Staphylococcus aureus Overcomes Anaerobe-Derived Short-Chain Fatty Acid Stress via FadX and the CodY Regulon

ABSTRACT

Chronic rhinosinusitis (CRS) is characterized by immune dysfunction, mucus hypersecretion, and persistent infection of the paranasal sinuses. While Staphylococcus aureus is a primary CRS pathogen, recent sequence-based surveys have found increased relative abundances of anaerobic bacteria, suggesting that S. aureus may experience altered metabolic landscapes in CRS relative to healthy airways. To test this possibility, we characterized the growth kinetics and transcriptome of S. aureus in supernatants of the abundant CRS anaerobe Fusobacterium nucleatum. While growth was initially delayed, S. aureus ultimately grew to similar levels as in the control medium. The transcriptome was significantly affected by F. nucleatum metabolites, with the agr quorum sensing system notably repressed. Conversely, expression of fadX, encoding a putative propionate coenzyme A (CoA)-transferase, was significantly increased, leading to our hypothesis that short-chain fatty acids (SCFAs) produced by F. nucleatum could mediate S. aureus growth behavior and gene expression. Supplementation with propionate and butyrate, but not acetate, recapitulated delayed growth phenotypes observed in F. nucleatum supernatants. A fadX mutant was found to be more sensitive than wild type to propionate, suggesting a role for FadX in the S. aureus SCFA stress response. Interestingly, spontaneous resistance to butyrate, but not propionate, was observed frequently. Whole-genome sequencing and targeted mutagenesis identified codY mutants as resistant to butyrate inhibition. Together, these data show that S. aureus physiology is dependent on its cocolonizing microbiota and metabolites they exchange and indicate that propionate and butyrate may act on different targets in S. aureus to suppress its growth.

IMPORTANCE Staphylococcus aureus is an important CRS pathogen, and yet it is found in the upper airways of 30% to 50% of people without complications. The presence of strict and facultative anaerobic bacteria in CRS sinuses has recently spurred research into bacterial interactions and how they influence S. aureus physiology and pathogenesis. We show here that propionate and butyrate produced by one such CRS anaerobe, namely, Fusobacterium nucleatum, alter the growth and gene expression of S. aureus. We show that fadX is important for S. aureus to resist propionate stress and that the CodY regulon mediates growth in inhibitory concentrations of butyrate. This work highlights the possible complexity of S. aureus-anaerobe interactions and implicates membrane stress as a possible mechanism influencing S. aureus behavior in CRS sinuses.

INTRODUCTION

Chronic rhinosinusitis (CRS) is an inflammatory condition of the sinuses that is characterized broadly by facial pain, mucus hypersecretion and accumulation, immune dysfunction, pathogen colonization, and persistent polymicrobial infection (1–6). Although CRS affects up to 15% of the population and represents a substantial economic burden, its complexity has slowed the development of new treatments and therapeutic strategies (7). CRS patients are frequently prescribed antibiotics and yet many do not respond and require functional endoscopic sinus surgery (FESS) to remove accumulated mucus and inflamed mucosa that prevent proper sinus drainage (5). Given the urgent threat of antimicrobial resistance among CRS microbiota, there is a critical need to better understand microbial community dynamics in the upper airways and how they may contribute to disease (8).

Staphylococcus aureus is a frequently isolated CRS pathogen and is aggressively targeted by antibiotic therapy; yet, this bacterium is also prevalent and abundant in the upper airways of asymptomatic healthy individuals (9, 10). This apparent paradox suggests that colonization by S. aureus is not sufficient to drive disease but rather that there may be important environmental cues in the upper airways that shift the lifestyle of S. aureus toward commensalism or pathogenesis. Indeed, in a genome-wide association study of S. aureus isolated from 28 CRS patients, few S. aureus genetic signatures were associated with CRS subtypes, suggesting that S. aureus pathogenesis in CRS is unlikely due to selection for increased production of a particular toxin (11).

Application of culture-independent genomics to the study of CRS has led to a paradigm shift from a small number of etiologic bacterial species toward a polymicrobial basis of disease (4, 5, 12). However, the role of the greater CRS microbiome in disease pathophysiology remains poorly understood. To address this knowledge gap, we recently surveyed 16S rRNA gene sequences in FESS-derived mucus from a cohort of CRS patients and found increased relative abundances of numerous anaerobic bacterial taxa, including many known to degrade mucin glycoproteins (6). CRS bacterial communities enriched on mucins as a sole carbon source converged on similar profiles, which were dominated typically by a combination of Streptococcus, Prevotella, Fusobacterium, and Veillonella. Interestingly, S. aureus had a variety of growth phenotypes and gene expression patterns when cultured in supernatants from these enrichment communities, indicating that nutrient usage and metabolite release by cocolonizing microbiota can profoundly affect S. aureus physiology (6). Enrichment supernatants that best supported S. aureus growth had low levels of short-chain fatty acids (SCFAs; acetate, propionate, and butyrate) and undetectable levels of Fusobacterium, of which members are known for producing SCFAs as amino acid fermentation by-products (13, 14). However, neither growth promotion nor inhibition could be ascribed to any one taxon or metabolite within these communities.

In this study, we extend our previous work by demonstrating that SCFAs impede S. aureus growth and lead to reduced expression of the accessory gene regulator (agr) quorum sensing system while inducing a putative fatty acid degradation operon (fadXEDBA). We confirm that the SCFAs propionate and butyrate are sufficient to impair S. aureus growth and alter gene expression, while acetate had relatively little effect. We show that growth of a ΔfadX mutant is significantly attenuated in the presence of propionate only, despite differing from butyrate by only one carbon. Spontaneous resistance to growth inhibition by butyrate arose frequently, while we failed to obtain propionate-resistant mutants. Genome sequencing of butyrate-resistant mutants identified premature stop codons and in-frame deletions in the gene encoding the nutrient-responsive global regulator codY, indicating a connection between derepression of the CodY regulon through nutrient limitation and SCFA resistance. These data suggest that certain anaerobes may influence CRS community structure by limiting S. aureus growth via propionate and butyrate production. In addition, they implicate the CodY regulon as a mechanism allowing S. aureus persistence in otherwise inhospitable anaerobic bacterial communities of the upper airways.

RESULTS

S. aureus growth is impaired by F. nucleatum.

In a previous study of upper airway (CRS) microbiota, we found that in patients (n = 27) with detectable Fusobacterium spp., relative abundances of Staphylococcus spp. were minimal or below the level of detection (Fig. 1A) (6). When S. aureus was grown in supernatants derived from anaerobic enrichment cultures of CRS sinus mucus, it exhibited slower growth in those that had Fusobacterium as a core constituent genus of the enrichment community. These supernatants also contained higher levels of the short-chain fatty acids (SCFAs) propionate and butyrate (6). Given these data, we hypothesized that Fusobacterium spp. might indirectly exert an antagonistic effect on S. aureus through the production of SCFAs. To test this hypothesis, we grew S. aureus USA300 LAC in filtered cell-free supernatants (CFSs) from F. nucleatum ATCC 25586 grown for 48 h in Brucella broth (BB) supplemented with hemin and vitamin K (Fig. 1B). S. aureus grew in F. nucleatum CFS, albeit at a slower rate than that in the control medium (BB) with an increased lag phase. However, both cultures reached approximately similar optical density at 600 nm (OD₆₀₀) values by 24 h, indicating that S. aureus was able to obtain sufficient nutrients over the course of the experiment. We reasoned that the extended lag phase was likely due to F. nucleatum-mediated depletion of an easily metabolizable nutrient source and/or the presence of an inhibitory metabolite(s) that S. aureus was able to adapt to over time. We tested the presence of an inhibitory metabolite(s) by measuring the acetate, propionate, and butyrate content of the CFS before and after S. aureus growth (Fig. 1C). All three SCFAs were detected in F. nucleatum CFS (∼5 mM acetate, ∼5 mM propionate, and ∼15 mM butyrate). After overnight growth (∼16 h) in F. nucleatum supernatants, S. aureus cultures had increased acetate levels, while propionate and butyrate remained the same as in CFS alone, indicating that S. aureus does not actively metabolize propionate and butyrate under these conditions. We interpret this finding to mean that S. aureus adapts to SCFAs by modifying its physiology rather than directly detoxifying them via degradation. The increased acetate levels are likely due to S. aureus utilization of glucose remaining in F. nucleatum CFS, as F. nucleatum preferentially ferments amino acids (15, 16). Given that S. aureus growth is similarly impaired when BB is supplemented with the sodium salts of acetate (5 mM), propionate (5 mM), and butyrate (15 mM) (Fig. 1B), these data suggest that SCFAs may be key factors driving bacterial interactions in the CRS sinus environment, providing a mechanism by which Fusobacterium and other anaerobes may restrict S. aureus growth in vivo.

FIG 1.

S. aureus growth is impaired in F. nucleatum supernatants. (A) Relative abundances of Fusobacterium and Staphylococcus in sinus mucus from patients with chronic rhinosinusitis are inversely correlated (6). (B) Representative growth curve of S. aureus USA300 in Brucella broth (BB; control); BB supplemented with 5 mM acetate, 5 mM propionate, and 15 mM butyrate; and cell-free supernatants from F. nucleatum (Fn CFS). (C) Production of SCFAs by F. nucleatum after 48 h (Fn CFS) and their levels after S. aureus (USA300) growth in Fn CFS. SCFAs were below the limit of detection in sterile BB. All data shown in B and C are the mean ± standard deviation of three biological replicates.

F. nucleatum metabolites significantly alter S. aureus gene expression.

We next determined how S. aureus modified its transcriptome in F. nucleatum CFS. To do so, we performed a targeted analysis using a custom NanoString code set (see Table S4 in the supplemental material) that included 34 genes encoding several known virulence factors, key metabolic genes, and master regulators of gene expression (Fig. 2A). Of these genes, we detected 13 differentially expressed transcripts (≥2-fold change in expression and adjusted P value of <0.05), of which many are in overlapping transcriptional networks; expression of fadX, cidA, icaB, and gltB increased while nanA, alsS, lrgA, narG, agrA, hla, hld, saeR, and ldh1 decreased in S. aureus grown on CFS relative to BB alone (Fig. 2B). A number of other transcripts (aur, fib, pgi, codY, opp3b, and arlR) were statistically significant but exhibited a less than 2-fold change in expression (Table S4). Decreased signaling through the quorum sensing response regulator agrA or the saeRS two-component system in F. nucleatum CFS may result in lower expression of the hla and hld genes, encoding alpha and delta hemolysins (17). Neuraminate lyase, encoded by nanA, is induced by the presence of sialic acids and exhibited lower expression in F. nucleatum CFS, indicating that F. nucleatum likely utilized sialic acids present in Brucella broth as a nutrient source (18, 19). Expression of the nutrient-sensing transcriptional regulator codY was reduced in F. nucleatum CFS by nearly 50% compared with that of the control medium, likely explaining the increase in the glutamate synthase subunit gene gltB (20). We selected three genes (fadX, agrA, and nanA) for validation via reverse transcription-quantitative PCR (qRT-PCR) and show that they were highly correlated with the NanoString results (see Fig. S1 in the supplemental material). That nearly one-half of the transcripts in the NanoString probe set are differentially regulated in F. nucleatum CFS, including a number of major transcriptional regulators important for integrating metabolic cues and virulence gene expression, highlights the scope of alterations to the S. aureus transcriptome. These data show that S. aureus physiology can be influenced significantly by the metabolic activity of a single anaerobic species and underscore the possible complexity of bacterial behaviors and interactions within a diverse CRS community.

FIG 2.

F. nucleatum metabolites significantly impact the S. aureus transcriptome. (A) Heatmap depicting log₁₀-transformed S. aureus gene expression in control media (BB) and F. nucleatum supernatant (CFS) as detected by NanoString. Genes were clustered with unsupervised hierarchical clustering. (B) MA plot representation of S. aureus gene expression in F. nucleatum CFS relative to the control medium. Genes were considered significant if they had a log₂-fold change of ≥1 and a Benjamini-Hochberg-adjusted P value of <0.05.

SCFAs significantly alter S. aureus gene expression.

SCFAs, especially propionate and butyrate, have been reported to impair S. aureus growth and attenuate murine skin infections (21). We therefore sought to determine if individual SCFAs were sufficient to drive some of the S. aureus gene expression patterns observed after growth in F. nucleatum CFS. To measure the effects of each SCFA on the agr quorum sensing system, we grew S. aureus carrying pAH1 (encoding P3_agr-mCherry) in LB or LB supplemented with acetate, propionate, or butyrate for 24 h and measured fluorescence intensity normalized to culture density (Fig. 3A). All three SCFAs led to decreased fluorescence, with propionate (P = 0.0035) and butyrate (P < 0.0001) significantly inhibiting reporter activity, while acetate (P = 0.203) had the smallest effect. Given these observations, CRS bacterial communities dominated by Fusobacterium or other taxa that produce propionate and butyrate would be predicted not only to impede the growth of S. aureus but also minimize the production of agr-regulated virulence factors.

FIG 3.

Propionate and butyrate repress the S. aureus agr system but fail to induce biofilm. (A) S. aureus carrying pAH1 (P_agr-mCherry) was grown for 24 h in LB supplemented with 100 mM sodium acetate, propionate, or butyrate (n = 3 biological replicates with n = 3 cultures per replicate). Fluorescence was measured and normalized to culture density for each replicate and then normalized to the LB controls. Significance was determined by. (B, C) Expression of cidA and lrgA from S. aureus in LB supplemented with SCFAs (n = 3). Copy number was determined via standard curve and normalized to the gmk housekeeping gene. (D) Crystal violet assay quantifying biofilm formation in LB, LB supplemented with glucose (positive control for increased biofilm formation), or LB supplemented with 100 mM each SCFA. All data are the mean ± standard deviation of three biological replicates. Significance was determined by Kruskal-Wallis one-way ANOVA with Dunn’s multiple-comparison test for A and ordinary one-way ANOVA with Holm-Sidak’s multiple comparison’s test for B to D. **, P < 0.01; ****, P < 0.0001.

Given the reduction in agr quorum sensing activity and thus a lack of repression of proteins involved in surface attachment, we hypothesized that SCFAs may be a probiofilm signal to S. aureus (22). We performed qRT-PCR on S. aureus grown to an OD₆₀₀ of ∼0.2 to 0.3 in LB or LB supplemented with 100 mM each individual SCFA to determine if biofilm-associated transcripts identified as differentially regulated in F. nucleatum CFS were affected. We found that the expression of cidA, encoding a holin-like protein involved in programmed cell death and extracellular DNA release during biofilm formation, was approximately 10-fold higher (P < 0.0001) in the presence of propionate but was relatively unaffected by acetate or butyrate (Fig. 3B) (23). Conversely, there was no effect of SCFAs on the expression of lrgA, which encodes a putative antiholin that is antagonistic to CidA (23). This finding suggests that decreased lrgA expression detected in F. nucleatum CFS is likely independent of the SCFAs tested here (Fig. 3C). Despite increased cidA expression, SCFA supplementation of LB had marginal effects on biofilm production, with acetate and propionate leading inducing modest but insignificant increases relative to LB alone and butyrate having no detectable effect (Fig. 3D). The lack of downregulation of lrgA under these conditions suggests that sufficient LrgA protein may be available to offset any increased CidA activity. Alternatively, other environmental cues may be needed to enhance biofilm formation under these conditions.

The fadX gene mediates propionate resistance.

The most highly induced transcript in S. aureus grown in F. nucleatum CFS was fadX, encoding a putative propionate CoA-transferase, the first in a five-gene operon predicted to be involved in fatty acid degradation. Given their annotation, we hypothesized that the fad operon may encode a component of the S. aureus SCFA stress response. The fad genes were induced after growth in F. nucleatum CFS and in the presence of propionate and butyrate; their status as an operon was confirmed by obtaining amplicons from cDNA using PCR primer sets that spanned each intergenic region (Fig. 4; see Fig. S2 in the supplemental material). We tested a fadX transposon mutant (obtained from the Nebraska Transposon Mutant Library) and its parental strain (JE2) for the ability to grow in 100 mM propionate and found that fadX::tn had a significant growth defect in propionate relative to the wild type (see Fig. S3 in the supplemental material). The mutant grew as well as the parent strain in LB alone, indicating that growth inhibition was specific to propionate in the medium. We then performed dose-response growth curves in six concentrations of sodium propionate, ranging from 100 mM to 3.125 mM in 2-fold reductions, and found clear growth differences between JE2 and fadX::tn, with the mutant having a defect in media with as low as 12.5 mM (see Fig. S4A in the supplemental material).

FIG 4.

The fad operon is induced by propionate and butyrate. (A and B) Reverse transcription-quantitative PCR was used to detect fad operon expression in control (BB) or F. nucleatum CFS (A) or in LB supplemented with 100 mM sodium propionate (B). Cultures were grown to an OD₆₀₀ of approximately 0.2 to 0.3 prior to RNA extraction. Data shown are mean ± standard deviation of three biological replicates. Significance was determined by two-way ANOVA with Siadak’s multiple-comparison test. **, P < 0.01; ****, P < 0.0001.

To determine if the transposon insertion in the fadX::tn mutant disrupted the entire fad operon, we performed qRT-PCR and confirmed that the three genes downstream of fadX had considerably reduced expression in LB (Fig. S4B). Transposon mutants in each of the fadEDBA genes did not exhibit the same reduced growth in propionate-containing LB that the fadX::tn mutant did; therefore, we did not pursue them further. We then constructed a ΔfadX deletion mutant in the USA300 LAC background and tested its growth in LB supplemented with each SCFA (Fig. 5). Relative to the wild type, there was no growth defect in acetate; however, there was modest inhibition of the mutant in propionate, with growth curves diverging after approximately 8 to 10 h and remaining consistent through the end of the experiments. Neither strain grew well in butyrate, although sporadic growth was detected after ∼15 h, irrespective of genotype and only in butyrate. Together, these data implicate FadX in ameliorating or resisting propionate stress, although the mechanism remains unclear. Furthermore, the occasional growth of either strain at later time points in butyrate, but not propionate, provides indirect evidence that these SCFAs may have different mechanisms of S. aureus growth inhibition.

FIG 5.

The ΔfadX mutant is more susceptible to growth inhibition by propionate than the wild type. Combined growth curves (n = 4) of wild-type USA300 or the ΔfadX mutant in 100 mM of the sodium salts of acetate (A), propionate (B), or butyrate (C). Data shown are the mean ± standard deviation of three biological replicates.

codY mutants are resistant to butyrate.

Although occasional growth was detected in LB supplemented with butyrate after ∼15 h of incubation, it consistently occurred in only one of three technical replicates of a given sample (either wild type or ΔfadX). These wells were plated onto LB and mannitol salt agar to confirm the absence of contamination and that the observed turbidity was due solely to S. aureus growth. To further investigate this phenomenon, we grew the ΔfadX mutant for 24 h in LB and then plated 10-fold serial dilutions onto LB agar + 200 mM sodium butyrate. Large colonies were observed occasionally in the 10⁻² dilutions (the lowest plated) after overnight incubation, which we assumed to be due to increased resistance to butyrate. Patching these colonies onto the same medium confirmed their resistance phenotype. We interpret these data to mean that butyrate-resistant mutants arose spontaneously in LB starter cultures and that the occasional turbidity in LB + butyrate cultures after ∼15 h represents growth after an extended lag phase resulting from their extremely low starting abundance.

Growth curves in each SCFA were then performed to determine if the large colonies had a growth advantage over the parental strain. We found that all four large colonies grew significantly faster in the presence of butyrate than the parental strain, and yet there were no differences in media supplemented with acetate or propionate (Fig. 6A to C). We then sequenced their genomes to identify genetic determinant(s) of butyrate resistance and found two independent mutations in the gene encoding the GTP- and branched-chain amino acid-sensing global regulator CodY. The first mutation resulted in a premature stop codon truncating the protein after 65 amino acids, while the second led to a 20-amino acid deletion from a conserved region of the protein at codons 171 to 190 (Fig. 6D). We repeated growth curves using a JE2 codY::tn mutant (with an intact fadX gene) and confirmed that codY mutation alone was sufficient to rescue growth in the presence of butyrate (Fig. 7A). Finally, we performed qRT-PCR on JE2 and the codY::tn mutant to assay for fad operon expression and found that while the operon was induced modestly in the mutant, none of the genes reached significance (Fig. 7B). These data, coupled with the fact that codY mutants have similar levels of growth impairment in propionate, are further evidence that propionate and butyrate may act on different targets to inhibit S. aureus growth.

FIG 6.

Spontaneous S. aureus codY mutants are not inhibited by butyrate. Combined growth curves of the ΔfadX mutant and butyrate-resistant derivatives in 100 mM sodium acetate (A), sodium propionate (B), or sodium butyrate (C). (D) Alignment of CodY protein sequences from diverse Gram-positive bacteria and S. aureus, including butyrate-resistant mutants (but^R1 and but^R2); but^R1 encodes a premature stop codon at position 66 (blue), while but^R2-4 mutants have a 20-amino acid deletion from positions 171 to 190 (red). The but^R3 and but^R4 mutants were omitted from the alignment as their codY mutations are identical to but^R2. All data shown in A to C are the mean ± standard deviation of three biological replicates.

FIG 7.

codY mutation is sufficient to escape butyrate growth inhibition, although not likely through Fad activity. (A) Growth of wild-type S. aureus and a codY::tn mutant in 100 mM sodium butyrate. (B) Expression of the fad operon from the same strains as in A, grown in LB to an OD₆₀₀ of approximately 0.2 to 0.3. All data shown are the mean ± standard deviation of three biological replicates.

DISCUSSION

Although not lethal, CRS remains a significant source of morbidity for a large percentage of the population (∼15%), and recalcitrance to antibiotic therapy often requires invasive surgical intervention (5). The recent advances in bacterial community profiling by 16S amplicon sequencing has revealed extensive colonization of CRS sinus mucus with oral anaerobes and other taxa not observed frequently via traditional culture-based methods (4, 6). While S. aureus is appreciated as a significant CRS pathogen, it coexists with these communities and must adapt to the sinonasal microenvironment shaped by both host and microbial processes. Many of the anaerobes associated with CRS release nutrients in the form sialic acid and other carbohydrates, peptides and amino acids, and by-products of mixed-acid fermentation (6). One such class of metabolites, short-chain fatty acids, are derived from amino acid-fermenting anaerobes, particularly members of the Fusobacterium genus (24). Our goals in this work were to (i) determine if Fusobacterium nucleatum, a model member of the Fusobacteria, could impair S. aureus growth; (ii) characterize the response of S. aureus to individual SCFAs; and (iii) identify potential mechanisms of SCFA stress.

We show that F. nucleatum produces millimolar amounts of the SCFAs acetate, propionate, and butyrate and that S. aureus has an extended lag phase and significant alterations to its transcriptome when grown in F. nucleatum supernatants or control medium supplemented with SCFAs. Consistent with a prior study, propionate and butyrate were both inhibitory to S. aureus growth, although we found that butyrate was more potent in this regard (21). Both SCFAs were sufficient to reduce the expression of the master regulator of virulence agrA and alter expression of metabolic pathways (cidA and the fad operon), in support of the hypothesis that SCFAs were responsible for altered gene expression and the delayed lag phase in F. nucleatum supernatants. Reduced agr-regulated virulence factor output may alter the inflammatory tone of the CRS sinus environment, with the host response instead being directed toward members of the anaerobic community rather than S. aureus. In support of this idea, CRS patients have been reported to have circulating antibodies targeting Fusobacterium and Prevotella, of which members were enriched in the CRS sinus mucus in our previous study, and that they show a decline in these antibodies after successful antibiotic therapy (6, 25). Alternatively, butyrate produced by anaerobes in the CRS sinus environment may occasionally select for S. aureus codY mutants, whose growth inhibition would be relieved (26). Such mutants overproduce numerous virulence factors and, as such, may exacerbate the inflammatory response (27). Whatever the case, the recent development of robust animal models of CRS will facilitate our ability to test these hypotheses in vivo (28, 29).

While the mechanisms of action of propionate and butyrate on S. aureus are still unclear, our data suggest that they may act on different targets. Propionate-induced expression of the fad operon and a fadX mutant exhibited worse growth in its presence than did the wild-type strain. The fad operon annotation suggests a role in fatty acid degradation, although it is unlikely that it acts directly on propionate, as S. aureus showed no evidence of metabolizing it over time (Fig. 1C). Another possibility is that propionate induces lipid membrane stress and the Fad proteins may act to degrade or repair damaged lipid species. This idea is consistent with recent findings from human gut commensal Bacteroides members, where butyrate (rather than propionate) induced membrane stress in a species- and context-dependent manner (30). Butyryl-CoA levels were increased by acyl-CoA enzymatic activity, suggesting that other CoA-regulated enzymes could be starved of an essential cofactor, likely impairing several metabolic processes. While butyrate also induced the expression of fadX, we did not detect a mutant phenotype in LB supplemented with 100 mM butyrate, suggesting that the other members of the fad operon may compensate for the loss of fadX under these conditions or that they are less important for the response to butyrate. Interestingly, we readily obtained spontaneous butyrate resistance in the form of codY mutations, while we did not obtain resistance for propionate. Furthermore, butyrate-resistant codY mutants were as sensitive to propionate as the parent strain, suggesting that each SCFA may have unique mechanisms of action on S. aureus. CodY is an allosteric transcriptional regulator whose affinity for specific motifs in promoter regions is dictated by the levels of GTP and branched-chain amino acids in the cell (31). As it regulates hundreds of genes in S. aureus, determining which CodY target(s) are responsible for bypassing butyrate-mediated growth inhibition are beyond the scope of this work, although transposon screens in a codY mutant background may prove fruitful in this regard.

In summary, we have identified a possible mechanism by which anaerobic bacteria in polymicrobial airway infections may influence S. aureus growth and physiology via the activity of the short-chain fatty acids propionate and butyrate and have identified the fad operon and the CodY regulon as possible mechanisms of resistance, respectively. Our study has some limitations, as the experiments were performed in F. nucleatum supernatants in vitro and under defined medium conditions rather than in the context of intact anaerobic communities or two-species cocultures. Additionally, our experiments lack the potential contributions of the host, such as reactive oxygen and nitrogen species or cationic antimicrobial peptides (32, 33). Despite these limitations, the genetic approach taken here is informative and amenable to translation into animal models for further dissection of the effects of CRS bacterial communities on S. aureus pathogenesis.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Bacterial strains used throughout this study are shown in Table S1 in the supplemental material. Plasmids and primers used for mutagenesis and complementation can be found in Table S2 and S3, respectively, in the supplemental material. Staphylococcus aureus strains USA300 LAC and JE2 and fadX::tn transposon mutant (obtained from the Nebraska Transposon Mutant Library) were routinely cultured aerobically at 37°C on LB agar (IBI Scientific; IB49020) or with shaking at 220 rpm in LB broth and supplemented as needed with 10 μg/mL chloramphenicol (Cm; Teknova; C0325) (6, 34). S. aureus was also grown in cell-free supernatants (CFSs) of Fusobacterium nucleatum ATCC 25586 that had been cultured anaerobically for 48 h in BBL Brucella broth (BD; 2011088) supplemented with 250 μg/mL and 50 μg/mL of hemin and vitamin K (Hardy Diagnostics; Z237), respectively. To test the effects of specific SCFAs on S. aureus growth and gene expression, the sodium salts of acetate (Fisher Scientific; S209), propionate (Sigma; P1880), or butyrate (Sigma; 303410) were added at various concentrations to LB and then passed through a 0.22-μm polyethersulfone (PES) filter prior to use. Escherichia coli strain One Shot TOP10 (ThermoFisher Scientific; C404010) was used for cloning the fadX mutagenesis plasmid, while E. coli DC10B was used for plasmid passaging to prevent cytosine methylation to facilitate easier transfer to S. aureus. E. coli strains were grown routinely on and in LB with 20 μg/mL Cm or 100 μg/mL ampicillin (Amresco; 0339) as needed.

Growth curves.

Overnight cultures of wild-type S. aureus and various mutants were diluted 1:100 in sterile phosphate-buffered saline (PBS) and then 5 μL was added to 195 μL of growth medium per well in a 96-well microtiter dish. Plates were incubated at 37°C for 24 h in a BioTek Synergy H1 microplate reader for 24 h, with 5 s of orbital shaking performed prior to hourly OD₆₀₀ readings.

agr promoter activity in response to SCFAs.

To determine how individual SCFAs affect the expression of the agr quorum sensing system, S. aureus USA300 LAC was electroporated with pAH1 (kindly gifted by Alex Horswill). This construct has the P3 promoter of agr fused to mCherry, and thus, red fluorescence is a reporter for RNAIII expression (35). Overnight cultures were added to fresh LB or LB supplemented with 100 mM of each SCFA and were grown at 37°C with shaking at 220 rpm for 24 h. A total of 200 μL of each culture was added to an individual well of a 96-well plate, and the OD₆₀₀ values were recorded in a BioTek Synergy H1 microplate reader. Excitation and emission at 580 nm and 635 nm, respectively, were used to measure red fluorescence, which was subsequently normalized to OD₆₀₀ values.

Biofilm quantification.

The crystal violet staining of biofilm material was performed according to Merritt et al. (36). Briefly, overnight S. aureus LB cultures were centrifuged at 5,000 rpm for 5 min and washed once with sterile PBS. They were subcultured 1:100 into fresh media in a 96-well microtiter plate and incubated statically at 37°C for 48 h, after which time the OD₆₀₀ values were recorded in a BioTek Synergy H1 microplate reader. Plates were then inverted to remove the cultures, washed three times in water, and allowed to dry. The wells were stained with 0.1% (wt/vol) crystal violet for 15 min at room temperature. The crystal violet was removed, and the plates were washed a further three times in water and allowed to dry. The dye was solubilized with 30% acetic acid for 15 min, and then the absorbance at 560 nm was recorded. The OD₅₆₀ was normalized to the OD₆₀₀ for each well to generate the final values.

RNA extraction.

For S. aureus growth in anaerobe cell-free supernatants and in LB with or without SCFA supplementation, 2 mL of growth medium was inoculated 1:100 with S. aureus overnight LB cultures and grown at 37°C with shaking at 220 rpm. The growth of each culture was monitored until they reached an OD₆₀₀ of ∼0.2 to 0.3, after which they were centrifuged for 1 min at 14,000 rpm. Supernatants were discarded, and pellets were suspended in 50 μL of fresh LB supplemented with 20 μg/mL of lysostaphin (Sigma-Aldrich; L7386). These samples were then incubated in a 37°C water bath for 15 to 20 min or until the suspension cleared (no longer than 30 min). One milliliter of TRIzol reagent (ThermoFisher; 15596018) was added to the lysate, pipetted gently until mixed, and incubated at room temperature (RT) for 5 min. A total of 200 μL of chloroform (VWR; 0757) was added per tube, and samples were shaken vigorously for 15 s and then incubated at RT for 5 min. Phase separation was performed by centrifugation at 12,000 rpm for 15 min at 4°C. A total of ∼500 μL of the aqueous phase was removed and added to 500 μL of 95% ethanol (Decon Laboratories, Inc.; UN1170), vortexed for 5 s, and incubated at RT for 5 min. RNA was then isolated using the Zymo RNA clean and concentrator-5 kit according to the manufacturer’s instructions, including an on-column DNase I treatment.

NanoString analysis of S. aureus gene expression.

A custom NanoString probe set (Table S4) was designed to target transcripts for several key S. aureus virulence factors, metabolic genes, and global regulators. The probe set also included six housekeeping genes for normalization. DNase I-treated RNA from S. aureus grown in triplicate to an OD₆₀₀ of ∼0.2 to 0.3 in control medium (Brucella broth [BB]) or 48-h cell-free supernatants from F. nucleatum (CFS) was submitted to the University of Minnesota Genomics Center (UMGC) for hybridization to the custom probe set. Raw data were imported into the nSolver Advanced Analysis software package for normalization and differential gene expression analysis using default settings. Transcripts were considered differentially expressed if their levels changed by 2-fold and the Benjamini-Hochberg-adjusted P value was less than 0.05. The heatmap was constructed using the pheatmap package (v.1.0.12) in R (v.4.1.0) (37).

Reverse transcription-quantitative PCR.

A total of 1.5 μg of RNA was reverse transcribed using M-MuLV reverse transcriptase (New England BioLabs [NEB]; M0253L) following the manufacturer’s protocol. cDNA was diluted 1:15 in sterile water prior to use in qRT-PCR using SsoAdvanced Universal SYBR green supermix (BioRad; 1725271). PCR products for each gene being assayed (see Table S3 for primer sequences) were used to construct standard curves for quantification. To determine relative copy number, transcript levels were normalized to the housekeeping gene guanylate kinase (gmk), which was confirmed to be consistent across growth conditions.

Construction of a S. aureus ΔfadX deletion mutant.

Next, ∼500-bp sequences flanking the fadX gene (SAUSA300_0229) were amplified by PCR using a Q5 DNA polymerase (NEB; M0491L). For cloning purposes, the upstream amplicon contained a 5′ KpnI restriction site and the downstream amplicon contained a 3′ SacI site, while the internal ends contained complementary overhangs to facilitate overlap extension PCR to fuse the fragments together. The final product was an ∼1-kb fragment encoding the first 11 codons and the stop codon of fadX. The amplicon was digested with KpnI-HF (NEB; R3142S) and SacI-HF (NEB; R3156S) and cloned into the temperature-sensitive, counterselectable mutagenesis plasmid pIMAY with T4 DNA ligase (NEB; M0202) (38). The ΔfadX plasmid was transformed into E. coli DC10B and then electroporated (2,900 V, 25 μF, and 100 Ω) in a 2-mm cuvette into S. aureus. The culture was grown at the plasmid replication permissive temperature of 28°C with shaking for 4 h, after which time it was plated onto LB + 10 μg/mL Cm and incubated on the benchtop for 48 h. The single colony that was obtained was streaked onto two LB + Cm plates. One was incubated on the benchtop for 48 h to generate a freezer stock and the other was incubated at 37°C. Cm-resistant colonies were screened via PCR for chromosomal integration of the plasmid. Positive colonies were grown in LB in the absence of selection at 37°C for 24 h, subcultured 1:1,000 into fresh LB for another 24 h, then plated onto LB + 1 μg/mL anhydrotetracycline hydrochloride (Sigma-Aldrich; 37919) for counterselection via induction of the TetR-regulated secY antisense RNA, and incubated overnight at 37°C. The resultant colonies were patched onto fresh counterselection agar and LB + Cm to screen the for loss of the plasmid. Cm-sensitive clones were screened by PCR for the loss of fadX coding sequence, and four clones were confirmed as mutants by Sanger sequencing.

HPLC method for extraction and measurement of organic acids in complex media.

Reversed-phase high-performance liquid chromatography (HPLC) was used for the targeted quantification of acetate, butyrate, and propionate in cell-free supernatants. Organic acids of interest were purified from complex medium components through a modified liquid-liquid extraction method (39). To account for analyte loss during extraction, 100 μL of 0.2 M succinate was added as an internal standard to 2 mL of each sample (9.5 mM final concentration) (39). After equilibration at room temperature for 5 min, 200 μL of 12 N HCl was added, and samples were vortexed for 15 s. A total of 10 mL of diethyl ether was then added to each sample before gently rolling them for a total of 30 min. After centrifugation for 5 min at 4,000 rpm, supernatants were transferred to a new extraction tube, and 1 mL of 1 M NaOH was added before gently rolling for another 30 min. The resulting aqueous phase was extracted and transferred to an autosampler vial followed by addition of 100 μL of 12 N HCl before being vortexed and stored at 4°C until analysis.

Samples were analyzed using a Dionex UltiMate 3000 ultra-HPLC (UHPLC; Thermo Fisher) system equipped with a reversed-phase Acclaim organic acid (OA) column (5 μm, 120 Å, and 4.0 by 250 mm). A total of 8 μL of each sample was injected, and separation was achieved using a 32-min isocratic instrument method (1.0 mL/min and 30°C) employing Na₂SO₄ (100 mM and pH 2.6) as the mobile phase. The column was allowed to equilibrate for 8 min prior to sample injection. Absorbance was monitored at 210 nm to identify compounds with a carboxylic acid functional group. Chromatograms were processed using Dionex Chromeleon 7 (Thermo Fisher) chromatography data system. Cobra Wizard was used to reproducibly identify and gate peaks of interest.

Isolation and genome sequencing of butyrate-resistant mutants.

The ΔfadX mutant was grown for 24 h in LB, serially diluted, plated onto LB supplemented with 200 mM sodium butyrate, and incubated overnight at 37°C. Four large colonies were restreaked onto the same media to confirm their ability to grow in the presence of butyrate. Genomic DNA was then isolated from our S. aureus USA300 LAC isolate and butyrate-resistant mutants using the PowerSoil Pro kit (Qiagen; 47014). Overnight cultures were pelleted and suspended in 200 μL of an enzymatic lysis buffer (20 mM Tris-HCl, 2 mM EDTA, 1.2% [vol/vol] Triton X-100, and 20 μg/mL lysostaphin) for 30 min at 37°C. Lysates were transferred to Power Bead tubes, and the manufacturer’s protocol was followed with no further alterations. Genomic DNA was processed and sequenced on the Illumina MiSeq platform at the Microbial Genome Sequencing Center (MiGS; Pittsburgh, PA). Raw paired-end fastq files were imported into Geneious v.2022.0.1 and trimmed for quality using BBDuk with the following settings: Set ORDERED to true, k = 27, mink = 6, maskMiddle=true, hamming distance = 1, right-ktrimming using 1 reference, quality-trimming both ends to Q30. Trimmed reads for our USA300 isolate were mapped to the S. aureus subsp. aureus USA300_FPR3757 reference genome (GenBank accession no. CP000255) using the Geneious Mapper on medium-low sensitivity with a minimum mapping quality of 20 and only mapping reads whose pair mapped appropriately nearby. The assembled USA300 genome was then used as the reference against which reads from the butyrate-resistant mutants were mapped using the same quality trimming and mapping strategy detailed above. Putative single-nucleotide polymorphisms (SNPs) and indels were detected using the “Find Variations/SNPS…” function within Geneious, requiring the occurrence of the variation in >90% of reads, with a minimum of 10 reads.

Data availability.

Raw sequences were deposited at NCBI Sequence Read Archive (SRA) with the BioProject identifier (ID) PRJNA798706.