Staphylococcus aureus is an important human pathogen and a significant global health concern^{1, 2}. The most common interaction with its human host is as an asymptomatic coloniser; however, it frequently transitions to a pathogenic state with the ability to cause a wide range of diseases, ranging from relatively minor skin and soft tissue infections (SSTI) to more life-threatening incidents of endocarditis or bacteraemia^{1, 2}. S. aureus is notorious for producing a plethora of virulence factors ranging from pore forming toxins to various immune evasion strategies³. The toxicity of S. aureus is generally accepted as playing an important role during infection, with high toxicity isolates typically causing more severe symptoms and disease progression^{4, 5}. However, for the more invasive diseases such as bacteraemia and pneumonia, it has been shown that the causative isolates are often impaired in their toxin production and instead rely on alternative virulence approaches to cause disease, such as being able to better survive exposure to host immune defences^{6, 7}.

As the most severe type of infection caused by S. aureus, bacteraemia has a mortality rate of between 20 and 30%^{8, 9}, and despite infection control measures that have reduced the incidence of MRSA bacteraemia, the incidence of MSSA bacteraemia continues to increase in several countries, suggesting we have much to learn about this disease process¹⁰. Entry into the bloodstream is the first step in the development of bacteraemia and represents a major bottle neck for the bacteria, whether they seed from an infection elsewhere in the body, or take a more direct route through an intravenous device. As a heavily protected niche, the bloodstream contains numerous humoral immune features with potent anti-staphylococcal activity, including antimicrobial peptides (AMPs) and host defence fatty acids (HDFAs)^11–13. However, the fact that cases of S. aureus bacteraemia occur demonstrates that these do not represent an impregnable force and that the bacteria can adapt to resist these defensive features.

In previous work we have pioneered the application of genome wide association studies (GWAS) to characterise bacterial virulence, where it has proven to be a powerful approach to define complex regulatory pathways^{7, 15–19}. In related work we have also begun to characterise a new category of S. aureus genes we refer to as MALs (mortality associated loci) and found that many of these are involved in serum survival²⁰, with serum containing the bulk of the humoral immune response present in blood, but without the added complexity of the cellular components. Given the outstanding research questions surrounding the development of bacteraemia we performed a GWAS on a collection of 300 clinical isolates with respect to serum resistance. We found significant variability in how well individual isolates survive exposure to serum and identified seven novel effectors of this activity. Of particular note was the TcaA protein, which we have further characterised and demonstrate that it is a critical protein for the bacteria involved in the remodelling of the bacterial cell wall during the development of bacteraemia.

Survival of S. aureus upon exposure to human serum is a variable trait. The human bloodstream is a highly protected niche, and so to establish an infection in this environment the bacteria must evade many aspects of host immunity, such as the bacterial membrane damaging antimicrobial peptides (AMP) and host defence fatty acids (HDFAs) found in serum^{11–13, 21–24}. To examine the level of variability that exists in serum-susceptibility in natural populations of S. aureus, we exposed 300 clinical isolates to pooled human serum and quantified the proportion of each culture that survived. These 300 isolates represent the two major clones, as defined by multi-locus sequence typing: clonal complex 22 (CC22) and clonal complex 30 (CC30), that are responsible for the majority of MRSA cases in the UK and Ireland¹⁶. The isolates were collected from 300 individual cases of bacteraemia. This collection is a unique resource as each strain has been sequenced, extensively phenotyped, and we have clinical metadata for many of the patients the bacteria were isolated from¹⁶. We found there to be similar levels of variability in the ability of the bacteria to survive exposure to human serum across the two CCs, both with a >10³ fold difference between the most and least susceptible isolate with respect to the number of bacteria that survived (Fig. 1a &b).

Survival of S. aureus upon exposure to human serum is a polygenic trait. As the genome sequence for each of the 300 clinical S. aureus isolates was available we performed a GWAS (genome wide association study) to identify polymorphic loci or single nucleotide polymorphisms (SNPs) that associated with the level of susceptibility of isolates to serum. For this, the data from the two distinct clones were analysed independently, with population structure within the clones being accounted for (Fig. 1c & d, Tables 1a & 1b). We applied both uncorrected and corrected (for multiple comparisons) significance thresholds to this analysis, as our previous work has demonstrated that the stringency of multiple correction approaches increases the likelihood of type II errors (i.e. false negative results). For the CC22 collection 12 loci were associated with serum survival and for the CC30s there were 32 associated loci (Fig. 1c & d, Tables 1a & 1b).

Functional validation of the GWAS results identifies seven novel genes that affect serum survival. Of the 44 loci associated with serum survival, there were transposon mutants available for 32 in the Nebraska library²⁵. To functionally verify our GWAS findings, each of these mutants were tested for their ability to survive exposure to human serum. We found seven mutants were significantly affected in this ability, and this effect was complemented by expressing the inactivated gene in trans (Fig. 2). The pRMC2 plasmid used to complement the mutated phenotypes requires the use of two antibiotics, one for selection of the plasmid and one for induction of gene expression²⁶. These antibiotics can often affect bacterial gene expression. As such validation of complementation is based on the comparison between the empty and target gene containing plasmid which can be grown under the same conditions. The seven genes were tcaA, a gene associated with resistance to the antibiotic teicoplanin²⁷; tarK, a gene involved in wall teichoic acid (WTA) biosynthesis²⁸; gntR, which encodes gluconate kinase²⁹; ilvC, which encodes ketol-acid reductoisomerase³⁰; arsB, which is an efflux pump involved in arsenic resistance³¹; yfhO, which is involved in the glycosylation of lipoteichoic acid³²; and pdhD, which encodes a lipoamide dehydrogenase³³.

The SNPs in the tcaA gene of the clinical isolates are associated with decreased serum sensitivity. The tcaA gene has previously been reported to be involved in conferring increased sensitivity to the antibiotic teicoplanin²⁷, although the molecular detail of how it achieves this has yet to be determined, and our findings suggest it also confers increased sensitivity to serum (Fig. 2a). To visualise the signal detected by the GWAS for this gene we have mapped the position of the clinical isolates with the tcaA SNPs onto a figure displaying the range of serum sensitivities of the collection (Fig. 3a). The majority of the isolates with the SNPs were less sensitive to serum, suggesting the SNPs negatively affects the activity of the protein. The most common SNP in the tcaA gene amongst the clinical isolates conferred a Phe (290) to Ser change in the protein sequence. Using AlphaFold³⁴ we generated a structure for TcaA where it is predicted to be membrane bound with a single membrane spanning domain; to have a zinc finger presented intracellularly; with the majority of the protein including the Phe290 displayed extracellularly (Fig. 3b). Given that phenylalanines are frequently involved in protein-protein interactions³⁵ we examined the effect this change has on the activity of TcaA by cloning the tcaA gene with this SNP into the complementing pRMC2 plasmid and comparing its ability to confer serum resistance relative to that of the tcaA gene without the SNP. The TcaA protein with Ser at position 290 instead of Phe were slightly more resistant to serum killing, but not as affected as when the entire protein was inactivated, suggesting this change has a subtle effect on the serum sensitising activity of the protein (Fig. 3c).

TcaA confers increased sensitivity to multiple antibacterial components of serum. To examine whether the effect of TcaA on serum resistance was specific to the JE2 background we transduced the tcaA::tn mutation from JE2 into S. aureus strains SH1000 and Newman. Attempts were made to transduce this mutation into a clinical CC22 isolate, but this lineage was refractory to any of the available transducing phage^{36, 37}. In all three backgrounds TcaA increased the sensitivity of the bacteria to serum (Fig. 4a) verifying that the effect was not limited to the JE2 background. To determine which aspects of human serum TcaA was conferring increased sensitivity to, we measured the relative ability of the wild type and mutant to survive exposure to some of the anti-bacterial factors found in serum: to the HDFA arachidonic acid, and to two AMPs: HNP-1 and LL37. The strains producing TcaA (i.e. the WT and complemented strain) were more sensitive to all three components of human serum (Fig. 4b-d). For the AMPs this sensitivity was significant after only 90 min exposure, whereas for arachidonic acid sensitivity the bacteria were incubated overnight, and differences in sensitivity determined by bacterial growth (OD₆₀₀).

Although none of the individual antibacterial components of serum tested here are involved in S. aureus cell clumping, it is a possibility that some of the effects seen in serum is as a result of this activity, artificially amplifying the effect. To examine this the wild type and tcaA mutant were visually examined by microscopy in both PBS and serum where no differences in clumping was observed (Supplementary Fig. 1). Of the other GWAS identified genes, the ilvC mutant was less sensitive to arachidonic acid, the yfhO mutant was more resistant to HNP-1 and LL-37, while the tarK, gntK, arsB and phdD mutants were all more sensitive to HNP-1 and LL37 when compared to the JE2 wild type strain (Supplementary Fig. 2).

Wall teichoic acids contribute to the arachidonic acid sensitivity of TcaA producing strains. The antibacterial properties of HDFAs and AMPs share some common features in that they both penetrate through the cell wall and attack the bacterial membrane^{11–13, 21–23}. AMPs are positively charged molecules that rely on the relatively negative charge across the bacterial cell envelope to penetrate, and thus resistance is frequently acquired by changing the charge across the cell wall such that the AMPs are repelled^{22, 23}. Less is known about how HDFAs penetrate the cell wall; however, resistance is associated with changes in the abundance of wall teichoic acids (WTA) in the bacterial cell walls²⁴. WTA are hydrophilic and when present in the cell wall they are believed to be protective by interfering with the penetration of the hydrophobic HDFAs through to the bacterial membrane²⁴. It has also been shown that when the ligation of WTA to the cell wall is affected, such that they are instead released into the environment, this also decreases the sensitivity of the bacteria to HDFAs¹³. With distinct modes of action and means of accessing their target, it is intriguing to consider how TcaA may be contributing to increasing the sensitivity of S. aureus to both these types of anti-bacterial molecules.

To examine whether the TcaA associated increase in sensitivity to arachidonic acid is a result of an increase in the abundance of WTA in the bacterial cell we extracted and quantified WTA from the wild type and tcaA mutant, where we found that the cell wall of the wild type had significantly more WTA when compared to the tcaA mutant cell extract (Fig. 5a & b & c)). We next quantified WTA in the bacterial supernatant, as recent work on a S. aureus lcpA mutant demonstrated that another WTA related means of altering sensitivity to arachidonic acid is to release the WTA the bacteria surface¹¹. We found significantly less WTA in the extract of the TcaA producing wild type strain compared to that of the tcaA mutant (Fig. 5a), which could explain the observed difference in sensitivity to HDFAs. It was unclear from the previous work on LcpA how the released WTA by the lcpA mutant affects arachidonic acid sensitivity, and here we demonstrate that this mutant, like the tcaA mutant, is also less sensitive to human serum (Fig. 5d). One hypothesis is that the WTA can either sequester or inactivate HDFAs in the environment surrounding the bacteria, thereby neutralising them. To test this, we harvested the supernatant of the wild type JE2, the JE2 tcaA mutant, and the JE2 lcpA mutant from overnight growth. These supernatants were used to supplement fresh broth (at 10%) into which the wild type JE2 strain was inoculated. In the absence of any arachidonic acid JE2 grew equally well regardless of the supernatant supplement (Fig. 5e). However, in the presence of the supernatant of both the tcaA and lcpA mutant (both of which contain an abundance of released WTA), JE2 was able to grow in the presence of arachidonic acid; whereas when the JE2 supernatant was used as the supplement the bacteria were unable to grow (Fig. 5e). To further verify that soluble WTA can neutralise arachidonic we performed WTA extractions from an isogenic wild type and mutant of the S. aureus strain LAC where the mutant has had the tarO gene deleted and consequently does not produce any WTA (Fig. 5e). The incorporation of the wild type LAC purified WTA extract to broth at 2% created an environment in which JE2 could grow in the presence of arachidonic acid, whereas the equivalent WTA extract of the tarO mutant, which contains no WTA, did not (Fig. 5e). Together these data demonstrate that WTA in the environment can neutralise arachidonic acid, and this provides a likely explanation for the increased sensitivity to HDFAs associated with TcaA production.

The reduced abundance of WTA in the bacterial cell wall contributes to the resistance of S. aureus to AMPs. In additional to being hydrophilic, WTA is also predominantly negatively charged²⁸. Given that a change in charge across the bacterial cell wall is frequently associated with resistance to AMPs^21–23, and that the wild type strain has more WTA in its cell wall, this may explain the increased AMP sensitivity of the TcaA producing strain. To examine this, we incubated the bacteria with cytochrome C, which is positively charged and through its electrostatic-related ability binds to the bacterial cells, providing a measure of the charge across bacterial cell walls. Using this we found that the wild type strain bound significantly more cytochrome C and is therefore more negatively charged than the tcaA mutant, which explains the increase in AMP sensitivity we have found associated with TcaA production (Fig. 5e). Together these results explain how TcaA production affects the sensitivity of S. aureus to both HDFAs and AMPs. By retaining WTA within the cell wall, there is less released to sequester or inactivate the fatty acids from the environment, and due to their negative charge, increased amounts of WTA in the cell wall affects the charge across the cell wall and results in increased electrostatic attraction of the positively charged AMPs.

Serum induced tcaA expression increases both the abundance of WTA in the cell wall and the sensitivity S. aureus to teicoplanin. As mentioned previously, an additional interesting feature of the TcaA protein is that is has also been associated with changing the sensitivity of the bacteria to the antibiotic teicoplanin, although this has been reported to vary between strains^{27, 38}. To examine this for our strains we determined the Minimum inhibitory concentrations (MICs) for three independent sets of isogenic wild type and tcaA mutants (ie in strains JE2, Newman and SH1000, all corresponding to CC8), where TcaA production resulted in increased sensitivity to this antibiotic across all three strains (Fig. 6a). To examine whether the clinical isolates with the SNPs in the tcaA gene associated with increased serum resistance also had altered teicoplanin sensitivity, we compared the ability of 12 clinical isolates containing the GWAS identified tcaA SNPs to grow in teicoplanin to that of 12 isolates from the same collection with the wild type gene sequence, i.e., without the polymorphism. The isolates with the polymorphism in the tcaA gene were on average less sensitive to teicoplanin (Fig. 6b). Using the pRMC2 complementing plasmid expressing either the wild type or SNP (conferring the Phe290-Ser substitution) containing tcaA gene, we demonstrated that this mutation significantly reduces the ability of the protein to sensitize the bacteria to teicoplanin (Fig. 6c), confirming that this mutation negatively affects the activity of the TcaA protein.

As TcaA has been shown to be upregulated when exposed to teicoplanin^{27, 38}, we examined whether serum would also induce its expression. The wild type JE2 strain was exposed to either sub-inhibitory (2.5% serum and 0.5μg/ml teicoplanin) or inhibitory (10% serum and 10μg/ml teicoplanin) concentrations of either serum or teicoplanin for 20 mins, total RNA was extracted and the transcription of the tcaA gene quantified by qRT-PCR. For both serum and teicoplanin, induction of expression of the tcaA gene was concentration dependent (Fig. 6d). Given the association of WTA abundance in the cell wall with tcaA expression described above (Fig. 5a-c) we next examined whether serum induced expression of tcaA would also result in an increase in abundance of WTA in the bacteria cell wall. Both the wild type and tcaA mutant were exposed to inhibitory concentrations of human serum for 90 mins and the WTA extracted from the cell walls, where serum exposure resulted in an increase in WTA for the wild type strain but not the tcaA mutant (Fig. 6e & f). To examine whether serum induced increased expression of tcaA would affect sensitivity to teicoplanin, we exposed the wild type bacteria to a sub-inhibitory concentrations of serum (2.5%) and then incubated these overnight in a subinhibitory concentration of teicoplanin (0.5μg/ml). Pre-exposure to serum induced increased sensitivity to teicoplanin suggesting that induction of tcaA expression by the initial serum exposure was sufficient to induce cross-sensitivity to the antibiotic (Fig. 6g). To put this sensitisation effect in the context of serum, the reverse experiment was also performed whereby the wild type strain was exposed to a sub-inhibitory concentration of teicoplanin (0.5μg/ml) overnight and a serum killing assay performed following this incubation (Fig. 6h). Pre-exposure to teicoplanin also induced increased sensitivity to serum again highlighting this cross-sensitivity effect.

To verify this induction effect was not specific to the JE2 lineage a tcaA::gfp reporter plasmid was constructed and electroporated into JE2, SH1000, Newman and a clinical CC22 isolate EMRSA15. In each background serum, arachidonic acid and LL37 reduced tcaA expression with the exception of LL37 and the EMRSA15 strain (Supplementary. Fig. 3). These findings suggest that upon entry into the bloodstream and exposure to many of the antibacterial components of serum, expression of the tcaA gene is induced and this sensitizes S. aureus to killing by these components of serum.

The inactivation of tcaA alters the antibiotic resistance profile of S. aureus. There is emerging evidence that WTA play a role in cell division and peptidoglycan biosynthesis³⁹. In support of this, it has been reported that a tarO mutant, devoid of WTA, confers sensitization to β-lactam antibiotics due to mislocalization of the PBPs⁴⁰. As the target for teicoplanin is peptidoglycan, more specifically the terminal D-Ala-D-Ala residue of lipid II, we hypothesised that the increased ligation of WTA into the cell wall of the TcaA producing strain may affect peptidoglycan biosynthesis and assembly, and that this may be responsible for the change in sensitivity to teicoplanin. To test this further, we compared the sensitivity of the wild type and mutant to a range of peptidoglycan attacking antibiotics, where we found the TcaA producing wild type strain to be more sensitive to vancomycin (as shown previously³⁸), oritavancin, but not dalbavancin (all alternative members of the glycopeptide class of antibiotics), and ramoplanin, a glycolipodepsipeptide antibiotic (Supp. Fig. 4). However, TcaA production conferred decreased sensitivity to oxacillin, a β-lactam antibiotic that targets penicillin binding protein 2 and moenomycin a phosphoglycolipid antibiotic that inhibits the transglycosylase PBP enzymes and prevents the formation of peptidoglycan polymers (Supp. Fig. 4). This suggests that TcaA production affects the overall composition and structure of peptidoglycan in the wildtype strain.

To examine whether the crosslinking of peptidoglycan is affected by TcaA production, we tested the susceptibility of the wild type and tcaA mutant to lysostaphin, which is a glycylglycine endopeptidase capable of cleaving the pentaglycine crosslinks of peptidoglycan⁴¹. We found that lysostaphin cleaved the cell wall of the wild type strain more slowly than for the tcaA mutant (Fig. 7a). We next performed a triton X100 induced autolysis assay where the wild type exhibited a slower rate of autolysis compared to the mutant, suggesting that the cell wall is more robust due to higher levels of crosslinking when TcaA is produced (Fig. 7b). Together these data suggest that the integrity of the peptidoglycan, in particular its crosslinking, is increased when TcaA is produced. This provides a likely explanation for why the mutant is more resistant to teicoplanin, reduced crosslinking in the cell wall increases the number of off target D-ala D-ala which teicoplanin will bind to instead of reaching lipid II in the membrane and causing disruption to this. With changes to the quantity and composition of WTA and peptidoglycan associated with TcaA production, we sought to visualise the cell wall using transmission electron microscopy. We observed a clear difference in the density and structure of the outer layer of the bacterial cell wall between the wild type and tcaA mutant, confirming a role for TcaA in the structural integrity of the S. aureus cell wall (Fig. 7c, additional images in Supp. Fig. 5)).

Once established, TcaA contributes positively to the development of bacteraemia. Our data shows that expression of tcaA is induced upon exposure to serum, that this can sensitize the bacteria to serum killing but also increase the abundance of WTA in the cell envelope. From a pathogenicity perspective these activities would appear to be in conflict, with one making it less likely for a bacterium to cause bacteraemia (increased serum sensitivity) and the other making it more likely for a bacterium to cause bacteraemia (increased WTA in the cell envelope^{42, 43}). To understand this apparent dichotomy we first examined whether TcaA production only affected survival in serum, whereas in whole blood where other components of the human immune system may be present (e.g. complement, antibodies and phagocytes) that this effect may be lost. Using human blood from three donors, we demonstrated that TcaA production increases the sensitivity of the bacteria to whole blood killing (Fig. 8a). We next interrogated our human clinical data, where in addition to the 300 bacteraemia strains that form the basis of this study we have an additional 176 S. aureus isolates, which has allowed us to include more genetic diversity to this analysis (i.e. from CC22 (n=138), CC30 (n=162), ST8 (n=132) and ST93 (n=44)). The source of these isolates was also available to us (i.e. bacteraemia (n=341), SSTI (n=80) or carriage (n=55)), as well as their genome sequences (Supplementary Table 1). We examined the distribution of nonsynonymous mutations in the tcaA gene across these, where it was present in 6% of the bacteraemia isolates compared with only 1.5% of non-bacteraemia isolates (n= 21/341 and 2/135 respectively; p = 0.03, in a Fisher’s Exact test). However, when we examined the 30-day mortality rates for the 113 bacteraemia patients for whom we had this clinical data, the mortality rate of the proportion of patients infected with a S. aureus strain with the wild type tcaA gene was significantly higher compared to those infected with a mutated tcaA gene (Fig. 8b & c) (2-tailed chi squared test: p = 0.03). So, although the incidence of mutations in tcaA appear to be relatively rare, that isolates with variant tcaA genes are enriched amongst those causing bacteraemia suggests that a functional TcaA may limit the propensity of S. aureus to cause a bloodstream infection, such that it acts as a selective force for mutants of this gene. However, once the bacteraemia has become established, TcaA positively contributes to the infection process, likely due to the increased abundance of WTA in the cell envelope, and the greater structural integrity of the cell wall.

To experimentally test these human findings we utilised a mouse model of bacteraemia where the tail vein of C57/Bl6 mice are inoculated with 2 x 10⁷ colony forming units (CFU) of bacteria to establish a bloodstream infection, where they seed from there into organs such as the liver spleen and kidneys. We performed mixed infections, where the wild type and mutant bacteria had to compete during the development of the infection within the mouse. The mice were infected with equal quantities of both bacteria and the change in ratio of the bacteria in the mouse blood, liver, spleen and kidney recorded after 48hr, which allowed the relative virulence ratio (RVR) of each bacterium to be calculated⁴⁴. The RVR of the TcaA producing wild type strain was significantly higher than that of the mutant at a systemic level (Fig. 8d). As such, while the TcaA associated self-sensitization of the bacteria was not evident here, possibly due to the size of the inoculum level needed, once an infection has become established, TcaA contributes positively to disease severity as demonstrated by the wild type strain outcompeting the tcaA mutant.

To develop novel therapeutic approaches for infectious diseases, we need to understand how microorganisms cause disease and evade the host’s immune system. To address this, we have applied a population-based approach to analyse the pathogenicity of S. aureus and identified seven genes that affect the bacteria’s ability to survive exposure to human serum, the first step in the development of bacteraemia. The dissection of the molecular detail of how these genes affect serum survival is underway to determine their potential for therapeutic intervention. However, our focus here has been on the tcaA gene where we show that its expression is induced upon exposure to serum, that it is involved in the ligation or retention of WTA in the cell wall, and that this renders the bacteria more susceptible to killing by both AMPs and HDFAs. We have also found that TcaA is associated with changes to the crosslinking of peptidoglycan, such that the bacteria are more susceptible to some antibiotics (e.g., teicoplanin) and less susceptible to others (e.g., oxacillin). It is interesting to note that our GWAS approach identified tcaA due to the presence of function altering SNPs amongst clinical isolates. Our human data suggests that bacteraemia and/or the antibiotics commonly used to treat it such as teicoplanin and vancomycin may impose a selective pressure on the bacteria to mutate this gene, due to the decreased sensitivity this confers to serum killing and to clinically relevant antibiotics. However, both our human and mouse data have confirmed, that once bacteraemia has become established, TcaA contributes positively to disease progression, likely due to its cell wall re-modelling activity. A graphical summary of this has been provided in Fig. 9.

The tcaA gene is part of a three gene locus alongside the tcaR gene, which is predicted to encode a transcriptional regulator, and the tcaB gene which is predicted to encode an efflux pump²⁷. This locus is also reported to be part of the S. aureus cell wall stimulon, following a number of studies that monitored the response of S. aureus to cell wall attack^{45, 46}. It is interesting to note that the tcaA gene was not associated with serum resistance for the CC30 collection of isolates, however, there were very few isolates with variation in this gene amongst that collection such that they would have fallen below the minimum allele threshold. While work to understand the activity of each of the proteins encoded by the tca locus is underway, our data suggests that TcaA play a role in the ligation or retention of WTA within the cell wall, perhaps alongside the LcpA protein¹³. How this is achieved is yet to be determined. TcaA may play a direct role in ligating WTA to the cell wall; it may play a role in cell wall turnover and the subsequent release of WTA and other cell wall macromolecules into the external environment; or it may interact and subsequently interfere with other cell wall biosynthetic processes that are responsible for WTA ligation. It is interesting to note that tarK, which is one of the other six genes associated here with serum survival is also involved in the biosynthesis of WTA. Additionally, yfhO which encodes an enzyme that glycosylates lipoteichoic acid (LTA) is also involved in serum resistance³², providing further evidence for the importance of teichoic acid associated molecules to this aspect of the biology and pathogenicity of S. aureus. For TcaR, we hypothesise that it likely controls the transcription of the co-transcribed tcaA and tcaB genes, and may directly respond to the external stimulus of teicoplanin and serum as reported here. What is less clear is what role TcaB plays, as efflux pumps are more typically associated with increasing the resistance of a bacterium to an antibiotic, or other toxic molecules.

In addition to tcaA this study has identified one other locus (yfhO) that negatively affects serum survival, and added a further five loci (tarK, gntR, ilvC, arsB, and pdhD) to the list of genes that contribute positively to serum survival by protecting the bacteria from HDFAs and AMPs. This leads us to the questions of why S. aureus has loci in apparent conflict with regards to the survival of the bacteria during bacteraemia. It is perhaps through the consideration of the other ways in which S. aureus interact with humans that we may understand this. The vast majority of interactions that occur between S. aureus and humans is as a commensal in the human nose from which it can readily transmit. The five identified protective genes may therefore also protect the bacteria during colonisation, which would confer a selective advantage, and it is just an unfortunate coincidence that they also confer increased survival in serum during the development of bacteraemia. While the focus of this work has been on the role of TcaA in serum, it is also worth considering what role it may have during colonisation of the nose, which involves the adherence of the bacteria to the nasal epithelium, to which WTA has a well-established role. It is therefore possible that TcaA enhances the ability of S. aureus cells to colonise the nose by retaining WTA within the cell wall. However, the role of TcaA in responsiveness to cell wall attack is as yet unclear, but perhaps it contributes to the competitiveness of S. aureus within the nasal microbiome in response to cell wall damage inflicted by competitors.

The evolution of the virulence of a pathogen is heavily dependent upon their mode of transmission, which for opportunistic pathogens raises interesting challenges. If within-host invasiveness limits between-host fitness by preventing them from transmitting to a new host, there are likely selective pressures acting on bacterial populations to evolve means of limiting their ability to cause invasive disease. However, if increased invasiveness facilitates enhanced within-host fitness, as the bacteria multiply and spread throughout the body, the ‘short-sighted evolution of virulence’ hypothesis proposed by Levin and Bull⁴⁷ could apply and may explain situations like S. aureus bacteraemia. We propose that TcaA may represent a system at the cusp of both these scenarios, in that it responds to cell wall attack by serum by making itself more susceptible to killing by serum, thereby potentially limiting its ability to cause bacteraemia, which could have long-term between host benefits. However, once established in the bloodstream and TcaA has remodelled the cell wall, the short-sighted view of virulence hypothesis applies where the bacteria survive and thrive, allowing them to cause a successful bacteraemia.

Entry into the bloodstream to establish an infection has long been considered a significant bottle neck for the bacteria, but until now, the stringency of this bottleneck was believed to be a result of the onslaught of all the immune mechanisms present in blood. Here we suggest that S. aureus may contribute to this bottleneck by responding to serum induced cell wall damage by further sensitizing itself to this attack as part of the pathway towards increasing their pathogenic capabilities by remodelling the cell wall. Our findings presented here opens up an entirely novel aspect to the biology of a major human pathogen and brings into question the appropriateness of the term ‘opportunistic’ when we refer to such highly effective bacterial pathogens.

C57/Bl6 mice were bred in-house in Trinity College Dublin and C57BL/6NCrl mice were purchased from Charles River Laboratories. Mice were housed under specific pathogen-free conditions at the Trinity College Dublin Comparative Medicines unit and maintained under pathogen-free conditions in an Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC)-accredited animal facility in NIAID, NIH, respectively. All mice were female and used at 6–8 weeks. All animal experiments were conducted in accordance with the recommendations and guidelines of the health product regulatory authority (HPRA), the competent authority in Ireland and in accordance with protocols approved by Trinity College Dublin Animal Research Ethics Committee and according to the regulations of NIAID’s Division of Intramural Research Animal Care and Use Committee (DIR ACUC), animal study proposal LB1E. For the human serum collection, all healthy volunteers provided written informed consent and all methods and experimental protocols were carried out in accordance with the recommendations of the University of Bath, Research Ethics Approval Committee for Health. The present study was approved by the University of Bath, Research Ethics Approval Committee for Health [reference: EP 18/19 108].

Paired two-tailed student t-test or One-way ANOVA (GraphPad Prism v9.0) were used to analyse the observed differences between experimental results. A p-value <0.05 was considered statistically significant and the Sidak method was used to correct for multiple comparisons.