A Specialized Polythioamide‐Binding Protein Confers Antibiotic Self‐Resistance in Anaerobic Bacteria

Abstract

Understanding antibiotic resistance mechanisms is central to the development of anti‐infective therapies and genomics‐based drug discovery. Yet, many knowledge gaps remain regarding the resistance strategies employed against novel types of antibiotics from less‐explored producers such as anaerobic bacteria, among them the Clostridia. Through the use of genome editing and functional assays, we found that CtaZ confers self‐resistance against the copper chelator and gyrase inhibitor closthioamide (CTA) in Ruminiclostridium cellulolyticum. Bioinformatics, biochemical analyses, and X‐ray crystallography revealed CtaZ as a founding member of a new group of GyrI‐like proteins. CtaZ is unique in binding a polythioamide scaffold in a ligand‐optimized hydrophobic pocket, thereby confining CTA. By genome mining using CtaZ as a handle, we discovered previously overlooked homologs encoded by diverse members of the phylum Firmicutes, including many pathogens. In addition to characterizing both a new role for a GyrI‐like domain in self‐resistance and unprecedented thioamide binding, this work aids in uncovering related drug‐resistance mechanisms.

A central contributor to the self‐resistance of an anaerobic antibiotic producer is identified. CtaZ, a new member of the GyrI‐like superfamily, confines the gyrase inhibitor closthioamide in a ligand‐optimized pocket. Genome mining uncovered CtaZ homologs encoded in the genomes of diverse bacteria, including numerous pathogens.

Introduction

Most natural environments harbor a remarkably complex collection of microorganisms, ^[1] where bacteria must compete with their neighbors for space and resources. ^[2] One way in which bacteria do this is by producing antibiotic natural products to inhibit susceptible competitors. ^[3] Antibiotic‐producing bacteria require a means of protection against suicide caused by the toxic effects of the antibiotic, which is most often mediated by self‐resistance genes. ^[4] The self‐resistance strategies employed by antibiotic‐producing bacteria include export and reduced influx of antibiotics, target modification, sequestration, and enzymatic inactivation,[ ^4b , ⁵ ] which are mirrored in the counter‐measures used by target organisms to defend themselves. This co‐evolution of resistance can occur spontaneously, for example by alteration of the antibiotic target through mutation, or by horizontal transfer of resistance genes from other bacterial species. ^[5] Unlike natural ecosystems where microbes produce low concentrations of multiple antibiotics, humans tend to deploy large concentrations of single antibiotics in medical and agricultural settings. ^[6] Unfortunately, the accompanying promotion of antimicrobial resistance (AMR) inevitably goes on to limit the health and economic benefits of antibiotic use.

Still, antibiotics are our most important weapon to fight infectious diseases, with the vast majority in clinical use being either microbial natural products or their derivatives. ^[7] While AMR has been an ongoing problem since the first clinical use of antibiotics, the rise in the number and diversity of resistant pathogens, including multidrug‐resistant strains, is raising world‐wide concern. ^[8] In the majority of cases, the origins of the genetic determinants of clinical resistance can be traced to the self‐protection mechanisms of antibiotic producers in natural environments. ^[9]

The GyrI‐like protein superfamily is of particular interest from the perspective of AMR since the conserved fold is functionally malleable, being capable of mediating various (self‐)resistance strategies by differing modes of action. ^[10] GyrI, the founding member of the superfamily, protects Escherichia coli against exogenously produced microcin B17 by inhibiting DNA gyrase, which is the cellular target of the bacteriocin. ^[11] Furthermore, one subgroup contains enzymes that inactivate cyclopropyl antibiotics by hydrolysis, with two of these enzymes involved in the self‐resistance of their producers. ^[10c] Lastly, the transcriptional regulator BmrR from Bacillus subtilis binds diverse toxic compounds with its GyrI‐like domain and subsequently activates the expression of efflux pumps to confer multidrug resistance (MDR). ^[12]

There is a pressing need for continued characterization of the molecular underpinnings of (self‐)resistance mechanisms since this understanding will both guide the prudent use of antibiotics to minimize the emergence of clinically problematic AMR, as well as aid in the development of effective antibiotics capable of circumventing AMR. ^[13] To this end, producers of unusual antibiotics could inform on novel self‐resistance strategies. For instance, Ruminiclostridium cellulolyticum DSM 5812 is a soil anaerobe that produces closthioamide (CTA, Figure 1A), an antibiotic with a unique polythioamidated structure and mode of action. ^[14] CTA is a potent DNA gyrase inhibitor that is active against a number of clinically relevant pathogens, such as methicillin‐resistant Staphylococcus aureus, vancomycin‐resistant enterococci and Neisseria gonorrhea.[ ^14a , ¹⁵ ] CTA has copper‐complexing properties ^[16] and is a promising drug candidate because bacteria that are resistant to other DNA gyrase inhibitors, such as ciprofloxacin, remain susceptible to CTA.[ ^14c , ¹⁵ ] Although the biosynthetic pathway to CTA has been completely deciphered, ^[17] the manner in which the producer is protected from self‐intoxication remains a mystery.

Figure 1.

Adjacent to the closthioamide (CTA) biosynthetic gene cluster is a gene (ctaZ) encoding a GyrI‐like protein. A) Structure of CTA. B) CTA biosynthetic gene cluster. C) Neighbor‐joining phylogenetic tree of GyrI‐like domains from multidrug resistance and self‐resistance proteins, and CtaZ homologs (1000 bootstrap replicates). Red triangles mark proteins encoded in biosynthetic gene clusters. For a more detailed phylogenetic tree, see Figure S1 and Figure S2 for the sequence alignment used to generate this tree.

Here, we report the identification of a GyrI‐like protein (CtaZ) as a central contributor to the self‐resistance of R. cellulolyticum to CTA, provide insight into thioamide binding in a ligand‐optimized hydrophobic pocket, and demonstrate the prevalence of CtaZ homologs, including among many pathogenic Firmicutes.

Results and Discussion

Since genes coding for self‐resistance factors are often located in the vicinity of antibiotic biosynthetic gene clusters (BGCs), ^[4a] we examined the flanking genes of the CTA BGC of R. cellulolyticum with the goal of identifying possible self‐resistance genes (Figure 1B). The gene Ccel_3263 (referred to hereafter as ctaZ) is encoded upstream of the CTA BGC and is annotated in the GenBank database as encoding a “transcription activator effector binding protein”. ^[18] Our in silico analysis of the protein product of ctaZ using BLASTp ^[19] and HHpred ^[20] indicated that CtaZ belongs to the GyrI‐like protein superfamily (PF06445). This finding drew our attention given that GyrI‐like domains occur in several proteins involved in MDR ^[10b] or self‐resistance of antibiotic producers. ^[10c] In fact, the only GyrI‐like proteins characterized to date that are encoded in the proximity of BGCs, C10R6 and YtkR7, have been implicated in self‐resistance to the cognate antibiotics. ^[10c] In view of the co‐occurrence of a gene encoding a GyrI homolog (CtaZ) with the CTA BGC, we deemed CtaZ as a promising candidate for the elusive self‐resistance factor.

Proteins possessing GyrI‐like domains engage in varied cellular functions, such as DNA gyrase inhibition (GyrI),[ ^11b , ²¹ ] transcriptional regulation (BmrR and Rob) ^[22] and hydrolysis of antibiotics (Lin2189, C10R6, YtkR7). ^[10c] We conducted a phylogenetic analysis to contextualize CtaZ to the characterized members of the GyrI‐like superfamily. Homologs of CtaZ, GyrI, BmrR, Rob and the three hydrolases Lin2189, C10R6, YtkR7 were retrieved by BLASTp to construct a neighbor‐joining phylogenetic tree (Figure 1C and Figure S1). The phylogenetic analysis reveals that CtaZ and its homologs branch separately from the other GyrI‐like superfamily members, including the clade containing the self‐resistance‐mediating hydrolases (C10R6, YtkR7). The localization of CtaZ to a distinct clade led us speculate that, if indeed CtaZ bestows CTA resistance on R. cellulolyticum, it likely does so by way of a different cellular function.

In order to investigate whether CtaZ acts as a self‐resistance protein, we sought to inactivate the corresponding gene in R. cellulolyticum. The CTA susceptibility of the resulting mutant strain could then be tested to verify whether CtaZ contributes to the CTA tolerance of R. cellulolyticum. R. cellulolyticum ΔctaZ was generated by targeting ctaZ for incorporation of an in‐frame nonsense mutation by CRISPR‐nCas9 genome editing and verifying by restriction analysis (Figure 2A and Figure S3). We noted that the growth of R. cellulolyticum ΔctaZ was impaired compared to the wild type only under conditions known to induce CTA production (Figure 2B), ^[17a] presumably due to endogenous CTA production in the absence of CtaZ. We circumvented this issue by inactivating ctaZ in R. cellulolyticum ΔctaA, a mutant devoid of production of CTA and all congeners, ^[17a] thereby generating R. cellulolyticum ΔctaA ΔctaZ (Figure 2A and Figure S3). Notably, growth of the double mutant was not inhibited under CTA induction conditions (Figure 2B).

Figure 2.

CtaZ contributes to the self‐resistance of R. cellulolyticum to CTA. A) Scheme depicting the CRISPR‐nCas9 strategy used to incorporate an in‐frame nonsense mutation in ctaZ (Ccel_3263) of R. cellulolyticum and R. cellulolyticum ΔctaA to generate the null mutants R. cellulolyticum ΔctaZ and R. cellulolyticum ΔctaA ΔctaZ, respectively. The simultaneous introduction of an EcoRV recognition site enabled successful genome‐editing to be detected by restriction digest of PCR products of the relevant region of ctaZ. B) Photos of cultures under CTA induction conditions demonstrate reduced growth of R. cellulolyticum ΔctaZ compared to R. cellulolyticum wild type and R. cellulolyticum ΔctaA ΔctaZ. Black and white contrast in the background of the cultures is used to visualize differences in optical density. C) Left: Examples of the inhibition zones observed in lawns of the listed R. cellulolyticum strains in an agar diffusion assay using 40 μM CTA. Dashed outlines show the zone borders. Red strikethroughs indicate inactivated genes. Right: Graphical representation of the minimum inhibitory concentrations (MICs) of CTA measured by agar diffusion method against the listed R. cellulolyticum strains. The MICs presented are averages determined from the analysis of three biological replicates (Figure S4), each consisting of two technical replicates, with error bars indicating ±SD (see Table S6 for data summary).

We proceeded to assess the effect of CtaZ deficiency on CTA susceptibility by in vivo trans‐complementation experiments. We transformed R. cellulolyticum ΔctaA ΔctaZ with pMTL‐ctaZ, an expression vector in which ctaZ is under the control of a constitutive promoter, and both R. cellulolyticum ΔctaA and R. cellulolyticum ΔctaA ΔctaZ were transformed with the relevant empty vector (pMTLØ) as controls. The susceptibility of the three stains to exogenous CTA was determined by the agar diffusion method (Figure 2C). ^[23] Briefly, CTA solutions of differing concentrations were applied to paper discs on the surface of agar plates seeded with the test strains. CTA diffusion into the medium suppressed growth in the vicinity of the discs, resulting in measurable zones that increased in size with increasing CTA concentration. Minimum inhibitory concentration (MIC) values were calculated by analyzing the inhibition data using the dissipative model of antibiotic diffusion (Figure S4). ^[24] We found that the double mutant (R. cellulolyticum ΔctaA ΔctaZ pMTLØ) is 35‐fold more sensitive to CTA than R. cellulolyticum ΔctaA pMTLØ (Figure 2C). This CTA‐sensitive phenotype could be partially complemented by the provision of a plasmid‐borne copy of ctaZ, as shown by the intermediate MIC of R. cellulolyticum ΔctaA ΔctaZ pMTL‐ctaZ (Figure 2C). The incomplete restoration of CTA resistance by trans expression of ctaZ can conceivably be explained by i) insufficient ctaZ expression, for example due to inadequate plasmid copy number or a weaker promoter compared to that in the CTA BGC, and/or ii) the separation of ctaZ from native control disrupting the tightly regulated timing of expression typically observed for self‐resistance genes. ^[4] Nevertheless, the sensitivity of the ctaZ‐deficient strain to CTA and its phenotypic complementation supports the proposal that R. cellulolyticum employs the GyrI‐like protein CtaZ in self‐resistance to CTA.

In light of our characterization of CtaZ as a self‐resistance‐conferring GyrI‐like protein, we were curious as to how CtaZ endows R. cellulolyticum with CTA resistance. The other known self‐resistance‐conferring GyrI‐like proteins, YtkR7 and C10R6, inactivate the products of their cognate BGCs by hydrolysis. ^[10c] We reasoned that a sequence comparison with these enzymes could shed light on the manner of self‐resistance imparted by CtaZ. Therefore, we created a sequence alignment and ascertained that CtaZ lacks the active site glutamate essential for hydrolytic activity of YtkR7 and C10R6 (M99, Figure 3A). ^[10c] Additionally, we recognized that another glutamate residue (E123), which is assumed to be crucial for binding of diverse ligands in GyrI‐like proteins,[ ^10a , ²⁵ ] is conserved in CtaZ. This bioinformatic analysis indicates that CtaZ potentially binds CTA, but it likely does not hydrolyze the antibiotic.

Figure 3.

CtaZ does not hydrolyze but putatively binds CTA. A) Sequence alignment of CtaZ and the hydrolases YtkR7 and C10R6. Icons highlight residues important for hydrolysis (Pac‐Man) and small molecule binding (spheres). B) HPLC profiles of CtaZ‐CTA modification assays after 24 h incubation; red strikethrough indicates heat‐inactivated protein. C) Left: HPLC profiles of CtaZ‐CTA binding assays. Binding assays were performed in aqueous solution, synthetic CTA standard was dissolved in methanol. Right: LC‐HRMS profiles of CtaZ‐CTA binding assays. Traces correspond to the extracted ion chromatogram of the [M+H]⁺ ionic species for CTA (695.1453) and are displayed with m/z values ±5 ppm from the calculated exact mass. See Figure S6 for an illustration of the influence of filtration during sample preparation.

We subsequently evaluated whether CtaZ acts as a CTA hydrolase by testing if CtaZ could modify CTA in vitro. To this end, we heterologously produced CtaZ in E. coli and purified the recombinant protein by Ni²⁺ affinity and size exclusion chromatography (Figure S5). After incubation of CTA with CtaZ, high‐performance liquid chromatography (HPLC) analysis of the assays indicated that CTA was neither modified nor degraded under the conditions tested (Figure 3B). This implies that in not hydrolyzing CTA, CtaZ has a different mode of action from that of YtkR7 and C10R6, in accordance with both the phylogenetic analysis (Figure 1C) and the sequence alignment (Figure 3A).

To address the question of whether CtaZ binds CTA, we turned to standard binding experiments such as isothermal titration calorimetry (ITC) and fluorescence quenching. Because the UV absorption spectra of CtaZ and CTA overlap, quenching of the intrinsic fluorescence of CtaZ was not a suitable method for testing CtaZ‐CTA binding. Additionally, due to the markedly poor solubility of CTA in aqueous solutions (2 μM), ^[16] it was not possible to robustly perform ITC experiments.

We instead developed an alternative approach based on liquid chromatography high‐resolution mass spectrometry (LC‐HRMS) to test for binding. Specifically, we prepared samples by incubating CtaZ and CTA individually and in combination in an aqueous buffer. However, we noted precipitation in the sample containing CTA alone, presumably due to saturation of the solution with CTA, while none was observed in that containing both CTA and CtaZ. Upon filtration and subsequent LC‐HRMS analysis, CTA was not detectable in the absence of CtaZ. In contrast, the presence of CtaZ apparently allowed CTA to pass the filter (Figure 3C and Figure S6). We interpret this outcome to mean that CtaZ binds CTA, although the putative CtaZ‐CTA complex is not discernable because it is disrupted in the process of HPLC separation.

Bolstered by the results of the in silico and in vitro experiments, we aimed to interrogate the proposed binding interaction by elucidating the protein structure of CtaZ alone and in the presence of its candidate ligand CTA. Therefore, we purified CtaZ to homogeneity, crystallized it and solved its structure by single‐wavelength anomalous diffraction techniques to 2.0 Å resolution (Protein Data Bank (PDB) ID: 7ZHE; Figure S7A). We found that CtaZ is monomeric and, despite sharing sequence identities of less than 20 % with known members of the GyrI‐like superfamily, it adopts the characteristic GyrI‐like tertiary structure.[ ^10a , ^10b ] We identified the transcriptional regulator EcmrR as the closest homolog in the PDB by a protein structure comparison search (PDB ID: 6WL5, Z‐score: 17.4, sequence identity: 19 %, root‐mean‐square deviation (RMSD): 2.3 Å for 144 residues; Figure S7B). ^[26] CtaZ displays a pseudo‐symmetrical duplicate of the sheet‐helix‐sheet‐sheet (SHS2) motif. This SHS2‐twin is arranged as a cupped hand that forms a groove‐like pocket, which is mostly composed of non‐polar residues and has putatively evolved for binding of hydrophobic small molecules. ^[10b] Thus, it seems that this pocket of CtaZ could provide an ideal site for binding the hydrophobic compound CTA.

Upon crystallizing CtaZ with its potential ligand, CTA, we successfully obtained a structure of CtaZ:CTA at 1.65 Å resolution (PDB ID: 7ZHD, Figure 4A). The complex structure not only confirms that CtaZ binds CTA, but allows us to evaluate key residues in CtaZ involved in CTA binding as well as the conformational changes induced by ligand binding. We found that CTA adopts an elongated conformation to occupy a pronounced groove (40 Å in length and 4 Å in width; Figure 4B) in the hydrophobic pocket of CtaZ, which we recognized in the CtaZ(apo) structure as likely capable of accommodating CTA. A comparative analysis of the structures of CtaZ(apo) and CtaZ:CTA provided a snapshot of the dynamics of CtaZ binding to CTA. Substantial structural rearrangements occur in four defined regions of CtaZ after ligand binding (Figure 4C). Secondary structure elements of regions I (residues 19–32) and II (residues 51–61) are displaced by up to 4 Å to wrap around CTA. Side chains in the neighboring region III (residues 94–100) are shifted by 4 Å to facilitate the formation of a defined binding cavity. The most striking conformational changes induced by CTA‐binding take place in the C‐terminal region IV (residues 128–140) of CtaZ. Compared to CtaZ(apo), side chains of region IV are displaced up to 6 Å in the CtaZ:CTA complex. Additionally, a loop that is disordered in CtaZ(apo) is rigidified in CtaZ:CTA as if to grasp one of the terminal p‐hydroxybenzoic acid (PHBA) moieties of the ligand.

Figure 4.

CTA is bound in a hydrophobic pocket of CtaZ. A) Ribbon diagram of CtaZ with its natural ligand CTA (carbon atoms in green; PDB ID: 7ZHD). B) Surface cross section of the CtaZ:CTA‐complex with colors indicating negative and positive electrostatic potentials. C) Structural superposition of the CtaZ:CTA complex (violet coils) and CtaZ(apo) (black, PDB ID: 7ZHE). Dots indicate a disordered loop region in CtaZ(apo). Conformational changes occur in four defined CtaZ‐regions upon ligand binding: I (residues 19–32), II (residues 51–61), III (residues 94–100) and IV (residues 128–140). D) Close‐up view of the CtaZ ligand binding site with depicted side chains engaged in binding. The 2F_O‐F_C electron density map (grey mesh, contoured to 1.0 σ) and the anomalous sulfur signal (cyan mesh, contoured to 5.0 σ) are displayed for CTA. Residues involved in ligand stabilization by π‐stacking and H‐bonding (grey) and hydrophobic interactions (black) are highlighted. E) X‐ray structure of Cu₂(tBu)₂CTA (Cu^I in brown spheres, CSD ID DIQPAW). ^[16]

Although the linear conformation of the ligand in the binding pocket is energetically unfavorable, CTA is stabilized by extensive protein interactions due to its hydrophobic nature. Surprisingly, despite the symmetric architecture of CTA and its six thioamide repeats, all interactions with monomeric CtaZ are unique. The aliphatic β‐alanine scaffold is buried in a narrow channel that minimizes ligand mobility. Furthermore, the terminal aromatic residues of CTA act as anchors: one PHBA undergoes strong π–π stacking with F35 whereas its counterpart is H‐bonded to E107 and stabilized by Y103, F121 and F146 (Figure 4D). A special feature of the CtaZ:CTA structure is the unique coordination pattern of the polythiocarbonyl scaffold. There is a defined specificity pocket for each thioamide group, which is mainly formed by methionine and aromatic side chains. The nitrogen atoms of CTA form polar interactions with residues of CtaZ including S28 and E123, and the carbonyl oxygen of P129. Notably, the previously mentioned glutamate residue (E123), thought to be essential for small molecule binding (Figure 3A), contributes to the binding of CTA.

In compiling and interpreting this comprehensive structural information on the CtaZ‐CTA interaction, we recognized two remarkable details in the CtaZ:CTA structure: i) The location of CTA in the binding pocket of CtaZ differs from previously reported structures of small molecules bound to GyrI‐like domains. For the MDR transcriptional regulators BmrR ^[27] and EcmrR ^[26a] and the hydrolase Lin2189, ^[10c] ligands are bound to one of three discrete binding sites distributed along the binding pocket (Figure S8), whereas CTA spans nearly the entire length of the binding groove with a contact area of 1034 Ų (Figure 4B). Therefore, the binding mode invoked by CtaZ to capture its ligand differs from that of other GyrI‐like proteins. ii) CTA orientates lengthwise and rotated within the CtaZ binding groove, which is drastically different from the observed helical conformation of a CTA surrogate in complex with Cu^I (Figure 4E). ^[16] From these contrasting conformations, we assume that CTA cannot complex Cu^I while bound to CtaZ. In conclusion, the collected crystal structures not only confirm that CtaZ is a binding protein of CTA, but also reveal a novel binding mode between a self‐resistance‐mediating GyrI‐like protein and its natural ligand.

Inspired by our designation of CtaZ as a self‐resistance‐conferring binding protein, we aimed to gain insight into the distribution of similar proteins in the GyrI‐like superfamily. To obtain an overview of the superfamily, we generated a sequence similarity network (SSN) based on members of the GyrI‐like protein superfamily (PF06445; 39345 sequences) and homologs of CtaZ (Figure 5A and Figure S9). ^[28] We used UniProt ^[29] to construct the SSN because it is a resource for well‐annotated and classified protein sequences and is integrated with the EFI‐EST web tool. Collapsing proteins with ≥50 % sequence identity over 80 % of the sequence length into a single node resulted in a network composed of 4318 nodes. The superfamily clusters into 91 groups containing at least three nodes (covering 2773 nodes), with the remaining 1545 nodes being organized as pairs or singles. The hydrolase proteins cluster together in a large group, whereas the inhibitory (GyrI) and binding proteins (BmrR, Rob), including CtaZ, form separate individual groups. The groups reflect the diversity of cellular functions performed by members of the GyrI‐like protein superfamily, which until now has comprised transcriptional regulation,[ ^12b , ³⁰ ] hydrolysis ^[10c] and gyrase inhibition. ^[21] CtaZ and its homologs form a group of previously undescribed proteins in the superfamily, alongside numerous uncharacterized groups (Figure S9). Overall, the SSN illustrates that the GyrI‐like superfamily is far more complex than previously appreciated. Similar functional complexity is observed in other large and diverse protein families with a conserved structural motif but various cellular activities, such as the AAA+ protein family. ^[31] Therefore, with the identification of the CtaZ group we contribute to a better understanding of the cellular functions performed by members of the GyrI‐like protein superfamily and lay the foundation for deciphering the cryptic functions of the remaining uncharacterized groups.

Figure 5.

Homologs of CtaZ form a group in the GyrI‐like protein superfamily and homologous genes are part of a putative resistance gene cassette. A) Sequence similarity network for CtaZ homologs and the GyrI‐like protein superfamily (PF06445; proteins that share ≥50 % sequence identity over 80 % of the sequence length are conflated; alignment score 40); only clusters containing members with characterized cellular functions are depicted (for full network see Figure S9). Nodes are colored based on a known resistance or self‐resistance (encoded in biosynthetic gene clusters (BGCs)) function. Node outline color indicates the protein domain architecture, while the node shape corresponds to protein activity. Node labels correspond to a representative protein and are colored based on the phylum of the organism harboring the GyrI‐like gene. B) Representative genetic loci that encode a CtaZ homolog (selected based on similarity to the CTA BGC or locus diversity). The CTA BGC (top) is shown for reference. Putative BGCs (middle) are separated by a dashed line from genetic loci not resembling BGCs (bottom). The putative resistance gene cassette is surrounded by a dashed box in each case.

In addition to assessing the distribution of CtaZ homologs within the GyrI‐like superfamily, we were interested in whether CtaZ homologs are used for self‐resistance by other potential antibiotic producers. Therefore, we mined the genomic neighborhoods of ctaZ homologs for biosynthetic genes. To this end, BLASTp was used to identify 200 homologs of CtaZ in GenBank ^[18] sharing minimum 44 % sequence identity. We chose the GenBank database instead of UniProt because it contains a larger number of sequences and therefore provides a more accurate picture of the available data. Notably, we found that approximately one third of the CtaZ homologs are encoded in pathogenic members of the phylum Firmicutes. We next surveyed the genomic neighborhood surrounding the ctaZ homolog for genes encoding enzymes involved in secondary metabolite biosynthesis. The local genomic neighborhoods are highly diverse, with 2 % of the ctaZ homologs found in loci similar to the CTA BGC (Figure 5B and Table S8). Surprisingly, most CtaZ homologs are encoded in loci that do not resemble typical BGCs. Therefore, it seems that CtaZ homologs are not commonly used in self‐resistance. Interestingly, we noticed that in 96 % of the cases the ctaZ homolog is co‐localized with an upstream regulator and two downstream transporter genes, resembling a gene cassette (Figure 5B and Table S8). Export processes of cytotoxic molecules are commonly induced by transcriptional regulators and facilitated by transporters. ^[32] Since genes encoding these types of proteins co‐occur with ctaZ homologs in the vast majority of cases, we inferred that they may represent a functional unit responsible for protection against toxic molecules. Taken together, our findings on CtaZ expand the current picture of the cellular functions mediated by members of the GyrI‐like superfamily to include antibiotic binding. In addition, our results on a putative resistance gene cassette may indicate that these genes are involved in (self‐)resistance to thioamide‐possessing antibiotics, including in pathogenic bacteria.

Conclusion

In summary, we have characterized CtaZ as a self‐resistance factor from R. cellulolyticum to the polythioamide antibiotic CTA. We demonstrated that CtaZ is a binding protein of CTA and establish it as the founding member of a new group in the GyrI‐like protein superfamily. We observed a different conformation of CTA in complex with CtaZ compared to a CTA surrogate bound to Cu^I (Figure 4A and Figure 4E), intimating that CTA cannot complex Cu^I while bound to CtaZ. The complex structure suggests that CtaZ‐bound CTA, in addition to being shielded from the intracellular copper pool, is prevented from interfering with the DNA gyrase of R. cellulolyticum. This work sets the basis for deciphering the molecular details of the CtaZ‐mediated self‐resistance mechanism and its regulation in the CTA producer R. cellulolyticum.

In light of our findings regarding the function of CtaZ, our current understanding of the role of GyrI‐like proteins in AMR can be updated: i) The GyrI‐like fold is crucial for binding of small molecules not only as part of multidomain transcriptional regulators (Rob and BmrR),[ ^12c , ^22b ] which have an additional DNA‐binding domain, but also in single domain proteins. ii) GyrI‐like single domain proteins provide resistance not only by hydrolysis (YtkR7, C10R6) ^[10c] or gyrase inhibition (GyrI), ^[21] but also by small molecule binding (Figure 6). The GyrI‐like fold is structurally distinct from that of other important multidrug‐binding domains within the AlbA and QacR classes.[ ^32b , ³³ ] Therefore, the detailed structural information on the CtaZ‐CTA interaction contributes to a better understanding of the molecular factors that determine drug recognition by binding proteins.

Figure 6.

Overview of resistance strategies mediated by members of the GyrI‐like protein superfamily. Proteins are colored according to Figure 1C. A) Inhibition of DNA gyrase by GyrI prevents the drug‐induced formation of lethal DNA double‐strand breaks. B) Members of the cyclopropanoid cyclopropyl hydrolase (CCH) group of GyrI‐like proteins, such as YtkR7 and C10R6, inactivate drugs by hydrolysis. C) Drug binding to the GyrI‐like domain of multidomain transcriptional regulators, such as BmrR and Rob, induces the transcription of genes encoding multidrug efflux pumps. D) The CTA binding protein CtaZ confers self‐resistance to the antibiotic producer R. cellulolyticum.

The requirement for self‐resistance in antibiotic producers provides a fertile ground for the evolution of AMR in non‐producers due to the selective pressure exerted by microbial competition. Mobilization of these resistance genes to pathogenic bacteria is prompted by widespread antibiotic use.[ ^9a , ^9c ] Our bioinformatic survey uncovered numerous CtaZ homologs, thereby illuminating uncharted territory of the antibiotic resistome, which encompasses all antibiotic resistance genes in microorganisms. ^[9c] It is essential that such antibiotic (self‐)resistance genes of potential clinical significance are subject to biochemical characterization and bioinformatic surveying, in addition to monitoring emerging AMR.

This knowledge is critical to develop resistance management plans for both established antibiotics and new lead compounds. Given the undeniable connection between environmental and clinical resistance, we should perhaps be mindful that the newly identified CtaZ homologs and associated putative resistance gene cassette could potentially facilitate the transfer of AMR. In conclusion, our results not only address the mystery of self‐resistance to CTA, but also serve as a starting point for the characterization of previously undetected resistance proteins.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

F. Gude, E. M. Molloy, T. Horch, M. Dell, K. L. Dunbar, J. Krabbe, M. Groll, C. Hertweck, Angew. Chem. Int. Ed. 2022, 61, e202206168; Angew. Chem. 2022, 134, e202206168.

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.