Staphylococcus aureus is a frequent colonizer of the human population and one of the foremost opportunistic bacterial pathogens of humans, causing major morbidity and mortality globally1. Any clinician, especially those working in infectious diseases, will attest to the frequency, complexity and the potential catastrophic nature of invasive infections caused by this pathogen. Although on a global scale human colonization and infection with S. aureus are forefront of mind, the organism has a long evolutionary history as a multihost opportunistic pathogen. S. aureus colonizes approximately 20–30% of humans persistently in the nose2 and frequently in other sites such as the skin, throat, axillae, groin and intestine3. The interplay between the potential pathogen and the host microbiota is being explored with increasing depth and complexity, highlighting roles for other commensals in shaping S. aureus colonization4. Colonization is harmless but it is a risk factor for developing subsequent infections (often caused by the colonizing strain5), which can range from mild skin and soft tissue infections to serious invasive infections, including osteomyelitis and septic arthritis, bacteraemia or septicaemia, pneumonia and endocarditis1. S. aureus infections can be acute, recurrent or chronic and persistent. To be best placed to develop effective therapies and clinical responses to these infections, we need to understand the complex pathophysiological relationships between S. aureus and its hosts6.

The breadth of immune evasion mechanisms in S. aureus reflects the longstanding nature of the interaction of bacterium with humans, characterized by persistent colonization and intermittent invasive infections. S. aureus produces a large array of virulence and immune evasion factors that hinder the human immune response7,8. Neutrophils represent the major defence against staphylococcal infections. Staphylococcal factors that perturb adaptive B cell and T cell responses and diminish protective immunity are well described8. Factors subverting innate immune response also dominate, such as inhibition of neutrophil chemotaxis and killing, inhibition of complement activation and phagocytosis, killing of host cells and staphylococcal agglutination6,8. Reflecting the presence of factors inhibiting adaptive immune responses, infection with S. aureus does not elicit a protective immune response, meaning that recurrent infections are common throughout life9,10. Decades of vaccine studies with promising preclinical results have failed to translate into effective vaccines for use in human, including the failure of two major phase III clinical trials to demonstrate a benefit11. The adaptation of some S. aureus clones to the human host means that preclinical animal models are limited in utility11.

In the colonization state, S. aureus is part of a polymicrobial community that primarily includes other staphylococci, corynebacteria and propionibacteria12,13. The roles of non-pathogenic commensals and interspecies interactions that influence S. aureus colonization and disease are largely unexplored14. However, advances in genomics-enabled microbial community analyses and metabolomics provide an opportunity to delve into these interactions (Supplementary Box 1 and Supplementary Fig. 1). In vivo S. aureus biofilms are more typically a monomicrobial community, in which surface-exposed proteins, referred to as microbial surface components recognizing adhesive matrix molecules (MSCRAMMs)15, initiate the attachment to surfaces, further consolidated by the subsequent release of polysaccharides, proteins and extracellular DNA, forming an extracellular matrix encasing the bacteria. Biofilm-associated infections, which have a propensity to chronicity through increased antimicrobial resistance and immune evasion, are frequently linked to implanted devices and contribute to the pathogenesis of S. aureus infections in native tissues15,16. The mechanisms of immune evasion afforded by S. aureus biofilms have not been clearly elucidated, but are thought to be in part caused by the masking of pathogen molecular signatures17.

Antimicrobial resistance acquired through acquisition of mobile genetic elements or chromosomal mutations has further complicated staphylococcal infections and therapeutics18. Although penicillin initially revolutionized the treatment of serious S. aureus infections, resistance through the acquisition of the β-lactamase encoding gene blaZ was soon widespread19. The acquisition of resistance to anti-staphylococcal penicillins through the mecA gene has resulted in the global dissemination of diverse lineages of methicillin-resistant S. aureus (MRSA), now considered a global public health threat19. Glycopeptides became the mainstay of therapy for invasive MRSA infections, but reduced susceptibility through adaptive mutations further hampered therapeutic efforts20, which results in the move to using ‘last-line’ antimicrobials such as daptomycin, and combination therapies19,20. Pathogen evolution in cases of antimicrobial resistance and persistent human infections occurs in the milieu of the host environment, which imparts an additional selective pressure potentially selecting for not just drug resistance but also immune escape21 (Supplementary Box 2 and Supplementary Fig. 2). The advent of accessible, high-throughput pathogen genomics technologies has revolutionized the study of staphylococcal resistance and adaptation, and the investigation of complex phenotypes linked to antimicrobial resistance, such as small colony variant (SCV) S. aureus22. The genomes of tens of thousands S. aureus isolates have been sequenced and are publicly available, and we are at an exciting stage in the utilization of this incredible resource for further understanding the pathogen23. Early use of the technology enabled the tracking of mutations that evolve during persistent clinical infections, which revealed an unexpected capacity for the pathogen to adapt to complex antimicrobial exposures and the host environment24,25, and uncovered S. aureus genetic heterogeneity that can exist within host26 (Box 1).

In addition to the growing list of newly discovered immune-modulating molecules, recent advances in our understanding of S. aureus pathogenesis and disease are highlighting the roles of biofilm formation, immunometabolic host interplay and immune evasion phenotypes (such as the SCV) to avoid host clearance. Added to this, S. aureus genomic studies are revealing a surprisingly high mutagenic potential that also materially contributes to immune evasion phenotypes. Finally, assessing bacterial adaptation processes occurring during switching between different vertebrate hosts is providing insights into host-specific disease mechanisms. In this Review, we examine and integrate recent key advances in understanding the mechanisms that S. aureus uses to cause infections. We explore our understanding of S. aureus immune evasion molecules and the role of biofilms, immunometabolism and S. aureus avoidance mechanisms of trained innate immunity. We also investigate S. aureus host specificity and adaptation and summarize current knowledge of the interplay between colonizing S. aureus and other bacteria with a summary of our emerging knowledge of S. aureus genomic plasticity, drug resistance and persistence. A detailed dissection of host responses, vaccines and other therapeutics is not covered in this Review.

Immune responses and immune evasion molecules

Staphylococci were first described in the 1870s, and staphylococcal leukotoxins were reported almost 100 years ago27. Despite intense study, the reasons for the apparent redundancy in these innate immune effector molecules secreted by S. aureus have remained uncertain, partly owing to a lack of detailed understanding of the molecular mechanisms of action7. Nonetheless, new immune evasion factors, including extracellular proteins and polysaccharides, are still being discovered and refined mechanistic insights continue to be gained (Fig. 1 and Supplementary Table 1). The large number of immune evasion factors found in S. aureus contrasts with many other bacterial pathogens that often produce a smaller number of specific factors28 and probably reflects the long-standing and complex relationship between S. aureus and its mammalian hosts.

Fig. 1: New insights into Staphylococcus aureus immune evasion.
figure 1

a, Recruitment of neutrophils and phagocytic cells to the site of infection. Vascular neutrophils and phagocytic cells, such as macrophages, are actively recruited by chemotaxis to Staphylococcus aureus infection site following chemokine gradients and sensing pathogen-associated molecular patterns and bacterial-derived ligands of the formyl-peptide receptors released by S. aureus. b, Recently described S. aureus immune evasion molecules. S. aureus secretes phenol-soluble modulins (PSMs) that cause neutrophil lysis through interaction with formyl-receptor 2 (FPR2). Secreted neutrophil serine proteases (NSPs) degrade PSMs, but this process is inhibited by S. aureus extracellular adherence protein (Eap; or orthologues EapH1 and EapH2). S. aureus secretes a range of superantigens including staphylococcal superantigen (SElW) and the bifunctional staphylococcal enterotoxin-like toxin X (SElX) that lead to T cell proliferation and immune dysregulation. SElX also binds to neutrophil surface receptors and inhibits phagocytosis. Within the phagosome, myeloperoxidase (MPO) contributes to phagosomal killing through the generation of reactive oxygen species (ROS). However, S. aureus secretes staphylococcal peroxidase inhibitor (SPIN) that inhibits this process. c, Disruption of macrophage efferocytosis. S. aureus can indirectly interfere with macrophage efferocytosis of neutrophils undergoing formation of neutrophil extracellular traps (NETosis), whereby S. aureus neutralizes the bactericidal activity of neutrophil extracellular traps (NETs) using staphylococcal nuclease (Nuc) and adenosine synthase A (AdsA). The activities of these enzymes generate deoxyadenosine (dAdo), which triggers macrophage apoptosis. d, Genetic heterogeneity in virulence factors can alter their function. Clonal complex-specific sequence heterogeneity in the leukocidin AB (LukAB) leukotoxin alters phagocyte receptor tropism and cytotoxicity profile. This is exemplified by the strains belonging to CC30(CC45), which gained the ability to bind human hydrogen voltage-gated channel 1 (HVCN1). e, S. aureus biofilms secrete immune evasion factors to promote and inhibit NETosis. Secreted factors from S. aureus biofilms include leukocidins Panton–Valentine leukocidin (PVL) and S. aureus γ-haemolysin (HlgAB) that promote NETosis as well as nucleases that degrade NET DNA. CD11b, cluster of differentiation molecule 11B; ECM, extracellular matrix; Mϕ, macrophage.

Neutrophils are critical in the defence against infection from S. aureus8, with chronic granulomatous diseases caused by defects in neutrophil killing and recurrent S. aureus infections being a key example29. S. aureus targeting of neutrophils spans multiple arms of bactericidal activities, including chemotaxis, opsonophagocytosis and neutrophil-mediated killing and staphylococcal killing of host cells, including neutrophils6,8.

Recently discovered immune evasion molecules

A large repertoire of well-characterized molecules contributes to the survival of S. aureus following phagocytosis, in which the bacterium is exposed to several toxic products such as reactive oxygen species, nitric oxide, host antimicrobial peptides and neutrophil serine proteases (NSPs)30,31. The extracellular adherence protein (Eap) and homologues (EapH1 and EapH2) secreted by S. aureus can non-covalently inhibit NSPs at low nanomolar concentrations31 (Fig. 1). The targeted inhibition of NSPs prevents bacterial killing, regulation of neutrophil extracellular traps (NETs) and the degradation of phenol-soluble modulins (PSMs)32. PSMs target the neutrophil formyl-receptor 2, which leads to degranulation and neutrophil lysis32. In addition to PSMs, NSPs can target, degrade and functionally inactivate a range of staphylococcal immune evasion factors, including chemotaxis inhibitory protein of S. aureus (CHIPS) and staphylococcal complement inhibitor-A. However, the activity of these proteases seems different against homologues of immune evasion proteins with similar functions, possibly providing some explanation for the redundancy in the large number of immune evasion factors secreted by S. aureus33. All three Eap proteins were shown to protect against the activity of NSPs against staphylococcal immune evasion factors32,33 (Fig. 1).

Within the phagosome, myeloperoxidase (MPO) uses H2O2 as a substrate to generate highly toxic reactive oxygen species that may act synergistically with intracellular proteases against S. aureus34. High-throughput screening recently identified an additional factor inhibiting neutrophil killing through specific binding and inhibition of MPO, called staphylococcal peroxidase inhibitor35. Expression of this molecule is upregulated following phagocytosis and contributes to evasion of MPO-dependent killing35 (Fig. 1). Like other human immune evasion factors secreted by S. aureus, staphylococcal peroxidase inhibitor does not bind to, nor inhibit MPOs from other species such as mouse, rabbit or cow. The host specificity adds additional challenges investigating in vivo activity of these virulence factors36.

Superantigens (sAgs) are bacterial toxins that result in T cell proliferation and immune dysregulation, contributing to the pathogenesis of invasive S. aureus infections and having a role in colonization. sAgs bind to a large repertoire of variable-β chains in the T cell receptor. Toxic shock syndrome toxin-1 (TSST-1) and staphylococcal enterotoxins B and C are the most well described of these molecules8 (Supplementary Table 1). However, the family of staphylococcal sAgs numbers over 26 (ref. 37), which facilitates binding to a large repertoire of variable-β chains in the T cell receptor. Large-scale genomic analyses are further refining the distribution and identification of staphylococcal sAgs37. Recently, staphylococcal enterotoxin-like toxin X (SElX) has been identified as an additional sAg that is encoded in the genome of 95% of S. aureus isolates examined. SElX has been shown to induce T cell activation, which contributes to necrotizing pneumonia in a rabbit model38 (Fig. 1). SElX has a mechanistically independent contribution to phagocytosis inhibition by binding neutrophils via neutrophil surface receptors, providing an example of a staphylococcal sAg with both innate and adaptive immune disruptive strategies39. Genomics recently identified the staphylococcal sAgs SElW in 97% of the isolates studied37. The gene-encoding SElW (selw) has been truncated in several staphylococcal clones but full length in others, including the human and livestock-adapted clone CC398. Functional analysis revealed that SElW is a potent T cell activator contributing to pathogenesis in a bacteraemia model and that this sAg is solely responsible for T cell mitogenic activity of CC398 (ref. 37).

New insights into previously described molecules

Many cell-wall-anchored adhesins are expressed by S. aureus, including those making up the MSCRAMM family40. The serine–aspartate repeat protein D (SdrD) is a major MSCRAMM protein that contributes to colonization and infection, through attachment to the nasal epithelium41, and abscess formation42. Recently, this protein has been shown to inhibit innate immune-mediated killing of S. aureus, while promoting bacterial survival in human blood; however, the molecular mechanisms supporting these activities have not been elucidated43. Another serine–aspartate repeat protein, SdrE, has also recently been found to bind complement factor H, which leads to complement evasion44 (Supplementary Table 1). The fibronectin-binding protein FnbpB has an essential role in binding to host ligands, platelet activation and invasion into non-phagocytic cells45. In addition to the binding of plasminogen that can then be converted to plasmin by host activators or endogenously expressed staphylokinase, a study recently identified a novel role for FnbpB in the binding of histones (which are a major antibacterial component of NETs), which leads to their degradation46.

Host deployment of NETs is a remarkably efficient defence mechanism against bacterial intrusions, albeit leading to neutrophil death (NETosis). In response, S. aureus has evolved a two-pronged counterattack. By secreting the staphylococcal nuclease Nuc, S. aureus degrades the NETs DNA backbone and uses adenosine synthase A (AdsA), which is a cell-wall-anchored enzyme that converts extracellular ATP and ADP into the nucleoside Ado, an anti-inflammatory molecule47,48. AdsA also converts dAMP into deoxyadenosine (dAdo), a potent inducer of caspase-3-dependent apoptosis in macrophages, thus AdsA has multiple roles blunting macrophage efferocytosis activity at the infection site (Fig. 1).

Another important class of immune evasion molecules in S. aureus consists of the bicomponent pore-forming toxins (leukocidins), which to date have been primarily linked to neutrophilia (also called neutrophil leukocytosis), although these toxins target a range of other immune cells7. Leukocidins share a highly conserved structure, with chromatography elution profiles differentiating fast (F) and slow (S) components that form octameric membrane-spanning pores (Supplementary Table 1). Different leukocidins bind to specific cell receptors, which cause cell lysis after pore formation7 (Fig. 1). There is substantial interest in further defining the action of these major toxins in disabling the axis between innate and adaptive immune responses to S. aureus infection. By deploying leukocidin AB (LukAB), S. aureus kills dendritic cells, thereby preventing antigen presentation necessary for the establishment of adaptive immunity49. However, the cell specificity of LukAB is still under investigation, which will be important for the future development of immune-mediated prevention strategies49.

To date, the leukocidins have been studied as a ‘class’, and there is increased interest in understanding if their amino acid sequence heterogeneity may contribute to differential action. Greater sequence variation in LukAB compared with other leukocidins has been described50, and recent genomic interrogation of lukAB highlighted sequence variations for this toxin, specific to S. aureus lineages51. Although binding of the LukAB toxin to phagocytes is mediated by cluster of differentiation molecule 11B (CD11b), it has also been shown that LukAB derived from two major S. aureus clonal complexes (CC30 and CC45) are cytotoxic to CD11b-depleted human monocytes51. The human hydrogen voltage-gated channel 1 was identified as the alternate target for LukAB produced by these clonal complexes51.

The resistance of mouse phagocytes to most leukocidins that target human phagocytes has hampered research into the activity of these toxins7. The S-component of the toxins is the major molecule that demonstrates human cell-receptor specificity, and for protein subunits S and F of Panton–Valentine leukocidin (LukSF-PV) this is the human complement C5a receptor 1 (hC5aR1)52. Use of high-resolution microscopy suggests that hC5aR1 may dissociate from the toxin ligand after pore formation providing an additional receptor for other ligand binding and potential amplification of the inflammatory response53. This observation has implications for the modes of action of other bi-component toxins if the process is generalizable. Although binding of the S-component of LukSF-PV to hC51R1 explains human phagocyte specificity, a CRISPR–Cas9 screen identified CD45 as a specific target of the F-component54. Thus, although mouse models have proven very useful in dissecting host–staphylococcal interactions, their limitation, particularly pertaining to human specificity of virulence factors, has led to the development of humanized mouse models with increased susceptibility to S. aureus infection55,56,57,58,59.

Although numerous studies have investigated the mechanisms of action of staphylococcal toxins and immune evasion factors, linking specific factors to disease manifestations has not always been straightforward. This area of research is complicated by the large repertoire of bacterial factors and apparent redundancy as described earlier. However, as access to bacterial genome sequence data has increased, statistical genomics approaches such as genome-wide association studies and machine learning have begun to link genome traits with S. aureus disease manifestations60. For example, the role of LukFS-PV in staphylococcal disease was heavily debated following the global emergence of community-MRSA clones that frequently carried genes encoding this toxin. Although the contribution of this toxin to purulent skin diseases is well established, the role of the toxin in other manifestations such as pyomyositis was disputed61. The use of genome-wide association studies linking bacterial genome data to clinical data on diseases manifestations has now shown that LukFS-PV is a critical toxin in the manifestation of pyomyositis in children61.

S. aureus also produces a subset of proteins called sAg-like proteins (SSLs) that have a range of immunomodulatory functions, including inhibition of chemotaxis and phagocytosis8, although the function and full range of biological activities for all SSLs are not known. Through a broad range analysis of staphylococcal molecules that inhibit matrix metalloproteinases, endopeptidases involved in inflammatory cell recruitment and migration, SSL1 and SSL5 were identified as broad range inhibitors of matrix metalloproteinases, hindering the motility and chemotaxis of neutrophils62. This represents the first identified role of SSL1 and identifies an additional role for SSL5 (ref. 63). Recently, a second function was identified for SSL1 exhibiting in vitro protease activity and in a rabbit ocular infection model64. The study of the contributions of SSLs to immune evasion and pathogenesis has been hampered because SSLs are not expressed in mouse infection models. Overexpression vectors have been used to study SSLs in these models and have confirmed a role for SSL3 in staphylococcal pathogenesis via toll-like receptor 2 inhibition65. Contrary to the understanding that SSLs function as immune evasion factors, it seems that one SSL at least (SSL13) induces neutrophil responses following infection via the formyl-peptide receptor 2 to stimulate attraction and activation of neutrophils66. This result further highlights the complex interactions of immune modulating molecules during S. aureus infection.

Antibodies are key to the human immune response against bacterial infections, through direct bacterial binding, which results in complement activation through the formation of IgG clusters67. All clinical S. aureus strains produce protein A (SpA), which is a sortase-anchored surface protein initially deposited in the cell envelope and released through activity of LytM, a cell wall hydrolase68, and inefficient sorting69. SpA binds the Fc region of IgG and the Fab domain of antibodies (Fig. 2). Binding to the Fc region of IgG hampers phagocytosis70, whereas binding to the Fab domains and cross-linking of B cell IgM receptors lead to B cell expansion and apoptosis, tempering adaptive immune responses against S. aureus8 (Fig. 2). More recently, the inhibition mechanism of complement activation by SpA has been described under highly purified conditions. Binding of soluble SpA to IgG was found to inhibit Fc–Fc contacts, thereby inhibiting IgG hexamerization and thus hindering complement activation71. Interestingly, spa expression and SpA production were more increased in a hospital-adapted clone of S. aureus (sequence type [ST]239 MRSA) than in a clone that is emerging in the community (ST398 MRSA), which suggests that higher expression in ST239 MRSA may contribute to the colonization and immune evasion phenotypes observed clinically72. Like Spa, the staphylococcal immunoglobulin-binding protein (Sbi) binds the Fc region of IgGs with high affinity via two surface exposed globular domains73. The release of Sbi exposes two additional domains that directly interfere with the activation of the alternative complement pathway74. Components of the complement cascade are also the targets of S. aureus serine protease-like B protein and the metalloproteinase aureolysin, which are active in human serum and shown to block neutrophils opsonophagocytosis75 (Fig. 2).

Fig. 2: Evasion of immunoglobulin and complement-mediated immunity.
figure 2

Binding of IgG Fc domain by secreted protein A (SpA) and staphylococcal immunoglobulin-binding protein (Sbi) protects Staphylococcus aureus from phagocytosis and inhibits complement activation. Protein A also functions as a B cell superantigen, binding to the Fab domain of IgM and cross-linking VH3-type B cell receptors (BCRs). In conjunction with aureolysin, the secreted protease serine protease-like protein B (SplB) degrades complement components, thus hampering complement activation and opsonophagocytosis. C1q, complement component 1q.

Access to large-scale human genome sequencing capacity is also starting to reveal host genetic factors linked to staphylococcal diseases (Box 2). How our deepening understanding S. aureus immune evasion is starting to inform new therapeutic exploration is discussed in Box 3.

Role of biofilms and immunometabolism

In addition to the host evasion factors described in the previous section, biofilm formation is another mechanism that contributes to the pathogenesis of S. aureus76. Biofilms on biomedical devices, such as catheters and medical implants, or on host tissues such as wounds or heart valves cause notoriously difficult-to-treat infections77. By embedding bacteria in an extracellular matrix consisting of proteins, polysaccharides and extracellular DNA, biofilms confer protection against antimicrobial therapeutics and host immune clearance77. Nonetheless, secretion of factors including haemolysins, nucleases and PSMs are also important mediators of host evasion during the biofilm lifestyle and could shift immune responses towards an anti-inflammatory state78,79.

The contribution of the leukocidins Panton–Valentine leukocidin (PVL) and S. aureus γ-haemolysin (HlgAB) to bacterial survival in biofilms was demonstrated using a porcine wound model17. High-level secretion of PVL and HlgAB in biofilms elicited the formation of NETs by neutrophils (Fig. 1). Although NETs trapped and killed planktonic bacteria, they had no activity against S. aureus in biofilm17. S. aureus nuclease (Nuc) degraded the DNA content of NETs, which is likely to have promoted the dispersal of staphylococcal cells from the biofilm, increasing the risk of metastatic infections76. Additionally, LukAB has been shown to contribute to evasion of phagocyte-mediated killing of S. aureus in biofilm76.

S. aureus biofilms skew the host immune response towards an anti-inflammatory state, thereby promoting bacterial persistence instead of clearance (Fig. 3). This is evidenced by the recruitment of myeloid-derived suppressor cells (MDSCs)80,81,82,83 and the polarization of anti-inflammatory M2-like macrophages84. MDSCs are typically prominent in tumour microenvironments and notorious for their immunosuppressive effect, inhibiting T cell activation and promoting tumour progression (reviewed elsewhere85). Analysis of purified MDSCs recovered from S. aureus biofilm-associated orthopaedic infection revealed increased expression of anti-inflammatory mediators such as arginase 1 and IL-10 (ref. 82). Furthermore, depletion of MDSCs promoted bacterial clearance82.

Fig. 3: Staphylococcus aureus biofilm shapes the phenotype of leukocytes.
figure 3

The left panel shows metabolic reprogramming, characterized by increased glycolysis, in macrophages infected with live planktonic bacteria or stimulated with pathogen-associated molecular patterns such as lipopolysaccharide (LPS) recognized by pathogen recognition receptor (PRR). This skews the macrophage towards a pro-inflammatory phenotype (that is, the production of the pro-inflammatory cytokine IL-1β), which promotes bacterial clearance. The right panel shows two different strategies through which staphylococcal biofilms evade innate immunity. These include either rewiring host metabolism or epigenetic changes. S. aureus biofilm, which is induced by host-derived itaconate, promotes mitochondrial oxidative phosphorylation (OXPHOS) in macrophages and an anti-inflammatory phenotype. The mechanism remains unknown (indicated by the question mark). Lactate produced by S. aureus biofilms is transported into macrophages and myeloid-derived suppressor cells (MDSCs) through the monocarboxylate transporter 1 (MCT1), inhibits histone deacetylase 11 (HDAC11) and promotes IL-10 production. Ac, acetylation; CoA, coenzyme A; TCA, tricarboxylic acid.

Typically, macrophages become activated on recognition of bacteria, with the ensuing inflammatory response driven by major cellular metabolic changes, a process termed immunometabolism86. These changes include increased host cell glycolysis, impaired tricarboxylic acid (TCA) cycle and mitochondrial oxidative phosphorylation (OXPHOS), all resulting in the cellular accumulation of succinate and itaconate, which are two TCA metabolites87,88. Succinate is a pro-inflammatory metabolite that stabilizes the transcription factor, hypoxia-inducible factor 1𝛼, which enhances the production of the pro-inflammatory cytokine IL-1β; whereas itaconate is an anti-inflammatory metabolite that restores homeostasis following succinate-mediated inflammation87,88. Itaconate exerts its immunomodulatory role by alkylating cysteine residues and inhibiting several targets including the NLRP3 inflammasome, glycolytic enzymes and succinate dehydrogenase (SDH) mitochondrial complex II88,89,90. Interestingly, rather than stimulating inflammatory macrophages, S. aureus biofilms induce the polarization of macrophages into an anti-inflammatory state, with a metabolic bias towards increased OXPHOS and decreased aerobic glycolysis84 (Fig. 3). Importantly, the delivery of nanoparticles containing oligomycin, which is an OXPHOS inhibitor, in a mouse model of prosthetic joint infection resulted in the polarization of inflammatory monocytes and enhanced bacterial clearance. This demonstrates the potential of immunometabolic therapy in treating persistent staphylococcal infections (Box 3).

The increased expression of anti-inflammatory IL-10 by monocytes and MDSCs recruited to biofilms has been attributed to staphylococcal lactate, a metabolite that inhibits histone deacetylase 11 (HDAC11), thus enhancing IL-10 gene transcription91. This example highlights the crucial contribution of active bacterial metabolism and its crosstalk with host metabolic activities in determining the outcome of infection. Although mechanisms promoting immune evasion in biofilm-associated infections are being revealed92, the factors driving S. aureus biofilm formation in vivo are poorly defined. A recent study exploring the role of host immunometabolism during S. aureus pulmonary infection identified the inhibitory effect of the host metabolite itaconate on bacterial glycolysis as a biofilm-promoting cue93.

Itaconate is produced by the enzyme ACOD1 (also known as IRG1) in myeloid cells, particularly in response to lipopolysaccharide stimulation94. Of note, despite lacking lipopolysaccharide, S. aureus induces the production of host itaconate via the stimulation of mitochondrial oxidant stress93. Itaconate in turn inhibits the staphylococcal glycolytic enzyme aldolase, mirroring one of its interactions in mammalian cells90. Given the preferential reliance of S. aureus on glucose consumption and glycolysis for growth from carbon catabolite repression, the inhibition of glycolysis by itaconate rewires S. aureus metabolism, directing carbon flux towards biofilm formation and a sessile lifestyle. In an infection context, the consequence of metabolic adaptation for biofilm formation was revealed in a longitudinal study of S. aureus isolates colonizing the airways of a patient with cystic fibrosis over 15 years. These isolates displayed increasing production of biofilms in the presence of itaconate, consistent with the role of this metabolite in driving bacterial adaptation via biofilm formation93. Although the mechanism through which itaconate promotes persistent staphylococcal pulmonary infection and its clinical relevance requires further exploration, this study provides an intriguing insight into the immunometabolic pressures that support pathogen immune evasion adaptive mechanisms, akin to antibiotic selective pressure.

S. aureus small colony variants

The subversion of host immunometabolism by pathogens profoundly influences the infection outcome, as highlighted in the previous section. The compounding effect of accumulated metabolites on the epigenetic landscape of immune cells adds a further level of complexity to host–pathogen interactions. This is best exemplified by trained immunity or ‘innate immune memory’, which refers to increased innate immune protection against a secondary challenge following primary infection. Trained macrophages undergo substantial metabolic changes including the hallmark increase in glycolysis, and they accumulate metabolites that induce epigenetic rewiring of cellular function95. Accumulated fumarate inhibits KDM5 histone demethylases, which promotes trimethylation at the fourth lysine residue of the histone protein H3 (H3K4me3) at the promoters of pro-inflammatory cytokine genes such as those encoding tumour necrosis factor and IL-6, thus inducing their rapid production on restimulation and secondary infection.

In a mouse skin infection model, priming by previous infection with wild type S. aureus USA300 isolate LAC reduced skin lesion severity and bacterial burden on secondary infection. This protection was highly localized and mediated by macrophages96,97. Importantly, adoptive transfer of wild type S. aureus-primed macrophages into naive mouse skin afforded protection in vivo. However, priming by an S. aureus SCV prototype, ∆hemB, or a host-adapted isolate from a patient with atopic dermatitis did not confer protection from secondary infection98,99. SCVs, which are often associated with chronic infections, were first characterized in 1995 as phenotypically distinct from normal S. aureus colonies100. SCVs are slow-growing, have pinpoint colony size and may carry mutations in genes associated with electron transport chain components such as menadione and haeme22. As a result of these mutations, SCVs often have altered bacterial metabolism with decreased TCA cycle or OXPHOS activity and increased glycolysis to meet their energy requirements101. The ability of SCVs to abrogate trained immunity and cause recurrent infection was mediated by increased expression of fumC, which encodes the enzyme fumarate hydratase to degrade fumarate99. Increased fumC expression was also observed in S. aureus isolates from patients with atopic dermatitis and cystic fibrosis98,102. The blunting of immunity by host-adapted S. aureus isolates remains to be explored in the persistently colonized organs, such as the lungs of patients with cystic fibrosis. Genomics-enabled studies are providing further insights into SCV genetic mechanisms such as chromosomal rearrangements103. Wider uptake of such approaches is likely to accelerate the discovery of other bacterial genomic modifications occurring in SCVs and arising in response to immune selective pressures and promoting persistent infections (Supplementary Box 2 and Supplementary Fig. 2).

S. aureus host specificity and adaptation

Understanding host adaptation is providing insights into host-specific immune evasion mechanisms and by extension enhances our understanding of disease at the species level. It is well recognized that S. aureus is a multispecies pathogen, able to cause disease in humans and livestock. The diseases in livestock include mastitis in cows, goats, sheep and rabbits, skin infections in pigs and rabbits and invasive infections in chickens104. Advances in agricultural practices are increasing transmission risks between humans and livestock. For instance, the threat to human health from livestock-associated S. aureus has been well described for the pig-associated clones, especially CC398 (ref. 105) as well as ST9, ST5 (ref. 106) and ST93 (ref. 107); however, key immune evasion factors (scn, chp, sak, sea, which are encoded by the immune evasion complex) are observed in isolates from humans but not from pigs. Other clones from livestock can also spread in humans108 and exemplified by the recent description of the emergence of MRSA in non-human mammals, in this case European hedgehogs, which subsequently spread to livestock and humans, before the use of methicillin as an antibiotic109.

Host adaptation can be shaped by genetic drift (a process of neutral diversification) or adaptive evolution, whereby beneficial mutations are selected or deleterious mutations removed in the new host110. Genetic drift is typically associated with reduction of the effective population size as it occurs in ‘bottlenecks111. For example, in the transition from colonizing to invasive pathogen, only a few cells of a more diverse S. aureus colonizing population survive within a macrophage or other phagocytic cell (bottlenecks) and then subsequently expand to cause invasive disease112. By contrast, adaptive evolution results in an increase in frequency of the beneficial variant. It is likely that both have a role during infection and adaptation113. Combining comparative population genomics including both human and animal isolates with functional genomic studies will illuminate the staphylococcal drivers of host adaptation and determine the host specific factors that support bacterial survival114. Such information will guide the design of strategies that prevent the emergence of new pandemic clones and cross-species transmission.

Over the past decade, genomics studies have provided a comprehensive mapping of S. aureus accessory genome elements that are specific to certain host species and enriched the understanding of S. aureus host adaptation115, revealing that immune evasion factors targeted to certain hosts were specifically produced by host-adapted S. aureus lineages. Larger scale analyses of S. aureus population genomics across multiple host species and humans demonstrated that the latter were the major donors for S. aureus host-switching events, and that cows represent a potential reservoir of new epidemic clones in humans104. Understanding the gene flow in different ecological niches highlights host-specific gene repertoires that are linked to host adaptation, with S. aureus accessory genome elements from isolates from the same host species highly correlated even across diverse S. aureus clonal complexes104. By contrast, core genome evolution can be traced through the assessment of adaptive mutations that result from diversifying selection, with some biological pathways found to be under positive selection in humans (Fig. 4 and Supplementary Fig. 3). Antimicrobial resistance elements are another component of the accessory genome which can reveal unique host species patterns. Unsurprisingly, human isolates are enriched for elements that provide resistance to antimicrobials used in humans such as β-lactams, whereas tetracycline and antiseptic resistance are significantly associated with pig isolates104.

Fig. 4: Staphylococcus aureus host species adaptation over time.
figure 4

a, Host-switching modelling of Staphylococcus aureus between species identified humans as a major transmission hub104. Overlayed are reported host-specific mobile genetic elements (MGEs) in pigs118,119, cows119,120, small ruminants121,122, horses123, birds119 and hedgehogs109; and their major associated clonal complexes (CCs). MGEs belonging to the non-dominant S. aureus lineages in pigs and cows have been reported, these CCs and sequence types (STs) are denoted in superscript. S. aureus infecting rabbits has no unique MGEs. A single nucleotide polymorphism (C338A) in chromosomal gene, dtlB, causes a non-synonymous mutation required for infection of rabbits by CC121 (see Supplementary Fig. 3c for all CC121 rabbit adaptations). No unique S. aureus MGEs have been described in rodents or carnivora. Globally disseminated, multihost lineage CC398 has been reported in rodents and carnivora, but no CCs are recognized as dominant in these species (see Supplementary Fig. 3d for the mechanism of broad host distribution for CC398). Human S. aureus strains possess the ϕSa3 prophage (β-haemolysin-converting phage) encoding a human-specific evasion cluster that integrates into hlb, causing loss of β-toxin production. Many animal-adapted S. aureus strains have lost ϕSa3, resulting in restored β-toxin production and increased fitness. Linewidth represents frequency of host jumps. b, Function of S. aureus genes carried on animal-adapted MGEs. S. aureus superantigens are virulence factors that cause nonspecific T cell activation and massive cytokine release. This excessive activation of the host immune system paradoxically promotes infection; antibiotic resistance is conferred by multiple resistance determinants; species-adapted leukodicins, encoded by luk, cause host neutrophil lysis. von Willebrand-binding protein (vWbp) causes host-specific plasma coagulation; adhesion protein (Bap) promotes S. aureus adhesion to bovine mammary mucosa and biofilm production resulting in colonization, contributing to the pathogenesis of bovine mastitis. Putative adhesin, LPXTG-surface protein promotes colonization (see Supplementary Fig. 3a for bovine-specific S. aureus adaptions); staphylococcal complement inhibitor (SCIN) encoded by scn blocks the activation of the host complement system; Avian-specific scpA encodes the thiol protease staphopain A, which is associated with enhanced pathogenicity through the inhibition of host neutrophil activation and chemotaxis (see Supplementary Fig. 3b for S. aureus CC5 adaptations in chickens). #Putative adhesin. ϕSabov-vSa, bacteriophage ϕ-S. aureus bovine in genomic island v-S. aureus; LPXTG, conserved peptide motif in surface linked proteins cleaved between the threonine and glycine residues and covalently attached to cell wall peptidoglycan; SCCmecC, staphylococcal cassette chromosome methicillin C resistance gene; TSST-1, toxic shock syndrome toxin-1; vSaα, genomic island v-S. aureus-α.

The exploration of host-specific adaptation in experimental models could provide robust validation of genetic signatures that have been identified by population genomics analyses. An exemplar of a targeted approach is the study of ST121 S. aureus, which causes serious infections in both human and rabbit hosts. Bayesian modelling of genomic data suggests a human-to-rabbit host jump about 50 years ago, with genome diversification between host species since then116. Although no specific mobile elements were identified in the genome of rabbit-associated S. aureus ST121, the functional interrogation of mutations between host strains identified a single point mutation in the dltB gene. Such mutation was sufficient to cause the rabbit disease phenotype in animal models116; however, the mechanisms by which dltB mutations result in this host adaptation are yet to be defined (Fig. 4 and Supplementary Fig. 3).

The ability of S. aureus to adapt to new hosts, and the capacity to adapt to population bottlenecks, has been further explored in a model of human-to-ovine host switching complemented by large-scale genomic analyses117. This revealed that S. aureus could overcome substantial population bottlenecks and acquire beneficial mutations over time. However, there was not a strong signal of convergence, which suggests that multiple pathways to adaptation were occurring in this model. The exploitation of host-switching models will further enhance the understanding of adaptive mechanisms underlying species-specific pathogenesis.

Conclusions and outlook

S. aureus remains a formidable multispecies pathogen and major challenge to human health. Fuelled by the major progress in omics, important advances have been made in the past 10 years to understand the duality of S. aureus as an asymptomatic colonizer and devastating invasive pathogen. Although new immune evasion molecules and novel ligands for well-described molecules have been identified, a holistic understanding of the in vivo dynamics and interactions within host will be unlikely revealed by a traditional one-by-one gene discovery approach. The variability in the immune evasion repertoire across S. aureus clones and the heterogeneity of host responses substantially complicate this issue, particularly in complex clinical contexts. Thus, systems-level experimental approaches that account for S. aureus strain variation, combined with improved laboratory infection models (perhaps even controlled human infection or colonization models) that consider immune response and microbiome context, are needed to better understand the roles of critical S. aureus factors during infection. Such advances could help address the issues that have challenged S. aureus vaccine development such as immune imprinting resulting from previous S. aureus exposure.

With its capacity to switch between host species, transition from colonization to invasion, or developing antibiotic resistance and persistence, S. aureus has the propensity to adapt to new environments. An integrated understanding of how S. aureus genomics, evolution and adaptation intersect with virulence and immune evasion strategies will guide the design of prevention strategies and therapeutic approaches. Other potential avenues could stem from the identification of human genetic signatures associated with S. aureus disease risk as well as from the microbiome composition of the skin, the nose and the gut.

It is also becoming increasingly apparent that immunometabolism has a key role in the pathogenesis of S. aureus infections. However, microbial metabolism has often been neglected, with many research groups instead using pathogen-associated molecular patterns as a proxy for bacterial stimulation. There is increasing evidence that staphylococcal metabolic flexibility is not only critical for bacterial persistence during infection but also in shaping the host immunometabolic response. Importantly, this metabolic interplay between host and S. aureus varies by infection site. Targeting immunometabolism as an alternative or complementary therapeutic strategy to enhance multidrug-resistant S. aureus clearance will become more realistic as our understanding of host–pathogen metabolic interactions deepens.

Ultimately, advancing the study of S. aureus host–pathogen interactions by acknowledging the roles of other microbiota and immunometabolism, and further understanding pathogen genetic variability and adaptation will generate holistic insights that lead to meaningful clinical impact on diseases caused by S. aureus.