Screening novel antiviral compounds to treat Clostridioides difficile infections Authors: Brice J. Stolz, Ahmed A. Abouelkhair, Mohamed N. Seleem Published: 2024 ABSTRACT Clostridioides difficile is a major cause of nosocomial infections, often associated with individuals who have gut dysbiosis from previous antibiotic therapies. C. difficile infections (CDI) have a high recurrence rate and impose significant financial and mortality burdens on the healthcare system. Therefore, novel anti-C. difficile drugs are urgently needed to treat and reduce the severity and recurrence of infection. In this study, we screened a library of 618 antiviral drugs to identify a potential candidate for repurposing as novel anti-C. difficile therapeutics. Following our preliminary screening, we identified 9 novel compounds that inhibited C. difficile at a concentration of 16 μM or lower. Among these, 4 antiviral compounds demonstrated the most potent anti-C. difficile activity against a panel of 15 C. difficile isolates, with minimum inhibitory concentrations (MICs) comparable to the drug of choice, vancomycin. These include rottlerin (MIC50 = 0.25 μg/mL), α-mangostin (MIC50 = 1 μg/mL), dryocrassin ABBA (MIC50 = 1 μg/mL), and obefazimod (MIC50 = 4 μg/mL). All exhibited minimal to no activity against representative members of the human gut microbiota. Interestingly, α-mangostin, a natural xanthone derived from the mangosteen fruit, exhibited strong bactericidal action, clearing a high inoculum of C. difficile in less than an hour. All other drugs exhibited bacteriostatic activity. Given their characteristics, these compounds show great promise as novel treatments for CDI. Introduction Healthcare-associated infections are one of the greatest burdens on healthcare systems. Most bacteria responsible for a part of this burden are associated with antimicrobial resistance, with one notable exception: C. difficile. Despite lacking a significant antimicrobial resistance profile, C. difficile is still considered an urgent threat and one of the deadliest bacterial infections responsible for approximately 30,000 deaths annually, primarily affecting older populations [1–5]. The incidence of C. difficile infections (CDI) in the United States has increased over the last two decades and has remained relatively stable from 2021 to 2022, with the COVID-19 pandemic seeing a drop in cases due to improved strategies preventing the spread of microorganisms [2, 3, 6]. Typically, C. difficile targets patients receiving antibiotic-based therapy. This can be attributed to that the use of antibiotics leads to gut dysbiosis, which increases the growth and colonization of C. difficile in the gut. As a consequence, colonizing C. difficile produces toxins that can cause damage of the tight junctions of the intestinal epithelium eliciting a variety of symptoms, ranging from mild diarrhea to intestinal tissue necrosis and pseudomembranous colitis [7–9]. Novel antibiotic therapies approved for CDI have been conspicuously lacking. In the past, the recommended drugs for mild to severe CDI were metronidazole and vancomycin, respectively. Fidaxomicin is the most recently approved new antibiotic for CDI, having been approved nearly 40 years ago. Due to the high rates of treatment failure and recurrence, metronidazole is no longer recommended; and vancomycin and fidaxomicin are now the main therapeutic options for CDI treatments [8, 10, 11]. Despite the potency of these antibiotics against C. difficile, recurrence rates can still be significant even if they are an improvement over metronidazole. Fidaxomicin and vancomycin have both encountered recurrence rates of 20% or higher, with subsequently increasing chances of recurrence after the first episode [12, 13]. Hence, there is an immediate and growing need for new antibiotics to address the shortcomings of the existing therapeutics and their impact on the healthcare system. Antivirals are one of the most prolific and growing categories of treatments since the first was approved in 1963 [14]. With the advent of COVID-19, antiviral research has become an area of increasing interest, and with it many antiviral compounds were approved for treatment or held back due to their pharmacokinetics [15]. Poor oral bioavailability is a challenge for many antivirals, which prevents them from being approved as new antivirals [16]. This might be advantageous for treating CDI, which requires drugs with low intestinal absorption to remain in the gut for a longer period of time [17–19]. Despite the need for more potent drugs for treating CDI with minimal intestinal absorption, screening antiviral drugs for possible activity against C. difficile has not been thoroughly investigated. Prior screens and clinical studies have demonstrated the potential of antiviral drugs, like nitazoxanide, a broad-spectrum antiviral and antiprotozoal drug, against C. difficile in comparison to vancomycin [20–22]. Moreover, antiviral compounds have been effective against other microorganisms. For instance, the antiviral hypericin increased the effectiveness of beta-lactam antibiotics against methicillin-resistant Staphylococcus aureus (MRSA) [23]. Given the potential of the aforementioned medications, the goal of this work is to screen an antiviral library to find antivirals with considerable potency against C. difficile. Screening an antiviral library revealed four antiviral compounds that had strong anticlostridial activity. These hits were assessed for potency and specificity against a panel of clinical isolates of C. difficile and the human gut microbiota, respectively. Antivirals also had their killing kinetics assessed to determine their killing kinetics over a 48-hour period. Materials and methods Bacterial strains & reagents Bacterial strains utilized in this study were sourced from the Centers for Disease Control and Prevention (CDC, Atlanta, GA), Biodefense & Emerging Infections Research Resources Repository (BEI Resources, Manassas, VA) and the American Type Culture Collection (ATCC, Manassas, VA). Phosphate-buffered saline (PBS) (Corning, NY), Brain heart infusion broth (BHI) and De Man—Rogosa—Sharpe broth (MRS) (Becton, Dickson and Company, Franklin Lakes, NJ), yeast extract (Fisher Scientific Global Solutions, Suwanee, GA), L-cystine (Thermo Fisher Scientific, Waltham, MA), and vitamin K1, resazurin, and hemin (Sigma-Aldrich, St. Louis, MO) were purchased commercially. Compounds and libraries The MCE antivirals library (Cat. No. HY-L027), which includes 618 unique compounds displaying antiviral activity, was purchased from MedChemExpress (Monmouth Junction, NJ). After an initial screening and confirmation, the active hits were purchased commercially as follows: rottlerin and dryocrassin ABBA (MedChemExpress; Monmouth Junction, NJ), α-mangostin (TargetMol Chemicals; Wellesley Hills, MA), and obefazimod (Ambeed; Arlington, IL). Vancomycin hydrochloride (Gold Biotechnology, Olivette, MO, USA) was included as a positive control. Screening assay against C. difficile To discover antivirals with anti-C. difficile action, the MCE antivirals library was screened using the broth microdilution technique, as previously described [24–27], against C. difficile ATCC BAA 1870 at a fixed concentration of 16 μM. In brief, bacteria were streaked and grown anaerobically at 37°C for 48 hours on brain heart infusion supplemented (BHIS) agar plates. A 0.5 McFarland solution of C. difficile was then prepared and diluted in BHIS to reach an inoculum of ~5 × 105 CFU/mL and placed in 96-well plates. The antiviral compounds were added at a concentration of 16 μM. The plates were then incubated anaerobically at 37°C for 48 hours. Using a BioTek Synergy H1 Microplate Reader with BioTek Gen 5 and Imager Software, the OD600 was determined. The compounds that inhibited >80% of the bacterial growth, as compared to the OD600 of the growth control wells (DMSO), were recognized as potential hits of interest and purchased commercially for further confirmation. GraphPad Prism version 10 was employed to illustrate the growth inhibition. Anticlostridial activity of the potent hits The broth microdilution technique was used to test the antiviral hits against C. difficile ATCC BAA 1870 to identify the minimum inhibitory concentrations (MICs) of the promising hits [28–30]. The hits and the control antibiotic, vancomycin, were serially diluted along the 96-well plates. A bacterial solution equivalent to 0.5 McFarland standard was diluted in BHIS broth to obtain a final bacterial concentration of 5 × 105 CFU/mL and added to the plates. These plates were then incubated anaerobically at 37°C for 48 hours. The lowest concentration at which agents could totally prevent bacterial growth was identified as the MIC. We selected hits with MIC ≤4μg/mL (rottlerin, dryocrassin ABBA, α-mangostin, and obefazimod) for further investigation. These four compounds were purchased commercially and examined against a panel of 15 clinical isolates of C. difficile in comparison to the control antibiotic, vancomycin. The test agent concentrations that inhibited 50% and 90% of the strains (MIC50 and MIC90, respectively), were determined. MICs were performed at least in two independent experiments, each containing the biological replicates. Antibacterial activity of potent hits against representative members of gut microbiota The most potent hits’ MICs against gut microbiota strains were ascertained in accordance with previous studies [31–33]. To achieve a final bacterial concentration of 5 × 105 CFU/mL, a 0.5 McFarland bacterial solution was diluted in BHIS broth (for Bifidobacterium and Bacteroides strains), and MRS broth (for Lactobacillus). After serially diluting this suspension with the hits and control antibiotics, it was incubated for 48 hours at 37°C in anaerobic conditions for Bifidobacterium and Bacteroides or in the presence of 5% CO2 for Lactobacillus before determining the MICs of the hits. Time-kill kinetics assay against C. difficile In a time-kill experiment, potential antiviral drugs were challenged against C. difficile to ascertain the bactericidal or bacteriostatic nature of inhibition, as previously described [24, 34]. C. difficile ATCC 43255 and C. difficile ATCC BAA-1870 were grown overnight, then diluted 1:150 into sterile BHIS, yielding a concentration of approximately 5 × 105 CFU/mL. C. difficile was treated with the hits and the control antibiotic, vancomycin, at a concentration of 5× MIC, and then cultured anaerobically at 37°C. DMSO was also included as an untreated control. At 0, 2, 4, 8, 12, 24, and 48 hours, aliquots were collected from each treatment, diluted, and plated onto pre-reduced BHIS agar plates. Plates were incubated in anaerobic conditions at 37°C before determining the bacterial CFU. A drug was deemed to be bactericidal if it reduced the initial inoculum by ≥3 log10 CFU/mL. Results Screening of the antiviral library against C. difficile To discover novel anti-C. difficile drugs, 618 antivirals (MCE antiviral library) were tested for possible anti-C. difficile action against the hypervirulent strain of C. difficile ATCC BAA-1870. The screening was done at an initial concentration of 16 μM using the broth microdilution method. A total of 24 compounds were found to possess anticlostridial action, effectively preventing the growth of C. difficile at the screening concentration (Fig 1, Table 1, and S1 Table in S1 File). Based on past literature, 15 hits that have previously been reported or recognized as unsafe or anticancer drugs [27, 28, 35–43] have been excluded (S1 Table in S1 File) [44–48]. The other unique 9 hits have been confirmed against C. difficile, and their MICs were determined. Out of these 9 hits, 4 antiviral compounds (rottlerin, α-mangostin, dryocrassin ABBA, and obefazimod) displayed the most potent activity against C. difficile ATCC BAA-1870 with MIC values of ≤4 μM (Fig 1). Due to their potency and lack of prior research, these promising hits were chosen for further investigation. Among the most promising hits in the library (Table 1), 3 were natural products (rottlerin, α-mangostin and dryocrassin ABBA), and 1 (obefazimod) is currently in clinical trials for HIV and ulcerative colitis treatments [49, 50]. MCE antiviral library screening of 618 antivirals at 16 μM against C. difficile ATCC BAA-1870. Compounds with a greater than 80% inhibition of bacterial growth were identified as hits (highlighted in green), while hits below 80% inhibition were excluded due to lack of activity (in gray). 24 total hits were identified with 9 novel hits and 15 excluded hits. The 9 novel hits are: rottlerin, dryocrassin ABBA, α-mangostin, obefazimod, baloxavir, 4’-O-methylbavachalcone, TMC647055 (choline salt), brefeldin A, and maslinic acid. The 15 excluded hits can be found in S1 Table in S1 File. 10.1371/journal.pone.0309624.t001Table 1Description and MICs (μM) of potent antivirals against C. difficile ATCC BAA-1870.No.Compound IDDescriptionMICs (μM) 1 RottlerinNatural product isolated from Mallotus phillippinesis and PKC-δ Inhibitor0.5 2 Dryocrassin ABBANatural Product isolated from Dryopteris crassirhizoma2 3 α-MangostinNatural Product isolated from Garcinia mangostana4 4 ObefazimodAnti-HIV & Colitis4 5 BaloxavirAntiviral8 6 4’-O-MethylbavachalconeAntiviral8 7 TMC647055 (choline salt)Antiviral16 8 Brefeldin AAntiviral lactone produced by fungi16 9 Maslinic acidNatural product isolated from Olea europaea16 Description and MICs (μM) of potent antivirals against C. difficile ATCC BAA-1870. Anticlostridial activity of the potent hits The anticlostridial activity of the potent 4 hits was further confirmed by testing them against 15 clinical isolates of C. difficile (Table 2) and calculating the MIC50 and MIC90 of each drug. Against the tested strains of C. difficile, the activity of rottlerin (MIC50 = 0.25 μg/mL and MIC90 = 0.5 μg/mL) and α-mangostin (MIC50 = 1 μg/mL and MIC90 = 2 μg/mL) was comparable to those of vancomycin (MIC50 = 0.5 μg/mL and MIC90 = 1 μg/mL). Likewise, dryocrassin ABBA (MIC50 = 1 μg/mL and MIC90 = 4 μg/mL) had similar activity to vancomycin with a 2-fold difference in its MIC90. Conversely, obefazimod needed a 3- fold greater concentration to inhibit C. difficile (MIC50 = 4 μg/mL and MIC90 = 8 μg/mL) in comparison to vancomycin. 10.1371/journal.pone.0309624.t002Table 2MIC values of the most potent antivirals against clinical isolates of C. difficile.C. difficile strain IDMIC (μg/mL)Antiviral hitsControl antibioticRottlerinα-MangostinDryocrassin ABBAObefazimodVancomycin ATCC-BAA 1870 0.51241 ATCC 1871 0.52482 ATCC 43255 0.250.50.520.5 ATCC 43598 0.51180.5 BAA 630 0.250.5140.5 NR-49288 0.52481 NR-49302 0.50.5240.5 NR-49308 0.250.50.540.5 NR-49313 0.2510.520.5 NR-49319 0.51140.25 CD-12 0.2510.541 CD-16 0.250.5120.5 CD-20 0.250.5140.5 CD-21 0.52481 CD-28 0.251181 MIC 50 0.25 1 1 4 0.5 MIC 90 0.5 2 4 8 1 MIC values of the most potent antivirals against clinical isolates of C. difficile. Antibacterial activity of the most potent hits against representative members of the human gut microbiota To test the activity against gut microflora, selected antiviral compounds were screened against a panel of representative commensal gut microbiota including Lactobacillus, Bacteroides, and Bifidobacterium, starting at a concentration of 256 μg/mL. The selected hits and vancomycin had little to no effect on Lactobacillus strains apart from fidaxomicin (MIC values of 8 and 32 μg/mL against L. rhamnosus and L. brevis strains, respectively) (Table 3). 10.1371/journal.pone.0309624.t003Table 3MIC (μg/mL) values of selected antivirals and control antibiotics against representative members of human normal gut microbiota.Bacterial strain IDMinimum inhibitory concentration (μg/mL)Antiviral hitsControl antibioticsRottlerinObefazimodα-MangostinDryocrassin ABBAVancomycin L. brevis 14864 >256>256>256>256>256 L. rhamnosus 53103 128>256>256>256>256 L. casei ATCC 334 128>256>256>256>256 B. fragilis HM-709 328≤23216 B. fragilis HM-714 648≤26432 B. fragilis HM-710 16128≤23216 B. adolescentis HM-633 ≤2>25628≤2 B. breve HM-411 88≤28≤2 B. breve HM-1120 48≤28≤2 MIC (μg/mL) values of selected antivirals and control antibiotics against representative members of human normal gut microbiota. Rottlerin, obefazimod, and dryocrassin ABBA showed activity at higher concentrations (MIC values ranged from 8 to 128 μg/mL) against Bacteroides fragilis strains in comparison to vancomycin (MIC values, 16–32 μg/mL); however, α-mangostin displayed potent activity at lower concentrations (MIC values ≤2 μg/mL) (Table 3). Rottlerin, obefazimod, and dryocrassin ABBA showed activity at lower concentrations (MIC values, 4–8 μg/mL) against Bifidobacterium breve strains, whilst α-mangostin showed activity at lower concentrations (MIC values, ≤2 μg/mL) similar to the control antibiotic, vancomycin (Table 3). Overall, antiviral compounds showed similar or less activity against gut microflora when compared to drugs of choice for CDI, with the greatest difference in activity being against Bifidobacterium strains being at least 2–4 folds less active. Time-kill kinetics assay against C. difficile To assess whether the antivirals have bactericidal or bacteriostatic killing kinetics activity against C. difficile, a time-kill assay was performed against C. difficile ATCC-BAA 1870 and ATCC 43255. As demonstrated in Fig 2A, and 2B, α-mangostin exhibited strong bactericidal activity within 2 hours, significantly better than vancomycin, which demonstrated bactericidal activity at 12 hours for both strains. In contrast, dryocrassin ABBA, rottlerin, and obefazimod displayed bacteriostatic activity with approximately 1–1.5 log10 CFU reduction within 12 hours and remained static up to 48 hours (Fig 2A, and 2B). Time-kill assay of most potent hits and control antibiotics against C. difficile. The growth of C. difficile ATCC 43255 (A) or ATCC BAA-1870 (B) at was observed as log10 CFU/mL over a 24-hour period. The compounds dryocrassin ABBA (purple), obefazimod (blue), rottlerin (red), α-mangostin (brown), vancomycin (green), and DMSO (orange) were tested at 5x their MIC. Discussion C. difficile is the leading cause of antibiotic-associated diarrhea in the United States, accounting for nearly half a million infections annually [7, 51]. This is in part due to currently available treatment options, vancomycin and fidaxomicin, being associated with significant rates of treatment failure (20–35%) for the initial antibiotic treatment with 40–60% of those cases experiencing recurrence again [13, 52–54]. Moreover, both of the approved drugs continue to pose the risk of C. difficile acquiring antibiotic resistance, which emphasizes the urgent need for a unique, reliable, and efficient medication or pharmacological scaffold to treat CDI [55]. Many new antivirals have poor oral bioavailability, which is a considerable challenge for these compounds to be approved as drugs or antivirals [56, 57]. Therefore, finding antivirals with potent anti-C. difficile activity is a fruitful strategy because these drugs could remain longer in the gut where C. difficile grows and colonizes due to their limited absorption. In this study, a whole cell-based screening of 618 of antiviral compounds was conducted against C. difficile. In the initial screening, 24 compounds were found to have anti-C. difficile activity. After excluding anticancer compounds and previously reported hits, we pursued 9 antivirals for further study against C. difficile. These results were confirmed and refined, ultimately narrowing down to 4 agents (rottlerin, α-mangostin, dryocrassin ABBA, and obefazimod) that displayed the most potent activity against C. difficile ATCC BAA-1870 with MIC values ≤4 μM. Due to their unique nature and lack of prior research regarding their use as antimicrobials, these 4 agents were chosen for further investigation. To further confirm their activity, these hits were purchased commercially and have been tested against a panel of 15 clinical isolates of C. difficile and their MIC50 and MIC90 were determined. Rottlerin, also known as mallotoxin, is a natural product isolated from Mallotus phillippinesis and is a known protein kinase Cδ (PKCδ)-selective inhibitor and mitochondrial uncoupler that is typically utilized as the base for substrate phosphorylation studies [58, 59]. However, there has been some debate on whether it is selective for protein kinase C [59]. Rottlerin has displayed activity targeting quorum sensing and biofilm activity in Pseudomonas aeruginosa [60]. It has also shown activity against mycobacterium spp., including M. tuberculosis and M. smegmatis (IC50 range from 9 to 74 μM), potentially interfering with shikimate kinase, a promising target for antimicrobials, which is essential for many bacteria [61]. It also has been reported to target Chlamydia species with MIC values of 1 μM [62]. Recently, rottlerin was found to have some fungicidal activity in a study where MIC and MFC values against Candida species were as low as 7.81 μg/mL [63]. Here, we report that rottlerin exhibited strong antibacterial activity against C. difficile (MIC50 = 0.25 μg/mL and MIC90 = 0.5 μg/mL), which was comparable to the control antibiotic, vancomycin. α-mangostin is a xanthone derivative extracted from the edible fruit of Garcinia mangostana, commonly known as mangosteen [64]. Interestingly, α-mangostin has faced issues as a treatment due to its lack of absorption and oral bioavailability, leading researchers to modify α-mangostin to improve its pharmacokinetics [65]. Interestingly, α-mangostin has been found in high concentrations within the small intestine, which makes it an intriguing molecule to further pursue for CDI treatment. Both primary treatments for moderate to severe CDI, vancomycin and fidaxomicin, have minimal systemic absorption when taken orally [66, 67]. α-mangostin showed potent anti-C. difficile activity (MIC50 = 1μg/mL and MIC90 = 2μg/mL), equivalent to that of the drug of choice, vancomycin, with little to no activity against the microbiota. In addition, α-mangostin has shown potent and rapid bactericidal activity both in this study and against methicillin-resistant S. aureus (MRSA) [68–70]. Dryocrassin ABBA is a phloroglucinol derivative that is isolated from Dryopteris crassirhizoma [71] and has previously been reported to have antimicrobial activity against both H5N1 avian influenza virus and S. aureus virulence factors [72, 73]. Particularly, it has potent activity against virulence factors of MRSA such as Von Willebrand factor-binding proteins (vWbp) and Sortase A [73, 74]. Otherwise, it does not have activity against S. aureus itself (MIC >1024 μg/mL) [73]. Herein, dryocrassin ABBA showed activity similar to that of vancomycin against C. difficile, with an MIC50 and MIC90 of 1 μg/mL and 4 μg/mL. Obefazimod, also known as ABX464, is a novel and first-in-class compound developed by Abivax and has shown antiviral, antirheumatic, and anti-colitis properties [49, 75, 76]. It is currently in phase 2 trials for its antiviral and anti-rheumatic properties, while in phase 3 trials for the anti-colitis effect [49, 50, 77, 78]. There have been no previous reports of antibacterial activity published so far, making this compound of great interest as a potential novel class of antimicrobials on top of its other therapeutic effects. This compound had an MIC50 of 4 μg/mL and an MIC90 of 8 μg/mL, and displayed bacteriostatic activity against C. difficile. Its activity against human microbiota was lower across the board when compared to control antibiotics apart from Bacteroides strains. C. difficile is an opportunistic pathogen that is intimately tied with gut dysbiosis. Administration of broad-spectrum antibiotics can disrupt commensal gut microflora, which are considered the first line of defense against C. difficile colonization [79]. Due to this, selectivity against C. difficile could be beneficial in preventing recurrence and reducing the severity of CDI [80]. Therefore, it is necessary to take into consideration whether these compounds have an impact on beneficial bacteria typically found in the human gut microbiota. Overall, the selected compounds had very little effect on gut microbiota and, in some cases, had MICs higher than those of drug of choice. Next, we aimed to monitor the killing kinetics of the selected antiviral compounds to determine whether they had bactericidal or bacteriostatic activity. We found that 3 compounds, rottlerin, dryocrassin ABBA, and obefazimod had bacteriostatic activity. However, α-mangostin displayed rapid bactericidal activity far more potent than both fidaxomicin and vancomycin, achieving 100% inhibition within the first 2 hours. Remarkably, α-mangostin’s potent activity against C. difficile (MIC50 = 1μg/mL and MIC90 = 2μg/mL) did not translate to other bacteria with little to no activity against microbiota and a far more rapid mechanism of killing highlighted the potential of this compound for CDI treatment. To conclude, the goal of this work was to identify the antivirals with notable anti-C. difficile activity, such as obefazimod, rottlerin, α-mangostin, and dryocrassin ABBA, that could be thus classified as anti-C. difficile drugs. Most notably, α-mangostin’s killing kinetics were superior to those of the currently approved drugs. Additionally, nearly all drugs have shown little to no efficacy against the commensal microbiota, which is particularly favorable in preventing recurrent CDI. Therefore, these antivirals could function as lead structures for further development of anti-C. difficile drugs. Supporting information (DOCX) (XLSX) (XLSX)