-Ph)CH2-piperidin-1-yl	33
35	Piperazin-1-yl	90
36	4-Ph-piperazin-1-yl	151
37	4-PhCH2-piperazin-1-yl	115
38	4-Ph(CH2)2-piperazin-1-yl	118
39	-N(Me)-(CH2)3Ph	42
40	-N(Me)(CH2)3-3Me-4-F-Ph	11
41	-N(Me)CH2-4(3-Me-4-F-Ph)-Ph	135

^aValues are means of three experiments; standard deviation is <10%

Collected in Table 1 is a selection of data based on 3,5-dimethyl-4-fluoroaniline core structure derived from R-lysine which is representative of the SAR observed in the one atom linking series. The synthetic versatility associated with a primary amine led to an early study on how different nitrogen containing functional groups would impact intrinsic potency. The primary amine 13 was found to be a potent inhibitor of LF activity with a K_i = 8.2 nM. The corresponding amide 14, sulfonamide 15 and urea 17 derived from this amine and bearing the same R-group were potent inhibitors but less active compared to the sub-nanomolar benzyl amine derivative 16 (K_i=0.05 nM), a compound >500× more potent than 4 (K_i=46 nM). As a result, our focus was directed towards further exploration of the benzylamine series.¹⁵ Positioning one or more small groups anywhere on the phenyl ring (18 to 20) afforded potent LFIs. Further exploration at the C4-position was investigated with the preparation of inhibitors 21 to 26. Little or no effect was seen with variation in the nature of the R-group at C4 with all of the compounds displaying single digit to sub-nanomolar inhibition of LF activity. This suggested that these R-groups either extend into a large cavity or that they are directed into the solvent. Finally, extension of the chain length to give the phenethyl (27) or phenpropyl (28) analogs lead to a slight decrease in potency relative to the benzyl analogs.

Shown in Table 2 are data for a representative set of LF inhibitors based upon the γ-ether core structure. Simple symmetric dialkyl and cyclic amine derivatives (29 to 32) afforded LF inhibitors with moderate activity. The 3-substituted piperidine derivatives 33 and 34 displayed slightly higher potency relative to the unsubstituted compounds, with the individual (R) and (S) isomers in each case having K_i values essentially unchanged from the mixture shown. Examination of 4-substituted piperidines showed a similar profile having K_i values in the high double digit nanomolar range (data not shown).

Preparation and testing of 4-substituted piperazine analogs (35 to 38) resulted in inhibitors with K_i values in the 100 nM range. Somewhat surprisingly, the size of the 4-substitutent had no effect on LF inhibitory activity. In contrast, preparation of unsymmetrical tertiary amines where one R-group was methyl (39 to 41) provided analogs with improved potency. Of the many analogs made, those having a total C2 length equivalent to the N-substituted lysine derivatives discussed above (Table 1) exhibited the highest potencies, with compound 40 displaying a K_i of 11 nM.

Kinetic studies conducted with representatives of both the aniline and γ-ether series showed them to be slow tight-binding competitive inhibitors (Figure 4); a result in line with the finding of the Merck scientists and 4.^6a

Figure 4.

Kinetic inhibition data for 40 versus anthrax LF. Global nonlinear fitting (GraFit v5.0) to multiple inhibition models and statistical results indicate competitive behavior.

After identifying potent LFIs in two unique structural series, our next goal was to demonstrate their efficacy in an in vivo Lethal Toxin (LT) model of anthrax.¹⁶ Based upon selection criteria in our LFI screening cascade which included intrinsic potency (K_i < 100 nM), selectivity for the target (> 500× vs. MMP-1, 3, 9, 12, 14) drug-like properties (ElogD < 3, S_k > 5 mg/mL 50% DMSO/PBS) and low cytotoxicity (CC₅₀ > 10 μM), five compounds were selected for an in vivo study. Fischer 344 rats were dosed (iv) with three aniline and two γ-ether analogs at 10 mg/kg in DMSO/PBS (1:1, v/v) followed 20–30 minutes later with a 10 μg dose (iv) of LT (10 μg PA + 10 μg LF). Using the extension of Median Survival Time (MST) and percent survival as endpoints, the results in Figure 5 show that with the exception of compound 14, all of the compounds extended the MST relative to the control group, and one compound, the aniline 16, was capable of providing 100% protection at this dose. Repeating this experiment with lower doses of 16 indicated that this compound was fully protective at both 5.0 and 2.5 mg/kg, although at the lowest dose the animals became ill approximately 3 hours after treatment with LT but appeared fully recovered by 24 hours post exposure (data not shown).

Figure 5.

Survival curves for Fischer 344 rats (n = 4) dosed (iv) with LFIs followed by LT (10 μg PA + 10 μg LF) dosed iv.

In summary, we have identified sub-nanomolar inhibitors of anthrax lethal factor with potent in vivo efficacy. Part 2 of this work will describe the initial steps taken to optimize these compounds with the goal of developing LFIs to treat anthrax in humans.