Synthesis and antimicrobial evaluation of amixicile-based inhibitors of the pyruvate-ferredoxin oxidoreductases of anaerobic bacteria and Epsilonproteobacte

Article abstract of DOI:10.1128/AAC.00670-16

Amixicile is a promising derivative of nitazoxanide (an antiparasitic therapeutic) developed to treat systemic infections caused by anaerobic bacteria, anaerobic parasites, and members of the Epsilonproteobacteria (Campylobacter and Helicobacter). Amixicile selectively inhibits pyruvate-ferredoxin oxidoreductase (PFOR) and related enzymes by inhibiting the function of the vitamin B₁ cofactor (thiamine pyrophosphate) by a novel mechanism. Here, we interrogate the amixicile scaffold, guided by docking simulations, direct PFOR inhibition assays, and MIC tests against Clostridium difficile, Campylobacter jejuni, and Helicobacter pylori. Docking simulations revealed that the nitro group present in nitazoxanide interacts with the protonated N4′-aminopyrimidine of thiamine pyrophosphate (TPP). The ortho-propylamine on the benzene ring formed an electrostatic interaction with an aspartic acid moiety (B456) of PFOR that correlated with improved PFOR-inhibitory activity and potency by MIC tests. Aryl substitution with electron-withdrawing groups and substitutions of the propylamine with other alkyl amines or nitrogen-containing heterocycles both improved PFOR inhibition and, in many cases, biological activity against C. difficile. Docking simulation results correlate well with mechanistic enzymology and nuclear magnetic resonance (NMR) studies that show members of this class of antimicrobials to be specific inhibitors of vitamin B₁ function by proton abstraction, which is both novel and likely to limit mutation-based drug resistance.

Full text of DOI:10.1128/AAC.00670-16

AAC Accepted Manuscript Posted Online 18 April 2016
Antimicrob. Agents Chemother. doi:10.1128/AAC.00670-16
Copyright © 2016, American Society for Microbiology. All Rights Reserved.
1
2
Synthesis and Antimicrobial Evaluation of Amixicile-Based Inhibitors of the Pyruvate:
Ferredoxin Oxidoreductases of Anaerobic Bacteria and Epsilon Proteobacteria
3
4
Andrew J. Kennedya, Alexandra M. Brucea, Catherine Ginesteb, T. Eric Ballarda, Igor N.
Olekhnovichb, Timothy L. Macdonalda and Paul S. Hoffmanb*.
5
6
7
Departments of Chemistrya and Medicine, Division of Infectious Diseases and International
Healthb, University of Virginia, Charlottesville, Virginia,
8
9
10
11
12
Running title: Amixicile analogues
*Corresponding author:
13
14
15
16
17
18
19
20
Paul S. Hoffman
Division of Infectious Diseases and International Health
MR4 Bldg, Room 2146
409 Lane Road
Charlottesville, VA 22908
Phone: (434) 924-2893
Email: psh2n@virginia.edu
21
22
23
24
25
26
27
28
29
30
31
32
Current addresses:
Andrew Kennedy
Bates College
Department of Chemistry
2 Andrews Road
Lewiston, Maine 04240
Eric Ballard
Pharmacokinetics, Dynamics and Metabolism,
Pfizer, Inc., Groton, CT 06340, USA
33
34
35
1

36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
Abstract
Amixicile is a promising derivative of nitazoxanide (antiparasitic therapeutic) developed to treat
systemic infections caused by anaerobic bacteria, anaerobic parasites and by members of the
epsilonproteobacteria (Campylobacter and Helicobacter). Amixicile selectively inhibits pyruvate:
ferredoxin oxidoreductase (PFOR) and related enzymes by inhibiting the function of the vitamin
B1 cofactor (thiamin pyrophosphate) by a novel mechanism. Here we interrogate the amixicile
scaffold, guided by docking simulations, direct PFOR inhibition assays and by minimal inhibitory
concentration (MIC) tests against Clostridium difficile, Campylobacter jejuni and Helicobacter
pylori. Docking simulations revealed the nitro group present in nitazoxanide to interact with the
protonated N4’-aminopyrimidine of thiamine pyrophosphate (TPP). The ortho-propylamine on
the benzene ring formed an electrostatic interaction with an aspartic acid moiety (B456) of
PFOR that correlated with improved PFOR inhibitory activity and potency by MIC tests. Aryl
substitution with electron withdrawing groups and substitutions of the propylamine with other
alkyl amines or nitrogen-containing heterocycles both improved PFOR inhibition and in many
cases biological activity against C. difficile. Docking simulation results correlate well with
mechanistic enzymology and NMR studies that show this class of antimicrobials to be specific
inhibitors of vitamin B1 function by proton abstraction which is both novel and likely to limit
mutation-based drug resistance.
2

62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
Introduction
Infectious diseases are a leading cause of death worldwide and antibiotics developed to
combat these infections are being lost to the rapid emergence of drug resistance. For most
antibiotics that are derivatives of natural products, resistance mechanisms often pre-exist
clinical usage and these resistance genes over time tend to accumulate in pathogens via lateral
transfer of genetic elements (1, 2, 3). In contrast, synthetic antimicrobials developed against
new drug targets where no natural product inhibitor exists are prone to mutation based drug
resistance (4, 5). Strategies employing empiric approaches to target identification and high
throughput screening of chemical libraries have also failed to deliver new therapeutics to the
clinic (1, 2, 6, 7). Several studies have suggested that the number of druggable targets in
microbial pathogens is quite small when essentiality, selectivity, catalytic mechanism and
chemical space are factored in (7, 8). Thus, identifying new druggable targets is one of the
major challenges to the development of next generation antimicrobials. One potential target not
found in humans or mitochondria, but common in many human pathogens, is pyruvate:
ferredoxin oxidoreductase (PFOR) (9). We discovered that nitazoxanide (NTZ), a synthetic
antiparasitic nitrothiazolide therapeutic (Figure 1), inhibits this enzyme by a novel mechanism
that does not involve nitroreduction (10, 11).
NTZ completely inhibits the production of acetyl-CoA and CO2 from pyruvate by out-
competing pyruvate for binding to the vitamin B1 thiamine pyrophosphate (TPP) cofactor of
PFOR (Ki value of ~5 x 10-6M, which is roughly two orders of magnitude lower than the Km value
for pyruvate Km ~3 x 10-4M) (11). The NTZ amide is unusually acidic and is deprotonated at
biological pH, producing an anion that is delocalized throughout the thiazole ring. The nitro
group of NTZ interacts with and abstracts a proton from the activated N4’ aminopyrimidine of
TPP, thus inactivating the catalytic cycle (10, 11, 12). The kinetics of protonation can be tracked
spectrophotometrically (10). Protonated NTZ is no longer biologically active, but the anion is
readily regenerated under physiological conditions (pKa ~ 6.18) (11). Unlike many nitro-
3

88
89
containing drugs, the nitro group of NTZ, a weak acid, is not susceptible to nitroreduction or
otherwise chemically modified (11, 12, 13). Conceptually, therapeutics that target the function of
a vitamin cofactor, itself a small molecule, are unlikely to be amenable to mutation-based drug
resistance (11, 14, 18).
90
91
92
NTZ is largely retained in the intestine where it is used for the treatment of infections
caused by Cryptosporidium and Giardia (15). Based on several in vitro studies (16, 17) and
recognizing the potential for a systemic derivative, we chemically interrogated the NTZ scaffold
and from ~350 derived analogues, identified an attractive candidate amixicile (Figure 1) that
both retained potency and selectivity for PFOR targets and possessed good pharmacokinetic
properties (10, 11, 13, 14, 15). In preclinical studies, amixicile showed equivalence with
vancomycin and other mainline therapeutics in treatment of Clostridium difficile infections (CDI)
and similarly with metronidazole in treatment of Helicobacter pylori infections in mouse models
(14,18). Importantly, amixicile did not accumulate in the mouse cecum or alter the gut
microbiome of healthy animals (18). Based on serum binding, it has been suggested that
amixicile most likely concentrates in areas of mucosal inflammation via serum leakage where it
is active locally against offending susceptible microbes (18). Amixicile differs from NTZ by
replacement of the acetoxy group on the benzene ring with propylamine (see Figure 1). We
have used a combination of in silico PFOR docking simulations and validation via direct PFOR
inhibition assays and minimal inhibitory concentration (MIC) determinations to direct lead
optimization of the amixicile scaffold. Here we report on several modifications to the amixicile
scaffold that improve in vitro activity against several susceptible pathogens.
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
4

114
115
116
117
118
119
120
Materials and Methods
Determination of MIC values for H. pylori and C. jejuni (micro dilution). H. pylori
stain 26695 was grown over night at 37 °C under microaerobic conditions in either Brucella
broth (BB) or Brain Heart infusion (BHI) medium supplemented with 7.5% serum (4). C. jejuni
strain H840 was grown in BB medium without supplementation (11). For the microdilution
assay, bacterial cultures were diluted to a final OD600 of 0.03 for H. pylori and 0.01 for C. jejuni
and 100 μL was dispensed into wells of a 96 well microplate (12, 13). Analogues, either in
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
dimethyl sulfoxide (DMSO) or water, were diluted serially starting at 32 μg/mL in the microtiter
plate and the DMSO concentration was always less than 1%. DMSO and NTZ served as
controls. Plates were incubated with gentile shaking at 37 °C in a microaerobic incubator (7%
O2 and 10% CO2). The turbidity in the wells was read visually at 27 h or with a plate reader
(Molecular Dynamics). MIC is defined as the concentration of drug that produced no detectable
bacterial growth (12).
Determination of MIC Values for C. difficile (agar dilution). C. difficile strain VPI
10463 was grown anaerobically overnight in chopped meat medium (Anaerobe system) from
stock, and it was subcultured to a new chopped meat medium for 5 h at 37 °C (13, 14). It was
standardized to an optical density of 0.1 at OD600. Analogues were then diluted into the agar
media at concentrations ranging from 0.125 - 8 μg/ml. Ten-microliter volumes of the
standardized inoculum were delivered to the surface of the agar plates. The number of viable
bacteria contained in each inoculum was approximately 7 x 104 and 3.5 x 104 organisms. The
plates were incubated for 18 h in an anaerobic chamber and were read visually for growth or no
growth. Anaerobic plates containing no compound were used as controls.
5

138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
PFOR enzyme assay. H. pylori PFOR enzyme was overexpressed and purified from E.
coli as described previously (11, 13). Enzymatic assays were carried out at 25 °C in 1-mL
cuvettes in a modified Cary-14 spectrophotometer equipped with an OLIS data acquisition
system (On Line Instrument Co., Bogart, Georgia). PFOR (EC 1.2.7.1) was assayed under
anaerobic conditions with 100 mM potassium phosphate (pH 7.4), 10 mM sodium pyruvate, 5
mM benzyl viologen (BV; ε=9.2 mM-1 cm-1 at 546 nm), 0.18 mM CoA, and 1 mM MgCl2. The
reaction was started by addition of enzyme, in the presence or absence of inhibitor (NTZ or its
derivative in concentration of 40 μM) and the reduction of redox-active BV dye was monitored at
546 nm. Inhibition of PFOR was expressed in percent with NTZ set at 50%.
Docking simulations. Docking simulations (MOE; molecular operating environment
20010.0 release by Chemical Computing Group) with the 1.87 Å crystal structure of PFOR from
Desulfovibrio africanus (PDB 1B0P) (19, 20, 21) were performed to rationalize the proposed
mechanism of action of NTZ and AMX (11, 12). Anionic NTZ, TZ, and amixicile were docked
into the PFOR crystal structure using the triangle match algorithm, biasing the nitro group to
remain with 5 Å of TPP, and potential modes of binding were assessed by estimating the free
energy of binding using the Merck molecular force field and the London dG scoring function,
which estimates enthalpic interactions within the binding pocket, the energy of desolvation, and
the cost of rigidifying freely rotatable bonds. The docking studies indicated that the 5-nitro group
of 2ATdirectly interacts with TPP and the binding pocket’s residues Arg B114 and Thr B31.
Additionally, the amide carbonyl accepted a hydrogen bond from Asn B996. The in silico studies
suggest that improved binding of amixicile might result from electrostatic interactions of the
amine tail with the acidic side chain of Asp B456 that mediates acetylation of coenzyme A in the
pocket, taking advantage of unexplored chemical space and leading to better selectivity. This
knowledge, combined with the previous studies of benzene ring substitution, allow for further
derivatives of AMX to be explored.
6

164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
General chemistry methods, chemical synthesis and analysis of analogues.
Details on chemistry, synthesis and purity of compounds can be found in the supplemental
materials (S1).
Results
Our strategy for further development of the amixicile scaffold is based on previous
unpublished work associated with docking simulations with nitazoxanide (NTZ) and its active
metabolite tizoxanide (TZ, the phenol) with the crystal structure of PFOR from Desulfovibrio
africanus (20, 21). Previous mechanistic and NMR studies indicated that NTZ interferes with the
PFOR enzymatic reaction as a competitive inhibitor of pyruvate binding to TPP (11). Therefore,
the PFOR crystal structure in which pyruvate is bound (but not reacting) was chosen for the
development of the homology model. Since the carboxylate moiety of pyruvate forms a
hydrogen bond to the thiamine pyrophosphate (TPP) cofactor of PFOR (19), the electron-rich,
isosteric nitro group of NTZ was placed in the active site in a similar conformation. The head
group, amide linker and benzene ring of amixicile were then built into the PFOR pocket and a
library of conformers was assembled and screened. Those conformations that had the lowest
calculated binding affinity in a rigid enzyme model were minimized and re-scored, this time
allowing both the ligand and amino acid side-chains in the active site to resolve torsional strain
and ligand collisions, while holding the protein backbone and non-active site residues rigid. The
best scoring orientation became the template for second-generation NTZ analogues. With this
as a guide, various tail groups and benzene linker substitution patterns were built, screened and
scored to predict their efficacy in vitro. Figure 2A depicts the interaction of TPP with pyruvate
showing the carbonyl oxygen interacting with the aminopyrimidine. The carbanion of the thiazole
then attracts the positively charged carbonyl of pyruvate to form the lactyl-TPP intermediate. As
7

190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
seen in Figure 2B, the anion of the tizoxanide is predicted to interact with TPP via the N4’-
aminopyrimidine. In this model, TPP forms a V shape that enables the thiazole carbanion to be
relatively stable and the N4’ to be essentially protonated in the activated state (19). Amixicile
appears to bind and function mechanistically similar to NTZ, but the steric bulk and cation of the
propyl amine better fills the substrate pocket and forms a weak ionic interaction with Asp B456.
As seen in Figure 2D (docking with amixicile), the chemical space of the catalytic cavity is rather
large (30 Å), suggesting room for further chemical development. Alignments of other PFOR
amino acid sequences (BLASTP) in key regions show that several active site amino acids, such
as ThrB31, GluB64 and ArgB114, are highly conserved. Docking simulations with amixicile
show probable interactions, not only with the N4’ of the aminopyrimidine, but also hydrogen
bonds with ArgB114 and ThrB31 with the nitro group oxygens, AsnB996 with the keto group of
the amide bond and AspB456 with the protonated amine of the propylamine moiety (see Figure
3A). Figure 3B shows the relationship of TPP, amixicile and the three 4Fe:4S redox centers
which show considerable chemical space in which to drive further development. Moreover, the
benzene ring fits into a trough in the enzyme (Figure 3B) that might be amenable to SAR
development to improve binding efficiency.
A conclusive SAR was required to set rational structural limits and propose compounds
that would optimize binding interactions at PFOR. Previous studies by our group and those of
others have focused on the 2ANT moiety and benzene ring (12, 13, 14). The addition of a
propylamine tail to the NTZ scaffold introduced a new region for optimization that could lead to
more potent inhibitors in whole cells. Some thought was also directed to the possibility that the
reaction of analogues with TPP might be augmented by additional substituents on the ring that
might kinetically improve the inhibitory action. To design such inhibitors, the amixicile scaffold
was broken down into three parts for structural analysis (Figure 4). Studies then focused on the
linker and tail regions for further development.
8

216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
Tail region optimization. To begin optimizing the structure of amixicile, the length of the
tail region was analyzed. The tail in the ortho- position was systematically shortened,
lengthened and rigidified to determine the optimal length and presentation allowed in the PFOR
pocket. Compounds were prepared by first coupling the commercially available methyl
benzoate or aldehyde to the desired tail region coupling partner under Suzuki conditions (Figure
5) (22). Once the tail group was added, the resulting ester or aldehyde was hydrolyzed or
oxidized to the respective carboxylic acid via saponification or Pinnick oxidation (13, 23).
Finally, the 2-amino-5-nitrothiazole (2ANT) head group was appended using EDC coupling
methods previously described and the amine was deprotected to form the hydrochloric acid salt
(12, 13). This provided a logical series of derivatives to assess the spatial tolerance of the
PFOR pocket. The computational study indicated a potential electrostatic interaction between
the amine tail and Asp B456 that could be amplified by adjusting the length and structural fluidity
of the tail group. The small library was subjected to direct PFOR inhibition assays to assess
preliminary activity at the enzyme (Figure 6) (11,12).
Modifying the amine tail from zero to four carbons allowed for the area of the PFOR
pocket facing away from the active site to be analyzed. Preliminary PFOR inhibition assays
revealed that activity drops when the spacer is less than three carbons. When the spacer was
four carbons, PFOR inhibition increased but this did not translate into lower, more potent, MIC
values against target pathogens. Due to the high cost of the coupling partner required for 4.3
and 4.4, further optimization studies were performed using the propylamine tail of amixicile for
preliminary analysis.
Optimization of the benzene ring of amixicile. A new series of amixicile derivatives
were synthesized using the propylamine tail in the ortho-position (Figure 7). Previous SAR
studies, while largely inconclusive, provided clues to substituents that could further increase
inhibitory activity of the PFOR enzyme and in whole cells (20). Electron-donating and
9

242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
withdrawing groups (EDG and EWG, respectively) were added to the benzene ring linker region
of amixicile to establish a correlation between the spatial and electronic properties of the ring in
the pocket and activity. The inhibitory activities of these analogues at 40 μM are listed in Figure
7. All of these compounds showed greater inhibitory activity than did NTZ and amixicile.
Importantly, these compounds retained selectivity for PFOR containing bacteria and showed no
inhibitory activity against non-PFOR-containing bacteria like E. coli and S. aureus (>32 μg/mL,
data not presented). Similarly, CC50 determinations with HeLa cells revealed none of these
derivatives was toxic at concentrations below 100 μg/ml (data not presented). In accordance
with previous trends, R2 and R3 -substituted compounds 4.12 – 4.17 were comparable to
amixicile in potency against C. difficile in vitro by MIC tests, with R3 CH3 4.12 showing the
greatest improvement in activity (0.06 μg/ml versus amixicile 0.125 μg/ml). Despite this
enhanced activity, the compounds were not as active as NTZ against C. difficile (Figure 7).
Interestingly, meta-substituted (R2) compounds 4.10 and 4.11, had the second lowest MIC value
of 0.0625 μg/ml against C. difficile and much greater percent inhibition of PFOR inhibition of
97% and 93% respectively. MIC results are often a function of drug uptake or efflux, factors that
do not influence IC50 tests. These data show a marked improvement compared to previous
studies of compounds without the propylamine tail (12, 13). Of all the substitution patterns,
electron withdrawing groups in the R2 (meta-) position have shown the greatest increase in
activity and potency.
The meta-substitution pattern of 4.8, 4.10 and 4.11 corresponds well with in silico
docking studies of amixicile, which shows a cavity that can accommodate substituents in the R2
position. This pocket, while small, was calculated to be large enough to accept small
substituents, such as fluorine, chlorine, methyl, and trifluoromethyl (CF3). Even though these
groups have different electron withdrawing effects, all four substituents improved potency,
suggesting that steric effects drive the observed improvements in performance.
10

267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
Despite the enhanced potency, there continue to be inconsistencies between in vitro
enzyme and in vivo whole cell data. The percent PFOR inhibition has nearly doubled from NTZ,
yet the MIC values for respective compounds remain only two or three fold those of NTZ (Figure
7). This may not be detrimental to the overall goal, however, as amixicile still outperforms NTZ
in a CDI mouse model even with this discrepancy (14). While hydrophobicity can affect drug
uptake or efflux, evaluation of cLogP values for the analogues did not correlate with biological
activity.
Discussion
We have used a combination of docking simulations, enzymology and MIC testing of
selected susceptible bacteria to direct development of second generation PFOR inhibitors
based on the amixicile scaffold. Initial mechanistic studies indicated that these 2ANT
derivatives exhibit extensive electron delocalization of the negative charge generated by amide
deprotonation and our docking simulations indicated that the resonance contributor with an
anionic 5-nitro group is the most likely electronic resonance form to interact with and abstract a
proton from the N4’ amino pyrimidine group of TPP. Docking simulations also supported
previous conclusions that the nitro group is essential to the biological activity and that this
activity can be influenced by coupled substituents (12, 13, 14). Docking simulations also
confirmed that the chemical space within PFOR was large and therefore can accommodate
large bulky substituents (12, 13). We determined that the propylamine tail chain length was
optimized and is predicted to interact with Asp B456. Since protonated amixicile and analogues
are rapidly deprotonated by water (pKa 6.2), these secondary interactions within this chemical
11

293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
space might maintain drug proximity to TPP and thereby improve efficiency against the
reactivated enzyme. In this regard, we have shown that a large variety of heterocycles similarly
coupled to 2ANT exhibit increased inhibitory action in direct PFOR assays and in potency
against a wide range of PFOR containing bacteria and anaerobic parasites (11).
The second premise driving SAR was that on the phenyl ring strong electron
withdrawing groups would enhance biological activity at PFOR and electron-donating ones
would not. However, our studies found that halide substitutions at position (R2) were only
incrementally improved over methyl and trifluoromethyl groups as indicated by PFOR inhibitory
testing and by MIC for the selected susceptible pathogens. We also found that rigidifying the
propylamine tail of amixicile into the cyclized piperidine (Figure 6) also improved potency
against PFOR. However, these compounds were less soluble in water which might affect
bioavailability for systemic use. The structures are clearly amenable to further substitution by
halides, propylamine and perhaps other heterocycles in a directed SAR strategy. Similarly and
based on previous studies, the 2ANT head group is amenable to further development as
dinitrothiophene derivatives show potency against PFOR by enzyme assay and by direct MIC
tests (12,13). These derivatives have not been analyzed in docking simulations that might infer
additional interactions within the catalytic pocket.
We noted variation in potency (MIC) of compounds against PFOR expressing bacteria,
with C. jejuni exhibiting higher MICs when compared with C. difficile and H. pylori. Pyruvate
metabolism is far more diverse in C. jejuni suggesting that PFOR might be a less robust target
in this species (11, 14). Overall, our SAR strategy led to more potent analogues based on in
vitro studies. While docking simulations directed synthesis we found that potency for inhibition of
PFOR did not necessarily correlate with MIC. The greater potency of some analogues for Gram
positive C. difficile versus H. pylori or C. jejuni is likely due to penetrability through two
membranes in the Gram negatives. The length of the alkyl tail also affected aqueous solubility
as cLogP values increased with carbon length, again with the propylamine being optimal for
12

319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
both MIC and cLogP (amixicile). Whether these more potent leads also retain the excellent
DMPK, cytotoxicity metrics and efficacy of amixicile will require further study. Preliminary testing
of selected analogues in a mouse CDI model showed that analogue 4.11 (2-trifluoromethane
derivative) was well tolerated in a lethal challenge mouse CDI model (20 mg/kg/day for 5 days)
with minimal weight loss and no diarrhea. However a related analogue 4.10 (2-chloro) was not
well tolerated and infected animals exhibited greater weight loss and diarrhea. These
preliminary studies indicate that animal studies will be required to delineate toxicity versus
efficacy. Since the vast majority of new antimicrobials developed for the CDI market are poorly
absorbed, we cannot rule out the possibility that poorly absorbed analogues might be good
candidates for this indication.
In this study, the homodimeric PFOR from Desulfovibrio africanus was used in modeling
studies. PFORs can also be comprised of subunits ranging from 3 to 4 that form the
homodimeric oligomer that reconstitutes the catalytic site and in which the key amino acids are
conserved (9). For example, the H. pylori PFOR is the product of a 4 gene operon as is the
oxoacid: ferredoxin oxidoreductase (10, 11, 24). In C. jejuni and C. difficile, PFOR is expressed
from a single gene and forms a homodimer. Remarkably, the core structure, positioning of TPP
and the catalytic mechanism are functionally conserved. A QSAR study of the PFOR from
Entamoeba histolytica in docking simulations with NTZ and benzylthiazoles concluded that
these compounds could not replace TPP (25), confirming earlier studies (26). However, studies
with the E. histolytica enzyme (homodimer), while confirming our previously reported
mechanism, did not provide specific interactions of NTZ with the vitamin cofactor of PFOR.
Since the study did not model pyruvate, it is possible that the enzyme was not in a catalytically
activated state as set up in our modeling studies. Since considerable conformational change
occurs during activation and catalysis (19), we believe our docking simulations are a good fit to
the previously reported experimental data (19). Moreover, we confirmed experimentally with
13

344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
purified PFOR from E. histolytica that NTZ was both a substrate of and an inhibitor of the
enzyme which supports the proton abstraction mechanism (11).
One of the greatest challenges to the development of new antimicrobials is finding new
drug targets, especially ones with mechanisms that might escape or delay the inevitable
development of antimicrobial resistance. Our modeling simulations and SAR strategy show that
vitamin B1 is the target of NTZ, amixicile and analogues. Consistent with this notion, we and
others have not been successful using laboratory conditions to generate mutants resistant to
NTZ or amixicile (10, 11,14, 18). Similarly, there are no reports of resistance to NTZ from its use
clinically. While we anticipate that the analogues will also escape mutation based drug
resistance, this has not been confirmed experimentally. Our studies do not rule out second-site
mutations or changes in gene expression (efflux genes) that might contribute to increased
resistance to these analogues. In this regard, we do know that metronidazole-resistant strains of
H. pylori show some cross-resistance to NTZ, amixicile and analogues (18). The noted cross-
resistance appears to result from incremental increases in PFOR gene expression and to
compensatory metabolic activities that increase tolerance, rather than to mutations within PFOR
or vitamin that can be transferred genetically to susceptible strains (10). This noted incremental
tolerance can be overcome by drug concentrations above MIC that are readily achievable
clinically (18).
In summary, we have used docking simulations with amixicile together with direct
enzyme assay and MIC testing to identify analogues of amixicile with increased inhibitory action
against PFOR containing bacterial pathogens. Our studies further show that the PFOR target is
amenable to further space filling chemical development, as long as the anionic nitro group of the
thiazole is maintained. This interaction with the protonated N4’ aminopyrimidine is required for
inactivation of the PFOR catalytic cycle. To our knowledge, this is the first example of an
antimicrobial therapeutic that specifically targets the mechanism of action of a vitamin cofactor
14

369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
and not the enzyme directly. Perhaps similar drug discovery strategies can be used to probe
other unique microbial enzymes which rely on a vitamin cofactor for biological activity.
Acknowledgements
We thank Jennifer L. Tomlinson and Xia Wang for technical assistance. This project was funded
by National Institute of Allergy and Infectious Diseases, National Institutes of Health,
Department of Health and Human Services grants U01 AI075520 and R21AI111604 to PSH.
References
1. Silver LL. 2011. Challenges of antibacterial discovery. Clin Microbiol Rev. 24:71-109.
2. Devasahayam G, Scheld MW, Hoffman PS. 2009. Newer antimicrobials for a new century.
Expert Opinion on Investigational Drugs, 19:215-34.
3. Boucher HW, Talbot GH, Bradley JS, Edwards JE, Gilbert D, Rice LB, Scheld M,
Spellberg B, Bartlett J. 2009. Bad bugs, no drugs: no ESKAPE! An update from the
Infectious Diseases Society of America. Clin Infect Dis 48:1-12.
4. Sisson G, Jeong JY, Goodwin A, Bryden L, Rossler N, Lim-Morrison S, Raudonikiene
A, Berg DE, Hoffman PS. 2000. Metronidazole activation is mutagenic and causes DNA
fragmentation in Helicobacter pylori and in Escherichia coli containing a cloned H. pylori
RdxA(+) (Nitroreductase) gene. J Bacteriol. 182:5091-5096.
5. O'Dwyer K, Spivak AT, Ingraham K, Min S, Holmes DJ, Jakielaszek C, Rittenhouse S,
Kwan AL, Livi GP, Sathe G, Thomas E, Van Horn S, Miller LA, Twynholm M, Tomayko
J, Dalessandro M, Caltabiano M, Scangarella-Oman NE, Brown JR. 2015. Bacterial
15

394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
resistance to leucyl-tRNA synthetase inhibitor GSK2251052 develops during treatment of
complicated urinary tract infections. Antimicrob Agents Chemother. 59:289-298.
6. Payne DJ, Gwynn MN, Holmes DJ, Pompliano DL. 2007. Drugs for bad bugs: confronting
the challenges of antibacterial discovery. Nature Rev. Drug Discovery. 6:29-40.
7. Durrant JD, Amaro RE. 2015. Machine-learning techniques applied to antibacterial drug
discovery. Chem Biol Drug Des. 85:14-21.
8. Chalker AF, Minehart MW, Hughes NJ, Koretke KK, Lonetto MA, Brinkman KK, Warren
PV, Lupas A, Stanhope MJ, Brown JR, Hoffman PS. 2001. Systematic identification of
selective essential genes in Helicobacter pylori by genome prioritization and allelic
replacement mutagenesis. J. Bacteriol. 183:1259-1268.
9. Horner DS, Hirt RP, Embley TM. A single eubacterial origin of eukaryotic
pyruvate:ferredoxin oxidoreductase genes: Implications for the evolution of anaerobic
eukaryotes. Mol. Biol. Evol. 16:1280-1291.
10. Sisson G, Goodwin A, Raudonikiene A, Birks NJ, Han H, Mukhopadhyay A, Berg DE,
Hoffman PS. 2002. Enzymes Associated with the Reductive Activation and Action of
Nitazoxanide, Nitrofurans and Metronidazole in Helicobacter pylori. Antimicrob. Agents and
Chemother. 46: 2116-2123.
11. Hoffman PS, Sisson G, Croxen MA, Welch K, Harman WD, Cremades N, Morash MG.
2007. Antiparasitic drug nitazoxanide inhibits the pyruvate oxidoreductases of Helicobacter
pylori and selected anaerobic bacteria and parasites, and Campylobacter jejuni. Antimicrob
Agents Chemother. 51:868-876.
12. Ballard TE, Wang X, Olekhnovich I, Koerner T, Seymour C, Hoffman PS, Macdonald
TL. 2010. Biological activity of modified and exchanged 2-amino-5-nitrothiazole amide
analogues of nitazoxanide. Bioorg Med Chem Lett. 20:3537-3539.
13. Ballard TE, Wang X, Olekhnovich I, Koerner T, Seymour C, Salamoun J, Warthan M,
Hoffman PS, Macdonald TL. 2011. Synthesis and antimicrobial evaluation of nitazoxanide-
16

420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
based analogues: identification of selective and broad spectrum activity. ChemMedChem.
6:362-377.
14. Warren CA, van Opstal E, E. Ballard TE, Kennedy A, Wang X, Riggins M, Olekhnovich
I, Warthan M, Kolling GL, Guerrant RL, Macdonald TL, Hoffman PS. 2012. Amixicile: A
novel inhibitor of pyruvate: ferredoxin oxidoreductase shows efficacy against Clostridium
difficile in a mouse infection model Antimicrob. Agents Chemother. 56:1403-1411.
15. Gilles HM, Hoffman PS. 2002. Treatment of intestinal parasitic infections: A review of
nitazoxanide. Trends in Parasitol. 18:95-97.
16. Dubreuil L, Houcke X, Mouton Y, Rossignol JF. 1996. In vitro evaluation of the activities
of nitazoxanide and tizoxanide against anaerobes and aerobic organisms. Antimicrob
Agents and Chemother. 40:2266-2270.
17. Pankuch GA, Appelbaum PC. 2006. Activities of tizoxanide and nitazoxanide compared to
those of five other thiazolides and three other agents against anaerobic species. Antimicrob.
Agents Chemother. 50:1112-1117.
18. Hoffman PS, Bruce AM, Olekhnovich I, Warren CA, Burgess SL, Hontecillas Viladomiu
RM, Bassaganya-Riera J, Guerrant RL, Macdonald TL. 2014. Preclinical studies of
amixicile, a systemic therapeutic developed for treatment of Clostridium difficile infections
that also shows efficacy against Helicobacter pylori. Antimicrob Agents and Chemother.
58:4703-4712.
19. Chabriere E, Vernede X, Guigliarelli B, Charon M-H, Hatchikian EC, Fontecilla-Camps
JC. 2001. Crystal structure of the free radical intermediate of pyruvate:ferredoxin
oxidoreductase. Science. 294:2559-2563.
20. Alvarado O, Jana G, Delgado EJ. 2012. Computer-assisted study on the reaction between
pyruvate and the ylide in the pathway leading to lactyl-ThDP. J. Comput. Aided Mol. Des.
26:977-982.
17

445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
21. Charon M, Volveda A, Chabriere E, Pieulle L, Fontecilla-Camps JC. 1999. Structure and
electron transfer mechanism of pyruvate: ferredoxin oxidoreductase. Curr Opin. Struct Biol.
9:663-669.
22. Kamatani A, Overman LE. 1999. A Suzuki coupling method for directly introducing a
protected β-aminoethyl group into arenes and alkenes. convenient synthesis of phenethyl
and homoallylic amines. J. Org. Chem. 64: 8743-8744.
23. Li Y, Combs A, Yue E, Li H. 2014. Substituted fused aryl and heteroaryl derivatives as
PI3K inhibitors. US Patent No. 8,680,108.
24. Hughes NJ, Chalk PA, Clayton CL, Kelly DJ. 1995. Identification of carboxylation
enzymes and characterization of a novel four-subunit pyruvate:flavodoxin oxidoreductase
from Helicobacter pylori. J Bacteriol. 177:3953-9.
25. Scior T, Lozano-Aponte J, Ajmani S, Hernández-Montero E, Chávez-Silva F,
Hernández-Núñez E, Moo-Puc R, Fraguela-Collar A, Navarrete-Vázquez G. 2015.
Antiprotozoal nitazoxanide derivatives: Synthesis, bioassays and QSAR study combined
with docking for mechanistic insight. Curr Comput Aided Drug Des. 11:21-31.
26. St Maurice M, Cremades N, Croxen MA, Sisson G, Sancho J, Hoffman PS. 2007.
Flavodoxin:quinone reductase (FqrB): a redox partner of pyruvate:ferredoxin oxidoreductase
that reversibly couples pyruvate oxidation to NADPH production in Helicobacter pylori and
Campylobacter jejuni. J Bacteriol. 189:4764-4773.
18

471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
Figure 1. Structures of nitazoxanide and amixicile. Tizoxanide is the phenol of NTZ.
Figure 2. Docking simulations and mechanism of action. A. Interaction of pyruvate (blue)
with the protonated N4’ aminopyrimidine of TPP. B. Projected interaction of tizoxanide (TZ, red)
with the N4’ aminopyrimidine of TPP. C. Spatial orientation of nitazoxanide interacting with TPP
based on docking simulation. D. Docking simulation showing amixicile in the pocket of PFOR.
19

497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
Figure 3. Amixicile presentation in the PFOR pocket. A. In silico docking studies suggest
that the propylamine tail of amixicile forms an electrostatic interaction with aspartic acid residue.
Note also the interactions with TPP and with Thr B31, Asn B996 and Arg B114. Image
generated with MOE software. B. A crevice in the R2 position of amixicile explains increased in
vitro activity of substitutions at the R2 position on the benzene ring.
20

523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
Head
O
N
S
NO2
N
Linker
H
+
NH3
Tail
Figure 4. Inhibitor design from amixicile. The scaffold of amixicile was divided into three
sections for second-generation SAR studies: the head group region (blue), linker region (pink)
and tail region (green).
21

549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
O
Y
O
R
O
R
N
X = OMe: a
X = H: b
NO2
Suzuki
S
X
X
N
H
c, d
X = OMe, H
Y = I, Br
R Group
Coupling Partner
Conditions
R Group
Steps
Compound
+
NH3
Purchasable
NHBoc
c; d
4.1
B
+
PdCl2(dppf),
3 M K3PO4, THF
NH3
4.2
NHBoc
a; c; d
a; c; d
b; c; d
NHBoc
NHBoc
+
B
NHBoc
NHBoc
NH3
See above
AMX
B
+
PdCl2(dppf),
3 M K3PO4, DMF
NH3
NHBoc
NBoc
4.3
O
B
Pd(PPh3)4, CsCO3,
PhMe, MeOH
a; H2, Pd/C;
c; d
+
NH2
4.4
O
NBoc
NBoc
O
B
+
See above
a; c; d
NH2
4.5
O
NBoc
a) X = OMe: LiOH.H2O, H2O/THF/MeOH; b) X = H: NaH2PO4, 2-methyl-2-butene, THF/H2O/tBuOH;
c) 2-amino-5-nitrothiazole, EDC, HOBt, DMAP, THF, Δ; d) TFA/DCM, HCl.
Figure 5. Chemical strategy for tail assemblies. Suzuki coupling was performed to append
the desired amine tail to the methyl benzoate or benzaldehyde linker region, after which a
Pinnick oxidation or saponification was performed to convert the carbonyl to the carboxylic acid.
An EDC aminde coupling was performed to attach the 2ANT head group followed by
deprotection of the amino group.
22

575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
O
R
N
NO2
S
N
H
PFOR Inhibition
MIC (μg/ml)
Compound
R Group
cLogP
a
μ
[40 M], %
H. pylori C. difficile
+
0.46
0.79
4.1
4.2
NH3
51
4
n.d.
>10
.
+
n.d.
n.d.
NH3
AMX
4.3
1.17
1.70
1.81
75
91
88
95
0.5
2
0.125
+
+
NH3
NH3
1
4
1
4.4
8
+
+
NH2
NH2
4.5
2.81
4
Figure 6. Biological activity of R group tail constructs. The cLogP was inferred using
ChemDraw. PFOR inhibition in triplicate was determined as the percent inhibition by
compounds at 40 μM and indexed to NTZ which is 50% inhibition. MIC tests for H. pylori and C.
difficile (μg/ml) and the results of triplicate determinations. N.d. means not determined.
596
597
598
599
23

600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
R1
O
N
R2
R3
NO2
S
N
H
+
NH3
R1 = F;
4.6
R
R
R
R
R
3 = CH3
3 = OCH3 4.13
3 = CN
3 = CF3
3 = Cl
4.12
R
R
2 = CH3 4.7
2 = F
4.8
4.14
4.15
4.16
4.17
R2,3 = F 4.9
2 = Cl
4.10
R
R2 = CF3 4.11
R3 = F
μ
MIC ( g/mL)
PFOR Inhibition
Analoguea cLogP
H. pylori
C. difficile
C. jejuni
μ
[Drug] = 40 M (%)
NTZ
AMX
4.6
2.2
50
75
1
1
0.125
0.25
1
4
8
1.17
0.91
1.67
1.32
1.40
1.89
2.08
1.67
1.25
0.67
2.08
1.89
1.32
95
0.5
1.5.
0.125
0.5-1
1
0.25
4.7
n.d.
93
0.06.
0.125
0.125
0.0625
0.0625
0.06
n.d.
4.8
4
4
4
4
4.
8
4.
8
4
8
4.9
90
4.10
4.11
4.12
4.13
4.14
4.15
4.16
4.17
97
93
0.5-1
1
n.d.
n.d.
n.d.
n.d.
94
0.5
4
0.125
n.d.
1
0.125
0.164
0.125
0.5
1
86
Figure 7. Ring substitutions of the amixicile scaffold. The cLogP was determined with
ChemDraw, PFOR inhibitory activity as described in Figure 6 and MIC determinations (μg/ml)
for H. pylori, C. difficile and C. jejuni. CC50 determinations with HeLa cells (Alomar blue) were
>100 μM and are not depicted. Not determined (n.d.). All MIC tests are the mean of 3 to 6
experiments performed in triplicate and for some analogues a range is presented.
aR = H unless otherwise noted.
24