Repurposing Birabresib to target Gram‑positive bacteria

Abstract

The threat of antibiotic-resistant bacteria is rapidly escalating, creating an urgent need for new types of antimicrobials. Although most antibiotics act by disrupting essential cellular processes, such as cell division, replication, transcription, and translation, few novel agents targeting these pathways have been discovered in recent years. So, we pursued a different strategy by searching for molecules that may hyperactivate the ClpP protease, leading to bacterial cell death. ADEP4, a peptide antibiotic, exemplifies this approach by inducing uncontrolled protein degradation via ClpP. Based on key binding sites of ADEP4 on ClpP, we conducted in silico screening using molecular docking, structural chemistry, and pharmacokinetic parameters, primarily evaluating binding affinity (kcal/mol). We identified Birabresib as a promising candidate using this approach. Here, we demonstrate that Birabresib effectively inhibits Gram-positive bacteria, exhibits synergy with the antibiotic rifampicin, and has limited cytotoxicity on the human cell line tested.

Introduction

Antibiotics fight bacterial infections by either killing bacteria or hindering their ability to grow and reproduce. However, bacterial cells have developed a range of defense mechanisms to survive antibiotic treatment, primarily through resistance and tolerance. Some bacteria are naturally resistant, while others acquire resistance through gene transfer or by accumulating mutations in their genome, which allow them to counteract antibiotics^1–5. According to the US Centers for Disease Control and Prevention (CDC), over 2.8 million antibiotic-resistant infections occur annually in the USA, resulting in more than 35,000 deaths⁶. Current projections estimate that in less than 25 years, antibiotic-resistant infections could cause 10 million deaths annually^6–8. The growing threat of antibiotic resistance is alarming. For comparison, cancer, the second leading cause of death worldwide, claims approximately 9.5 million lives each year^9,10. Without new tools to fight pathogenic bacteria, antibiotic-resistant infections could result in a death toll comparable to cancer in just 25 years. To mitigate this looming crisis, it is essential to develop new types of antimicrobials to combat this rising threat.

Bacterial proteases represent a largely untapped class of targets for the development of novel antimicrobial agents¹¹. Bacterial proteases are important in stress response, virulence, and protein homeostasis¹². Among these, ClpP (caseinolytic protease proteolytic subunit ClpP) has gained attention as a therapeutic target due to its role in bacterial survival and pathogenicity¹³. ClpP is a highly studied protease present in most bacteria, but it cannot degrade large peptides without the assistance of a chaperone¹⁴. On its own, it can only degrade short peptides (7–10 amino acids), but degrades full-length proteins when associated with Clp-ATPases chaperone, including ClpX, ClpA, or ClpC^15–17. Recent studies have identified acyldepsipeptides (ADEPs)^18,19 Sclerotiamide²⁰ that target the ClpP, hyperactivating ClpP degradation activity. These compounds bypass the need for chaperones, forcing ClpP into an “open” state, causing uncontrolled protein degradation and leading to cell death²¹ (Fig. 1a). For example, ADEP4 is effective against vancomycin-resistant Enterococci²² and biofilms^22,23.

Fig. 1. ClpP Activation and Dysregulation.

a ClpP & ADEP4-mediated protein degradation. ClpP requires ATP-dependent chaperones to unfold proteins and feed peptides into its proteolytic core. ADEP4 bypasses this need by expanding the ClpP core, leading to uncontrolled protein degradation. b Molecular docking with pharmacokinetic screening: identified Birabresib as a small molecule binding to ClpP at a site similar to ADEP4 and Sclerotiamide with high affinity.

Pharmaceutical investment in antibiotic development has declined due to high cost, lengthy timeline, and high risk of failure²⁴. However, one of the most promising, fastest, and least costly ways of new antibiotic development could be drug-repurposing²⁵. For example, halicin, a compound originally discovered for managing diabetes, has been shown to have antimicrobial activity against various bacterial pathogens^26–28. In silico drug screening accelerates drug repurposing by simulating interactions between the target protein and compound²⁹. High-performance computing-based docking studies and drug-repurposing could save both time and money in the hit identification process and identify less harmful compounds, which can also increase the chance of success in finding new and safe antibiotics³⁰.

In this study, we aimed to identify antimicrobial molecules based on how ADEP4 and Sclerotiamide bind to ClpP. Building on our previous approach that led to the identification of AGI-6780 as a repurposed antimicrobial³¹, we performed a similar virtual screening of candidate molecules against the crystal structure of the ClpP protease using in silico approaches. Compounds were prioritized based on stringent criteria, including market availability, cost-effectiveness, and high binding affinity to the ClpP active site. Birabresib emerged as a promising candidate from this screening process and was subsequently evaluated for its antibacterial efficacy. Our results indicate that Birabresib effectively inhibits the growth of 4 Gram-positive bacterial species, Bacillus subtilis, Priestia megaterium, Staphylococcus aureus, and Staphylococcus epidermidis, while exhibiting little to no bactericidal activity against Gram-negative bacteria, including Escherichia coli, Chromobacterium violaceum, and Pseudomonas aeruginosa. These findings suggest that it is a narrow-spectrum antimicrobial that targets specific bacterial groups and is less likely to harm beneficial bacteria^32,33. Furthermore, we demonstrated using a ClpP mutant that ClpP is likely a target of Birabresib. We also used a combination assay with rifampicin (Rif) and Birabresib to demonstrate synergistic killing. The combination of Rif and Birabresib was effective at killing B. subtilis, even at their sub-MIC levels, underscoring the therapeutic potential of combining a protease activator with an RNA polymerase-targeting antibiotic. We then showed, using a cytotoxicity assay, that Birabresib had minimal adverse effects on human cell lines at concentrations that were effective against bacteria. Collectively, these results underscore the potential to repurpose Birabresib as an antimicrobial agent, either as monotherapy or in combination with established antibiotics such as Rif.

Results and discussion

Computational discovery of Birabresib as a ClpP activator via binding affinity profiling, followed by pharmacokinetic and drug-likeness evaluation

We conducted a molecular docking study to evaluate the binding affinity between ClpP’s active-site residues and various ligands, focusing on identifying key nonbonding interactions. As controls, we employed ADEP4 (a peptide antibiotic)²² and Sclerotiamide (a small molecule known to hyperactivate ClpP)¹⁶. To this end, we virtually screened a library of molecules using AutoDock Vina³⁴, guided by the ADEP4-ClpP co-crystal structure³⁵, structural chemistry analysis, and pharmacokinetic parameters³⁶. The pharmacokinetic properties and drug-likeness screen were based on Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET)³⁷. ADMET studies are crucial assessments in the early phase to decrease the failure rate of new drug candidates in clinical trials³⁸. Prior to in vitro ADMET analysis, there were some established predictive models that could be used to predict many ADMET parameters³⁹. Among them SwissADME server is one of the most popular free-access tools⁴⁰. Although Lipinski’s Rule of Five (Ro5) is not relevant for all types of drug development, we used it as a good starting point for preliminary screening⁴¹. In our screen, the target did not need to meet all 5 rules. While no single parameter can definitively exclude a candidate, and all predictions should be interpreted in light of potential limitations and uncertainties, the goal is to pinpoint novel molecules that satisfy many of these benchmarks. We were only interested in small molecules that are structurally distinct from current antibiotics. Birabresib emerged as a primary candidate for further assessment (Fig. 1b). Birabresib meets many of these criteria, similar to ADEP4 and Sclerotiamide, underscoring its potential as a viable therapeutic agent (Table 1).

Table 1.

Evaluation of physicochemical, pharmacokinetic, and drug-likeness profiles

Screening properties	ADEP4	Sclerotiamide	Birabresib	Goal
Molecular weight (g/mol)	770.86	463.53	491.99	<500
H-bond acceptor	10	5	5	≤10
H-bond donor	3	3	2	≤5
Log p	4.88	2.76	3.01	<5
Bioavailability	0.17	0.55	0.55	≥0.20
Lipinski’s violations	2	0	0	0
P-gp substrate	Yes	Yes	Yes	No
CYP1A2 inhibitor	No	No	No	No
CYP2C19 inhibitor	No	Yes	Yes	No
CYP2C9 inhibitor	No	No	Yes	No
CYP2D6 inhibitor	No	Yes	No	No
CYP3A4 Inhibitor	Yes	No	Yes	No
Sum	7	9	8

Standard ADMET (absorption, distribution, metabolism, and excretion) screening criteria^91–95 were employed to identify drug candidates with optimal activity and safety profiles for human use. Underlined: Indicates that meets expectations. For CYP1A2, CYP2C19, CYP2C9, CYP2D6, and CYP3A4 inhibitors, the “Yes” or “No” scores serve as indicators of the potential for drug-drug interactions. A “Yes” denotes that the compound's predicted binding energy meets or exceeds a defined threshold, suggesting it could significantly interact with and potentially inhibit each respective enzyme. Conversely, a “No” indicates that the compound exhibits weak binding, thereby reducing the likelihood of enzyme inhibition and associated drug-drug interactions. For bioavailability, higher values are generally preferred. While there is no strict cutoff, a threshold of 0.20 is commonly used as a reference point.

In our initial rapid screening, we assessed binding energies by docking ligands to a single ClpP chain (chain A) of E. coli, B. subtilis, and S. aureus, wherein Birabresib demonstrated a binding energy of −7.8 kcal/mol, −7.8 kcal/mol, and −7.9 kcal/mol, respectively. Although this approach reduced computational demands, ADEP4 binds at the interface of two ClpP chains. Therefore, after we identified Birabresib as a candidate, We extended our analysis by docking the compound to two adjacent ClpP subunits from B. subtilis, E. coli, and S. aureus, which resulted in improved binding energies of −8.9, −8.8, and −8.4 kcal/mol, respectively (Supplementary Table S1), with the corresponding ligand–protein interactions illustrated in Supplementary Fig. S1. If Birabresib modulates ClpP in a similar manner as ADEP4 and Sclerotiamide, its binding interactions should involve many of the same key amino acid residues. As expected, Birabresib, ADEP4, and Sclerotiamide bind to similar amino acid sites on two adjacent ClpP subunits, suggesting that they may exert comparable modulatory effects on ClpP activity (Table 2). Based on these in silico results, Birabresib emerged as an excellent candidate for further investigation as a ClpP modulator, prompting subsequent assessments of its antimicrobial activity against several bacterial strains.

Table 2.

Molecular docking predicts that ADEP4, Sclerotiamide, and Birabresib selectively bind to specific amino acid residues (Supplementary Fig. S1) at the interface of two adjacent ClpP subunits from Bacillus subtills

ClpP residues	ADEP4	Sclerotiamide	Birabresib	Birabresib matching
Ala107	0	1	0	1
Ala45	1	0	1	2
Arg15	1	1	0	2
Arg22	1	1	1	3
Asp18	1	1	1	3
ASP71	1	0	0	1
Gln46	1	0	1	2
Gln81	0	1	0	1
Glu14	0	1	0	1
Glu53	0	1	0	1
His156	1	1	0	2
Ile19	1	1	1	3
Ile29	0	0	1	1
Ile83	1	1	0	2
Leu24	1	1	1	3
Leu42	0	0	1	1
Lys108	0	1	0	1
Lys25	1	1	1	3
Lys84	0	1	0	1
Met98	1	0	0	1
Phe17	1	1	1	3
Phe30	1	0	1	2
Phe49	1	1	1	3
Phe82	0	1	0	1
Pro56	1	0	0	1
Ser16	1	1	0	2
Ser21	0	0	1	1
Ser21	1	0	1	2
Ser22	0	1	0	1
Ser45	1	0	0	1
Tyr20	1	0	1	2
Val03	0	0	1	1
Val5	1	0	0	1
Sum	21	19	16

Shaded numbers 2 or 3 indicates when Birabresib binds to the same ClpP amino acid residues as ADEP4 and/or Sclerotiamide.

Birabresib selectively inhibits the growth of Gram-positive bacteria

Small molecule inhibitors and activators are typically first evaluated in vitro (i.e., protein-small molecule binding assays) to demonstrate direct interaction with their target proteins. Although informative, many compounds that exhibit activity in vitro fail in cellular contexts due to barriers such as limited uptake, restricted permeability, active efflux, intracellular degradation, and other processes that reduce the effective intracellular concentration or prevent access to the target^42,43. Recently, we were able to identify a small molecule with antimicrobial activity against Gram-positive bacteria³¹ without doing biochemical analysis by first testing it against bacterial strains. We attempted to identify another molecule in a similar manner.

We evaluated the antibacterial potential of Birabresib using agar disk diffusion assays (Fig. 2a), a widely accepted method for determining the antimicrobial susceptibility of various bacterial strains⁴⁴. B. subtilis and E. coli were initially selected for mechanistic studies due to its well-established genetic tractability. In this assay, clear zones, indicative of bacterial killing or growth inhibition, developed around disks impregnated with 0.1 mg/mL Birabresib for four Gram-positive strains: B. subtilis, P. megaterium, S. epidermidis, and S. aureus. In contrast, no clearing zones were observed with Gram-negative bacteria, including E. coli, C. violaceum, and P. aeruginosa. Negative control disks containing only the solvent (1% PEG400 and 3% EtOH) produced no clear zones for either Gram-positive or Gram-negative strains, confirming that the antibacterial effect was attributable solely to Birabresib. These findings show that Birabresib possesses selective antibacterial activity, with efficacy limited to Gram-positive bacteria under the tested conditions.

Fig. 2. Birabresib Inhibits Gram-Positive Bacteria.

a Disc-diffusion assay. Birabresib (0.1 mg/ml) was tested against 4 Gram-positive and 3 Gram-negative bacteria. Top row: solvent control, 1% PEG400 and 3% EtOH. Bottom row: Birabresib-treated plates resulted in a clear zone of inhibition against all Gram-positive bacteria (B. subtilis, P. megaterium, S. aureus, and S. epidermidis) tested, but no antimicrobial activity was observed against any Gram-negative bacteria (C. violaceum, P. aeruginosa, and E. coli) tested. Assays were performed in triplicate (Supplementary Fig. S2). b Liquid assay. Birabresib (0.1 mg/ml) was tested against 3 Gram-positive and 1 Gram-negative bacteria. Bacterial growth was monitored by OD_{600 nm}. Birabresib caused a statistically significant reduction in all Gram-positive strains (p < 0.001) and resulted in a slight reduction in the growth of E. coli. Each experiment was performed with 3 biological replicates (n = 3). The y-axis is log₂ scale.

To corroborate the solid media results, we tested 3 Gram-positive bacteria (B. subtilis, P. megaterium, and S. aureus) and one Gram-negative bacterium (E. coli) for Birabresib activity in liquid culture. When treated with 0.1 mg/mL Birabresib, all Gram-positive cultures exhibited significant growth inhibition as measured by optical density at 600 nm (Fig. 2b). Notably, P. megaterium demonstrated the highest susceptibility to Birabresib among the tested Gram-positive strains. A slight but statistically significant reduction in the optical density (OD) of Gram-negative E. coli was observed in liquid culture (p = 0.01); however, this effect was not apparent on solid agar (Fig. 2b). The decrease in OD may reflect cell death, growth inhibition, or changes in cell size, any of which can influence OD measurements. We did not further investigate the cause of this minor reduction in liquid culture, as no corresponding effect was observed on solid media.

This pattern of selective activity is reminiscent of ADEP4, which preferentially targets Gram-positive and not Gram-negative bacteria¹⁹. No antimicrobial activity was also observed in the negative control culture containing only the solvent (1% PEG400 and 3% EtOH), indicating that the antibacterial effect was exclusively due to Birabresib. The lack of efficacy against Gram-negative bacteria may be due to inherent differences in ClpP, the extra protective outer membrane barrier of Gram-negative bacteria relative to Gram-positive bacteria, or other factors yet to be determined.

Overall, these results indicate that Birabresib is a narrow-spectrum antimicrobial agent. Narrow-spectrum antimicrobial agents are often preferred because they selectively target bacteria (in this case, Gram-positive), sparing beneficial species (in this case, Gram-negative). This targeted approach can potentially reduce the risk of collateral damage to beneficial microbial communities^32,33, such as those in the gut or on the skin.

Birabresib likely targets ClpP

Molecular docking correctly identified Birabresib as an antimicrobial, and we wanted to obtain an indication of whether ClpP is the cellular target of Birabresib. So, we performed a disk diffusion assay and a liquid growth assay using both the wild-type (B. subtilis) and its clpP mutant. The solvent (1% PEG400 and 3% EtOH) was used as a control in both the disk diffusion and liquid growth assays. This experimental setup allowed us to directly compare the sensitivity of bacteria with and without functional ClpP when exposed to Birabresib. Both solid and liquid experiments were conducted using the same concentration of Birabresib (0.1 mg/mL). In wild-type B. subtilis cells, the activators likely trigger uncontrolled ClpP-mediated protein degradation, leading to cell death. However, since the mutant lacks ClpP, the activator should have no effect, allowing the mutant strain to survive better than the wild type. When Birabresib was added, cells lacking ClpP survived better than those with it in both the disk diffusion assay (Fig. 3a) and liquid growth assay (Fig. 3b). These results provide genetic and phenotypic evidence consistent with ClpP-dependent activity and support a proposed mechanism in which Birabresib may function as a ClpP activator, leading to uncontrolled proteolysis and cell death in wild-type cells. A limitation of these assays is that no biochemical analysis was performed to test whether Birabresib directly interacts with ClpP.

Fig. 3. Birabresib likely Targets ClpP.

a Disc Diffusion Assay. Comparing Birabresib (0.1 mg/mL) activity against wild-type B. subtilis and its clpP- mutant. Top row: Solvent control (1% PEG-400, 3% EtOH) shows no inhibition for either strain. Bottom row: Since the mutant lacks ClpP, the activator should have no effect, allowing the mutant strain to survive better than the wild type. The wild-type plate (left) exhibits a clear inhibition zone, whereas the plates containing the clpP- mutant (right) showed no zone of inhibition, indicating ClpP may be a potential target of Birabresib. Assays were performed in duplicate (n = 2; see Supplementary Fig. S3). b. Liquid Assay. Birabresib (0.1 mg/ml) was tested against wild-type B. subtilis and its clpP- mutant. Bacterial growth was monitored by measuring OD_{600 nm}. Birabresib caused a statistically significant reduction in growth of the wild-type strain (p < 0.001) but had no significant effect on the clpP- mutant (p = 0.29). Each experiment was performed with 3 biological replicates (n = 3). Results are expressed as mean ± SEM. Statistical significance was assessed using a T-test. The y-axis is log₂ scale.

Birabresib exhibits a synergistic effect when used in combination with the antibiotic rifampicin (Rif)

While drug cocktails are highly effective in cancer treatment⁴⁵, antibiotics are typically administered individually because some antibiotic combinations have the unintended consequence of enhancing bacterial survival^46–48. However, ADEP4 has been shown to be effective in combination with other antibiotics, such as Rif, resulting in enhanced bacterial killing²³. Given that ADEP4 targets a protease, a mechanism distinct from most antibiotics, we hypothesized that other protease-targeting agents might similarly synergize with existing antibiotics. We therefore evaluated this specific combination to determine whether Birabresib can also synergize with Rif, like ADEP4 did. Using a 96-well high-throughput solid media assay, we determined the minimal inhibitory concentration (MIC) (Supplementary Fig. S4), defined as the lowest concentration required to inhibit bacterial growth. Birabresib synergizes with Rif (Fig. 4). This synergy was observed even at concentrations below the MIC, highlighting the potency of combining a protease activator with Rif, an RNA polymerase-targeting antibiotic. Combining ¼ MIC Rif and ½ MIC Birabresib effectively killed B. subtilis on solid media (Fig. 4). This synergistic effect was also observed with other sub-MIC combinations (Supplementary Fig. S4). The fractional inhibitory concentration index (FICI) is a commonly used metric to assess drug synergy, with values ≤0.5 indicating synergistic interactions⁴⁹. The lowest combination tested that was effective was ¼ MIC Rif and ⅛ MIC Birabresib (Supplementary Fig. S4), yielding a calculated FICI value of 0.375, consistent with synergy.

Fig. 4. Birabresib synergizes with the Antibiotic Rif to kill B. subtilis.

Bacterial survival was evaluated on solid LB agar within 96-well plates. Colony formation occurred only at compound concentrations below the bactericidal threshold for B. subtilis. TTC was added to assess NADH activity and to better visualize the results. TTC is a colorless redox indicator that is reduced by NADH (produced during bacterial metabolism) into a red formazan product. Minimum inhibitory concentrations (MIC) for Birabresib and Rif were established (Supplementary Fig. S4), and combination treatments were assayed in triplicate (n = 3). Checkerboard analyses at sub-MIC concentrations confirmed a synergistic interaction across multiple dose pairs (Supplementary Fig. S4.

To assess metabolic activity, triphenyltetrazolium chloride (TTC) was added, a redox indicator commonly used as an indirect measure of NADH production⁵⁰. In its oxidized form, TTC is colorless; upon reduction by NADH, it yields red formazan. As expected, colonies treated with ¼ MIC rifampicin or ½ MIC Birabresib turned red, whereas those exposed to both agents at these sub-MIC levels remained colorless (Fig. 4). A similar lack of color change (indicative of synergistic metabolic inhibition) was observed for other sub-MIC combinations (Supplementary Fig. S4).

Birabresib exhibits minimal cytotoxicity on human cell lines at concentrations that are effective against B. subtilis

To evaluate its potential toxicity, we employed the MTT assay. It is a widely used method for assessing eukaryotic cell viability⁵¹ by measuring mitochondrial dehydrogenase activity, which is particularly relevant given the presence of a distantly related ClpP homolog in mitochondria. In this study, the MTT assay was performed using the Caco-2 cell line, with the IC₅₀ value serving as a critical parameter for cytotoxic potential. An IC₅₀ near 100% indicates no significant toxicity, whereas values below 50% suggest adverse effects on cell viability. At a concentration of 26 µg/mL, Birabresib maintained 62% cell survival in human cells while effectively inhibiting B. subtilis (Fig. 5). These results underscore Birabresib’s promise as a therapeutic candidate and warrant further evaluation in additional cell lines and animal models.

Fig. 5. Cytotoxicity assay and antimicrobial assay of Birabresib.

Top: MTT Cell Viability Assay. At 26 µg/mL Birabresib, 62% human Caco-2 cells survived, indicating low cytotoxicity at this dose. The human cell lines were also tested with other Birabresib concentrations (Supplementary Fig. S5). The IC₅₀ (marked by the red dashed line) corresponds to the concentration at which 50% of cells survive, falling within an acceptable safety threshold. Bottom: Disc Diffusion Assay. At 26 µg/mL Birabresib demonstrated a clear zone of inhibition, whereas the solvent control (1% PEG-400, 3% EtOH) shows no antibacterial activity. The experiment was performed in duplicate (n = 2) (Supplementary Fig. S5). Together, these results indicate that Birabresib possesses potent antimicrobial activity against B. subtilis while exhibiting minimal toxicity toward human cells at equivalent concentrations.

Birabresib has demonstrated promising efficacy in treating specific solid tumors, with clinical studies indicating it is generally well tolerated. In these trials, Birabresib was administered using an intermittent dosing schedule (80 mg for 14 days on, followed by 7 days off) to mitigate adverse effects, the most notable being thrombocytopenia (low number of platelets in the blood)^52,53. In a follow-up study involving 46 patients with castrate-resistant prostate cancer, testicular midline carcinoma, and non–small-cell lung cancer, dose-proportional exposure and a favorable tolerability profile were observed, with thrombocytopenia being reversible and self-limiting, though sometimes requiring dose modification⁵⁴. These findings suggest that intermittent scheduling is critical for minimizing potential toxicities associated with continuous dosing⁵⁴.

Building on these clinical insights, there is a strong rationale to explore Birabresib as a repurposed antimicrobial agent. Although side effects were noted during cancer treatment due to prolonged exposure, antimicrobial therapy typically involves much shorter treatment durations. Furthermore, our research demonstrated that Birabresib, when combined with Rif, exhibits synergistic activity even at low concentrations, thereby enhancing its antibacterial potency. Together, these results highlight the potential of Birabresib (or its derivatives) as a novel antimicrobial agent worthy of further investigation.

Implications, future work, and unanswered questions

Our study was limited to testing Birabresib in a single human cell line, and we did not evaluate its effects in animal models or in vivo hosts. The primary objective of this study was to identify and characterize antimicrobial activity, rather than to fully resolve the biochemical mechanism of target engagement. While direct biochemical validation of Birabresib-ClpP interactions would provide additional mechanistic detail, the genetic and phenotypic data presented here provide evidence of a link between ClpP and the observed antibacterial activity. Accordingly, comprehensive biochemical characterization, such as purified protein assays and binding studies, represents an important next step, particularly if Birabresib demonstrates low cytotoxicity across additional human cell lines.

A technical hurdle was Birabresib’s poor water solubility. Although we tested common solubilizing agents (e.g., DMSO, PEG derivatives, ethanol, etc.), only a 1% PEG-400 and 3% ethanol mixture effectively solubilized Birabresib at concentrations that were non-toxic to both bacteria and human cells. However, even this mixture proved unstable in liquid over time, likely due to ethanol evaporation during shaking, leading to precipitation of Birabresib. In contrast, solid media retained solubility, presumably due to reduced ethanol loss, enabling reliable plate-based assays. This aligns with previous findings that gels such as agarose-based gels slow evaporation and better retain volatile compounds^55–57. While salt forms of Birabresib may improve solubility and stability, we did not pursue this due to prohibitive cost and time constraints. Now that its antimicrobial activity has been demonstrated, developing a salt form or systematically screening additional solvents is a logical next step.

Solubility limitations also complicated efforts to classify Birabresib as either bactericidal (kills bacteria) or bacteriostatic (inhibits growth). The MIC (minimum inhibitory concentration) was 18 µg/mL in solid media (Supplementary Fig. S4), and no bacterial growth was observed after two weeks. This suggests Birabresib has bactericidal activity; the general rule is that if the minimum bactericidal concentration (MBC)/MIC (MBC/MIC) ratio is less than or equal to 4, it is bactericidal^58–60. In this case, the MIC and MBC appear equivalent with a MBC/MIC ratio of 1. However, in liquid culture, the MIC was 6.25 µg/mL, a discrepancy that is not uncommon, as MIC values often differ between solid and liquid formats^61,62. The MBC in liquid was 0.1 mg/mL, yielding an MBC/MIC ratio >4, typically interpreted as bacteriostatic.

Further complicating classification is the synergy observed between Birabresib and Rif, a known bacteriostatic antibiotic⁶³. Their combined effect occurred below each drug’s MIC, yielding an extremely low MBC/MIC ratio consistent with bactericidal behavior. However, the standard MBC/MIC cutoff (≤4 = bactericidal; >4 = bacteriostatic) is arbitrary, was originally developed for monotherapies, and is not well suited to evaluating synergistic drug interactions. Given these limitations and the fact that an antibiotic can be bactericidal against one species but bacteriostatic against another, we have refrained from assigning a definitive classification to Birabresib.

Beyond its synergy with Rif, we have not yet explored whether Birabresib interacts synergistically with other antibiotics. Future work could investigate such combinations, as well as test the addition of additives/adjuvants (EDTA, N-acetylcysteine, etc.) known to enhance antimicrobial activity^64–66, or improve tissue targeting^67–70. These could further potentiate Birabresib’s effects against Gram-positive pathogens. B. subtilis and E. coli were used in initial experiments because they are well-established model organisms with robust genetic tools, including gene knockout systems (e.g., clpP mutant). We extended the B. subtilis findings by testing additional species; however, a more comprehensive evaluation using clinically relevant pathogens will be necessary to better assess the therapeutic potential of this compound. Looking ahead, promising directions include identifying Birabresib analogs or derivatives with improved solubility, efficacy, and pharmacological profiles. Moreover, the use of Birabresib, alone or in combination therapies, against hard-to-treat Gram-positive bacteria warrants further investigation.

Methods

Screening and optimization of the ligand structures

Compounds from the APExBio database were first evaluated using a series of structural and physicochemical filters designed to identify potential protease modulators. These filters incorporated features associated with protease activation motifs, compatibility with ClpP binding-site geometry, and residue-level interaction potential. Molecules meeting these criteria (to ~4000 ADEP-like candidates) were subsequently advanced to molecular docking. Candidate prioritization considered three main factors: (i) similarity to small molecules previously reported to modulate the target protease, (ii) geometric complementarity within the active-site pocket, and (iii) predicted stabilizing contacts with key amino acid residues. For shortlisted compounds, 3D structures were obtained from PubChem⁷¹ in SDF format and geometry-optimized using Gaussian 09 (Revision D.01⁷²) at the B3LYP/6-31G(d,p) level in the gas phase prior to docking.

Protein preparation, optimization, and molecular docking analysis

Molecular docking analysis was performed using PyRx 0.8⁷³ program, which explores the binding affinities of the control compounds (ADEP4 and Sclerotiamide) and the optimized molecules. During the initial docking process, the structures of the screened molecules and the ClpP protein were uploaded and converted to the PDBQT format for further analysis. The grid box was created around each ligand and protein. The grid centers were positioned at 9.2 (x), 35.6 (y), and 17.5 (z) for Birabresib; 38.60 (x), 37.33 (y), and 21.01 (z) for Sclerotiamide; and 37.60 (x), 38.28 (y), and 22.60 (z) for ADEP4. Corresponding grid sizes were set to 130 × 84 × 91 Å, 35 × 84 × 86 Å, and 101 × 96 × 87 Å along the x-, y-, and z-axes, respectively (Supplementary Fig. S1), providing sufficient space to allow ligand movement and flexibility during docking. The crystal structure of the ClpP protein (PDB ID: 3MT6) was retrieved from the Protein Data Bank (PDB)⁷⁴ and prepared for docking using PYMOL v.1.1⁷⁵ and Swiss PDB viewer software (v.4.1.0)⁷⁶ programs. Published research articles and the CASTp⁷⁷ server was used to confirm the binding pocket of the protein. Docking was guided by available ADEP-bound ClpP structures, including S. aureus (6PMD) (Supplementary Fig. S1 iii), while E. coli (3MT6) was used as a representative structure.

Pharmacokinetic parameters analysis

The pharmacokinetics and drug-likeness profiles of the compounds were evaluated based on Lipinski’s Rule of Five (RO5) using the SwissADME⁴⁰ server. The canonical simplified molecular input line entry system (SMILES) for each compound was used as input taken from the PubChem database.

Chemical compound: Birabresib

The chemical compound used in this study was purchased from AchemBlock Advanced Chemistry Inc. (CAS No.: 202590-98-5; PubChem CID: 9936746 and ChemSpider: 8112374). A mixture of 99.99% ethanol (Sigma) and PEG 400 (TCL, lot RAWMG-WO) was used to dissolve Birabresib to prepare the stock solution.

Preparation of drug stock solutions

Typically, 10 mg of Birabresib drug was weighed and placed into a 5 mL Falcon tube. To dissolve the compound, 2 mL of a 3:2 ethanol and PEG 400 solution was added, followed by shaking at 300 rpm for 10 minutes at 37 °C. All chemicals used in this process were >98% pure. The resulting stock solution was transferred to a microcentrifuge tube and stored in a −20 °C freezer.

Bacterial strains and media

The bacterial strains used in this study included Escherichia coli DH5alphaZ1^78–81 (a commonly used E. coli strain in antibiotic studies^82–86), Bacillus subtilis 1A1 (168) (a commonly used strain in antibiotic studies⁸⁷), Priestia megaterium (ATCC 14581) (a commonly used strain in antibiotic studies⁸⁷), Staphylococcus aureus, Staphylococcus epidermidis, Pseudomonas aeruginosa, and Chromobacterium violaceum. Mutant strains, such as clpP- (BKE34540), were derived from the Bacillus subtilis 1A1 (168)⁸⁸ wild-type strain and obtained from Ohio State University. Wild-type strains were cultured in fresh LB medium (Merck, Germany), while the mutant strain clpP- (called clpP- throughout), was grown in LB medium supplemented with erythromycin (1 μg/mL) and kanamycin (7 μg/mL), respectively. All cultures were incubated at 37 °C with agitation at 300 rpm.

Disk diffusion assay

The Kirby-Bauer disk diffusion susceptibility test⁸⁹ was used to assess the antimicrobial activity of Birabresib against both Gram-positive and Gram-negative bacteria. The mutant clpP- strain of B. subtilis was grown in LB media with 1 μg/mL and 7 μg/mL of erythromycin and kanamycin, respectively, for the disk diffusion assay. The cultures were grown overnight at 37 °C with shaking. Overnight bacterial cultures were sub-cultured in fresh LB media and used for testing when the optical density (OD) reached 0.5 during the log phase. The bacterial strains were evenly spread on Mueller-Hinton agar (MHA) plates using a sterile cotton swab (Fisherbrand), rotating the plate at 45° to ensure uniform distribution. Sterile blank disks (Remel™, Thermo Fisher) were placed on the agar surface using sterile forceps, loaded with 50 μL of a drug solution (100 µg/mL), and let sit for ~15 min to allow the liquid to properly absorbed. The plates were incubated at 37 °C for 24 h to observe the zone of inhibition.

Liquid assay

A liquid growth inhibition assay was performed using a 96-well microtiter plate to evaluate the antibacterial activity of Birabresib. Four bacterial strains were tested: three Gram-positive species, B. subtilis, B. megaterium, and S. aureus, and one Gram-negative species, E. coli. Each bacterial culture was grown to mid-log phase (OD 0.5). In each well, a final volume of 300 µL was prepared, containing bacterial culture, 1% polyethylene glycol 400 (PEG400), 3% ethanol, and Birabresib at a final concentration of 100 µg/ml (Stock: 5 mg/ml). Control wells contained the solvent (1% PEG400 and 3% ethanol) with bacterial culture. The plate was incubated at 37 °C for 24 h in a microplate reader, and bacterial growth was monitored by measuring absorbance at 600 nm. All experiments were conducted in triplicate to ensure reproducibility.

Microdilution assay

The minimum inhibitory concentration (MIC)⁹⁰ assays were performed using a standard microdilution assay in a 96-well round-bottom microtiter plate. To perform the experiment, log-phase bacterial cultures were grown in LB medium from overnight cultures, and the inoculum was adjusted to an OD of 0.4–0.5. Different drug concentrations such as 50, 40, 30, 20, 19, 18, 17, 16, 15, 14, 13, and 10 µg/ml, were added in each well of a 96-well round bottom microtiter plate (270 µl of LB agar in each well), kept the plates on the bench for 2–3 h to solidify. A serial dilution of the bacterial culture was performed in 1X Minimal Media Broth (MMB), starting with a 10⁻¹ dilution and proceeding sequentially to 10⁻⁵, 5 µL of the serially diluted bacterial culture was added to each well, and the plate was incubated for 15–20 min until the liquid absorbed properly. The plates were incubated at 37 °C for 24 h to observe the results.

Human cell lines and cell culture

Human Caco-2 cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Sigma -D6429) complemented with 10% v/v fetal bovine serum (FBS) (Gibco) and 1% Penicillin–Streptomycin (Gibco). Cell cultures were maintained at 37 °C in a humidified incubator (Fischer brand Isotemp) supplied with 5% CO₂. These conditions were utilized for subsequent cell culture and MTT assay. Cells were grown to confluency in 96-well plates, and the assay was performed 1–2 days after seeding.

Human cell lines viability assay

The preliminary safety profile of Birabresib was evaluated using an MTT colorimetric assay to assess its effects on mammalian cells. Caco-2 cells at a 4 × 10⁴ cells/well were seeded with 100 µL culture medium into a 96-well plate (NEST®, Rahway, NJ). Cells were treated (100 µL) with the compound at different concentrations (~250 µM to 0.5 µM) in 100 µL of DMEM (Sigma -D6429) on the next day. Media (untreated) was used only as a negative control, and a mixture of PEG400 (1%) and ethanol (3%) was used as a positive control. After incubation for 24 h, the media containing the compound were removed, and cells were treated with 20 µL of MTT (Invitrogen-M6494) dye (5 mg/mL) and incubated at 37 °C for 4 h. In each well, 100 µL of DMSO (Fischer Scientific) solution was added. Cell viability was determined using a microplate reader (BioTek Instruments, EL808). The wavelength to measure the absorbance of the formazan product was 570 nm. Results were calculated as percentages and presented compared to the negative control values. Three biological replicates were used to perform this experiment.

Synergistic assay

The synergistic assay was performed using a 96-well plate to evaluate the combined effect of Birabresib and Rifampicin. The bacterial culture Bacillus Subtilis 1A1 was grown in Luria-Bertani (LB) broth until it reached an optical density (OD₆₀₀) of 0.5. Serial dilutions were prepared from 10⁻¹ to 10⁻⁵, with dilutions from 10⁻² to 10⁻⁵ used in the experiment. Different concentrations of Birabresib and Rifampicin were tested individually and in combination. Different drug concentrations were prepared in LB agar, and 270 µl of each concentration was added to each well in three replicates. The plate was left on the bench for 1 hour to solidify the medium. 5 µL of serially diluted (10⁻² to 10⁻⁵) bacterial culture was added to each well. The plates were incubated for 24 h at 37 °C. After incubation, 5 µL of 2,3,5-triphenyltetrazolium chloride (TTC) solution (100 µg/mL) was added to each well to assess bacterial growth, as TTC produces a color change (red color) in the presence of metabolically active bacteria.

Author contributions

H.K.K. and R.R. contributed equally to this work. H.K.K. designed the *in silico* experiment and conducted the initial wet-lab bacterial experiments, which R.R. later completed. H.K.K. also performed human cell line experiments. T.H. assisted H.K.K. in developing the *in vitro* experimental design. T.H. and N.C.B. initiated the project, with N.C.B. overseeing its planning and direction. H.K.K. and R.R. drafted the manuscript, with all authors contributing to discussions and revisions.

Data availability

All data supporting the findings of this study are included in the manuscript and its supplementary information files. Additional raw data are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s44259-026-00215-6.