Impact of Bicarbonate-β-Lactam Exposures on Methicillin-Resistant Staphylococcus aureus (MRSA) Gene Expression in Bicarbonate-β-Lactam-Responsive vs. Non-Responsive Strains

Methicillin-resistant Staphylococcus aureus (MRSA) infections represent a difficult clinical treatment issue. Recently, a novel phenotype was discovered amongst selected MRSA which exhibited enhanced β-lactam susceptibility in vitro in the presence of NaHCO3 (termed ‘NaHCO3-responsiveness’). This increased β-lactam susceptibility phenotype has been verified in both ex vivo and in vivo models. Mechanistic studies to-date have implicated NaHCO3-mediated repression of genes involved in the production, as well as maturation, of the alternative penicillin-binding protein (PBP) 2a, a necessary component of MRSA β-lactam resistance. Herein, we utilized RNA-sequencing (RNA-seq) to identify genes that were differentially expressed in NaHCO3-responsive (MRSA 11/11) vs. non-responsive (COL) strains, in the presence vs. absence of NaHCO3-β-lactam co-exposures. These investigations revealed that NaHCO3 selectively repressed the expression of a cadre of genes in strain 11/11 known to be a part of the sigB-sarA-agr regulon, as well as a number of genes involved in the anchoring of cell wall proteins in MRSA. Moreover, several genes related to autolysis, cell division, and cell wall biosynthesis/remodeling, were also selectively impacted by NaHCO3-OXA exposure in the NaHCO3-responsive strain MRSA 11/11. These outcomes provide an important framework for further studies to mechanistically verify the functional relevance of these genetic perturbations to the NaHCO3-responsiveness phenotype in MRSA.


Introduction
The rise of antibiotic-resistant bacteria has increasingly threatened healthcare initiatives worldwide [1][2][3][4]. In particular, methicillin-resistant Staphylococcus aureus (MRSA) is a major clinical pathogen, causing a wide variety of severe infections, with relatively limited treatment options, poor patient outcomes, and increased healthcare-associated costs [5][6][7][8]. MRSA strains are traditionally considered as "resistant" to nearly all β-lactams (the treatment of choice for methicillin-susceptible S. aureus [MSSA]) [9,10]. Standard treatment options for MRSA, such as vancomycin or daptomycin, tend to be less effective, less well tolerated (in terms of side effects and toxicities) and more expensive than standard-of-care β-lactams as used for MSSA infections [11][12][13][14].
Recently, a novel β-lactam susceptibility phenotype was discovered among MRSA strains, termed 'bicarbonate [NaHCO 3 ]-responsiveness', wherein certain MRSA strains become sensitized to β-lactam antibiotics in vitro in the presence of NaHCO 3 [15,16]. This phenotype was relatively widespread among a small collection (n = 58) of US bloodstream isolates (ranging from 33-75% for oxacillin and cefazolin, respectively [16]). Importantly, the translatability of this phenotype has been verified both ex vivo (in simulated endocarditis vegetations [SEV] models) and in vivo (in experimental infective endocarditis models) [15,17]. To-date, the potential mechanisms of this phenotype have focused on the impact of NaHCO 3 on mecA expression/PBP2a protein production, as well as upon genes associated with PBP2a maturation and functionality [15,18]. However, these mechanistic studies have predominantly centered on targeted investigations of well-characterized genes already known for their involvement in β-lactam resistance in MRSA, rather than a more unbiased and global consideration of genes that are selectively impacted among NaHCO 3 -responsive vs. non-responsive MRSA strains.
In the present investigation, we have utilized RNA-sequencing (RNA-seq) methods to identify a broader repertoire of genes likely involved in the NaHCO 3 -responsive vs. non-responsive phenotypes. We concentrated on identifying genes which were selectively expressed during NaHCO 3 -β-lactam co-exposures, comparing a prototype NaHCO 3responsive strain (MRSA 11/11) to a prototype non-responsive strain (COL) [15,17,18]. These RNA-seq analyses revealed a number of differentially and selectively expressed genes in the NaHCO 3 -responsive strain, with several noteworthy gene cadres, including: (i) genes within the sigB-sarA-agr regulatory axis; (ii) cell wall-anchored/peptidoglycanassociated genes; and (iii) genes involved in either global or divisome-specific autolysis.

RNA Isolation, Library Construction, Sequencing and Analysis
To obtain RNA, strains were grown in CA-MHB 100 mM Tris ± 44 mM NaHCO 3 overnight (O/N), then diluted 1:50 into 25 mL of either the same O/N growth medium or O/N growth medium supplemented with 2% NaCl and 1 2 x OXA MIC concentrations (as indicated above). Cultures were then grown for 4 h at 37 • C with aeration (to reach an OD 600 0.5), and total RNA was harvested by mechanical disruption by FastPrep disruption (MP Biomedicals, Irvine, CA, USA), followed by column isolation (Qiagen, Germantown, MD, USA), and treatment with Turbo DNAse (Invitrogen, Waltham, MA, USA) to remove DNA from sample [15]. Ribosomal  , are currently in-progress and will be made available by contacting the lead Author (S.C.E.), at the following email address (selvi.ersoy@lundquist.org). "Differentially and selectively expressed genes" in our study represented those genes which exhibited a significant level of differential expression in the presence of NaHCO 3 ±OXA exposure compared to media without NaHCO 3 for the responsive strain, MRSA 11/11 vs. the non-responsive COL strain (≥ 2-fold). Thus, this definition explicitly eliminated genes which were differentially and concordantly expressed under these conditions in MRSA 11/11 as well as in COL. For selected analyses, we catalogued genes which were differentially expressed by ≥ 5-fold. Kyoto Encyclopedia of Genes and Genomes (KEGG) metabolic pathways and annotations for differentially abundant genes were mapped using KOBAS v3.0 [24].
Following RNA-seq analyses, we selected five genes from the overall repertoire of differentially expressed genes for qRT-PCR validations, representing broad categories of bacterial factors, including: i) peptidoglycan synthesis (pbp2 and ddh); ii) autolysis (atl and sceD); and iii) virulence (fnbA). The primers listed in Table S1 were used to amplify each gene-of-interest to determine their relative expressions. The gene gyrB was used as a housekeeping gene to normalize transcript abundance. The qRT-PCR was carried out on a StepOne thermocylcer (ThermoFisher, Waltham, MA, USA) and analyzed with StepOne Software. Relative gene expression was calculated using the 2 −∆∆C T method from two independent biological replicates performed in triplicate on at least two separate runs for each strain/condition. Relative gene expression in CA-MHB 100 mM Tris+2% NaCl+ 1 2 × MIC OXA+44 mM NaHCO 3 was normalized to CA-MHB 100 mM Tris+2% NaCl+ 1 2 × MIC OXA for each gene, with expression in the latter condition set equal to 1.0. statistical significances related to relative expression profiles were determined by a Student's t-test, with a p value of ≤0.05 considered 'significant'.

Results and Discussion
3.1. Differentially Expressed Genes by RNA-seq in NaHCO 3 -responsive vs. Non-responsive Strains RNA-seq analysis was carried out on two prototypical MRSA strains, MRSA 11/11 (NaHCO 3 -reponsive) and COL (NaHCO 3 -non-responsive), for two distinct comparison groups: Group 1-CA-MHB Tris vs. CA-MHB Tris + NaHCO 3 ; and Group 2-CA-MHB Tris + 2% NaCl + OXA vs. CA-MHB Tris + NaHCO 3 + 2% NaCl + OXA. RNA-seq analysis of the prototypical NaHCO 3 -responsive strain MRSA 11/11 revealed that exposure to NaHCO 3 alone (Group 1) led to the differential change in abundance (≥ 2-fold change) of 21 transcripts ( Figure 1A); in contrast, NaHCO 3 -OXA combination (Group 2) yielded a differential change in abundance of 84 transcripts ( Figure 1B). Following initial analysis of differentially abundant transcripts for Groups 1 and 2 in MRSA 11/11 and COL, those genes whose expression was altered in the same direction for both MRSA 11/11 and COL in each group were subtracted from subsequent analyses, as presented in Table 1. Volcano plots of the differential expression of genes in the presence of: (A) NaHCO 3 exposure alone; or (B) NaHCO 3 + 1 2 xMIC OXA co-exposures in NaHCO 3 -responsive strain MRSA 11/11. NaHCO 3 alone led to an observed 21 differentially expressed transcripts. NaHCO 3 and OXA co-exposures together yielded 84 differentially expressed transcripts. These numbers represent the total number of transcripts identified, featuring genes of both known and unknown functions. Table 1. Overview of gene functional classes altered ≥ 2-fold in two comparison groups, (Group 1) CA-MHB Tris vs. CA-MHB Tris + NaHCO 3 , and (Group 2) CA-MHB Tris + 2% NaCl + OXA vs. CA-MHB Tris + NaHCO 3 + 2% NaCl + OXA. Genes/transcriptional units indicated in this tables are those altered by NaHCO 3 exposure alone (Group 1, n = 11), NaHCO 3 and OXA co-exposure (Group 2, n = 57), or common to both groups (Group 1 and 2, n = 10) for a total number of functionally identifiable genes/transcriptional units of 78. "Increase" and "decrease" refer to expression being increased or decreased by exposure to NaHCO 3 compared to media without NaHCO 3 in each comparison group. NaHCO 3 exposure, in the presence and absence of β-lactam co-exposure, impacted the expression of genes across a range of functional groups, with a large proportion relating to either cell wall synthesis, cell membrane transporters, or cellular metabolism (Table 1). Intriguingly, across most functional categories, NaHCO 3 exposure resulted in significant repression of gene expression, particularly for those genes relating to virulence, cell wall biosynthesis, and membrane transporters (Table 1, Table S2A-C). Of the genes that were upregulated by NaHCO 3 exposure, these tended to be related to cellular metabolism, protein translation, and/or stress response, indicating that NaHCO 3 may mirror the exposure of the responsive MRSA strain to an 'inhospitable' host environment.

Classifications of Gene Functions
Oxidative and membrane stress have been implicated in the formation of persister cells [25], which make up approximately 1% of the total cell population; such cells have increased antibiotic tolerance and reduced metabolism [26][27][28]. The RNA-seq data above indicate that, while NaHCO 3 may be stimulating a stress response in NaHCO 3 -responsive MRSA, NaHCO 3 also evoked increased expression of many metabolic genes as compared to cells grown in the absence of NaHCO 3 . This latter finding (i.e., enhanced metabolic activity) would not support the notion of NaHCO 3 -mediated entry of responsive MRSA cells into a persister state (Table 1, Table S2A-C).
We then interrogated only those differentially abundant transcriptional units that were specifically altered in MRSA 11/11, and for which a specific function is known and expression was altered by ≥ 5-fold. NaHCO 3 alone significantly impacted the expression at this magnitude of only a relatively small number of functionally-identifiable genes/transcriptional units (n = 7) as compared to a larger number of genes/transcriptional units differentially impacted by NaHCO 3 +OXA (n = 14) in the NaHCO 3 -responsive strain MRSA 11/11 ( Figure 2). Additionally, eight transcriptional units were differentially expressed in both growth conditions (NaHCO 3 alone and NaHCO 3 +OXA) (Figure 2).
Of particular interest, the autolysin sceD and the global virulence regulator agr were both substantially repressed (≥ 5-fold change in expression) by NaHCO 3 , with or without β-lactam co-exposure ( Figure 2, Table 2). The agr locus is a major virulence regulator in S. aureus [29][30][31], although its role in antibiotic resistance in vivo remains controversial [30,32]. Previously, we observed by qRT-PCR that NaHCO 3 is capable of repressing the global virulence regulons sarA and sigB (both agr-regulated) [15,18]. These RNA-seq data corroborate the downstream effects of sarA and sigB repression, as observed by the repression of agr and several other genes known to be part of the sigB-sarA-agr regulon, including fnbA, fnbB, cap8, clfA, and clpL ( Table 2, Table 3) [33][34][35][36]. Of note, one of the few genes upregulated by NaHCO 3 exposure in MRSA 11/11 was clfB, a cell wall-anchored (CWA) protein whose function is linked to nasal colonization and virulence in skin and soft tissue infections [37]. Additionally, NaHCO 3 repressed expression of sdrH, another CWA gene whose repression is associated with decreased biofilm formation [38,39], an impact that corresponds to the observed decrease in biofilm formation during NaHCO 3 exposure in MRSA 11/11 [40]. Taken together, these data indicate a broad impact of NaHCO 3 (especially in combination with OXA) on genes associated with cell surface modifications, as well as cell wall biosynthesis or remodeling. A Fold change comparisons are made by comparing the gene expression for the indicated strain and condition (± OXA) exposed to NaHCO 3 to the same strain/condition without NaHCO 3 exposure. Negative values indicate that exposure to NaHCO 3 for the given strain/condition decreases expression of the indicated gene, whereas positive values indicate that NaHCO 3 increases expression of the indicated gene. B OXA concentration equal to 1/2 × MIC in CA-MHB Tris ± NaHCO3 for the indicated strain. A Fold change comparisons are made by comparing the gene expression for the indicated strain and condition (± OXA) exposed to NaHCO 3 to the same strain/condition without NaHCO 3 exposure. Negative values indicate that exposure to NaHCO 3 for the given strain/condition decreases expression of the indicated gene, whereas positive values indicate that NaHCO 3 increases expression of the indicated gene. B OXA concentration equal to 1/2 × MIC in CA-MHB Tris ± NaHCO3 for the indicated strain.
It should be underscored that NaHCO 3 did increase the expression of several genes in the non-responsive strain, COL, whose expressions were down-regulated in MRSA 11/11, including the cap8 operon, the virulence-associated gene vraX, and the potassium transport operon kdpABCDF (Table 3). Despite these observed differences in such putative virulence gene expressions between COL and MRSA 11/11, we have not observed any differences in the intrinsic virulence of these two strains in vivo [15]. Further, a large cadre of genes was specifically impacted by NaHCO 3 , with or without OXA exposure in strain COL, but not in MRSA 11/11. However, the analytics required to determine the role of these specifically and differentially impacted genes in generating the non-responsive MRSA phenotype is beyond the scope of this paper.
Of primary interest in regard to NaHCO 3 -mediated susceptibility to β-lactams was its impact on the expression of genes related to cell wall synthesis. NaHCO 3 (with or without OXA co-exposure) repressed the expression of two key autolysins, atl and sceD; in addition, the expressions of several other key genes involved in cell wall synthesis were repressed by NaHCO 3 -OXA co-exposures (isaA, fmtA, ddh, pbp2) ( Table 2). These latter six genes encompass those involved in the formation of nascent (new) peptidoglycan and cell wall restructuring (pbp2, ddh, fmtA) [41][42][43][44], as well as peptidoglycan hydrolases, involved in cell wall turnover and division (atl, sceD, isaA) [45][46][47]. Previous studies have demonstrated that the deletion or inactivation of isaA and fmtA results in increased susceptibility to βlactams, an event independent of impacts on the expression of PBP2a [45,48]. Furthermore, the regulation of sceD and isaA transcription is linked, since deletion in isaA results in enhanced expression of sceD [45,46]. We observed a modest decrease in isaA transcription in the presence of NaHCO 3 -OXA, but a substantial decrease in sceD transcription in the presence of NaHCO 3 (with or without OXA). Although previous reports did not find that the deletion of sceD alone resulted in increased β-lactam susceptibility [45], it is possible that within the context of the gene expression profile stimulated by NaHCO 3 , reduced sceD expression may have a larger impact on β-lactam susceptibility than previously understood.

qRT-PCR Validation
The qRT-PCR validation of specific genes identified by RNA-seq revealed that sceD expression was strongly repressed by NaHCO 3 in MRSA 11/11, while being substantially upregulated in the non-responsive strain COL ( Figure 3); this finding further supports the notion that altered sceD expression may be of key importance to the NaHCO 3 -responsive phenotype. The qRT-PCR also validated the NaHCO 3 -mediated repression of pbp2, atl, and fnbA in responsive strain MRSA 11/11 as seen by RNA-seq, while NaHCO 3 stimulated increased expression of these genes in COL (Figure 3).

Limitations of Study
This study was limited by only using a single NaHCO 3 -responsive/non-responsive strain pair for RNA-seq analyses. A larger number of strains exhibiting each phenotype would increase the likelihood of identifying a more consensus cadre of differentially expressed genes involved in the NaHCO 3 -responsiveness phenotype. Further, only one βlactam (OXA) was tested for its impact on gene expression in the presence of NaHCO 3 . Our previous work has demonstrated that NaHCO 3 -induced sensitization to cefazolin (CFZ) is much more frequent among MRSA strains than sensitization to OXA [16]. This implies that CFZ and OXA may be impacting somewhat different genetic targets in NaHCO 3 -responsive strains, which could be causing the increased frequency of sensitization to CFZ vs. OXA. Additionally, only one concentration of NaHCO 3 (44 mM) was examined for its impact on gene expression. Investigating the impact of NaHCO 3 at blood concentrations (~25 mM NaHCO 3 ) may reveal genes that are particularly important to NaHCO 3 -β-lactam-induced changes in expression within that specific host microenvironment.
Although KEGG analysis was carried out, the relatively small number of differentially expressed genes in MRSA 11/11 limited the number that could be functionally categorized to a selected, definable metabolic or biosynthetic pathway. The relative lack of descriptive KEGG pathways to ascribe to the differentially transcribed genes in these analyses underscores the need for separate functional characterizations of those genes we have identified by RNA-seq. Furthermore, additional investigations with a larger collection of NaHCO 3 -responsive strains may help in determining specific, consensus KEGG pathways associated with the responsive phenotype. For each gene, expression is normalized to that observed for the indicated strain grown in CA-MHB Tris+2% NaCl + 1 2 × MIC OXA (CA-MHB Tris + 2% NaCl + 1 2 × MIC OXA expression set at 1.0). Statistics were calculated by Student's t-test (*** P < 0.001, **** P < 0.0001).

Conclusions
NaHCO 3 (especially in combination with OXA) impacts a variety of genes specifically and differentially expressed in the NaHCO 3 -responsive strain MRSA 11/11 relating to: cell wall synthesis, turn-over and division; and cell surface-associated genes regulated by the sigB-sarA-agr regulatory axis. These analyses point to a mechanism of β-lactam sensitization by NaHCO 3 that involves global impacts on the cell surface (where β-lactams initially bind), as well as altered expression of genes required to synthesize the cell wall and maintain its homeostasis. Moreover, further work is needed to validate and quantify the specific impact of those genes identified herein on the NaHCO 3 -responsive phenotype, using strategic gene knockouts. Finally, future studies will be designed to look at the differential impacts of NaHCO 3 upon protein-level expression profiles in NaHCO 3 -responsive vs. non-responsive MRSA.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are openly available in the NCBI Sequence Read Archive (SRA) as indicated in the manuscript materials and methods.

Conflicts of Interest:
The authors declare no conflict of interest.