Acute Infection with a Tobramycin-Induced Small Colony Variant of Staphylococcus aureus Causes Increased Inflammation in the Cystic Fibrosis Rat Lung

ABSTRACT

Cystic fibrosis (CF) disease is characterized by lifelong infections with pathogens such as Staphylococcus aureus, leading to eventual respiratory failure. Small colony variants (SCVs) of S. aureus have been linked to worse clinical outcomes for people with CF. Current studies of SCV pathology in vivo are limited, and it remains unclear whether SCVs directly impact patient outcomes or are a result of late-stage CF disease. To investigate this, we generated a stable menadione-auxotrophic SCV strain by serially passaging a CF isolate of S. aureus with tobramycin, an aminoglycoside antibiotic commonly administered for coinfecting Pseudomonas aeruginosa. This SCV was tobramycin resistant and showed increased tolerance to the anti-staphylococcal combination therapy sulfamethoxazole-trimethoprim. To better understand the dynamics of SCV infections in vivo, we infected CF rats with this strain compared with its normal colony variant (NCV). Analysis of bacterial burden at 3 days postinfection indicated that NCVs and SCVs persisted equally well in the lungs, but SCV infections ultimately led to increased weight loss and neutrophilic inflammation. Additionally, cellular and histopathological analyses showed that in CF rats, SCV infections yielded a lower macrophage response. Overall, these findings indicate that SCV infections may directly contribute to lung disease progression in people with CF.

INTRODUCTION

In cystic fibrosis (CF), loss of appropriate cystic fibrosis transmembrane conductance regulator (CFTR) anion channel function leads to mucus stasis, chronic airway infections, and neutrophilic airway inflammation (1). Without healthy mucus transport, the CF airway is susceptible to chronic infection, hosting a wide variety of opportunistic pathogens. These infections result in the majority of morbidity and mortality in this population, as >85% will eventually succumb to respiratory failure from infection-associated damage (2).

Staphylococcus aureus is currently the most common microorganism isolated from the CF lung, traditionally appearing early in life and persisting throughout adulthood (3). In children, positive S. aureus cultures have been linked to increased airway inflammation and pulmonary damage (4–6). As patients age, infection with S. aureus remains frequent, but its pathogenic role has been historically harder to characterize (2). This may be due in part to the genetic diversity seen across infectious strains, as reports indicate a broad range of S. aureus variations exist within the CF population and multiple distinct genotypes can often be recovered from a single patient (7). Small colony variants (SCVs) of S. aureus have become an emerging concern for people with CF (pwCF). In vitro, SCVs are marked by slow growth rate and the presentation of colonies at least 10 times smaller than normal colony variants (NCVs) on solid media. S. aureus SCVs typically lack pigmentation, display increased antibiotic resistance, and have reduced hemolytic ability (8, 9). These phenotypic traits are also usually coupled with metabolic defects affecting either electron transport or thymidine production. As a consequence, SCVs are difficult to culture and diagnose in the clinic (10, 11). Documented cases of S. aureus SCV infection arise most often in older patients or those with prolonged antibiotic exposure. These infections have been associated with markers of heightened disease, including lower FEV₁% pred, lower partial pressure of arterial oxygen (PaO₂), lower weight, and higher exacerbation frequency (10, 12–15). These associations raise the question whether there is a causal relationship between SCV infection and observed lung function decline, warranting investigation to better understand the mechanisms of pathophysiology in pwCF colonized with S. aureus.

Previous studies of S. aureus SCVs have focused predominantly on in vitro experiments characterizing traits such as gene expression, phenotypic selection, and metabolic profile (16–20). Several studies have evaluated systemic and organ-specific SCV infections using laboratory strains of S. aureus with site-directed mutations in genes such as thyA, hemB, and menD to induce the SCV phenotype (21–24). Because SCVs isolates exhibit a broader range of mutations and expression profiles, this approach may not fully recapitulate characteristics of naturally occurring SCV infections (25, 26). Some SCV isolates have also been found to revert to an NCV phenotype upon subculture through the introduction of new mutations, which may not be replicated in genetic knockouts (8, 27). Additionally, interactions between S. aureus and other pathogens or therapeutics may contribute to SCV formation in the CF airway. Coinfection with Pseudomonas aeruginosa occurs frequently in pwCF infected with S. aureus, and coinfections have been associated with more severe disease outcomes (5, 28, 29). In pwCF coinfected with P. aeruginosa and S. aureus, studies have also reported higher S. aureus SCV infection frequency (10, 15). In vitro, exposure to either P. aeruginosa or tobramycin, a common anti-pseudomonal drug, can induce formation of S. aureus SCVs (16, 30). These findings suggest that exposure to P. aeruginosa or anti-pseudomonal agents could drive SCV formation in the CF airway, contributing to SCV-related disease.

We hypothesized that in the CF lung, infection with SCV S. aureus would lead to worse host outcomes. To assess this, we derived a SCV mutant by exposing a CF clinical isolate of methicillin-resistant S. aureus (MRSA) from a P. aeruginosa-coinfected patient to tobramycin, replicating an exposure that commonly occurs in the CF airway. We then used this strain to model SCV airway infection in the CFTR^−/− rat, which closely mirrors aspects of human disease, including mucus accumulation and functional microanatomy defects and is capable of supporting long-term infections (31–33). By comparing host response to NCV and SCV infections, our objective was to determine whether S. aureus SCVs directly induce more severe pathology in the lung.

RESULTS

SCV mutant SA0831^SCV was generated by serial passaging NCV S. aureus with the aminoglycoside antibiotic tobramycin.

To investigate the pathology of SCV infections in CF, we derived a S. aureus SCV strain from a clinical isolate. The CF isolate SA0831, obtained from a patient at the University of Alabama at Birmingham (UAB), was serially passaged on mannitol salt agar (MSA) plates supplemented with 50 μg/mL tobramycin until it achieved a consistent small colony phenotype, reached after 10 passages. A single colony was selected and cultured from the final passage, hereafter referred to as SA0831^SCV. Serial passaging of SA0831^SCV on MSA without tobramycin consistently yielded small colonies out to five or more passages, indicating phenotypic stability. Colonies of SA0831^SCV were first visible on MSA after approximately 48 h of growth, compared with ≤24 h for the NCV, with a significantly smaller colony diameter (Fig. 1A to C). This correlates with a >20× reduction in colony area (2.00 ± 0.37 mm² parent versus 0.08 ± 0.02 mm² mutant at 48 h). On Luria-Bertani (LB) agar, SA0831^SCV colonies are also less pigmented than the parent (Fig. 1D and E). To assess potential growth defects in liquid media, optical density was measured at 600 nm (OD₆₀₀) for subcultures of both strains in brain heart infusion (BHI) broth. The SA0831^SCV strain had a delayed time to exponential growth and grew to a lower final OD₆₀₀ compared with the SA0831 parent (Fig. 1F). As a final metric of SCV phenotype, we assessed hemolysis on Columbia blood agar. Though SA0831^SCV colonies were visible at 24 h on nutrient-rich blood agar compared with the required 48 h on MSA, it presented with less hemolytic activity than the SA0831 parent strain (Fig. 1G and H). When a single colony of each strain was used to inoculate a defined area on a blood agar plate, the surrounding zone of clearance was significantly smaller for the SCV at 48 h (3.50 ± 0.18 mm NCV versus 1.64 ± 0.11 mm SCV), confirming reduced hemolytic ability (Fig. 1I). Taken collectively, these traits indicate the successful generation of an SCV mutant.

FIG 1.

Repeat passaging of NCV S. aureus clinical isolate SA0831 with tobramycin yielded SCV mutant SA0831^SCV. Parent strain SA0831 (A) and its mutant SA0831^SCV (B) are shown at 48-h growth on MSA. Colony diameter from these plates were measured digitally (C). n = 4 to 10 colonies/plate. To highlight pigmentation differences, SA0831 (D) and SA0831^SCV (E) are shown at 48 h growth on LB agar. Growth defects in liquid culture were assessed by monitoring bacterial growth from 10⁶ initial CFU in BHI broth over the span of 12 h (F). n = 3 biological replicates/technical replicates. Representative images of hemolytic activity are shown at 24-h growth on Columbia blood agar for SA0831 (G) and SA0831^SCV (H). Hemolysis was quantified via zone of clearance surrounding a 415.3 mm² circular culture on Columbia blood agar at 48-h growth (I). ****, P < 0.001.

We next assessed potential nutrient auxotrophies for hemin, menadione, and thymidine using disc diffusion and liquid culture assays. Growth on MSA plates in the presence of menadione discs restored a normal growth phenotype at 24 h for the SA0831^SCV strain (Fig. 2A), while growth remained suppressed with hemin (Fig. 2B) and thymidine (Fig. 2C). Menadione supplementation, but not hemin or thymidine, also partially restored growth of the SA0831^SCV strain in liquid media, allowing the SCV strain to reach the same final OD₆₀₀ as the NCV parent strain (Fig. 2D). Together, these experiments establish SA0831^SCV as a menadione auxotroph.

FIG 2.

SA0831^SCV is a menadione auxotroph. SA0831^SCV was grown on MSA in the presence of menadione (A), hemin (B), or thymidine (C) disks to evaluate supplementation effect on colony formation rate. Growth rate was also evaluated for liquid cultures of SA0831^SCV grown in BHI supplemented with hemin, menadione, or thymidine as indicated (D). n = 3 biological replicates/technical replicates.

SA0831^SCV exhibits antibiotic resistance not seen in the parent strain.

As S. aureus SCVs are known to exhibit a broad range of antibiotic resistance phenotypes, we evaluated growth of the SA0831^SCV strain in the presence of several antibiotics. We first tested tobramycin, the compound used to drive SCV formation. Though SA0831 growth was inhibited by tobramycin (Fig. 3A, blue line), SA0831^SCV was resistant (Fig. 3B, blue line). We also investigated resistance to the commonly used anti-staphylococcal antibiotic trimethoprim-sulfamethoxazole (SXT). Both strains exhibited reduced growth in the presence of SXT at 240 μg/mL, and SA0831^SCV had decreased susceptibility compared to the SA0831 parent strain (Fig. 3A and B). As this suggested potential differences in SXT tolerance, growth of each strain in SXT was further evaluated at lower concentrations ranging between 1.975 and 240 μg/mL (Fig. 3C and D), and the final OD₆₀₀ reached at 24 h for each condition was normalized to the respective untreated control (Fig. 3E). Results of this assay showed that at concentrations at or below 3.75 μg/mL, the NCV parent grew to an equal or higher relative maximum OD₆₀₀ compared with the SCV. However, at higher concentrations of SXT (above 3.75 μg/mL), the SCV grew better. This suggests the SCV is better suited for growth in more concentrated SXT than the NCV parent strain.

FIG 3.

SA0831^SCV exhibits increased antibiotic resistance. S. aureus strain SA0831 (A) and its SCV mutant (B) were grown in BHI broth either alone, with 50 μg/mL tobramycin, or with 240 μg/mL SXT. 10⁶ CFU SA0831 (C) or SA0831^SCV (D) were grown in varying concentrations of SXT to evaluate differential response to drug concentration. Maximum OD₆₀₀ over 24 h was normalized against the untreated control for each concentration of SXT (E). SA0831 (F) and SA0831^SCV (G) were grown in BHI broth with 10 μM CFTR modulator ivacaftor to evaluate antibacterial effect. DMSO is used as vehicle control for SXT and ivacaftor. *, P < 0.05; **, P < 0.01. n = 3 biological replicates of 10 technical replicates for (A to B, F to G) and n = 4 biological replicates of three technical replicates for each concentration in (C to E).

Previous studies indicate that the CFTR potentiator ivacaftor exhibits anti-staphylococcal activity (34). Additionally, with the approval of new modulator therapies, approximately 90% of pwCF in the United States regularly receive ivacaftor, either alone or in combination with other CFTR modulators. Therefore, we evaluated S. aureus SCV and NCV growth in the presence of ivacaftor or dimethyl sulfoxide (DMSO) vehicle. We found that SA0831 and SA0831^SCV were equally susceptible to ivacaftor, as indicated by near complete suppression of growth over 12 h (Fig. 3F and G). These data matches previous reports for other S. aureus strains and indicates that ivacaftor susceptibility is not dependent on colony phenotype.

Tobramycin passage led to genetic changes in SA0831^SCV.

To determine genetic changes arising from the tobramycin passage, we performed whole-genome sequencing with variant calling between the SA0831 and SA0831^SCV strains. Analysis of the SA0831 parent strain genome via the PATRIC platform predicted this isolate is resistant to erythromycin, methicillin and penicillin, and susceptible to ciprofloxacin, clindamycin, gentamicin, tetracycline, and trimethoprim-sulfamethoxazole. This confirms earlier results showing sensitivity to SXT (Fig. 3) and prior clinical laboratory tests indicating resistance to oxacillin. Based on sequence analysis, the multilocus sequence type (MLST) of SA0831 matched to sequence type 8 (ST8). Results of variant calling analyses indicated single point mutations in three predicted genes and five intergenic regions of SA0831^SCV, as well as a 23-gene deletion beginning at position 2,058,467 on the bacterial chromosome (Table 1). These changes include a point mutation in the gene identified as aroA_1, reported to encode protein AroA(G). Further analysis via the Basic Local Alignment Search Tool (BLAST) of UniProt (35) indicates this gene is a 100% identical match to S. aureus bifunctional 3-deoxy-7-phosphoheptulonate (DAHP) synthase/chorismate mutase, with reported gene names, including aroA_1, aroA_2, and aroF. DAHP synthase has a known role as the first enzyme in the chorismate biosynthetic pathway (Fig. 4). Chorismate is required for the synthesis of menaquinone (36), the naturally occurring vitamin K analog of the synthetic compound menadione. Mutations in the men genes belonging to the menaquinone synthesis pathway have been identified in menadione-auxotrophic SCV isolates (25, 37) and accordingly, genes, including menD are often the focus of in vitro studies of SCVs (22, 38, 39). A mutation in the aro genes of the upstream chorismate synthesis pathway would likely contribute to menaquinone deficiency in the same manner. Notably, an aminoglycoside-driven SCV menadione auxotroph of S. aureus with a causal mutation in aroD has previously been described (40).

TABLE 1.

Summary of genetic changes in SA0831^SCV compared with the parent SA0831 strain, as determined by breseq

Positiona	Mutation	Annotationb	Genec (direction)	Description
254,296	C → A	A209E (GCA → GAA)	aroA_1 →	Protein AroA(G)
718,912	G → A	P103L (CCA → CTA)	alsT ←	Amino-acid carrier protein AlsT
1,513,343	G → A	A580V (GCC → GTC)	fusA ←	Elongation factor G
2,039,568	C → T	Intergenic (−318/+298)	---- ←/← ----	Monoacylglycerol lipase/hypothetical protein
2,039,590	C → T	Intergenic (−340/+276)	---- ←/← ----	Monoacylglycerol lipase/hypothetical protein
2,039,602	G → A	Intergenic (−352/+264)	---- ←/← ----	Monoacylglycerol lipase/hypothetical protein
2,039,647	G → A	Intergenic (−397/+219)	---- ←/← ----	Monoacylglycerol lipase/hypothetical protein
2,039,821	G → A	Intergenic (−571/+45)	---- ←/← ----	Monoacylglycerol lipase/hypothetical protein
2,058,467	Δ23,686 bp	Deletion	---- - rlmH	23 gene deletion, including 18 unidentified genes and gloB_2, mecR1, mecA_2, ugpQ, and rlmH

^aPosition indicates the genomic sequence location where the mutation begins.

^bMutations in intergenic regions are reported with two relative positions, reporting the distance upstream (+) or downstream (−) of the two neighboring genes.

^cDashed line indicates an unidentified gene.

FIG 4.

Schematic diagram of the chorismate and menaquinone biosynthetic pathways in S. aureus. Callout indicates the location of the aroA_1 (aroA_2/aroF) gene, mutated in the SA0831^SCV strain. Pathways are modified from the Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathways, https://www.genome.jp/kegg/pathway.html. References used are ubiquinone and other terpenoid-quinone biosynthesis, map00130; biosynthesis of amino acids, map01230.

Our analysis also identified a mutation in the fusA gene, encoding the elongation factor G (EF-G) protein. Exposure of S. aureus to aminoglycosides has previously been seen to cause mutations in fusA, leading to increased aminoglycoside resistance. A subset of fusA mutants also display the SCV phenotype but are typically hemin auxotrophs (41, 42). Notably, a fusA mutant auxotrophic for menadione was previously reported, but further analysis indicated this auxotrophy was likely driven by a secondary mutation in menE (42). Because SA0831^SCV is only auxotrophic for menadione, the fusA mutation in this strain likely contributes to its observed tobramycin resistance (Fig. 3), but not to the SCV phenotype.

We identified an additional mutation in the gene alsT, reported to encode an amino acid carrier protein. Recent work has established that AlsT is the main glutamine transporter in S. aureus (43). Analyses of gene expression in menadione auxotrophic menD mutants showed altered metabolic gene expression, including upregulation of genes associated with glycolysis, fermentation, and anaerobic respiration and downregulation of genes associated with the tricarboxylic acid (TCA) cycle (39). Accordingly, altered glutamine uptake through mutated AlsT may help regulate amino acid metabolism or osmotic stress in SA0831^SCV.

SA0831^SCV also contains an ~24-kbp deletion leading to the loss of 23 genes. Reported genes located in this region include mecR1 and mecA_2. Loss of these genes indicates the loss of a portion or entirety of the methicillin-resistance cassette SCCmec, which is typically 20 to 60-kb in size (44). In an environment without selective pressure from β-lactam antibiotics, loss of SCCmec may be favorable for S. aureus, as other studies have shown that in vitro, strains with missing SCCmec or inactivated mecA outcompete isogenic S. aureus with intact SCCmec, implicating energetic cost (45, 46). With regard to SA0831^SCV, it is possible that the fitness cost of mecA is unfavorable due to its already stunted growth profile, resulting in the loss of SCCmec genes from the genome.

SA0831^SCV caused more severe acute infection than NCV S. aureus.

To evaluate host response to acute SCV infection, wild type (WT) and CFTR^−/− (CF) rats were intratracheally inoculated with 10⁹ CFU of SA0831 or SA0831^SCV and sacrificed after 3 days. Upon sacrifice, a similar bacterial burden of S. aureus, regardless of phenotype, was observed from lung homogenate of NCV- and SCV-infected animals, indicating no difference in persistence in the lungs (Fig. 5A). Phenotypic analysis of colonies recovered from a representative subset of these animals showed no spontaneous generation of SCV mutants in NCV-infected rats. Out of four SCV-infected rats, a heterogenous population including SCVs and NCVs was recovered from one animal, while exclusively NCVs were recovered from the other three animals evaluated (Fig. 5B). To assess overall disease severity, rats were weighed daily. For both WT and CF animals, SCV infection led to increased weight loss compared with NCV infection, indicating worse clinical outcomes (Fig. 5C and D).

FIG 5.

At 3 days postinfection, SA0831 and SA0831SCV are equally retained in the lung despite differing host response. WT and CF rats >6 months of age were infected intratracheally with 10⁹ CFU SA0831 or SA0831^SCV and sacrificed after 3 days. Bacterial CFU in the lung were determined by plating homogenate on MSA (A). Percent of animals returning SCVs after sacrifice was determined by the presence of small colonies from homogenate with reduced hemolytic activity upon subculture (B). n = 3 to 4 animals/group. Disease severity was measured for WT (C) and CF (D) rats via weight loss and recovery over time. n = 21 to 25 animals/group. P represents overall effect of group. *, P < 0.05; **, P < 0.01.

SA0831^SCV caused increased lung inflammation.

To assess the effects of SCV infection on host outcome, lungs were sectioned and stained with hematoxylin and eosin (H&E) to visualize inflammation. Under this method, dense patches of purple staining in the lung tissue indicate areas of immune cell infiltrate. Though general inflammation is apparent for each experimental group, these concentrated areas of immune cell infiltration are more pronounced in SCV-infected animals (Fig. 6A). Greater magnification shows NCVs provoke an immune response that is dominated by macrophages in CF, but not WT, rats (Fig. 6A, inset). However, when CF rats are infected with SA0831^SCV, the dominant immune cell in the lung switches to neutrophils. Quantification of immune cells in the bronchoalveolar lavage fluid (BALF) indicated that in both host backgrounds, total immune cell counts were not different for NCV- and SCV-infected animals (Fig. 6B and C). These analyses also showed that compared with NCV infections, SCV infections correlated with increased neutrophils in the BALF for both WT and CF (Fig. 6D and E). Additionally, macrophages in the BALF of CF rats were decreased in SCV infections compared with NCV infections in this background (Fig. 6F and G). We investigated potential causes of inflammation by measuring levels of circulating pro- and anti-inflammatory cytokines in the BALF, including TNF-α (Fig. 7A and B), IL-1β (Fig. 7C and D), IL-6 (Fig. 7E and F), and IL-10 (Fig. 7G and H). We observed a decrease in IL-6 for SCV-infected CF rats compared with NCV-infected CF rats (Fig. 7F). In WT animals, IL-10 was increased in SCV infection compared with NCV (Fig. 7G).

FIG 6.

Infection with SCV S. aureus alters acute disease pathology in WT and CF rats. WT and CF rats >6 months of age were infected intratracheally with 10⁹ CFU of SA0831 or SA0831^SCV and sacrificed after 3 days. Left lobes were paraffin-embedded and stained with H&E (A). Scale bar represents 200 μm. Images are representative of n = 7 to 9 animals/group. Immune cells were collected from the lungs via bronchoalveolar lavage and evaluated by differential staining for overall cell counts (B to C), % neutrophils (D to E), and % macrophages (F to G). *, P < 0.05; **, P < 0.01.

FIG 7.

Disease severity is independent of common anti- and proinflammatory cytokines. WT and CF rats >6 months of age were infected intratracheally with 10⁹ CFU SA0831 or SA0831^SCV and sacrificed after 3 days. Molecular lung contents were collected by bronchoalveolar lavage, and total TNF-α (A to B), IL-1β (C to D), IL-6 (E to F), and IL-10 (G to H) content was measured via ELISA. Values are relative to total protein content of the BAL as determined by BCA protein assay. *, P < 0.05.

DISCUSSION

In the CF lung, S. aureus SCVs are associated with advanced disease, lung function decline, and coinfection with pathogens such as P. aeruginosa. In this study, we used the common anti-pseudomonal drug tobramycin to derive the S. aureus SCV strain SA0831^SCV from an NCV clinical isolate. To examine the role SCVs play in CF disease progression and determine whether there is a causal relationship between SCV infection and worse clinical outcomes, SA0831^SCV was used to model respiratory infection in WT and CF rats.

In the experiments presented in this study, we used a clinical isolate of S. aureus to investigate characteristics of SCV infections in a CF context, and directly derived our SCV mutant from this isolate. SA0831^SCV was established by repeated in vitro exposure to the aminoglycoside antibiotic tobramycin, a known stimulus for SCV formation (16), in contrast to previous approaches which have primarily used genetic knockouts such as ΔthyA or interruptions to genes such as menD and hemB in laboratory strains (21–24, 47). Though these targeted mutations produce colonies with an SCV phenotype, they may not fully capture the genetic variability of SCVs seen in the clinic, which present with a broad range of mutations (25, 26). SA0831^SCV was generated without direct genetic manipulation and exhibits hallmark SCV phenotypic traits, including smaller colony size, slower growth rate, altered pigmentation, reduced hemolytic activity, and nutrient auxotrophy for the electron transport component menadione. Comparisons between the SA0831 and SA0831^SCV genomes indicated changes in several genes after tobramycin passage, with the most likely cause of the SCV phenotype being a point mutation in aroA_1 (also called aroA_2 or aroF). The bifunctional DAHP synthase/chorismate mutase protein encoded by this gene plays an essential role in the first step of the biosynthetic pathway for chorismate, a branch point metabolite required for menaquinone biosynthesis (36). Because the aro genes involved in chorismate biosynthesis are upstream of the men genes involved in the latter pathway, SCVs with aro mutations may recapitulate hallmark SCV traits such as menadione auxotrophy while also displaying defects in the biosynthesis of additional metabolites. Mutations in aroA and aroD genes have been associated with SCV phenotypes in Salmonella enterica serovar Typhimurium, and these mutants were also defective in chorismate production and respiratory metabolism (48). Interestingly, a prior study found a S. aureus aroA transposon insertion mutant was attenuated and showed decreased persistence in a mouse model (49). This aroA insertion mutant had a low reversion frequency, and it is possible the ability of the SA0831^SCV strain to revert to a NCV phenotype was key to its heightened virulence and ability to persist in the rat lung. Though these findings are preliminary and will be validated with rigorous genetic assays moving forward, they may highlight the aro family as an important focus for future SCV investigations. Additional mutations, including changes to fusA and alsT and the deletion of full or partial SCCmec, may also contribute to SA0831^SCV survival fitness, but require additional investigation.

SCVs generated without genetic intervention have been historically challenging to characterize, as clinical and laboratory isolates are often prone to spontaneous reversion to the NCV phenotype at highly variable rates, caused by mutation back to WT genes or the generation of new point mutations (27, 50). Conversely, SCV strains generated using full gene knockouts do not allow for the study of these mutations. Repeated culture of the SA0831^SCV strain on MSA did not lead to phenotypic reversion, indicating genetic stability in vitro. We were also able to retrieve colonies with the SCV phenotype in 25% of animals infected with SA0831^SCV in CF or WT infections, while the remaining 75% yielded only NCV colonies. This demonstrates that SA0831^SCV has both an ability to persist as an SCV during in vivo passage, and the capability to revert to NCV phenotype similar to what has been seen for other S. aureus isolates from animal models (50). Regardless of the recovered bacterial phenotype, animals infected with SA0831^SCV returned the same bacterial burden from lung homogenate as those infected with NCV SA0831, indicating this strain is equally suited to survival in vivo. Collectively, these traits make SA0831^SCV an appropriate candidate for modeling in vivo SCV infection in the context of CF.

Prior research on SCVs in the CF lung has primarily focused on variants driven by exposure to SXT, an anti-staphylococcal antibiotic which is commonly administered to pwCF colonized with S. aureus (17, 51, 52). Previous reports indicate a high prevalence of thymidine-auxotrophic S. aureus SCVs recovered from patients with a history of SXT treatment (53, 54). However, SCVs driven by exposure to other antibiotics commonly used in pwCF have not been thoroughly investigated. Aminoglycoside exposure has been shown to drive the formation of S. aureus SCVs auxotrophic for hemin or menadione both in vitro and in vivo (16, 30, 55). Aminoglycoside-driven SCVs have been studied in-depth in osteomyelitis and device-associated infections, and their use has been previously tied to SCV colonization in CF (8, 54, 56). In the clinic, the aminoglycoside tobramycin is administered to patients infected with P. aeruginosa, commonly found in coinfections with S. aureus in the CF lung. Therefore, it is important to also consider the role of aminoglycoside-driven S. aureus SCVs in CF disease. Here, we used tobramycin to generate a S. aureus SCV strain from a clinical isolate collected from a person with CF who was also coinfected with P. aeruginosa. In addition to being resistant to tobramycin, SA0831^SCV showed increased tolerance to SXT at higher drug concentrations. In our genetic analyses, SA0831^SCV was also found to carry a mutation in fusA, which has been shown to cause fusidic acid (FA) resistance in other S. aureus strains (41, 42). This implies that in the clinic, treatment administered for other pathogens coinfecting the airway with S. aureus may result in the formation of S. aureus variants better suited to survive anti-staphylococcal antibiotics. As overcoming antibiotic resistance is one of the most difficult challenges for treating SCVs in the clinic, these results indicate that focusing on other SCV types in addition to SXT-driven variants alone may be important to improve our understanding of SCV biology (8).

Animals infected with SA0831^SCV experienced more severe disease, as evidenced by increased weight loss and higher levels of neutrophilic inflammation compared with their NCV-infected counterparts. In previous studies, results of SCV virulence in vivo have been mixed, with reports of increased or decreased disease severity with SCV infection depending on model and infectious strain (21, 23, 55). However, this is the first time S. aureus SCV infection has been modeled in the rat respiratory system or in an animal model of CF. The rat respiratory anatomy is similar to that of humans, and the rat airway develops extensive submucosal glands, facilitating mucociliary transport and clearance (57). Additionally, the CFTR^−/− rat model accurately replicates clinical CF pathology, including defective airway microanatomy and aberrant mucus transport (31, 32). Because of these characteristics, this model may better capture the effects of S. aureus airway infection. In WT and CF rats, SCV infections caused increased neutrophil levels in the BALF and lung tissue. Interestingly, proinflammatory cytokines TNF-α, IL-1β, and IL-6 were not increased for these infections, and in WT animals, SCV infection was accompanied by an increase in the anti-inflammatory cytokine IL-10. These results indicate that neutrophilic inflammation in these animals is not driven by common inflammatory cytokines and may instead rely on a different mediator of inflammation. There was also a distinct population of macrophages present in lung tissue from NCV-infected CF rats that is not seen in SCV-infected CF rats. Macrophages from CF mice and humans are hyper-responsive toward microbial stimuli (58), yet are less effective at bacterial eradication. Therefore, the switch from macrophages to neutrophils as the primary innate immune responder in the CF lungs is interesting. Whether this is a defect in the host response to SCVs, a result of differing time to inflammatory resolution, or a result of a specific bacterial change in toxicity is unclear, and will be the focus of future investigations. The decreased macrophage response would also explain the reduced concentration of IL-6 in the BALF of SCV-infected CF rats compared to that of NCV-infected rats.

This study investigates SCV infection outcomes at a 3-day acute infection time point. Though an acute model of infection is a practical first step to understanding SCV pathology in vivo, future studies investigating these outcomes at chronic time points will also be necessary to fully capture infection dynamics affecting pwCF, as CF is characterized by primarily chronic infections (2). Though it is possible that SCV pathology is indistinguishable between WT and CF animals, extending these studies to chronic time points may also reveal differences arising during late-stage infection. This work also examines the pathology of a single SCV strain derived from one S. aureus CF clinical isolate. Examining SCVs derived from diverse clinical isolates in the same manner could provide insight on potential differences in virulence and mutations associated with SCV phenotypes. For instance, this study focused on SCVs arising from tobramycin exposure, which is a common clinical scenario for pwCF. However, many SCVs recovered from CF lungs are thymidine auxotrophs (10, 14, 15). Future studies will compare the menadione-dependent SCV with thymidine-dependent SCVs to determine how nutrient auxotrophies influence disease in pwCF. Despite these limitations, these studies provide a basis for understanding how SCV infection impacts CF disease. Results of this work show that an SCV created by exposure of a CF clinical isolate to tobramycin, an aminoglycoside that S. aureus may encounter in the coinfected CF airway, caused more severe disease in the CF rat. These findings are consistent with worse clinical outcomes reported for patients with positive SCV cultures, highlighting the importance of investigating SCVs and their role in CF lung disease progression.

MATERIALS AND METHODS

Bacterial strains and isolates used.

The MRSA strain SA0831 was collected from an adult inpatient with cystic fibrosis at UAB, chronically infected with S. aureus and P. aeruginosa. Initial culture on MSA (MilliporeSigma, Burlington, MA) was followed by antibiotic resistance testing against oxacillin and vancomycin discs (Becton, Dickinson and Company, Franklin Lakes, NJ) using the standard antibiotic disk diffusion assay, determining that this strain is methicillin-resistant and vancomycin sensitive.

Growth and maintenance of bacteria.

S. aureus strains SA0831 or SA0831^SCV were inoculated onto MSA or MSA + 50 μg/mL tobramycin (Sigma-Aldrich, St. Louis, MO), respectively, and incubated overnight at 37°C. Liquid cultures were grown at 37°C with 250 rpm shaking, either overnight in BHI broth (MilliporeSigma, Burlington, MA) for SA0831 or for 2 days in BHI supplemented with 50 μg/mL tobramycin at 37°C with 250 rpm shaking for the slower-growing SA0831^SCV. Prior to use in each assay, SA0831^SCV cultures were washed with phosphate-buffered saline (PBS; Thermo Fisher Scientific, Waltham, MA) and resuspended in BHI without tobramycin. CFU were determined by serially diluting liquid culture with BHI and plating on MSA. To confirm bacteria used in each assay had a uniform phenotype, aliquots of inocula were plated on MSA and monitored visually.

Generation of a S. aureus SCV strain.

An overnight culture of SA0831 was grown in LB broth (Sigma-Aldrich, St. Louis, MO) and 100 μL was spread onto MSA containing 50 μg/mL tobramycin. After 48 h growth at 37°C, a single small colony was chosen and inoculated onto MSA with 50 μg/mL tobramycin. This process was repeated with subsequent single colonies passaged on MSA with tobramycin every 48 h to 72 h until all colonies were visibly uniform and sampled colonies maintained SCV phenotype out to five passages on MSA not supplemented with tobramycin.

Phenotypic analysis.

To measure colony size, S. aureus strains were struck onto MSA to obtain isolated colonies and photographed at 48 h growth at 37°C. Diameters of single colonies in pixels were converted to diameters in mm with the ImageJ-based image processing software Fiji (59), using plate diameter as reference. Assessments of hemolysis were made on Columbia blood agar with 5% sheep blood (Thermo Fisher Scientific, Waltham, MA). To quantify hemolytic activity, a single bacterial colony was inoculated in a 415.3-mm² circular area, and distance was measured digitally from edge of culture to edge of clearance following 48 h growth at 37°C. To assess coloration differences during growth, S. aureus strains were inoculated onto LB agar (Supelco, Inc., Bellefonte, PA) and photographed 48 h following incubation at 37°C. Auxotrophy was determined by a paper disk method modified from the method reported by Kahl et al. (54) Briefly, auxotrophy for menadione and thymidine were tested using paper discs impregnated with 15 μL of 10 μg/mL menadione (Supelco, Inc., Bellefonte, PA) or 100 μg/mL thymidine (Sigma-Aldrich, St. Louis, MO), and for hemin with standard paper disks (MilliporeSigma, Burlington, MA). SA0831^SCV was struck to cover the entire surface of a MSA plate, and discs were placed onto the center of the plate immediately following. Plates were incubated at 37°C and resulting growth observed at 24 h. Auxotrophy was confirmed by growth in a circular area surrounding the disc, matching that of the nonsupplemented parent strain. These data were complemented by additional growth testing in liquid culture supplemented with menadione, hemin, or thymidine, as described.

Genetic sequencing.

SA0831 and SA0831^SCV strains were grown as described above and treated with an enzymatic lysis buffer containing recombinant lysostaphin (Ambicin L, Ambi Products LLC, Lawrence, NY) prior to genomic DNA extraction using the DNeasy blood and tissue kit spin-column protocol (Qiagen, Hilden, Germany). Gene sequencing was performed by the microbial genome sequencing center (MiGS; Pittsburgh, PA) using a combination of long and short reads. Short-read sequencing was performed with an Illumina NextSeq2000 to generate 151 bp paired-end reads with ~4,000,000 reads per sample. Long-read sequencing was conducted using Oxford Nanopore Technologies (ONT) sequencing, with ~300,000 long reads per sample, with an average length of ~5 kb each. Quality control and adapter trimming was performed by MiGS using bcl2fastq (v. 2.20.0.445) (60) for Illumina sequencing and porechop (v. 0.2.3_seqan2.1.1) (61) for ONT sequencing prior to genome assembly and annotation. Hybrid assembly of Illumina and ONT reads was performed by MiGS using the Unicycler package (v. 0.4.8) (62), with statistics recorded with QUAST (63). Assembly annotation was performed by MiGS using Prokka (64). Variant calling between genomes was performed by MiGS using breseq (v. 0.35.4) (65) to align and compare sequencing data. Further sequence analysis was performed using the PATRIC platform (66), and multilocus sequence type was determined using the MLST app in OmicsBox (version 2.1.14) (67).

Bacterial growth curves and antibiotic exposures.

To generate growth curves, liquid cultures of S. aureus strains were washed with PBS, aliquoted on a 96-well plate to a final volume in growth media of 200 μL, and incubated at 37°C with continuous shaking for 24 h with absorbance readings at OD₆₀₀ every 20 min. Inoculum contained approximately 10⁶ CFU of bacteria per well. CFU were confirmed by plating 1:10 serial dilutions of inoculum on MSA and manually counting colonies after incubation at 37°C for up to 3 days. Growth in the liquid assay was measured in BHI alone or BHI supplemented with 50 μg/mL tobramycin, 0.25 μg/mL hemin (Sigma-Aldrich, St. Louis, MO), 2.0 μg/mL menadione, 100 μg/mL thymidine, 240 μg/mL SXT (sulfamethoxazole: Supelco, Inc., Bellefonte, PA; trimethoprim: Sigma-Aldrich, St. Louis, MO) or 10 μM ivacaftor (Vertex Pharmaceuticals, Boston, MA), as indicated. Growth was also measured in BHI supplemented with SXT at 2× dilutions ranging from 1.875 to 240 μg/mL under the same growth conditions. Compounds insoluble in water were reconstituted in DMSO (Sigma-Aldrich, St. Louis, MO) prior to addition to BHI, with final DMSO concentrations to be no greater than 1% of the total media. In conditions when DMSO was needed, BHI with 1% DMSO was used as an additional control.

CF rat model.

All animal experiments at UAB were conducted in accordance with UAB Institutional Animal Care and Use Committee (IACUC) approved protocols. All animal experiments used Sprague-Dawley CFTR^−/− rats (31) (termed CF) or their littermate wild-type controls (termed WT). Heterozygote CFTR^+/− male and female rats were paired to generate WT and CF pups. Litters remained with lactating dams until weaning at 21 days of age. Animals were bred and housed in standard cages with a 12-h light/dark cycle in temperatures between 71°F to 75°F with ad libitum access to food and water. WT and CF rats of the same sex were cohoused from time of weaning to study conclusion. From weaning, rats were maintained on a standard rodent diet supplemented with DietGel 76A (Clear H₂O, Westbrook, ME) and water containing 50% Go-LYTLEY (Braintree Laboratories, Inc., Braintree, MA) to reduce mortality from gastrointestinal obstruction (31). Animals used in this study were ≥6 months of age to ensure full CF phenotype (32). Experimental groups were composed of even numbers of males and females.

Verification of infectious dose.

Cultures of SA0831 and SA0831^SCV, respectively, were grown as previously described in 30 mL liquid culture, washed twice with PBS, and resuspended to an OD₆₀₀ of 6.0 and OD₆₀₀ of 6.6 in PBS, respectively, to adjust the concentration to 10⁹ CFU in 300 μL for each strain. Appropriate concentration was confirmed by plating 1:10 serial dilutions on MSA and counting colonies after incubation at 37°C for up to 3 days. To confirm the reliability of the colony counting method to verify infectious dose, a DNA copy number analysis was also performed on cultures prepared to these specifications. Each 300 μL of prepared culture was treated with an enzymatic lysis buffer containing lysostaphin prior to genomic DNA extraction using the DNeasy blood and tissue kit spin-column protocol. Raw DNA products were diluted 1:2 with nuclease free water (Thermo Fisher Scientific, Waltham, MA, USA) and a quantitative PCR (qPCR) assay was performed using the QuantStudio 3 real-time PCR system (Applied Biosystems, Waltham, MA). TaqMan Fast Advanced Master Mix (Applied Biosystems, Waltham, MA) was used in combination with a PrimeTime Standard qPCR Assay (Integrated DNA Technologies, Coralville, IA) designed for the S. aureus gene nuc, which has been previously used to quantify S. aureus through similar assays (68). Assay primers and internal probe were designed using previously published sequences for a 269-bp fragment of the nuc gene (69). The resulting cycle threshold (Ct) values, defined as the number of cycles required for a fluorescent signal to exceed the background readout, were compared to quantify the number of nuc copies in each sample. Because the S. aureus genome contains a single copy of nuc, this comparison can be extended to the total amount of genome copies in each bacterial preparation. Each 300 μL dose of NCV and SCV was determined to contain the same number of genomic copies by this method.

S. aureus infection and tissue collection.

Overnight and 2-day cultures of SA0831 and SA0831^SCV, respectively, were grown as previously described in 30 mL liquid culture. Prior to inoculation, S. aureus was collected by centrifugation, washed with PBS, and resuspended to a 10⁹ CFU/300 μL dose in PBS. On day 0, rats were anesthetized by isoflurane and bacteria were instilled intratracheally. Inoculating CFU for all groups were confirmed by plating 1:10 serial dilutions onto MSA and manually counting colonies after incubation at 37°C for up to 3 days. Weight was monitored once daily over the study period. On day 3 postinfection, rats were euthanized via intraperitoneal injection of 500 μL pentobarbital sodium (390 mg/mL) followed by exsanguination via the hepatic portal vein. To collect BALF, tracheae were cannulated and lavaged with 5.0 mL cold PBS. BALF was kept on ice before cells were separated via centrifugation at 4°C. For tissue retrieval, the thoracic cavity was exposed, and blood perfused from the tissue by injecting 50 mL PBS into the heart. Tracheae and lungs were removed and immersion fixed in 10% phosphate-buffered formalin (Thermo Fisher Scientific, Waltham, MA, USA) and stored for no longer than 1 week at 4°C before preparation for histological analysis. In a representative subset of animals, after BALF retrieval, lung tissue was collected and homogenized in Ham’s F-12 nutrient mix (Thermo Fisher Scientific, Waltham, MA, USA). Homogenate was plated on MSA and incubated at 37°C to assess CFU/lung and recovered S. aureus phenotype.

Phenotypic characterization of recovered S. aureus lung isolates.

MSA plates of lung homogenate were prepared as previously described and evaluated from a representative subset of animals for S. aureus phenotype. Plates were observed for new colony formation at 24 h, 48 h, and 72 h incubation. Time to colony appearance and colony size were evaluated, and any colonies appearing after 48 h with a colony area ≥10× smaller than lab-cultured SA0831 or appearing after 72 h at any size were subcultured by streaking onto MSA. Subcultures were incubated for 48 h at 37°C, and individual colonies were further assessed for hemolytic ability on Columbia blood agar with 5% sheep blood as previously described. Subcultures were reported as SCVs if hemolytic phenotype matched lab-cultured SA0831^SCV. Colonies not meeting these criteria were considered to be NCVs.

Immune cell quantification and histology.

Immediately following sacrifice, cells recovered from BALF were resuspended in PBS and centrifuged onto glass microscope slides by cytospin. Slides were subsequently dried and stained with Diff-Quik stain (Siemens Medical Solutions, Inc., Malvern, PA). Total cell count, macrophages, and neutrophils were counted manually via microscopy. To evaluate tissue histology, lungs fixed in formalin were sectioned into lobes and transferred to the UAB Pathology Core Research Laboratory to be embedded in paraffin, sectioned onto glass microscope slides, and stained with H&E.

Cytokine analysis.

Cytokine concentrations in the BALF supernatant were analyzed by enzyme-linked immunoassay (ELISA). Total protein content was measured using a bicinchoninic acid (BCA) protein assay kit (Millipore Sigma, St. Louis, MO). Levels of IL-10, IL-1β, TNF-α, and IL-6 were evaluated by precoated sandwich ELISA kits, per the manufacturer’s protocol (R1000 Rat IL-10 Quantikine ELISA Kit: R&D systems, Minneapolis, MN; ab255730 Rat IL-1 beta SimpleStep ELISA Kit, ab100785 TNF-α Rat SimpleStep ELISA Kit, ab100772 Rat IL-6 ELISA Kit: Abcam, Cambridge, MA). Cytokine amounts were normalized to total protein content as pg cytokine/μg total protein.

Statistics.

Statistical analysis was performed with GraphPad Prism version 9.3.1 (GraphPad Software, San Diego, CA). Normality of groups was determined by Shapiro-Wilk test. Inferential statistics (mean, SD, and SEM) and significance between groups were computed using two-way ANOVA, one-way ANOVA, or Kruskal-Wallis test as appropriate, or unpaired t test or Mann-Whitney test for data with nonnormal distributions. ELISA data were cleaned for outlier wells using the ROUT method (Q = 1%). P < 0.05 were considered significant. Inferential statistics referenced in the text are presented as mean ± SD. Statistics incorporated within figures are presented as mean ± SEM.

Study approval.

Collection of bacteria from patient sputum samples was approved by the UAB Institutional Review Board (IRB-160720008). The animal experiments were approved by the UAB IACUC committee (IACUC-20532 and IACUC-20092).

Data availability.

BioSample metadata for SA0831 are available in the NCBI BioSample database (http://www.ncbi.nlm.nih.gov/biosample/) under accession number SAMN30633109.