Sigma factor M induces an ESX-4-focused regulon in Mycobacterium abscessus

Introduction

Type VII secretion systems, commonly known as ESX secretion systems, are complex assemblies of conserved proteins that span the cell membrane, translocating proteins out of the cytoplasm. ESX secretion systems are encoded by conserved genes clustered in loci that encode structurally similar core apparatus (1–5). Each esx locus also expresses a pair of small WXG100-proteins that are co-secreted by the core machinery (6, 7). In Mycobacterium tuberculosis, the WXG100-pair, EsxA and EsxB, are antigenic heterodimeric proteins secreted by ESX-1 (8–12). Slow-growing mycobacteria, such as M. tuberculosis and M. marinum, have five distinct ESX systems that have non-redundant roles in different processes (13, 14). ESX-1, ESX-4 and ESX-5 each contribute to pathogen-host interactions (8, 15) and ESX-3 functions in iron and zinc acquisition (16–18). The rapid-growing model organism, M. smegmatis, encodes three ESX systems (14). ESX-1 and ESX-4 are required for a unique conjugative process between donor and recipient strains, and ESX-3 acquires environmental iron and zinc (16, 19).

M. abscessus is a rapid-growing mycobacterial species and, similar to M. smegmatis, has evolved in environmental niches (20). However, individuals with certain chronic lung diseases are susceptible to infection by M. abscessus from aerosolized soil or water (21–24). In contrast to the obligate pathogen M. tuberculosis, M. abscessus is capable of both saprophytic and pathogenic existence. Further, M. abscessus encodes only two ESX secretion systems, ESX-3 and −4 (14, 25), raising the question of how fewer ESX secretion systems support the growth of M. abscessus in both the environment and as an intracellular pathogen. ESX-4 is regarded as the progenitor ESX system in mycobacteria that gave rise to paralogous secretion systems through successive duplication events followed by evolved specialization (10). It is possible that M. abscessus has retained an ancestral ESX composition with a subset of its secretion substrates conserved in other mycobacteria but secreted by descendant ESX systems.The importance of ESX-3 and ESX-4 in M. abscessus infection has been experimentally demonstrated in diverse models (25, 26). In a mouse model of M. abscessus infection, strains with mutations in esx3 and esx4 core genes were completely attenuated. Characterization of infected macrophages indicated that ESX-3 and ESX-4 blocked phagosome acidification and instead promoted phagosome lysis, allowing bacilli access to the cytosol (25). In support of this, esx4 locus genes were identified in a screen of M. abscessus transposon insertion mutants for survival in phagosomes of macrophages or amoebae (27). While the ESX-4 secretion system clearly plays an important role in M. abscessus pathogenesis, the specific phagosomal survival activities provided by ESX-4 secretion have not been identified.

In other mycobacteria, an alternative sigma factor, SigM, co-induces ESX-4 secretion systems and putative secreted substrates. Microarray gene expression surveys of M. tuberculosis SigM identified the induction of esxU and esxT, which encode the WXG100 primary secretion substrate proteins for ESX-4, as well as other uncharacterized genes (28, 29). Clear functional roles for SigM and ESX-4 in M. tuberculosis remain elusive. In M. smegmatis, both SigM and ESX-4 were found to be essential in the recipient strain for conjugation (19). The SigM regulon in M. smegmatis includes six of seven esx4 core locus genes, indicating that SigM regulates the synthesis of the ESX-4 apparatus in this species. In addition, 15 genes at nine loci in the genome were also induced by SigM and many encode proteins with structures and motifs consistent with ESX-secretion (30). In support of this, some of these non-esx4 core genes were required for conjugation (30–33).

We hypothesized that in M. abscessus, SigM coordinates expression of genes that encode key components of the ESX-4 secretion system and those mediating phagosome disruption. To identify candidate genes that support survival in macrophages, we generated a sigM-rsmA deletion mutant in M. abscessus strain ATCC 19977 and a complement expressing sigM from a constitutive plasmid promoter. We identified differentially expressed loci and defined mRNAs regulated by SigM, each transcribed from a promoter with a consensus SigM-binding site, indicating direct regulation of the responsive genes. We further determined that proteins encoded in each locus have structural characteristics consistent with function as secretion substrates, annotated to have enzymatic activities that include hydrolase, kinase and toxin activities. Our data define a SigM regulon in M. abscessus that induces the primary ESX-4 secretion substrates EsxU and EsxT and co-induces putative secretion substrates that likely contribute to survival in infection.

Methods

Strains and plasmids

Mycobacterium abscessus ATCC 19977 strain (Mab) and its derivatives were grown in 7H9 medium supplemented with 10% OADC (Oleic acid, Albumin, Dextrose, Catalase) with 0.2% glycerol and 0.05% Tween-80 at 37 °C. The sigM-rs locus was deleted by a two-step process utilizing a suicide plasmid encoding resistance to zeocin, which contained segments flanking sigM from the M. abscessus genome (5050979-5051458 and 5053037-5053524) to promote recombination into the native locus. Recombinants were selected on 7H10 agar plates with 50 µg/mL zeocin and verified for the presence of the suicide vector by PCR amplification. Cultures of zeocin-resistant colonies were expanded through four serial passages in 7H9 broth without antibiotic selection to allow excisive recombination to occur. The plasmid also encoded sacB and passaged culture was spread on 7H10 agar plates supplemented with 5% sucrose. Zeocin-sensitive, sucrose-resistant colonies were identified by replica patching and were tested by PCR across the locus to verify the scarless deletion of 1580 bp spanning the sigM-rsmA genes. For complementation studies, sigM was expressed from the UV15/P_myc1-tetO promoter on an L5 attB integrating plasmid encoding apramycin resistance (50 µg/mL) (34, 35). Colony morphology and size of the sigM-rsmA deletion or complementation strains were identical to the smooth ATCC 19977 WT M. abscessus.

RNA-Seq

Replicate mid-log (O.D.₆₀₀ ∼0.7) cultures of WT, ΔsigM-rsmA, and complemented M. abscessus strains were grown at 37 °C with aeration and collected by centrifugation. Total RNA was extracted by bead-beating in TRIzol (Invitrogen). Residual genomic DNA was degraded by DNase digestion and tested for complete removal by PCR amplification. 500 ng of DNase-treated total RNA was depleted of rRNA using a mycospike probe mix with the NEBNext rRNA Depletion Kit for bacteria (NEB). Ribosomal RNA-depleted RNAs were processed for Illumina sequencing using the NEBNext Ultra II Kit (NEB). 10 million reads were obtained for each replicate and were trimmed (minimum read length of 40 nts and minimum Phred quality value of 30) by TrimGalore v.0.6.7 and mapped to the M. abscessus ATCC 19977 reference genome (CU458896.1). Gene expression was quantified by Salmon v.1.10.2 using the coding sequences for differential expression analyses by DESeq2. FastQ files used in this study are deposited in a NCBI Sequence Read Archive, BioProject accession number PRJNA1398377.

RNA-read depth plots were generated as previously described (33), with locus-specific adjustments made to accommodate varied locus lengths.

SigM binding site identification

A FASTA file of 200 bp from the 5’ end of each induced operon was analyzed by MEME (https://m-suite.org/meme/tools/meme) (36). The default settings for the MEME search tool returns the top three enriched motifs. Only one sequence motif with a high probability score (4.1e⁻⁰⁷⁹) was returned and a single instance was identified in each promoter. The consensus was visualized using a DNA logo tool, aligned at the −35 GGAACCG motif, with 18 bp upstream and 33 bp downstream (https://weblogo.threeplusone.com/create.cgi) (37).

Protein modeling

Predicted structures were provided by Uniprot or amino-acid sequence queries of Alphafold and Alphafold3 (38, 39). The models were visu, annotated and oriented using Chimera to produce images (40). A Chimera plug-in, ChopChop provided structural homology associations to supplement Uniprot homology annotations in Table 1 (41).

Table 1. The sigM regulon. SigM responsive genes (column 2) are organized by operon in genome order (column 1) and listing the Log₂ fold change in expression upon SigM complementation of the deletion strain (column 3). The encoded protein for each gene is described by its Uniprot identifier (column 4), annotations derived from sequence or structural homologies (column 5) and the presence of WXG or YXXXD/E sequences (columns 6 and 7). Gene IDs were derived from the standard ATCC 19977 reference genome (CU458896.1) except MAB_RS09705, which was derived from reference genome (NC_010397.1). A complete list of direct SigM target operons and genes.

Results

We generated a precise deletion of the sigM-rsmA locus (ΔsigM) of M. abscessus strain ATCC 19977 (Figure 1A). An integrating plasmid that expresses sigM from the P_myc1 promoter was introduced into this deletion strain for complementation (+SigM). We cultured the wild-type (WT), the deletion strain and the deletion strain with the complementing plasmid to mid-log growth and processed the harvested RNA for RNA-seq analyses. Steady-state mRNA levels over the sigM-rsmA locus are shown in a consolidated display for WT (gray shading), ΔsigM (red tracing) and + SigM (blue tracing) strains (Figure 1B). We used DESeq2 to identify genes that were differentially expressed in deletion and complemented strains (Figure 1C and Table 1). Transcriptional profiles over loci identified by DESeq2 delineated responsive genes at each regulon locus. The promoter sequences of all SigM-responsive loci were analyzed for enriched sequence motifs, defining a consensus −35 and −10 SigM-binding site that matched the site reported for M. smegmatis (30) Ten promoters precisely matched the consensus (Figure 1D) and the other five had near-matches. We conclude that these 15 loci, totaling 50 genes, are direct transcriptional targets of SigM (Table 1).

In the esx4 core gene locus, the esxU-esxT operon (esxUT) promoter includes a SigM binding site. The esxUT mRNA levels decrease in the ΔsigM samples relative to WT (Figure 1E, red tracing in lower left). The + SigM samples, conversely, expressed higher esxUT mRNA levels than WT (Figure 1E, blue tracing in lower left). Transcription through other genes in locus was not reduced in the ΔsigM samples but was slightly elevated in the + SigM samples (Figure 1E). This suggests that indirect effects of SigM expression may marginally enhance basal transcription of these esx4 mRNAs.

The transcriptional response at each of the remaining 14 loci is shown in consolidated profile panels in Figure 2. In each instance, complementation by SigM increased transcript levels relative to the WT (the blue tracing exceeding the gray shading) samples through the operon, beginning immediately adjacent to the identified SigM-binding site (dashed vertical line in each panel). These profiles individually and collectively demonstrate that SigM-binding to its cognate −35/-10 binding site in promoters induces transcription of the operon. As in the esx4 locus, deletion of sigM only slightly decreased expression of some operon mRNAs below that of WT (e.g., Figures 2D, F – indicating partial SigM dependence), while other loci were unaffected even though they are transcribed from sigM promoters (e.g., Figures 2B, G). A notable feature in the GC-rich mycobacterial genome is the observation of an A-rich region 5’ of the −35 site in the SigM binding site sequence logo (Figure 1D), which we speculate may play a role in regulon gene expression.

ESX secretion substrates described in other mycobacteria feature a four-helix bundle structure and a bipartite secretion signal sequence comprising tryptophan-X-glycine (WXG) and tyrosine-XXX-aspartic/glutamic acid (YXXXD/E) motifs (53). Substrates are divided into three primary classes (42–44). Esx proteins belong to the WXG100 class and are expressed as tandem genes in individual esx loci and elsewhere in the genome. A predicted (Alphafold 3) image of EsxU and EsxT, of the WXG100 heterodimer family (6, 7), illustrates the hallmarks of the four-helix bundle with juxtaposed WXG and YXXXD/E sequences (Figure 3A). Other predicted Esx proteins are modeled in Figure 3 (I, K, L and O). Mab_3112 and Mab_3113 (Figure 3L) are homologs of M. tuberculosis EsxF and EsxE, respectively.

The PE/PPE (so named for their proline and glutamic acid-rich N-termini) class of proteins include a ∼100 amino acid PE protein that dimerizes with a larger PPE partner to produce a four-helix bundle and secretion signal. PE/PPE proteins are a heterogenous family that can be further subdivided by size and function and have been described in ternary complexes with cognate EspG chaperones (45, 46). Most of the M. abscessus SigM regulon proteins are predicted to have PE/PPE-like structures. Mab_0010c is annotated as a PPE family protein and is shown paired with Mab_0012c, a predicted PE protein (Figure 3C). The modeled Mab_0012c and Mab_0010c structure is the first of several PE/PPE heterodimer predictions that have a large unstructured domain. Other predicted PE/PPE-class secretion substrates encoded in the SigM regulon are shown in Figure 3 (D, E, F, G, H, J, M and N). The Mab_0149c and Mab_0148c PE/PPE proteins are co-expressed with a predicted EspG-like chaperone encoded by Mab_0147c suggesting association as a ternary complex (45, 46).

A monomeric class of ESX secretion substrate encodes the four-helix bundle and secretion signal in a single protein. EspB represents this class and has been reported to interact with EspK, each of which are encoded in esx1 loci in other mycobacterial species (47, 48). Mab_0007 expresses an EspB-like protein (Table 1) (Figure 3B).

Discussion

We identified potential ESX-secretion substrates of M. abscessus by defining the SigM regulon as 15 loci encoding 50 genes, producing proteins or protein complexes predicted to function as ESX secretion substrates. Thus, as has been shown for M. smegmatis and M. tuberculosis, SigM also coordinates the induction of an ESX response in M. abscessus.

Our data also support annotation updates for two SigM-regulated genes. In one locus, the ATCC 19977 reference genome (CU458896.1) predicts that a positive-strand gene, Mab_1282, separates the negative strand genes, Mab_1283c and Mab_1281c. An alternative reference genome, (NC_010397.1), correctly identifies a gene it calls MAB_RS06640, that extends Mab_1281c to include 108 N-terminal amino acids. Our mRNA profile over this region shows continuous transcription consistent with MAB_RS06640 (Figure 2E, red outline arrow reflects the N-terminal extension). Thus, we conclude that the predicted Mab_1282 is likely incorrect. The reference genome, NC_010397.1, also correctly identifies a 534-bp gene that is co-transcribed with Mab_1896c – 1894c. This gene encodes a 178 amino acid protein, MAB_RS09705, annotated as a DUF3558 protein (Figure 2G and Table 1).

Each SigM-regulon locus encodes proteins consistent with ESX secretion. While it is premature to assign functions, there are recurrent patterns that suggest two structural themes. The first is the six instances of DUF3558 domain-containing proteins (Table 1). There are 11 loci encoding proteins with DUF3558 annotations in M. abscessus (NC_010397.1), showing that this domain is highly enriched in the SigM regulon. The high frequency of DUF3558 proteins that are co-expressed with ESX secretion substrates strongly suggests a fundamental functional link. Determining whether DUF3558 functions as a secretion chaperone, or as an effector protein will require directed study. DUF3558 domains range from 177 to 195 amino acids and are found throughout Actinomycetes, including Nocardia spp. and Gordonia spp.

A second emerging theme is that some genes encode catabolic activities that we speculate could facilitate envelope remodeling of the host bacterium or, conversely, damage neighboring cells (Table 1, e.g., those encoded in operons 4, 8, 9, 11, and 12). For example, operons 8 and 12 encode predicted NlpC/P60 peptidases that may function in cell wall hydrolysis and remodeling (49). A SigM-regulon NlpC/P60 protein in M. smegmatis is induced in the recipient upon donor contact and is required for conjugation (30). Another example is operon 11, which has a homologous operon in M. tuberculosis that co-transcribes the esxF-esxE-cpnT-ift genes encoding an ESX-secreted necrosis-inducing toxin that is required for survival in macrophages. We suggest that in M. abscessus, Mab_3114 toxin secretion is co-dependent on the Esx proteins encoded by Mab_3112 and Mab_3113. Operon 4 also encodes a predicted hydrolase. α/β hydrolase folds are found in glycopeptide lysin proteins expressed by mycobacteriophages (50) and have been previously associated with PE/PPE proteins (51), consistent with roles in mycobacterial cell wall hydrolysis. Twelve of the 15 operons are present in 90 or more of a 96 M. abscessus cohort that served to provide a pangenome survey of the species, indicating that most are conserved and that our findings from ATCC19977 have broad relevance (52). Despite their conservation in distant mycobacteria, operons 8 (NlpC/P60) and 11 (CpnT), are present in only 39 and 72 isolates, respectively. Specific assays will be needed to determine the biochemical activities and biological roles of these proteins in the environment and in infection.

Several observations from this study support the evolutionary model of ESX-4 as a progenitor secretion system for specialized, spinoff ESX systems formed by subsequent esx locus duplication and divergence. Basal monoculture expression of esx4 genes in M. smegmatis is nearly undetectable until donor contact induces the recipient SigM regulon, which includes multiple mRNAs within the esx4 core locus (30). By contrast, in M. abscessus, SigM directly induces only esxUT above the marked basal transcription of genes in the esx4 locus. While each regulon operon we identified is directly induced by SigM, none are entirely dependent on SigM, leading us to speculate that a partially redundant sigma factor also activates SigM promoters. Collectively, these observations indicate that M. smegmatis requires ESX-4 function specifically in conditions that activate SigM, whereas M. abscessus has additional roles for ESX-4 secretion in situations separate from those that activate SigM. Further, we note that most M. abscessus SigM operons encode PE/PPE class substrates, which are predominantly secreted by ESX-5 secretion systems (2, 45, 46). Other loci encode EspB and EspG homologs, which are associated with ESX-1 and ESX-5 in other mycobacteria (45–48). This is consistent with M. abscessus representing an ancestral ESX-4 system that performs a broad range of functions that have subsequently evolved to be accomplished by derivative paralogous systems in other mycobacteria.

The conservation of this extended regulon in M. abscessus suggests that it is key for bacterial survival in its natural environment and during opportunistic infection. A basal level of regulon transcription provides steady expression suitable for rapid stimulus-responses, while specific contacts and/or environmental conditions activate SigM in M. abscessus to boost expression of ESX-4 secreted proteins. The identification of these SigM-regulon proteins is a significant step toward understanding these interactions and, potentially, to therapeutically disrupting them.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Laurent Kremer, Institut de Recherche en Infectiologie de Montpellier, France

Reviewed by: Shoshana Zha, University of California, San Francisco, United States

Daniel Barkan, Hebrew University of Jerusalem, Israel