Prevalence of the SigB-Deficient Phenotype among Clinical Staphylococcus aureus Isolates Linked to Bovine Mastitis

Phenotypic adaptation has been associated with persistent, therapy-resistant Staphylococcus aureus infections. Recently, we described within-host evolution towards a Sigma factor B (SigB)-deficient phenotype in a non-human host, a naturally infected dairy cow with chronic, persistent mastitis. However, to our knowledge, the prevalence of SigB deficiency among clinical S. aureus isolates remains unknown. In this study, we screened a collection of bovine mastitis isolates for phenotypic traits typical for SigB deficiency: decreased carotenoid pigmentation, increased proteolysis, secretion of α-hemolysin and exoproteins. Overall, 8 out of 77 (10.4%) isolates of our bovine mastitis collection exhibited the SigB-deficient phenotype. These isolates were assigned to various clonal complexes (CC8, CC9, CC97, CC151, CC3666). We further demonstrated a strong positive correlation between asp23-expression (a marker of SigB activity) and carotenoid pigmentation (r = 0.6359, p = 0.0008), underlining the role of pigmentation as a valuable predictor of the functional status of SigB. Sequencing of the sigB operon (mazEF-rsbUVW-sigB) indicated the phosphatase domain of the RsbU protein as a primary target of mutations leading to SigB deficiency. Indeed, by exchanging single nucleotides in rsbU, we could either induce SigB deficiency or restore the SigB phenotype, demonstrating the pivotal role of RsbU for SigB functionality. The data presented highlight the clinical relevance of SigB deficiency, and future studies are needed to exploit its role in staphylococcal infections.


Introduction
Chronic, persistent Staphylococcus aureus infections are difficult to treat with antibiotics and often recur. Although the role of S. aureus virulence factors in acute infections is well established, bacterial factors contributing to persistence are far less understood. The combined effects of insufficient host clearance mechanisms, bacterial immune evasion and ineffective antibiotic therapies are thought to allow the pathogen to survive for prolonged periods in the host [1]. In particular, intracellular survival by switching to the smallcolony variant (SCV) phenotype is suggested to be a common reservoir for persistent infections in humans and bovines [2,3]. This mechanism of intracellular survival and SCV formation requires the upregulation of Sigma factor B (SigB), a major stress regulator in S. aureus [4]. However, frequent isolation of SigB-deficient strains during S. aureus infections suggests an advantage of this phenotype in certain niches [5][6][7]. We recently hypothesized that strains lacking SigB expression might better adapt to the extracellular niche, enabling long-term persistence in chronic mastitis [8]. SigB deficiency is typically associated with high production of proteases and toxins [6,9], suggesting that in bovine (G122D) within the C-terminal phosphatase domain causes a loss in the RsbV-P-specific phosphatase activity of RsbU, rendering SigB dysfunctional. In the present study, we aimed to determine the prevalence of SigB deficiency among S. aureus bovine mastitis isolates, which is, to our knowledge, unknown so far. Here, we employed several phenotypic assays to search for isolates presenting the SigB-deficient phenotype and evaluated the functional status of SigB by asp23-transcriptional expression. To determine the genetic basis of SigB deficiency, we further constructed isogenic mutants introducing or reverting the SigB-deficient phenotype.

Asp23-Transcriptional Expression (RT-qPCR)
Since SigB positively controls asp23 transcription, we expected low levels of asp23 expression (reverse transcription-quantitative polymerase chain reaction, RT-qPCR) for all the SigB-deficient phenotypes (n = 8). We detected an average asp23 mRNA expression level relative to housekeeping genes of 69.8, ranging from 1.0 to 176.0, while the SigB-functional reference strain SH1000 had a relative expression of 266.0 (Supplementary Figure S2). Most relative asp23 mRNA expression levels were below 100, indicating minor SigB activity. Interestingly, relative expression levels between 100 and 200 were found for two isolates (ID17 and ID67), indicating moderate SigB activity.

Positive Correlation between asp23-Transcriptional Expression and Pigmentation
As asp23 gene transcription indicates the functional status of SigB [9], it was interesting to establish whether asp23-transcriptional expression (RT-qPCR) levels correlate with staphylococcal pigmentation (extracted pigments, AUC). Pearson correlation analysis revealed a strong positive correlation between asp23-expression and carotenoid pigmentation (r = 0.6359, p = 0.0008; Figure 3). Notably, many isolates (5/11) were not assigned to the SigB-deficient phenotype because no increased secretion of α-hemolysin and exoproteins could be detected, although they expressed low levels of asp23-mRNA and pigments and showed proteolytic activity.

Sequence Analysis of the sigB Operon and Construction of Mutants to Confirm the SigB-Deficient Phenotype
To search for the genetic basis for SigB dysfunction, we sequenced the sigB operon (mazEF-rsbUVW-sigB) of all isolates exhibiting the SigB-deficient phenotype (n = 8). We did not identify substitution mutations or insertion/deletion mutations creating prema-

Sequence Analysis of the sigB Operon and Construction of Mutants to Confirm the SigB-Deficient Phenotype
To search for the genetic basis for SigB dysfunction, we sequenced the sigB operon (mazEF-rsbUVW-sigB) of all isolates exhibiting the SigB-deficient phenotype (n = 8). We did not identify substitution mutations or insertion/deletion mutations creating premature stops of translation. However, we have determined several synonymous and nonsynonymous nucleotide substitutions by comparing the sequences of the sigB operon with known SigB-functional strains (Table 1). In particular, we noticed changes in the rsbU gene. We identified the SNP rsbU(G395A → S132N) in strain ID31 and rsbU(G431T → G144V) in strain ID56, which both were located within the phosphatase domain that has been shown to cause SigB dysfunction when inactivated [9]. For isolate ID18, we obtained three synonymous nucleotide substitutions in mazF, rsbW and sigB, and one SNP in the noncoding region between the genes rsbV and rsbU, the latter known to bind the feed-forward sigB-dependent P B promoter [13,14]. However, for the isolates ID17, ID37, ID48, ID54 and ID67, we could not track possible genetic differences in the sigB operon, although presenting the SigB-deficient phenotype.
We further constructed isogenic mutants introducing or reverting the SigB-deficient phenotype, serving as a proof-of-concept to determine the causal relationship of SigB deficiency. Indeed, the revertant ID56:rsbU(T431G) of the SigB-deficient isolate ID56 displayed increased pigmentation accompanied by increased asp23-transcriptional expression (RT-qPCR), both typical for higher SigB activity ( Figure 4A). In addition, exchanging the SNP from G to A at position 368 in rsbU in the SigB-functional strains-IN and the human reference strain Newman-induced the SigB-deficient phenotype as shown by abrogating pigment production and asp23 expression ( Figure 4B). Thus, we show in two examples that only a single nucleotide substitution within the phosphatase domain of rsbU caused the SigB-deficient phenotype.

Discussion
In this study, we report a prevalence of 10.4% (n = 77) of bovine mastitis isolate hibiting the SigB-deficient phenotype. To our knowledge, this is the first time tha prevalence of SigB deficiency among S. aureus isolates has been studied. Sequencing sigB operon (mazEF-rsbUVW-sigB) from each of the eight SigB-deficient isolates rev

Discussion
In this study, we report a prevalence of 10.4% (n = 77) of bovine mastitis isolates exhibiting the SigB-deficient phenotype. To our knowledge, this is the first time that the prevalence of SigB deficiency among S. aureus isolates has been studied. Sequencing of the sigB operon (mazEF-rsbUVW-sigB) from each of the eight SigB-deficient isolates revealed several single nucleotide substitutions potentially causing the SigB-deficient phenotype. We identified the phosphatase domain of the RsbU protein as a repeated target of mutations, as two SigB-deficient strains had non-synonymous SNPs in this part of the rsbU gene (G395A → S132N; G431T → G144V). Notably, we could confirm the causal relationship of the single SNP in rsbU(G431T), as well as the SNP in rsbU(G368A) from our previous study demonstrating within-host evolution towards the host-adapted, SigB-deficient variant (HA) [8]. Consequently, a loss in the RsbV-P-specific phosphatase activity of RsbU can be assumed, which is a crucial step in regulating SigB via a partner-switching mechanism, and activation of SigB is abolished. In line with this, we obtained reduced levels of asp23 expression and pigmentation for both SigB-deficient bovine isolates compared to their isogenic SigB-functional counterpart. Mutations in the rsbU gene causing SigB deficiency were known from human staphylococcal infections. Most prominently, the laboratory prototype strain NCTC8325 (RN1) was shown to harbour an 11-base-pair (bp) deletion and the high-protease-producing prototype strain V8 to include the insertion of a 1073 bp IS element in rsbU [5][6][7]. Moreover, an 18-bp deletion in rsbU was detected in an isolate derived from a chronic cystic fibrosis infection, and a stop codon (TGA) at AA252 was found in the clinical isolate KS26 [5][6][7]28]. Mutations in rsbU (IS256 insertion, early stop codon occurrence, substitutions A230T and A276D) were identified after exposing S. aureus human clinical isolates to phototoxic conditions (photoantimicrobial chemotherapy-PACT), suggesting a role of the RsbU-dependent SigB activity against reactive oxygen species (ROS) [29]. Our findings from bovine mastitis infections indicate that genetic variation in the rsbU gene is important independent of the genetic background or the host. However, in many isolates, we could not identify genetic features within the sigB operon that may cause the SigB-deficient phenotype. We assume that the SigB-deficient phenotype may be caused by many different underlying mutations. This requires care when inferring the causative genotype-phenotype relationship at the genetic and (post-)transcriptional level.
About 14% of the isolates from our mastitis strain collection were non-pigmented. However, studies on the prevalence of non-pigmented S. aureus are scarce. A survey conducted in the Shanghai region revealed that 41% (54/132) of isolates were non-pigmented, including clinical and food-related samples [30]. While pigment-producing strains can be considered SigB-functional, reduced pigmentation may be due to SigB deficiency or other factors. For example, knock-out mutations of crtNM genes in the STX-producing operon, of the two-gene regulatory operon-yjbIH and of genes that are part of the de novo purine biosynthesis pathway have been reported to reduce pigment production [30][31][32][33]. The strong positive correlation between asp23-expression (SigB activity) and pigmentation of isolates in our mastitis collection was therefore intriguing and suggested only minor SigB-independent regulatory perturbations on staphylococcal pigmentation. Based on the correlation analysis, it can therefore be assumed that many isolates that produce no amount, or intermediate amounts, of carotenoid pigments (approx. 40%) were likely SigBdysfunctional or at least SigB-compromised. However, it remains unclear why they do not show all the typical features of the SigB-deficient phenotype.
The co-occurrence of high production of proteases and a-hemolysin has been reported for many phylogenetically unrelated SigB-deficient strains [6,8,9]. Given the substantial proteolytic activity observed in the isolates in our bovine mastitis collection, we anticipate the secretion of proteases at a higher level associated with a slower transition towards the expression of toxins (including α-hemolysin). SarA is known to increase the expression of the agr/RNAIII system and to repress gene expression of proteases independently of agr [11]. SigB directly inhibits expression of the agr/RNAIII system but induces expression of sarA [24]. The lower SigB activity may help counterbalance the high expression of agr/RNAIII and decrease the expression of α-hemolysin in the specific context of these isolates. Moreover, low SarA levels increase the expression and secretion of proteases that were shown to degrade secreted α-hemolysin, possibly contributing to the loss of the α-hemolysin phenotype [34,35]. Furthermore, we recently did not detect secretion of proteases but strong α-hemolysin secretion of a SigB-deficient strain under growth-limiting conditions, indicative of differential regulation of the two virulence factors downstream of the sigB operon [10]. One could even speculate that the expression of proteases is evolutionarily favoured over toxins in bovine-adapted S. aureus, as milk proteins are readily accessible as a nutritional source in the mammary gland. Nevertheless, regulatory cross-talks might exist between the major regulators sigB, agr/RNAIII and sarA, which fine-tune S. aureus gene expression and secretion of proteolytic and cytotoxic virulence factors under certain niche-specific conditions. Thus, repression of SigB could better adapt the pathogen to the extracellular microenvironment. In contrast, SigB activation allows intracellular survival, and the continuous switch between the two lifestyles could then further promote long-lasting, persistent infections.
This work highlights the clinical relevance of SigB deficiency in the pathogenesis of S. aureus infections. To our knowledge, this is the first report evaluating the prevalence of SigB deficiency among S. aureus isolates. We demonstrate that carotenoid pigmentation correlates with asp23 expression, underpinning the role of pigmentation as a valuable predictor of the functional status of SigB. Further, we discuss that SigB activity might be fine-tuned rather than completely on or off, allowing an adaptive shift between intracellular and extracellular niches. However, future studies are needed to fully understand the nichespecific adaptation dynamics within the host to clarify the selection pressure acting on SigB during long-lasting, persistent infections.

Visualization and Determination of Carotenoid Pigmentation
To visualise S. aureus carotenoid pigmentation, one millilitre of overnight culture was precipitated by centrifugation (10,000× g for 1 min; Eppendorf, Hamburg, Germany) at the bottom of the tube. To quantify S. aureus carotenoid pigmentation (incl. STX and intermediate carotenoids), the absorbance of methanol-extracted pigments was measured using an adapted protocol [46]. Overnight cultures were inoculated in 5 mL TSB (Thermo Fisher Scientific, Oxoid, Hampshire, UK) in glass tubes and incubated at 37 • C while shaking at 120 rpm. After 18 h, OD 600 was measured (BioSpectrometer, Eppendorf, Hamburg, Germany), and samples were normalised to cell mass (8 mL at OD 600 = 1). Next, cultures were centrifuged at 10,000× g for 1 min to harvest the cells. The bacterial cell pellets were resuspended in 800 µL methanol (Carl Roth, Karlsruhe, Germany) and incubated at 55 • C for 3 min to extract the pigments. After removing the cell debris by centrifugation, the supernatant was transferred into new tubes, and 300 µL per sample was added in duplicates to a 96-well plate (Greiner Bio-One, Kremsmünster, Austria). The extracted pigments were quantified by measuring the OD using the SpectraMax M3 (MolecularDevices, San Jose, CA, USA) at an interval of 300 to 750 nm in 1 nm steps. Carotenoid pigmentation was calculated using the area under the curve (AUC) at an OD ranging from 390 to 520 nm, with the baseline adjusted as a line through the OD values at 390 and 520 nm (OriginPro 2022, OriginLab, Northampton, MA, USA) (Supplementary Figure S1). We calculated the cut-off value (AUCc) based on the average mean and standard deviation (SD) of the four SigB-deficient reference strains (SH1000∆sigB, 6850∆sigB, 8325-4 and HA), AUCc = average AUC + 3 x average SD (Supplementary Figure S1). The isolates were further divided into the following categories based on the calculated average SD value of the negative controls: (i) no pigmentation, AUC ≤ AUCc, (ii) intermediate pigmentation, AUC ≤ AUCc + 3 × SD, and (iii) full pigmentation, AUCc + 3 × SD < AUC (adapted from [47]). Each sample was measured in technical duplicates on at least two separate days.

Proteolytic Activity
Staphylococcus aureus proteolysis of non-and intermediate-pigmented (n = 30) mastitis isolates was conducted, as reported earlier [8]. Briefly, 1% skim milk broth was prepared using 10 g/L skim milk powder (Thermo Fisher Scientific) in MQ-water (Merck Millipore, Darmstadt, Germany). The solution was autoclaved for 5 min at 121 • C and stored at 4 • C until use. To assess proteolytic activity, 15 µL of a resuspended pellet derived from a 1 mL TSB overnight culture (5000× g for 2 min) was added to 3 mL 1% skim milk broth and cultivated at 37 • C without shaking. Results were recorded after 24 h.

Reverse-Transcription (RT)-qPCR
Isolates exhibiting the SigB-deficient phenotype were tested for asp23-expression (RT-qPCR) to assess the level of SigB activity [26]. Total RNA extraction and asp23-RT-qPCR were performed on bacterial strains grown at 37 • C in TSB (120 rpm and aerobe) according to Marbach et al., 2019 [8] using the indicated primers for asp23 [4] normalised to the geometric mean of three reference genes (rpoD, rho and dnaN). Relative expression levels were calculated using the REST method [49]. For experiments validating SNPs (Section 2.3.), strains were grown until OD 600 = 5 in three independent experiments with technical duplicates. For screening SigB activity (Sections 2.1.3 and 2.2), we used a sample of 24 isolates from our mastitis strain collection, exhibiting all pigmentation phenotypes, including the eight isolates presenting the SigB-deficient phenotype. Strains were grown until OD 600 = 3 in two independent experiments with technical duplicates. Mean relative expression and standard deviations were calculated from independently grown samples and their technical duplicates.

Genetic Manipulation
Genetic manipulation of S. aureus was carried out using the thermosensitive plasmid pIMAY [50] and engineered E. coli strains to simplify the transformation of S. aureus isolates [51]. For the introduction of rsbU(G368A) in various strains, the rsbU(G368A) allele was amplified using the chromosomal DNA of the host-adapted isolate (HA) and cloned into pIMAY. The recombinant plasmid was used to transform the initial isolate (IN) and S. aureus Newman (NM). Allelic replacement was used to exchange plasmid and chromosomal alleles [50]. Successful replacement was verified by Sanger sequencing. To exchange nucleotide rsbU:431T in strain ID57 to rsbU:431G, chromosomal DNA of ID57 was used, and two primer pairs (AB and CD) were used to amplify two fragments of rsbU. The desired point mutation was integrated into complementary primers B and C. The two fragments were fused by spliced extension overlap PCR (using primers A and D), and the created mutant allele was cloned into pIMAY. The recombinant plasmid was used to transform S. aureus ID57, followed by allelic replacement as described above. Oligonucleotides used for genetic manipulation in this study were summarized in Supplementary Table S1.

Statistics
Differences in relative asp23 expression were tested by an unpaired Student t-test (two-tailed) of log-transformed data. Minimum statistical significance was set to p < 0.05. A correlation analysis (Pearson r, correlation p-value) was performed on log-transformed relative asp23-expression levels and the area under the curve (AUC) of carotenoid pigmentation. GraphPad Prism software (San Diego, CA, USA) version 7.0 was used for all statistical comparisons and visualisations.  Data Availability Statement: The authors confirm that the data supporting the findings of this study are available within the article and its Supplementary Materials.