Metronidazole-triazole conjugates: Activity against Clostridium difficile and parasites

Metronidazole has been used clinically for over 50 years as an antiparasitic and broad-spectrum antibacterial agent effective against anaerobic bacteria. However resistance to metronidazole in parasites and bacteria has been reported, and improved second-generation metronidazole analogues are needed. The copper catalysed Huigsen azide-alkyne 1,3-dipolar cycloaddition offers a way to efficiently assemble new libraries of metronidazole analogues. Several new metronidazole-triazole conjugates (Mtz-triazoles) have been identified with excellent broad spectrum antimicrobial and antiparasitic activity targeting Clostridium difficile, Entamoeba histolytica and Giardia lamblia. Cross resistance to metronidazole was observed against stable metronidazole resistant C. difficile and G. lamblia strains. However for the most potent Mtz-triazoles, the activity remained in a therapeutically relevant window.


Introduction
For over 50 years the 5-nitroimidazole antibiotic metronidazole 1 has been in clinical use as a broad-spectrum agent for treatment of Gram-positive and Gram-negative anaerobic bacterial infections as well as parasitic infections [1]. Reduction of the nitro group to the nitro radical anion by electron carriers in an anaerobic environment leads to decomposition to form toxic metabolites, which cause DNA damage and nonspecific macromolecular damage leading to cell death [2]. However, metronidazole resistance has been observed in both parasites [3,4] and anaerobic bacteria [5,6]. Over the last ten years metronidazole has also been extensively used to treat Clostridium difficile infection (CDI), an intestinal infection that causes life-threatening severe diarrhea, abdominal pain and fever [7]. The spores produced by C. difficile are resistant to heat and alcohol based disinfectants, complicating eradication and promoting hospital-based outbreaks. Metronidazole is one of three antibiotics routinely prescribed to treat the disease [8]. Metronidazole is effective against mild to moderate C. difficile infection, but, while not wide-spread, resistance has been observed in clinical isolates [9]. To this end, the development of next generation metronidazole analogues that can overcome resistance is therapeutically important. Beena et al. previously reported the synthesis of metronidazoletriazole conjugates (Mtz-triazoles) with potent activity (reported as IC 50 values 0.06e0.35 mg/mL) against the facultative anaerobic bacteria Staphylococcus aureus, Staphylococcus epidermis, Escherichia coli and Pseudomonas aeruginosa [10]. Metronidazole is generally not active against Gram-negative and Gram-positive facultative anaerobic bacteria, so the potent activities of Mtztriazoles observed by Beena et al. are suggestive of a different mode of action to metronidazole. Given that the active 5nitroimidazole core ("warhead") in metronidazole is maintained in Mtz-triazoles, and that the hydroxyl group of metronidazole is amenable to modification [11] (Fig. 1), we reasoned that this class might also exhibit potent activity against the anaerobic bacteria and parasites for which metronidazole is used as a treatment. Therefore, we believed Mtz-triazoles warranted further investigation for both antimicrobial and antiparasitic activity under anaerobic conditions.
Independently, during the course of this work, Miyamoto et al. [12] reported a similar approach to Mtz-triazoles, prepared by reaction of six different 5-nitroimidazole azides with a library of alkynes. The compounds were prepared for testing by dilution of the crude reaction mixtures (>85% purity by LCMS) with dimethyl sulfoxide. This methodology bypassed the bottleneck of compound purification and characterization and allowed for the rapid assessment of activity. The solutions of crude material were tested against the parasites Giardia lamblia and Trichomonas vaginalis, the microaerophilic bacteria Helicobacter pylori, the anaerobes C. difficile and Bacteroides fragilis and the facultative anaerobic bacteria E. coli. It is possible that the crude mixtures contained unreacted starting azide 3, which we have shown in this study retains activity, and may influence the reported crude compound potency. A key finding by Miyamoto was that Mtz-triazoles were often active against metronidazole resistant (MtzR) strains of T. vaginalis (47% of compounds) and G. lamblia (100% of compounds), but not against H. pylori frxA and rdxA double mutant (1.4% of compounds) [12]. From this result, we hypothesized that Mtztriazoles may possess activity against metronidazole resistant C. difficile. More recently Beena et al. described the activity of Mtztriazoles against the anaerobic protozoan Entamoeba histolytica [13]. They tested a library of 10 Mtz-triazoles and found 4a, 4h, 4p and 4q to have potent activity (IC 50 ¼ 0.008e2.36 mM) against E. histolytica HM1: IMSS [13].
We now report an expansion and advancement of these approaches with the synthesis of a small library of purified Mtztriazoles 4a-t (>95% purity, Scheme 1), including ten novel derivatives (4b, 4c, 4f, 4g, 4i, 4j, 4l, 4o, 4s and 4t). We also explore the antimicrobial and antiparasitic properties of the set of Mtztriazoles and the parent azide 3 (>95% purity) and determine their activity against additional microbial targets that have not previously been examined. We evaluated the activity of Mtztriazoles against the parasites E. histolytica and G. lamblia, multiple pathogenic strains of the anaerobe C. difficile, the microaerophile H. pylori and the facultative anaerobic bacterial ESKAPE pathogens E. coli, methicillin resistant S. aureus (MRSA), Klebsiella pneumoniae, Acinetobacter baumanni and P. aeruginosa. The antibacterial activity was measured against a stable MtzR C. difficile strain (CD26A54_R), the parent C. difficile strain with elevated metronidazole minimum inhibitory concentration (MIC) (CD26A54_S) and MtzR H. pylori clinical isolates, while cytotoxicity of the compounds was evaluated against mammalian liver (HepG2) and kidney (HEK293) cell lines.

Synthesis of Mtz-triazole library
Mtz-triazoles were synthesized from metronidazole via an azide intermediate by activation of the hydroxyl group with methanesulfonyl chloride (Scheme 1). Displacement of methanesulfonate 2 with sodium azide provided the desired azide 3 by nucleophilic substitution. The azide substituent was then reacted with a library of alkynes to give 4a-t by copper-catalysed Huisgen 1,3-dipolar cycloaddition with copper sulfate and sodium ascorbate in methanol, with heating to 45 C and/or additional reagents added if monitoring indicated the reaction was incomplete. All alkynes utilized were commercially available with the exception of pyrazole alkyne 6. Pyrazole alkyne 6 was prepared by reacting pyrazole 5 with propargyl bromide in the presence of potassium carbonate and t-butylammonium bromide as a phase transfer catalyst (Scheme 1) [14]. All compounds were characterized by 1 H and 13 C NMR, LCMS and HRMS and detailed experimental procedures and characterization are provided in the supplementary information.

MIC assays
All compounds were tested for activity against a set of anaerobic and VRSA NARSA VRS10) using broth microdilution. C. difficile strains were grown at 37 C in a COY anaerobic chamber (5% H 2 , 10% CO 2 , 85% N 2 ). H. pylori strains were grown at 37 C under a microaerobic atmosphere (5% O 2 , 10% CO 2 ). All facultative anaerobic bacteria were grown at 37 C with normal atmospheric oxygen levels. All experiments were performed in duplicate with metronidazole, vancomycin, linezolid or colistin as positive controls for relevant strains (see Table 1). Positive growth control of bacteria and DMSO as well as a negative control of only media were included for every plate. Full assay details are provided in the supplementary information.

Cytotoxicity
Compounds were tested for cytotoxicity against mammalian HepG2 and HEK293 cell lines, as detailed in the supplementary information.

Design of metronidazole-triazole conjugate library
The Mtz-triazole library was designed to contain a variety of structural groups with some compounds identical to those reported by Beena et al. (4a, 4h, 4p and 4q) [10]. Initial results against C. difficile showed that hydrophobic 4a (R ¼ phenyl) maintained activity relative to metronidazole 1 while more hydrophilic 4p (R ¼ hydroxyethyl) and 4q (R ¼ hydroxymethyl) lost activity. Therefore the subsequent library was biased towards exploring variance of the aromatic ring substitutions, extension of the position of the aromatic ring relative to the triazole core and replacement of the phenyl group with various heterocycles, with several of the latter selected due to their rating in terms of medicinal chemistry 'developability' [19]. Amine 4t, and acids 4r and 4s were included to explore the structure activity relationships of nonaromatic ionisable groups. Compounds 4a, 4d, 4e, 4h, 4k, 4m, 4n, 4q and 4r were reported by Miyamoto [12], but 4a, 4h and 4q were not tested against C. difficile. Compounds 4b, 4c, 4f, 4g, 4i, 4j, 4l, 4o, 4s and 4t are described for the first time.

Antimicrobial and antiparasitic activity of metronidazoletriazole conjugates
The antimicrobial and antiparasitic spectrum of action of Mtztriazoles was assessed against the anaerobic bacteria C. difficile and the anaerobic parasites E. histolytica and G. lamblia (Table 1). Variations to the phenyl group of 4a were well tolerated and activity of 4b-e was maintained against C. difficile, E. histolytica and G. lamblia (Table 1). Phenyl derivatives 4a-e were several fold more active against G. lamblia than metronidazole. Pyridine 4h and thiophene 4n heterocycles also possessed broad-spectrum activity against C. difficile, E. histolytica and G. lamblia. However, we found that 4h was not as potent compared to metronidazole in the ATPbioluminescence parasite assay in contrast to the eosin-stain method used by Beena et al. [13] The pyrazole 4g and pyrimidine 4i-j maintained activity against C. difficile within one to two 2-fold dilutions of metronidazole. Compounds 4g and 4i-j were still active against G. lamblia compared to metronidazole, but were several fold less potent than the phenyl derivatives 4a-e. The pyrazole 4g and pyrimidines 4i-j did not inhibit E. histolytica at 25 mM, showing Scheme 1. Synthesis of metronidazole-triazole conjugate library. i) MsCl, TEA, DCM, 2 C to rt, 80 min, 85% yield ii) NaN 3 , DMF, 50 C, 21 h, quant. yield iii) CuSO 4 , sodium ascorbate, MeOH, rt or 45 C, 20 mine48 h, 9e97% iv) propargyl bromide (80% w/v in toluene), K 2 CO 3 , TBAB, rt, 3 days, 16%.
that the R group can be used as a handle to tune the selectivity of this class towards different organisms. Compound 4k (benzyl) and 4m (CH 2 NMe-benzyl) maintained broad-spectrum activity but 4l (CHOH-phenyl) was inactive against E. histolytica at 25 mM.
Several compounds were inactive, or weakly active, against all three microorganisms at the highest concentration tested. These included the non-aromatic thiomorpholine dioxide 4o, amine 4t and the carboxylic acids 4r and 4s. The methyl hydroxyl 4q and ethyl hydroxyl 4p were inactive against E. histolytica and G. lamblia and weakly potent against C. difficile, continuing the trend of reduced activity with more polar substituents.
We also demonstrate that the parent azide 3 possesses potent activity against C. difficile, E. histolytica and G. lamblia. Therefore, when assessing combinatorial-like libraries of crude material for biological activity, the activity of any unreacted parent compounds is an important consideration, particularly when the compound warhead is maintained.
The MICs of 1, 3 and 4a-t were >32 mg/mL against the representative ESKAPE pathogens (MRSA, E. coli, K. pneumoniae, A. baumannii and P. aeruginosa) tested under aerobic conditions. In addition, the MIC values of a subset of compounds (metronidazole 1, azide 3, 4a (phenyl) and 4q (CH 2 OH)) were all >64 mg/mL against a panel of 8 additional drug resistant S. aureus strains, vancomycin resistant E. faecalis, vancomycin resistant E. faecium and multidrug resistant S. pneumoniae. This lack of activity is contrary to Beena's report, but consistent with the reported inactivity of 378 Mtztriazoles against E. coli in Miyamoto's study, and the inactivity of metronidazole against facultative anaerobic bacteria.

Activity of metronidazole-triazole conjugates against C. difficile strain panel
The activity of the Mtz-triazoles did not vary significantly against multiple strains of C. difficile, including two major pathogenic strains of NAP1/027 and a VPI10463 strain associated with epidemics ( Table 1). The MICs of individual compounds typically remained within one 2-fold dilution against the four strains of C. difficile tested. This is important, as viable drug candidates must possess appropriate broad strain coverage. A review of antibiotic drug candidates in development found that activity against multiple strains of C. difficile could vary substantially [20]. This can be explained since C. difficile has a highly mobile, mosaic genome and there is wide strain diversity between isolates [21,22]. Isolates can be divided across five main phylogenetic clades, multiple ribotypes and toxinotypes [23]. Recently, genomic epidemiology studies examining and tracking C. difficile outbreaks in Europe found that a high proportion of strains causing infection were not related to prior infectious strains [22]. Therefore, there is a large pool of genetically diverse strains in the community and selection and spread of intrinsically resistant strains could occur. However, the broad-strain activity of Mtz-triazoles against C. difficile in this study encourages further development of this class.

Activity of metronidazole-triazole conjugates against metronidazole resistant C. difficile
While metronidazole resistance has been reported in the clinic [9,24], performing antimicrobial susceptibility testing against resistant C. difficile isolates is problematic since the resistance phenotype is unstable and often reported to be lost on freeze thaw cycles or on passaging of the isolates [5]. However, Lynch et al. have reported a stable metronidazole resistant (MtzR) C. difficile clone [25]. The activity of Mtz-triazoles was assessed against this stable MtzR strain of C. difficile (CD26A54_R) and the parent metronidazole sensitive (MtzS) (CD26A54_S) strain, which lost the metronidazole resistance on freeze thawing, but still retained a slightly elevated MIC value to metronidazole compared to the control MtzS NAP1/027 strain. The MIC to metronidazole in BHIS broth at 48 h  Table 1). Although several compounds (4a, 4e and 4n) were active against MtzR C. difficile at 1 dilution lower than metronidazole (MIC ¼ 2 mg/mL vs 4 mg/mL), none of the Mtz-triazoles were active against MtzR C. difficile at levels comparable to the activity of metronidazole against MtzS strains (MIC ¼ 0.5 mg/mL). Reduced activity against the parent C. difficile strain CD26A54_S with intermediary metronidazole susceptibility was also observed.

Activity of metronidazole-triazole conjugates against additional G. lamblia strains including metronidazole resistant G. lamblia
Selected compounds (4a-f, 4h, 4k-n) were tested against two additional strains of MtzS G. lamblia (106 and 713) and one MtzR strain (713-M3) derived from the parent MtzS strain 713 ( Table 2). The Mtz-triazoles displayed similar activity against the additional MtzS strains (106 and 713) compared to G. lamblia strain WB. Encouragingly, the Mtz-triazoles were all more potent than metronidazole against MtzR G. lamblia 713-M3. However, with the exception of 4d, all of the Mtz-triazoles lost activity against the MtzR G. lamblia 713-M3 when compared to the parent MtzS strain 713, similar to the loss of activity against MtzR C. difficile. Mtztriazoles of scaffold 4 were also all observed by Miyamoto et al. to lose activity against MtzR G. lamblia 713-M3 and 106-2ID10. However, while activity diminished against the MtzR 713M strain, the Mtz-triazoles were often still much more potent than metronidazole itself, meaning that the EC 50 remained within a therapeutically meaningful window.
The activities of purified Mtz-triazoles (>95% purity) against G. lamblia strains 106, 713 and 713M were compared with the literature activities [12] determined with crude reaction mixtures (Supplementary Table 3). The activities were generally in close agreement, supporting the methodology used by Miyamoto [12], although a 3.7e8 fold difference was observed for Mtz-triazoles 4d, 4e and 4m against 1 strain of G. lamblia.

Activity of metronidazole-triazole conjugates against metronidazole resistant H. pylori
Selected compounds (4f-h, 4m and 4s) were tested against a panel of H. pylori strains using the CLSI agar dilution susceptibility method (Table 3) [27]. These strains included the reference strain 26695 and three clinical isolates 13/25 (metronidazole E-test MIC ¼ 2 mg/mL), 98/285 (E-test MIC ¼ 24 mg/mL) and 13/61 (E-test MIC ¼ 256 mg/mL) with a range of resistance levels to metronidazole determined by the E-test. The susceptibilities of the strains to metronidazole determined by the agar dilution method differed from the susceptibilities obtained from the E-test as has been described previously [28]. The methyl ester 4f was weakly active against H. pylori, while the carboxylic acid 4s was inactive against all H. pylori strains as observed with E. histolytica and G. lamblia as well. Compounds 4g, 4h and 4m were more active than metronidazole against the MtzS strains (26695 and clinical isolate 13/25) and 1e2 dilutions more active than metronidazole against the MtzR strains (clinical isolates 98/285 and 13/61) but they could not completely overcome metronidazole resistance, similar to the loss of activity against metronidazole resistant C. difficile.

Cytotoxicity
All compounds, including metronidazole 1, azide 3, and 4a-t showed no cytotoxicity against HepG2 and HEK293 cells at concentrations up to 100 mM (CC 50 ). This is consistent with the findings by Miyamoto et al. which did not observe cytotoxicity against HeLa cells at 50 mM [12].

Relationship of compound activity with compound properties
In order to understand the relationship between compound properties and activity, compound properties including logP (or logD at pH ¼ 7.4), molecular weight, topological polar surface area (tPSA), and predicted solubility (logS) were examined for correlation with activity against MtzS strains of C. difficile, E. histolytica and G. lamblia, with the activity expressed as pMIC (-log 10 MIC) or pEC 50 (-log 10 EC 50 )) (Supplementary Table 4). Inactive compounds were included in the analysis at the highest concentration of compound tested. A correlation can be observed between the logD values and the activity against G. lamblia (with R 2 (linear regression) ¼ 0.84, Supplementary Fig. 1) (Fig. 2). The logD of the most potent compounds (4a-e, 4h, 4k and 4n) against G. lamblia is between 1.5 and 3.0, with activity decreasing with lower logD. In contrast, only a mild correlation could be detected between logD and the activity against C. difficile (R 2 ¼ 0.68, Supplementary Fig. 1) and no correlation for activity against E. histolytica (R 2 ¼ 0.48, Supplementary  Fig. 1). This might explain why the compounds with aromatic phenyl, thiophene and pyridine groups were more potent against G. lamblia than E. histolytica and C. difficile. Increasing the hydrophobicity (logD) of compounds to improve potency is generally considered undesirable in drug development due to the tendency for increased metabolism [29] and increased promiscuity [30]. However, in this case the antibiotic metronidazole has a low logD to begin with, such that the relatively higher logD of the compounds explored in this study does not place the compounds in an undesirable chemical space. On the other hand, the site of infection of G. lamblia is the small intestine, where low (<1) or high (>3) logD values [29], associated with reduced systemic uptake, might be beneficial for efficacy. However, metronidazole is 100% oral bioavailable [31] and so the importance of bioavailability for in vivo efficacy against G. lamblia infections is unclear.

Conclusion
Mtz-triazoles were synthesized and assessed for activity against the anaerobe C. difficile, microaerophile H. pylori, the parasites E. histolytica and G. lamblia as well as facultative anaerobic Grampositive and Gram-negative bacteria. While no activity was observed against the facultative anaerobic bacteria, nine compounds were identified with potent broad spectrum activity against anaerobic organisms, while having no cytotoxicity against mammalian cell lines. Hydrophobic R groups such as the benzyl 4k, varyingly substituted phenyl derivatives 4a-e and heterocyclic R groups such as pyridine 4h and thiophene 4n were well tolerated and favored broad spectrum anaerobic activity. In contrast, polar R groups including pyrimidines 4i-j, carboxylic acids 4p and 4r and methyl amine 4t either resulted in loss of broad spectrum activity or were inactive against all organisms tested.
While active Mtz-triazoles displayed a narrow MIC range against multiple strains of C. difficile and G. lamblia, there was metronidazole cross resistance against the stable MtzR C. difficile strain CD26A54_R, MtzR G. lamblia strain 713M and clinical isolates of MtzR H. pylori. Mtz-triazoles 4a-t generally lost activity against MtzR C. difficile, MtzR G. lamblia and MtzR H. pylori. However, since the Mtz-triazoles were often much more potent than metronidazole against G. lamblia, the EC 50 remained within a therapeutically meaningful window. It is possible that surveying a more diverse chemical space of R group substituents or using alternative 5nitroimidazole scaffolds could provide access to Mtz-triazoles that are even more potent and therefore active against MtzR strains of C. difficile and G. lamblia. Future work will focus on identifying such compounds and explore their in vivo efficacy against anaerobic pathogens.