The prevalence and implications of single nucleotide polymorphisms in genes encoding the RNA polymerase of clinical isolates of Staphylococcus aureus

Abstract Central to the regulation of bacterial gene expression is the multisubunit enzyme RNA polymerase (RNAP), which is responsible for catalyzing transcription. As all adaptive processes are underpinned by changes in gene expression, the RNAP can be considered the major mediator of any adaptive response in the bacterial cell. In bacterial pathogens, theoretically, single nucleotide polymorphisms (SNPs) in genes that encode subunits of the RNAP and associated factors could mediate adaptation and confer a selective advantage to cope with biotic and abiotic stresses. We investigated this possibility by undertaking a systematic survey of SNPs in genes encoding the RNAP and associated factors in a collection of 1,429 methicillin‐resistant Staphylococcus aureus (MRSA) clinical isolates. We present evidence for the existence of several, hitherto unreported, nonsynonymous SNPs in genes encoding the RNAP and associated factors of MRSA ST22 clinical isolates and propose that the acquisition of amino acid substitutions in the RNAP could represent an adaptive strategy that contributes to the pathogenic success of MRSA.

encoding the β′ subunit of the RNAP, after the adaptation of E. coli strain MG1655 to growth in minimal media. More recently, studies by Rajaraman et al. (2016) and Liu et al. (2017) identified SNPs in rpoA, encoding the RNAP α subunit, and rpoS, encoding the stress response σ factor σ S , in E. coli, after adaption to growth on acetate as the sole carbon source, and acclimatization to the laboratory environment, respectively. Overall, these laboratory-based observations suggest that the reprogramming of bacterial transcription by specific adaptive aa substitutions, that change the kinetic and regulatory performance of the RNAP, allows for optimal growth in new and challenging environments.
As the RNAP β subunit contains the binding site for the antibiotic rifampicin, adaptive aa substitutions in this subunit have been the focus of extensive study. Adaptive aa substitutions within this region, termed the rifampicin resistance determining region (RRDR), confer resistance to rifampicin in a variety of bacterial species (reviewed in Goldstein (2014)). Adaptive aa substitutions within the RRDR, that lies close to the RNAP active site, have long been recognized to elicit pleiotropic effects on gene expression (Jin & Gross, 1989).
With regard to the highly versatile pathogen Staphylococcus aureus, Gao et al. (2013) and Baek et al. (2015) reported the identification of adaptive aa substitutions within the region encoding the RRDR that, along with conferring resistance to rifampicin, decreased susceptibility of S. aureus to host immune effectors and several antibiotic agents, respectively. Further, Villanueva et al. (2016) identified two SNPs within the region encoding the RRDR that enabled S. aureus to survive in the absence of an essential oxidative stress-response protein Spx. Interestingly, both of these reported adaptive aa substitutions in the RRDR region have also been identified in Salmonella enterica and were associated with enhanced bacterial survival in aging cultures (Wrande, Roth, & Hughes, 2008).
Collectively, these reports suggest that nonsynonymous SNPs that give rise to adaptive aa substitutions often confer new properties on the bacterial transcriptional machinery, consequently eliciting pleiotropic changes in gene expression. The subsequent phenotypic alterations can contribute to the pathogenic success of bacteria, which could have further implications on both prevalence and treatment options in the clinic. However, SNPs in genes encoding the RNAP in clinical isolates of pathogenic bacteria have never been systematically analyzed. Hence, we undertook an unbiased analysis of the sequences of α, β, β′, ω, δ, and ε and σ A , σ B , σ H , and σ S in a large collection of whole-genome sequences of S. aureus isolates recovered from patients with bacteremia (Donker et al., 2017;Harris et al., 2013;Holden et al., 2013;Hsu et al., 2015;Koser et al., 2012;Reuter et al., 2016) to identify SNPs that potentially give rise to adaptive aa substitutions. The collection consisted of 1,429 MRSA isolates of sequence type 22 (ST22), a globally successful MRSA lineage, that represents both global and UK isolates. SNPs that resulted in aa substitutions were identified as described by Hsu et al. (2015) and then mapped onto a structural model of the S. aureus RNAP (using the E. coli RNAP; PDB id: 4XSY; (Bae et al., 2015) as the template) to gauge the potential implications of the aa substitutions on structure-function relationships in the RNAP. clinical isolates (~10% of the collection) harbored at least one SNP in genes encoding the core RNAP subunits or the σ factors that were absent in the reference strain HO 5096 0412. As shown in Figure 1, 55 unique SNPs were identified across all core genes, except that encoding the ω subunit of the core RNAP. A further 25 unique SNPs were identified in genes encoding the four σ subunits. Some isolates contained SNPs in genes for both the core RNAP and σ subunits (Table 1) giving rise to RNAP variants with different combinations of aa substitutions. Thus, these variants could conceivably impact RNAP performance and the transcription program in a promoter and environment-dependent manner. One hundred and twelve isolates (~8% of the collection) contained at least one SNP in genes encoding the major housekeeping form of the RNAP (αββ′ωδε + σ A ). We mapped all the resulting aa substitutions on to a homology model of the structure of the S. aureus σ A -RNAP (except for SNPs in the δ and ε subunits that were excluded from the structural model, which is based on the equivalent Gram-negative RNAP of E. coli containing the σ 70 factor where the δ and ε subunits are absent) ( Figure 2). We did not identify any SNPs in the gene encoding the ω subunit, which perhaps underscores the important role the ω subunit has in the overall structural stability of the RNAP complex ( Figure 1).
Although we expected that SNPs and resulting aa substitutions would be clustered in certain regions of the RNAP, which may point to selective pressure for adaptive mutations, strikingly, no discernible "hot-spots" for SNPs and resulting aa substitutions were noted.
Only aa substitutions within the β subunit RRDR (S. aureus aa residues 463-550) and a single SNP, which resulted in the β′-M455I sub- in the α carboxyl-terminal domain (α-CTD). The α-NTD is responsible for α subunit dimerization and forms a platform for subsequent binding of β and β′ subunits for the assembly of the core enzyme (Kimura & Ishihama, 1996;Murakami, 2015). The α-CTD binds to specific regions upstream of the start-site distal, −35 consensus promoter element at some promoters and can serve as a contact site for transcription regulatory factors, which stabilize the initial RNAPpromoter complex (Browning & Busby, 2016;Murayama, Ishikawa, Chumsakul, Ogasawara, & Oshima, 2015). Although the α subunit of E. coli has been subjected to intensive mutagenesis analysis, none of the aa residues affected by the SNPs in our collection has been previously described. Thus, future experimental testing will reveal whether these SNPs result in aa substitutions that affect the assembly and stability of the RNAP, DNA binding, initial promoter complex formation, and interaction with transcription regulatory factors.
β subunit: The highest occurrence of SNPs (32) was found in the gene encoding the β subunit. Unsurprisingly, eight of the SNPs resulted in aa substitutions that fell within the RRDR and many led to known substitutions that confer resistance to rifampicin (Figure 1).
The RRDR β-H481Y substitution reported by Gao et al. (2013) occurred at the highest frequency and was present in eleven isolates ( Figure 1). This aa substitution, in addition to conferring resistance to rifampicin, was reported to promote S. aureus immune evasion and increase the likelihood of persistent infections (Gao et al., 2013).
We also noted the occurrence of the β-A477D substitution, which was reported to decrease the susceptibility of S. aureus not only to rifampicin but also to vancomycin, daptomycin, and oxacillin (Baek et al., 2015). Substitution of the RRDR β-A477 residue to valine was identified in three additional S. aureus isolates (Figure 1).
Interestingly, this aa substitution only occurred in the presence of another β subunit substitution β-T622A (Table 1) identified. Since small deletions or single aa substitutions within or surrounding the equivalent region in the E. coli β subunit (Ecoβi4) results in unstable transcription-competent promoter complex formation (Nechaev, Chlenov, & Severinov, 2000), it is possible that the SNPs and resulting aa substitutions within Sauβi5 could similarly influence the S. aureus RNAP.
The β flap is an independent structural domain that defines the RNA exit channel. It is within this channel that nascent RNA forms hairpin structures that promote transcription termination (Figure 3).
Three SNPs resulting in aa substitutions (β-N822S, β-H880R and β-V886I) were identified at the base of the β flap domain, which may therefore influence the transcription termination ability of the RNAP at certain promoters. Interestingly, one SNP in the β flap domain, which resulted in a conserved aa substitution β-H880R, was only present in combination with two additional aa substitutions in the RRDR (Table 1). Based on the proximity of the β-H880R substitution to the active site, we suspect that this aa substitution may compensate for the potentially restrictive conformational changes in the RNA exit channel introduced by the β-A477D and β-R484H substitutions ( Figure 3).
Further, two SNPs that resulted in β-G1116D and β-D1160N TA B L E 1 List of the ST22 MRSA clinical isolates carrying more than one SNP within genes encoding the RNAP subunits and/or associated σ factors subunit ( Figure 3). This region makes extensive interactions with the β′ subunit and, in combination with the β′ clamp, forms the "pincers"  (Toulokhonov & Landick, 2006). Apart from SNPs localized to the region encoding the RRDR, only a single SNP resulting in the β′-M455I substitution was found proximal to the RNAP active site. As this region of the RNAP interfaces with the ω subunit, which aids in the assembly of the core RNAP (Murakami, 2015), it is possible that the β′ M455I substitution may indirectly affect RNAP complex stability by affecting the β′-ω interface.
Ribonucleotide substrates gain access to the RNAP active site via a secondary channel in the β′ subunit ( Figure 3); the set of helices lining the rim of this channel are referred to as the secondary channel rim helices. Only a single SNP, resulting in the β′-V706I F I G U R E 2 Location of aa substitutions resulting from SNPs identified in a collection of 1,429 ST22 MRSA clinical isolates mapped onto a homology model of the Staphylococcus aureus RNAPσ A holoenzyme viewed from the (a) downstream and (b) upstream faces with respect to the incoming DNA (not shown) and (c) directly at the active site. Amino acid substitutions are shown in red in space-filling mode; those identified within the active site channel are indicated in the accompanying zoomed-in panel. The σ A subunit is colored in orange and the core RNAP subunits in gray, and the conserved NADFDGD active site residues are indicated in green in space-filling mode substitution, was identified in the rim helices, which may alter access of RNA substrates and also the interaction of transcription regulatory factors such as DksA and Gre-factors, that are known to bind to this region (Hochschild, 2007;Laptenko, Lee, Lomakin, & Borukhov, 2003).
Interestingly, one of the most frequently identified SNPs in the gene encoding the β′ subunit was one which resulted in the β′-G602D substitution (Figure 1), present in 11 isolates. While this SNP has been reported previously in a rifampicin-resistant S. aureus isolate in combination with the β-D471Y substitution in the RRDR (O′Neill, Huovinen, Fishwick, & Chopra, 2006)-in the collection of bacteremia isolates studied here, the SNP resulting in the β′-G602D substitution was never associated with any mutations in the RRDR.
This suggests that the β′-G602D substitution may represent a "precursor" for the acquisition of a rifampicin resistance conferring substitution in the RRDR or may confer additional properties to the RNAP that warrant further investigation.  (Nagai & Shimamoto, 1997;Paget, 2015). We identified a single SNP within region 2.1, which resulted in the σ A -A138T substitution that we predict could impair the σ A -RNAP interface (Figure 4), with potential ramifications on the intracellular competition for core RNAP by σ factors in response to certain cues. Regions 2.4 and 4.2 are responsible for recognition of the consensus −10 and −35 promoter elements, respectively, and region 4.2 also serves as a contact site for transcription regulatory factors (Paget, 2015;Zuo & Steitz, 2015). While no aa substitutions were identified within region 2.4, one was identified within region 4.2, resulting in the σ A -A350V substitution. Since this aa position is highly conserved among primary bacterial σ factors (Lonetto, Gribskov, & Gross, 1992), it is possible that this  (Figure 4). In addition to the −10 and −35 consensus promoter elements, some bacterial promoters contain a so-called extended −10 element, that is recognized by region 3.0 of primary σ factors (Zuo & Steitz, 2015).

CO N FLI C T O F I NTE R E S T
None declared. F I G U R E 4 A close-up view of the σ A -RNAP interface from the homology model of the Staphylococcus aureus RNAP-σ A holoenzyme relating to the aa substitutions discussed in the accompanying text. The structural model is color coded as in Figure 2