Inhibitors of cytochrome c biogenesis pathways

ABSTRACT

The unique composition of bacterial electron transport chains (ETCs), distinct from the human mitochondrial ETC, presents opportunities for selective antibacterial targeting. Recently, inhibitors of bacterial ETCs have shown promise as potential antimicrobials, particularly against mycobacteria. Notably, bacterial cytochrome c (cyt c) biogenesis systems—Systems I and II—are absent in mammalian mitochondria, which utilize System III. This divergence suggests that cyt c biosynthesis pathways may serve as novel antibiotic targets. In this study, we screened a library of 1,760 FDA-approved compounds for inhibitors of in vivo bacterial cyt c biogenesis. We identified seven compounds that inhibit either System I and/or System II pathways. Among these, artemisinin and its analogs (ARTs) were found to directly inhibit in vitro cyt c biogenesis using purified CcsBA (System II) and human HCCS (System III). ARTs disrupt heme at the active sites of cyt c synthases, indicating their potential as direct inhibitors. However, therapeutic use of ARTs would likely require co-administration with inhibitors targeting alternative bacterial respiratory chains that do not involve cyt c. Additionally, three members of the 8-hydroxyquinoline (8-HQ) class—clioquinol, chloroxine, and broxyquinoline—inhibited in vivo cyt c biogenesis at low micromolar concentrations, below those needed to suppress bacterial growth. Results suggest that 8-HQs act through mechanisms beyond simple metal chelation (e.g., iron, magnesium), likely targeting multiple bacterial processes. Consistent with this hypothesis, 8-HQs did not directly inhibit in vitro cyt c synthases.

IMPORTANCE

This manuscript establishes bacterial cytochrome c biogenesis as a viable and previously underexplored antibacterial target that is fundamentally distinct from human mitochondrial pathways. Through systematic screening of 1,760 FDA-approved compounds, the study identifies two chemically and mechanistically distinct inhibitor classes—artemisinins and 8-hydroxyquinolines—that disrupt cytochrome c maturation via Systems I and II. The work moves beyond phenotypic growth inhibition by directly linking compound activity to heme degradation using complementary in vivo assays and in vitro experiments with purified cytochrome c synthases. Demonstrating that artemisinins directly target heme within these enzymes provides mechanistic insight with broad relevance to bacterial bioenergetics and drug-heme interactions. Importantly, the manuscript highlights the metabolic flexibility of bacterial respiratory chains and shows that inhibition of cytochrome c biogenesis alone is insufficient for robust killing. Together, these findings argue that cytochrome c biogenesis inhibitors will be effective in combination therapies and advance understanding of bacterial respiration.

INTRODUCTION

The urgent need for new antibiotics to combat bacterial pathogens has been well documented over the past few decades, primarily due to the growing prevalence of antimicrobial resistance (e.g., references 1–4). Since the discovery of penicillin nearly a century ago, numerous strategies for antibiotic discovery have emerged (e.g., references 5–9). Natural product screening from soil and other environments has yielded many important antimicrobial agents (reviewed in references 10–13). In addition, targeted approaches focusing on specific bacterial processes, such as transcription and translation, have proven successful (1, 14–17), along with broader searches for compounds that impede bacterial growth (18, 19).

More recently, components of bacterial energy metabolism—particularly electron transport chains (ETCs) and ATP synthesis—have emerged as promising targets. For example, Q203, an inhibitor of cytochrome bc₁, and aurachin D, which targets cytochrome bd, have been shown to suppress the growth of various mycobacterial species (20–24). Many bacteria possess ETCs that are structurally and functionally distinct from the mitochondrial ETC of their hosts (25–33), providing opportunities for selective inhibition. Most bacteria possess c-type cytochromes as components of ETCs. Some ETCs in bacteria are like the classic aerobic ETC in mitochondria, but other ETCs (and their specific c-type cytochromes) are used in anaerobic metabolisms. For c-type cytochromes, the heme group is covalently attached at a conserved CXXCH motif, a process that requires a cytochrome c synthase for biogenesis.

We have proposed that inhibiting cytochrome c (cyt c) biogenesis could disrupt multiple ETCs that rely on c-type cytochromes (34). Bacterial cyt c biogenesis occurs via pathways known as Systems I and II, which are mechanistically distinct from the simpler mitochondrial pathway (System III) (Fig. 1A through C) (reviewed in references 35–41). Selectively targeting Systems I and II could therefore enable bacteria-specific inhibition. These pathways operate in the periplasm, akin to peptidoglycan synthesis, making them accessible to drugs (Fig. 1A and B). In both systems, heme is exported from the cytoplasm to the periplasm, where it is covalently attached via thioether bonds to apocytochrome c at a conserved CXXCH motif before folding occurs. Recent structural advances, including high-resolution Cryo-EM and X-ray crystal structures of most proteins involved in Systems I and II, have laid the foundation for rational inhibitor design (Fig. 1A and B) (34, 42–46).

Fig 1.

Three cyt c biogenesis pathways: Systems I, II, and III. (A) System I is composed of eight Ccm proteins; CcmABCD exports heme to CcmE, which then delivers heme to the CcmF/H cyt c synthase. System I is in many gram-negative bacteria. (B) System II is a bifunctional CcsBA heme exporter and cyt c synthase. It is present in many gram-positive bacteria, cyanobacteria, mycobacteria, bacteroides, and some gram-negative bacteria. (C) System III is composed of HCCS, present in the mitochondrial intermembrane space of eukaryotes. Taken from (34).

In this study, we use recombinant Systems I and II to identify inhibitors of cyt c biogenesis. We screened the Pharmakon-1760 compound library—comprising 1,760 structurally diverse, biologically active drugs approved in the United States, Europe, or Asia (http://www.msdiscovery.com/pharmakon.html)—to identify inhibitors of cyt c maturation. Two compound classes, artemisinins (ARTs) and 8-hydroxyquinolines (8-HQs), were found to inhibit cyt c synthesis at low micromolar concentrations. We evaluate their inhibitory effects in vivo using recombinantly expressed Systems I, II, and III in Escherichia coli, and in vitro with purified Systems II and III. We employ mitochondrial System III since, ultimately, it may be necessary to find compounds that preferentially inhibit the bacterial pathways. Additionally, we assess their impact on bacterial growth and discuss their potential mechanisms of action and implications for future development of ETC-targeting antibiotics.

RESULTS

In vivo screening of recombinant E. coli strains for inhibitors of cytochrome c biogenesis Systems I and II

The Pharmakon-1760 compound library, described in the Materials and Methods section, was used to screen for inhibitors of cyt c biogenesis Systems I and II. The screen was conducted using two independent sets of 96-well microtiter plates, each containing a total of 1,760 compounds at a final concentration of 7.5 μM. One set was used to identify inhibitors of System I, and the other for System II.

As illustrated in Fig. S1, E. coli strains lacking the endogenous ccmA-H operon (System I) were complemented with plasmids expressing either the E. coli System I genes (ccmA-H) or the Helicobacter hepaticus System II (ccsBA) genes (47, 48). These recombinant strains were optimized for cyt c biosynthesis by induction with IPTG and co-transformed with a separate plasmid carrying a gene encoding the cyt c₄ reporter protein, also under inducible control.

Cultures were grown overnight, back-diluted, and induced with IPTG, and 200 μL aliquots were transferred into wells containing individual compounds. Arabinose was added to induce expression of cyt c₄ in each well, enabling cyt c biogenesis to proceed overnight. Inhibitors affecting transcription, translation, or the maturation pathways were expected to yield decreased cyt c₄ levels and were thus identified as positive hits.

Following incubation, cultures were centrifuged, and cell pellets were lysed using B-PER bacterial lysis buffer. Samples were analyzed by SDS-PAGE, followed by heme-specific staining, which selectively detects heme covalently attached to cyt c₄ via Systems I or II (Fig. S1). All c-type cytochromes contain heme that is covalently attached to the cytochrome c polypeptide. This attachment is catalyzed specifically by the cyt c synthase of each biogenesis system (Fig. 1). Because the covalently bound heme remains associated with the polypeptide after denaturing SDS-PAGE, it can be detected using a heme-specific stain. Accordingly, the bands shown in Fig. 2 and 3 represent a sensitive readout of cyt c production and cyt c synthase activity.

Fig 2.

Response of a subset of Pharmakon-1760 compounds in cytochrome c₄ assays. Heme stains of SDS-PAGE gels are shown, with the heme-stained band detecting heme attachment mediated by the indicated cyt c biogenesis system. Candidate inhibitors are shown in red. (A) Recombinant E. coli Δccm with System I producing cyt c₄ in the presence of the listed drugs. (B) Recombinant E. coli Δccm with System II producing cyt c₄. Rifampicin (green font) is a transcriptional inhibitor in the Pharmakon-1760 library that inhibits protein expression. kD refers to the molecular weights, with 20 kDa and 25 kDa markers shown.

As described below, Fig. 2 shows heme-stained bands (following SDS-PAGE) from whole E. coli cells expressing either System I or System II; thus, these experiments serve as in vivo assays for cyt c biogenesis. To determine whether the compounds act directly on cyt c synthases, we also performed in vitro activity assays using purified synthases. In these assays, covalent heme attachment was again assessed by heme staining after SDS-PAGE (Fig. 3).

Fig 3.

Inhibition of cyt c maturation by artemether and artemisinin in vitro. Either the purified System II CcsBA or human System III HCCS was used in cyt c biogenesis assays as described in Materials and Methods. (A and B) In vitro biosynthesis of cyt c by CcsBA was carried out in an anaerobic chamber. The heme stain visualizes heme-containing proteins (cyt c or 56-mer). Reactions performed in the presence of artemether and artemisinin (panel A, lanes 25 and 27, and panel B, lanes 23, 24, 26, 27) show inhibited cyt c maturation. In contrast, no inhibition was observed for pyrithione zinc (ZnP) (panel A, lanes 6, 14), disulfiram (panel A, lane 22), or the quinolines: chloroxine (panel B, lanes 8, 9), clioquinol (panel B, lanes 11, 12), or broxyquinoline (panel B, lanes 14, 15). (C–E) Titrations of artemether and artemisinin with CcsBA (C) or HCCS (D and E) showed a concentration-dependent inhibition. (F) A plot of the data presented in panels A–E ImageJ was used to measure the heme stain density. The signal in a given lane was normalized by the reaction signal in the absence of drug on a particular gel. Artemether-CcsBA (n = 3); HCCS-artemether (n = 2); HCCS-artemisinin (n = 1); CcsBA-ZnP (n = 1), CcsBA-disulfiram (n = 3); CcsBA-quinolines (n = 3). Error bars are the standard error of the mean (SEM). Determination of IC₅₀ values for in vitro inhibition of cytochrome c biogenesis by artemether and artemisinin. Dose-response curves were fit using a three-parameter nonlinear regression model in GraphPad Prism.

Compounds that resulted in reduced cyt c₄ levels were selected for further validation. B-PER lysates were reanalyzed by SDS-PAGE and heme staining, and the intensity of the cyt c₄ heme signal was quantified (Fig. 2; Table 1). Compounds with known antibacterial targets, such as tetracyclines (translation inhibitors) and rifampicin (transcription inhibitor), were excluded from further consideration.

TABLE 1.

Compounds from the Pharmakon-1760 collection with indicated cyt c biogenesis inhibition^a

Gel panel	Drug name	System I % heme signal/WT	System II % heme signal/WT
i	Acriflavinium hydrochloride	41	60*
ii	Artemether	47*	15
iii	Artenimol	6	36*
iv	Artemisinin	6	43*
v	Bleomycin	59*	31
vi	Broxyquinoline	28	92*
vii	Chloroxine	4	22
viii	Clioquinol	7	61*
ix	Cefotetan	41	15
x	Disulfiram	73*	35
xi	Nitazoxanide	7	85*
xii	Oxyclozanide	18	111*
xiii	Oxyphencyclimine	99*	11
xiv	Phenylmercuric acid/acetate	10	2
xv	Pyrithione zinc	14	12

^a An asterisk indicates that the heme signal was calculated based on the original Pharmakon-1760 gel; all others are calculated based on those in Fig. 2. Shading indicates compounds with no previously characterized targets in bacteria were prioritized for follow-up studies.

From the screen, 15 compounds that produced a reduction in heme signal to 42% or less of the control were selected for further analysis (Table 1; compound structures shown in Table S1). Of these, seven compounds with no previously characterized targets in bacteria were prioritized for follow-up studies (shaded in gray in Table 1).

Drug effects on in vitro cytochrome c biogenesis via purified System II (CcsBA) and System III (human HCCS)

To evaluate the specific effects of selected compounds on cyt c biogenesis, we investigated their impact on in vitro cyt c synthesis using purified System II (CcsBA) and System III (human HCCS) enzymes (49). These assays employed either full-length apocytochrome c or a previously characterized 56-amino-acid peptide containing the CXXCH motif as the heme acceptor (Fig. 3) (49).

Among the tested compounds, pyrithione zinc (ZnP), disulfiram, and the three 8-HQs—chloroxine, clioquinol, and broxyquinoline—did not inhibit cyt c synthase activity, even at concentrations exceeding 50 μM (Fig. 3A and B). In contrast, both artemisinin (ART) derivatives, artemether and artemisinin, exhibited significant inhibitory effects, with approximate IC₅₀ values of ~20 µM (Fig. 3C through F).

To further explore the mechanism of inhibition by ARTs, we analyzed the spectroscopic signatures of heme during the in vitro reactions catalyzed by CcsBA and HCCS (Fig. S2). Both synthases harbor b-type heme at their active sites, which initially yields characteristic absorption maxima near 560 nm (Fig. S2Aii, Bii, Ci, and Di). Formation of c-type heme, involving thioether linkages, produces a distinct absorption maximum in the 550–555 nm range, distinguishable from the 560 nm peak of b-type heme (Fig. S2Ai, Bi, Cii, and Dii) (46, 48–51).

In the absence of ARTs, the expected spectral transitions and heme levels were consistent with normal enzymatic activity—that is, a shift from b-heme to c-heme characteristics as the reaction proceeded. However, in the presence of ARTs, a dose-dependent degradation of heme was observed for both CcsBA and HCCS reactions (Fig. S2C and SD). Consistent with this, heme staining of cyt c and the 56 mer peptide following SDS-PAGE revealed minimal heme attachment in ART-treated samples (Fig. 3). This aligns with the marked reduction of the 550 nm peak (compare Fig. S2Ai through vi and vii).

Additionally, the Soret band (~420 nm), which reflects both b-type and c-type heme species, showed decreased intensity with increasing artemether concentrations (Fig. S2Avi, Bvi, and Cvi), indicating an overall reduction in heme content. These findings suggest that ARTs directly interact with the heme bound within the active site of the cyt c synthases.

This observation parallels the known mechanism of ARTs in Plasmodium species, where they inhibit hemozoin formation by forming covalent adducts with heme (e.g., references 52–66). Based on these results, we propose that ARTs are activated by heme within CcsBA and HCCS, leading to heme adduct formation, heme degradation, and ultimately the inactivation of cyt c synthase activity. We do not observe ARTs-dependent protein degradation (e.g., Fig. S3B, Coomassie-stained gel 1, compare *CcsA:His, CcsB* bands in lane 5 to lanes 10 and 12; gel 2, compare lane 5 to lanes 9 and 11; gel 3, compare lane 3 to lanes 5 and 6).

Analysis of growth inhibition of E. coli and determination of IC₅₀ values in vivo for selected drugs targeting cyt c biogenesis

To investigate the potential of selected compounds that inhibit cyt c biogenesis in vivo, we focused on two representatives from each drug class, ARTS (artemisinin and artemether) and 8-HQs (chloroxine and clioquinol). We assessed their effects in recombinant E. coli strains engineered to express cytochrome c maturation Systems I, II, or III (Fig. S1).

To further evaluate the specific impact on cyt c biogenesis of each drug, we selected concentrations below levels that inhibited growth (confirmed via OD₆₀₀ measurements; Fig. S4). Thus, OD₆₀₀ measurements indicate that protein synthesis and stability are not impacted by the drugs. Because most apocyt c proteins are degraded if heme is not attached (39), assaying for heme attached to the cyt c protein using heme stains of SDS-PAGE is a direct measure of cyt c synthase activity.

To determine IC₅₀ values for in vivo cyt c biogenesis, E. coli strains lacking the eight endogenous System I ccm genes were transformed with plasmid-borne maturation Systems I, II, or III. Note that E. coli Δccm strains are incapable of producing c-type cytochromes unless complemented with a System but remain viable due to alternative terminal oxidases (cyt bd and cyt bo). E. coli does not require c-type cytochromes for aerobic growth. Cultures were induced following the Pharmakon screening protocol (Fig. S1) with the volume of 5 mL, with compounds added after IPTG induction and 30 minutes prior to arabinose addition. Cyt c₄ levels were assessed by SDS-PAGE followed by heme staining (Fig. S5). A summary of IC₅₀ values is presented in Fig. 4.

Fig 4.

Dose-response analysis of inhibitor activity against cytochrome c biogenesis Systems I–III in vivo in E. coli Δccm. Heme stain assays for cyt c with the indicated concentrations of drugs were performed as shown in Fig. S5. (A) Nonlinear regression curves showing the response of Systems I (red circles), II (green squares), and III (blue triangles) to increasing concentrations of candidate inhibitors. Data points represent the mean ± SD from independent replicates, with n shown in the table. Each curve was fitted with a three-parameter logistic model to determine half-maximal inhibitory concentrations (IC₅₀). (B) Summary of the fitted IC₅₀ values (with 95% confidence intervals).

Artemether and artemisinin inhibited cyt c biogenesis via Systems I and II with IC₅₀ values ranging from 2.1 to 2.7 μM. Notably, artemether also inhibited System III (human HCCS) with an IC₅₀ of approximately 0.7 μM. In contrast, artemisinin showed inhibition for System III about 15-fold weaker at 12.2 μM. This differential activity may result from limited membrane permeability of artemisinin compared to artemether. Unlike Systems I and II, which function in the periplasm (Fig. 1), the human HCCS catalyzes heme attachment in the cytoplasm. For Systems I and II, cyt c₄ is equipped with a signal sequence for periplasmic targeting, whereas human cyt c lacks such a sequence and remains cytoplasmic.

Our results demonstrate that ARTs inhibit cyt c biogenesis in vivo at low micromolar concentrations—consistent with their in vitro inhibitory effects on both CcsBA and HCCS (Fig. 3).

Chloroxine and clioquinol inhibited bacterial Systems I and II with IC₅₀ values in the submicromolar range (0.43 μM–1.0 μM), well below concentrations that impair E. coli growth. In System III, clioquinol exhibited a higher IC₅₀ of 2.8 μM—approximately threefold higher than for the bacterial systems. These findings suggest that while both ARTs and 8-HQs interfere with cyt c maturation, their potency and mechanisms of action may differ depending on the maturation system and subcellular localization of the target enzyme.

Inhibition of bacterial growth as a proxy to identify potential targets and antibiotics

Eight compounds were selected for further evaluation of antibacterial activity, including two artemisinin derivatives (ARTs: artemisinin and artemether), three 8-HQs: clioquinol, chloroxine, and 8-bromoquinoline), and three additional compounds from Table 1. Antibacterial activity was assessed using disk diffusion assays to measure zones of inhibition (Fig. 5).

Fig 5.

Zones of inhibition produced by eight candidate cytochrome c biogenesis inhibitors on six bacterial strains with distinct respiratory systems. (A) Bar graph shows the mean inhibition diameter (mm) from n = 4 independent biological replicates. (B) Representative agar plates showing the primary data used to measure zones of inhibition for the indicated bacterial strains, as described in Materials and Methods. (C) Summary of the drug concentration, the mean zone of inhibition, and standard deviations. ZnP, pyrithione zinc; PMA, phenyl mercuric acetate.

We tested six phylogenetically diverse bacteria, chosen for their distinct respiratory systems and health relevance: two gram-positive species—Bacillus subtilis and Mycobacterium smegmatis (both utilizing cytochrome c biogenesis System II)—and four gram-negative species—Escherichia coli, Pseudomonas aeruginosa, Serratia marcescens, and Shewanella oneidensis (all using System I). Shewanella was included due to its unusually high c-type cytochrome content, which supports aerobic, anaerobic, and nanowire ETCs (e.g., references 67–70). All six bacteria were cultured aerobically, as detailed in Materials and Methods. Their ETC composition and cytochrome usage are well-characterized (e.g., references 29, 33, 71).

Figure 5 shows representative inhibition zones for all eight compounds across the six bacterial species; full data sets are presented in Fig. S6. All compounds except the ARTs exhibited zones of inhibition in at least one bacterial species. Disulfiram, pyrithione zinc (ZnP), phenyl mercuric acetate (PMA), and all three 8-HQs demonstrated antibacterial activity, suggesting broad-spectrum or selective inhibitory effects.

Given that 8-HQs are established metal chelators (e.g., references 72–75) and that cytochrome c biogenesis requires iron for heme synthesis, we further investigated whether their mechanism of action is primarily via metal chelation. To assess this, we compared their activity profiles with known metal chelators, including the iron chelator dipyridyl and the general chelator EDTA (Fig. S7). Our rationale was that if the mechanism of action of the 8-HQs is as specific metal chelators, the zones should be similar relative to the known chelators among the six bacteria.

Interestingly, the antibacterial profiles of the 8-HQs did not align with those of dipyridyl or EDTA. For example, none of the 8-HQs produced inhibition zones against P. aeruginosa or S. marcescens, whereas dipyridyl yielded large zones in both species. Conversely, M. smegmatis showed sensitivity to all 8-HQs but was unaffected by dipyridyl. E. coli and S. oneidensis were highly sensitive to dipyridyl but not to 8-HQs. EDTA exhibited broad activity across all species.

We also determined IC₅₀ and IC₉₀ values for the inhibition of Mycobacterium tuberculosis (Mtb) by artemisinin, artemether, clioquinol, and chloroxine (Table 2; Fig. S8). Consistent with the M. smegmatis data, M. tuberculosis was sensitive to the 8-HQs but resistant to ARTs.

TABLE 2.

MABA^b IC₅₀ values

Compound	IC50 ± SD (µM)	IC90 ± SD (µM)
Artemether	299.034 ± 2.090	378.309 ± 2.449
Artemisinin	386.306 ± 86.230	501.350 ± 121.408
Clioquinol	9.580 ± 0.124	11.728 ± 0.196
Chloroxine	5.122 ± 0.602	6.561 ± 0.739
INH	0.284 ± 0.013	NDa

^a ND, unable to calculate INH IC₉₀ due to insufficient data points.

^b MABA, microplate alamar blue assay.

These results suggest that if the 8-HQs acted solely through metal chelation, their inhibition patterns would mirror those of other chelators. Since they do not, we propose that 8-HQs may act through multiple mechanisms, including inhibition of cytochrome c biogenesis (see Discussion).

In contrast, the ARTs exhibited minimal antibacterial activity, with only artemether showing a zone of inhibition—and only against S. oneidensis. Two possible explanations may account for this limited activity: (i) functional redundancy in bacterial respiratory chains, where multiple terminal oxidases (e.g., c-type, bd-type, or a-type) allow respiration to continue despite partial ETC inhibition (34); or (ii) poor compound permeability, limiting ARTs’ access to intracellular or periplasmic targets.

To explore this further, we tested whether artemether affects cytochrome c synthesis in S. oneidensis. Cells were grown aerobically in increasing concentrations of artemether, and c-type cytochrome levels were analyzed via SDS-PAGE and heme staining. Even at sub-inhibitory concentrations, artemether reduced c-type cytochrome levels (Fig. S9), indicating specific inhibition of the cytochrome c biogenesis pathway. This also implies that artemether reaches the periplasm, where System I components such as CcmF and CcmH are localized (see Fig. 1A).

DISCUSSION

We previously demonstrated that CcsBA requires only a CXXCH peptide as a substrate, whereas human HCCS requires a longer peptide that includes both the CXXCH motif and an upstream α-helix (49). These findings, based on in vitro reconstitution of bacterial CcsBA and human HCCS, align with in vivo data using HCCS substrates (76, 77). Additionally, we showed that short CXXCH peptides lacking the extended α-helix act as inhibitors of CcsBA activity in vitro, but not of HCCS (49). The proposed mechanism involves these short peptides serving as competitive substrates, thereby consuming heme that would otherwise be incorporated into full-length apocytochrome c. Whether these short peptides also inhibit CcsBA in vivo remains uncertain, though peptide accessibility to the periplasm likely complicates such investigations. Another unresolved question is whether these peptides cause dead-end inhibition of the synthase. Modifying the CXXCH peptides with electrophilic “warheads” could potentially enhance inhibition, but again, periplasmic access and interference from cellular thiols present significant technical hurdles.

To expand the search for inhibitors of cytochrome c biogenesis, we conducted in vivo screens using a library of 1,760 compounds approved for human use across a broad range of biological targets. Among the hits (Table 1), we identified two distinct classes of compounds—ARTs and 8-HQs—each represented by three active members that inhibited cytochrome c biogenesis via Systems I and/or II. We first discuss the ARTs.

Our results show that ARTs inhibit cytochrome c biogenesis in vivo across Systems I, II, and III, with IC₅₀ values in the low micromolar range (Fig. 4). In vitro assays using purified CcsBA and human HCCS confirmed direct inhibition of synthase activity, likely via interaction of ARTs with the heme cofactors of the synthases. Drawing on a substantial body of literature describing ART-heme adduct formation in malaria parasites (e.g., references 52–66), our findings support the hypothesis that similar ART-heme adducts disrupt cyt c synthase activity. We have not observed degradation of the cyt c synthase proteins themselves in our in vitro assays. This suggests that the major reaction of ARTs is with the heme in the cyt c synthases, at least early on in reactions.

While our study was ongoing, a separate report described that ARTs may chemically rescue a respiratory-deficient Saccharomyces cerevisiae Bcs1 mutant by reducing cyt bc_₁ overaccumulation (78). This mutant accumulates inactive cyt bc_₁ complexes due to impaired Bcs1 chaperone function. Notably, ARTs reduced cyt bc_₁ levels in both yeast and human BCS1 mutants, suggesting a therapeutic potential for certain mitochondrial disorders. That study also reported covalent ART adducts with cytochrome c. Although we cannot exclude the possibility of ARTs targeting cytochrome c directly, our in vitro data point to the heme cofactors in cyt c synthases as the primary molecular targets.

A recent report showed that ARTs permeate both gram-positive and gram-negative bacteria and exhibit bactericidal activity against Vibrio cholerae but not E. coli, P. aeruginosa, or B. subtilis (79). Interestingly, copper was shown to sensitize bacteria to ARTs in that study. In our experiments, only Shewanella exhibited zones of inhibition in response to artemether. At micromolar concentrations, artemether led to decreased levels of c-type cytochromes in Shewanella. These findings suggest a potential for ARTs or related compounds to serve as inhibitors of bacterial Systems I or II in future antimicrobial development.

The second inhibitor class identified through our in vivo screens was the 8-HQs. While 8-HQs have long been known to inhibit bacterial growth, their precise mechanism of action has remained unclear (e.g., references 80, 81). Recent studies suggest that 8-HQ activity requires aerobic growth and is enhanced in the presence of zinc, with superoxide dismutase (SOD) implicated as a potential target (82). Given that 8-HQs are known metal chelators, we considered whether their antibacterial activity might result from sequestering essential metals such as iron, thereby indirectly inhibiting heme synthesis and, consequently, cytochrome c biogenesis.

To investigate this, we compared bacterial sensitivity to known iron chelators (e.g., dipyridyl) with sensitivity to 8-HQs across a diverse panel of bacterial strains (Fig. S7). No clear correlation was observed, suggesting that metal chelation alone does not account for the activity of 8-HQs. While we do not rule out metal chelation as a contributing factor, our results support a more complex mechanism involving multiple cellular targets. At varying concentrations, 8-HQs may act through a combination of iron chelation, SOD inhibition, and disruption of quinone-mediated electron transport. Importantly, our results show that 8-HQs inhibit cytochrome c biogenesis, although this is unlikely to be the sole mechanism of growth inhibition.

This conclusion is supported by our observation that E. coli, which lacks a functional cytochrome c biogenesis pathway and does not rely on a c-type cytochrome-based aerobic electron transport chain, remains sensitive to all tested 8-HQs. Thus, while cytochrome c biogenesis may be a relevant target, additional cellular processes are likely affected by 8-HQs.

Given the metabolic flexibility of bacteria, particularly in their respiratory chains, we propose that inhibitors of cytochrome c biogenesis alone are unlikely to exhibit broad-spectrum antibacterial activity. Many bacteria have evolved redundant and versatile electron transport chains (e.g., references 34, 83, 84), allowing them to bypass inhibition of any single component. Indeed, among the species tested, only Shewanella, which relies heavily on c-type cytochromes, exhibited clear susceptibility to ARTs. This highlights the challenge of targeting cytochrome c biogenesis as a standalone antimicrobial strategy.

As demonstrated in studies on mycobacteria (30, 85–87), effective targeting of bacterial respiration may require combination therapies that inhibit multiple components of the electron transport chain. The precedent set by inhibitors of cytochrome bc₁ (22, 30, 86, 88) and cytochrome bd (89–91) underscores the promise of bioenergetic targets in antimicrobial development. While most bacterial electron transport chains require heme, not all rely on c-type cytochromes. Thus, inhibition of heme biosynthesis itself, as proposed in recent studies (92), may offer a broader strategy. However, some pathogens can scavenge heme from their host, which may limit the efficacy of such approaches.

In conclusion, the diversity and adaptability of bacterial respiratory pathways necessitate multi-target strategies for the development of effective antibiotics. Inhibiting cytochrome c biogenesis could be one component of a broader approach targeting multiple respiratory enzymes and pathways.

MATERIALS AND METHODS

Pharmakon-1760 screen

The Pharmakon-1760 library (1.5 nmol of each drug) was tested for the ability to inhibit cyt c biogenesis by System I or System II in a volume of 200 μL (final drug concentration of 7.5 μM) (MicroSource Discovery Systems, Inc.).

E. coli with System I or System II was grown overnight to saturation in a rocking incubator. That culture was back-diluted into Luria Bertani broth in a 1:5 ratio and grown for 3 hours at 37°C and 200 rpm. The expression of System I or System II was induced with 0.1 mM IPTG and continued shaking for 1 hour. Two hundred microliters of E. coli was aliquoted into the plate wells. Note columns 1 and 12 do not contain any drugs. We used column 1 for a positive control (System I, II, and cyt c) and column 12 as a negative control (no cyt c plasmid in E. coli). The drug was allowed to interact with the E. coli for 30 minutes. Subsequently, cyt c expression was induced with 0.2% arabinose. The plate was incubated overnight with shaking. E. coli was harvested by centrifugation the following day. The supernatant was emptied from the plate and placed at −80°C.

To extract cyt c and test for the efficacy of the test drug, the E. coli was thawed and resuspended in 40 µL of B-PER Complete Bacterial Protein Extraction Reagent (Thermo Fisher Scientific).

In-gel heme stain

Samples were run on 4-20% or 15% gels purchased from Bio-Rad. An in-gel heme stain showed the presence of matured cyt c (band at 24 kDa) in the following way: The gel was rinsed with deionized water and then fixed with Coomassie Destain (40% water, 50% methanol, and 10% acetic acid) for 5 minutes. Forty-five milligrams of 3,3′,5,5′-tetramethylbenzidine (TMBZ) was dissolved in 22.5 mL of methanol with rocking. It was then diluted to 75 mL with 0.2 M acetate buffer pH 3. After fixing the gel, it was incubated in the TMBZ solution for 15 minutes. Finally, 450 µL of hydrogen peroxide was added to the gel to develop the heme-stainable bands for 10 minutes. The gel was then imaged with a Fuji camera. Hits were identified that had less cytochrome c expression compared to the positive control.

In vitro experiments

Experiments were performed as previously described (49). In brief, samples were put into an anaerobic chamber (N₂ [95%] and H₂ [5%] in a Coy anaerobic airlock chamber) and allowed to equilibrate for 30 minutes. CcsBA (or HCCS) was combined with apo cyt c or a 56-mer peptide that contains the heme attachment site. Samples were imaged with a UV spectrophotometer, and 10 μL was collected and placed in 10 μL loading buffer to run as T = 0 on an SDS-PAGE gel. The samples were then incubated at 37°C for 4 to 24 hours. Longer time allowed for the signal of cyt c to dominate the heme signature as CcsBA-heme signal is lost over time. The reaction was then imaged by UV-vis spectrometry. Ten microliters was collected and placed in gel loading buffer.

Gel samples were run on an SDS-page gel, rinsed, and imaged using the in-gel heme stain method with TMBZ.

IC₅₀ determination by heme stain of cyt c biogenesis pathways

In the follow-up experiments to determine the IC₅₀ of the candidate drugs, we followed the same protocol that we used for the plates. We back-diluted a saturated overnight culture into the number of test tubes that we required for the experiment at a ratio of 1:5 with a 5 mL volume. We allowed the culture to enter the log phase by growing it for 3 hours. Then we induced the expression of Systems I, II, or III with 0.1 mM IPTG. Unique to System III, the rotation speed was reduced after induction to 120 rpm from 200 rpm. Then the E. coli culture was added to the respective drugs in test tubes for 30 minutes, at which time 0.2% arabinose was added to the culture to induce the expression of cyt c. Cultures were grown overnight and then harvested by centrifugation.

Samples were stored at −80°C until processing. For lysis, samples were thawed at room temperature in the presence of B-PER Complete reagent for 15 minutes. The pellet was then resuspended by vortexing and incubated for an additional 15 minutes at room temperature. Insoluble material was removed by centrifugation at 16,000 × g for 10 minutes; this step was repeated three times. The final supernatant was collected and analyzed by SDS–PAGE.

Heme stain analysis by transfer

For the IC₅₀ determination, the protein from SDS-PAGE was transferred to a PVDF membrane. SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific) was used to detect the presence of heme. The chemiluminescence was detected using a LI-COR Odyssey (LI-COR).

Densitometry

Band densitometry was detected using the built-in software on the LI-COR software. A user-defined section of the gel was used as background. The band density was determined and exported. ImageJ was used in some instances to determine band densitometry, for example, Shewanella, and for in-gel heme stain quantitation from in vitro experiments (93).

Zones of inhibition

To obtain information about a range of bacteria that respond to our drugs of interest, we employed a zone of inhibition experiment. In this experiment, 50 μL of an overnight culture (37°C for E. coli-C43 Δccm, M. smegmatis-general laboratory strain (gls), P. aeruginosa-gls, 30°C for S. marcescens-ATCC 43862, S. oneidensis-gls, B. subtilis-gls) was plated onto a petri culture dish (size, 10 cm) of Luria Bertani. A filter disk was then placed on the plate, and 10 μL of the respective drugs was added to the disk. Plates were incubated at their respective temperatures until a zone was observed or a lack thereof. These methods have been widely used, and the low standard deviations of the measured zones (e.g., Fig. 5C) support their validity. As noted in the text, some bacterial species were selected because their cytochrome c compositions are well-characterized, even though they are not clinical pathogens.

IC₅₀ determination by microplate alamar blue assay (MABA) in Mtb

MABAs were performed with Mtb as previously described (88, 94, 95). Briefly, a culture of wild-type Mtb Erdman strain was initiated by inoculating 200 µL of frozen stock into 40 mL of Difco Middlebrook 7H9 liquid medium supplemented with 5 g/L BSA, 2 g/L dextrose, 0.003 g/L catalase, 0.85 g/L sodium chloride, 0.5% glycerol, and 0.05% Tween 80. The culture was grown in a 1 L roller bottle and incubated at 37°C while rolling for 5 days until it reached logarithmic growth phase at an OD₆₀₀ of approximately 0.6. It was then diluted in 7H9-T media (7H9 media with Tween 80) to an OD₆₀₀ of 0.0016. From this dilution, 100 µL was added to each well of a 96-well plate containing 100 µL of serially diluted test compounds prepared in 7H9-T. The final DMSO concentration was normalized to 1% across all wells. To minimize edge effects due to evaporation, the outermost wells were filled with 200 µL of 7H9-T to serve as a moat.

The microplates were incubated in a humidified incubator at 37°C and 5% CO₂ for 7 days. After incubation, 32.5 µL of a solution consisting of 0.6 mM resazurin and 20% Tween 80 in an 8:5 ratio was added to each well. Plates were returned to the incubator for an additional 24 hours. Fluorescence was then measured using a BioTek Synergy H1 Microplate Reader (excitation: 530 nm; emission: 590 nm).

Relative fluorescence units (RFU) for each well were converted to percent inhibition using the following formula:

$${\displaystyle {\displaystyle \mathrm{\%}\mathrm{i}\mathrm{n}\mathrm{h}\mathrm{i}\mathrm{b}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}=[1-(w-n)/(p-n)]\times 100\mathrm{\%}}}$$

where w is RFU of the well of interest, p is the average RFU of wells containing Mtb only (positive control), and n is the average RFU of wells containing media only (negative control).

Compound inhibition curves were generated using GraphPad Prism, and a non-linear regression model was manually fitted using Microsoft Excel Solver to estimate the IC₅₀ and IC₉₀.

In vivo assay of cytochrome response in Shewanella oneidensis

Overnight cultures of S. oneidensis were 10-fold diluted into 5 mL of LB with eight different concentrations of artemether (5 mM, 500 μM, 50 μM, 5 μM, 500 nM, 50 nM, 5 nM, 0 M). The cultures were grown at 200 rpm at 30°C for 5 hours and harvested. The bacteria were frozen overnight at −80°C. The protein was extracted with 200 μL of B-PER complete. The spectra were measured, and 10 μL were run on an SDS-PAGE gel from each sample. The protein was transferred to PVDF (Bio-Rad) and imaged with SuperSignal West Femto reagent (Thermo Fisher Scientific). The same blot was then imaged with Sypro Ruby (Bio-Rad) to see total protein levels. Data were collected from three independent experiments. The percent inhibition was plotted assuming complete cytochrome c inhibition at 5 mM. The data were fitted with a three-parameter logistic model to determine IC₅₀. The 95% confidence interval for the IC₅₀ is reported in Fig. S9C.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/mbio.00273-26.

Supplemental material. mbio.00273-26-s0001.pdf.

Table S1; Figures S1-S9.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.