The capacity of S. aureus to adhere to cells has been known since the early 1980s (Aly et al., 1980) and has been demonstrated for both primary cells and cell lines from various tissues (Ellington et al., 1999; Kerro Dego et al., 2002). Host cell adhesion mainly involves Fn forming a bridge between α5β1 integrin on the cellular side and Fn binding proteins (FnBPs, which are MSCRAMMs) on the bacteria (Tran Van Nhieu and Isberg, 1993; Sinha et al., 1999; Fowler et al., 2000; Grundmeier et al., 2004). This step is a prerequisite for any internalization into NPPCs; indeed, DU5883, an isogenic mutant of S. aureus NCTC 8325-4 defective in FnBP expression, cannot invade NPPCs (Fowler et al., 2000).

There are two FnBP isoforms in S. aureus, FnBPA and FnBPB, encoded respectively by the fnbA and fnbB loci, with very similar domain organizations and sequences (Jönsson et al., 1991; Burke et al., 2010). However, their presence varies across the population (Peacock et al., 2000). They consist of an amino-terminal secretion signal sequence followed by an A domain that is closely related to fibrinogen-binding protein clumping factor A (ClfA, Figure 1A). This domain can bind fibrinogen and elastin and is involved in Fn binding (Wann et al., 2000; Roche et al., 2004; Burke et al., 2011). The A domains of FnBPA and FnBPB only share 40% sequence identity (Burke et al., 2010). The A domain is followed by tandem repeats of Fn-binding regions (FnBRs, 95% identity between FnBPA and FnBPB), 11 in FnBPA versus 10 in FnBPB (Figure 1A). This additional FnBR, along with the higher Fn affinity of certain FnBRs, might explain FnBPA’s higher overall affinity for Fn and the fact that FnBA alone is sufficient for adhesion and cell invasion (Grundmeier et al., 2004; Testoni et al., 2011). Finally, the C-terminal peptidoglycan-binding motif (LPXTG) and the wall and membrane spanning domains anchor FnBPs to the cell wall.

In vivo, the deletion of either FnBP strongly attenuates the ability of strain SH1000 to colonize the kidney and multiply, and to cause fatal sepsis in mice. Although FnBPA contributes more to these symptoms than FnBPB does, both FnBPs appear necessary for the development of severe infections (Shinji et al., 2011). This observation suggests that cooperation between the two FnBPs may be required for strong cell adhesion or to efficiently trigger the internalization pathway in vivo.

The high affinity and specificity of FnBPs for Fn is conferred by the tandem β-zipper structure they form together (Fowler et al., 2000; Schwarz-Linek et al., 2003). Fn bridges S. aureus and the host cell through its binding to cellular α5β1 integrin (Fowler et al., 2003). The efficiency of this adhesion varies both between cell types—one high-affinity FnBR is sufficient for adhesion and invasion in endothelial cells (Edwards et al., 2010) but three are needed in keratinocytes (Edwards et al., 2011)—and S. aureus strains. Indeed, whereas studies have shown that isolates from infections of cardiovascular devices have amino acid changes in the FnBR domain that increase the affinity for Fn (Hos et al., 2015), this is not the case for isolates from prosthetic joint infections (Eichenberger et al., 2015). Likewise, certain methicillin-resistant S. aureus strains involved in endovascular infections harbor an additional FnBR that, together with a substitution in FnBR 11, reduces its affinity for Fn but promotes cell invasion. Although this fits with the hypothesis that the pathogenesis of endovascular infections involves the invasion of endothelial cells, it seems to go against the idea that better adhesion promotes internalization. However, it is estimated that one FnBP can bind 6–9 Fn molecules (Bingham et al., 2008) and thus could cluster α5β1 integrins on the cell surface to trigger the efficient intracellular signaling required for internalization. One hypothesis is that when affinity for Fn decreases, more interactions are needed to achieve sufficient cell adhesion, mobilizing more Fn and α5β1 integrins and thus leading to their clustering.

The signaling pathway of staphylococci internalization involves focal adhesion kinases (FAKs) and activated Src (Fowler et al., 2003; Agerer et al., 2005) that subsequently recruit cortactin to promote actin polymerization and mobilize the endocytic machinery (Figure 1B) (Agerer et al., 2005; Selbach and Backert, 2005). Downstream of the FAK-Src pathway, the activation of PI3K and Akt is also important for the internalization of S. aureus (Oviedo-Boyso et al., 2011; Wang et al., 2013), although the molecular pathway is still far from being fully understood. Previous studies have shown that internalization is inhibited by cytochalasin D (Ellington et al., 1999; Sinha et al., 1999) and is temperature dependent (Sinha et al., 1999, specifically, inhibited at 4 and 14°C and facilitated at 37°C versus room temperature). These results respectively show that the biological prerequisites for internalization are (1) a dynamic actin cytoskeleton and (2) a fluid host cell membrane. Since bacterial uptake can occur with heat-killed or fixed bacteria, this mechanism appears to be an active process on the cellular side only. However, it has also been shown that S. aureus can stimulate its own uptake by upregulating β1 integrin expression in the host cell by secreting α-hemolysin. In this case, the bacteria must be viable as the process requires both cellular and bacterial activity (Abel et al., 2011; Goldmann et al., 2016).

In summary, S. aureus adhere to cells via interactions between FnBPs, Fn, and α5β1 integrins. The resulting clustering of integrins may then be sufficient to trigger the signaling cascade involving FAK, Src, PI3K, and Akt. Finally, S. aureus mobilizes the actin cytoskeleton and possibly the endocytosis machinery to enter host cells (Figure 1).

The FnBP-Fn-α5β1 integrin pathway is widely acknowledged to be the main internalization process. However, other factors have been shown to affect the efficiency of this internalization. In epithelial cells indeed, the internalization efficiency is maximal when the bacterial FnBPs interact directly with Hsp60 at the cell surface (Dziewanowska et al., 2000). Whether Hsp60 acts as a co-receptor to strengthen the binding of FnBPs-Fn to α5β1 integrin or is involved in signal transduction has still to be investigated (Figure 1C). Another clue that the FnBP-Fn-α5β1 integrin pathway is not the only internalization mechanism is that blocking α5β1 integrin or the binding of Fn by FnBPs with an antibody does not fully prevent internalization by MG63 osteoblast cells (Testoni et al., 2011) and by primary keratinocytes (Kintarak et al., 2004). The latter even internalize the FnBP-defective strain DU5883. Together, these results suggest that alternative mechanisms are involved in the adhesion and internalization of S. aureus (Table 1).

These mechanisms mainly involve bacterial serine aspartate repeat-containing protein D (SdrD), clumping factor A (ClfA), autolysin (Atl), and serine-rich adhesin for platelets (SraP). These proteins are MSCRAMMs and (except for Atl) have the cell-wall anchoring sequence LPXTG located in their C-terminal portion. Their N-terminal part contains a signal peptide for their secretion, followed by a ligand binding region mainly consisting of repeated sequences often rich in Serine. The mechanisms involving these proteins are Fn-independent: SdrD binds directly to Desmoglein 1 on the cell surface of keratinocytes and desquamated nasal cells (Corrigan et al., 2009; Askarian et al., 2016); ClfA interacts directly with host cells or through fibrinogen bridges (McDonnell et al., 2016; Claes et al., 2017); Atl seems to mediate S. aureus internalization via direct interactions with Hsc70 (Hirschhausen et al., 2010); and SraP adheres to A549 cells through the salivary scavenger protein gp340 (Yang et al., 2014) (Figure 1C). Note that the SdrD and Atl mechanisms require both bacterial and cellular activity as their efficiency depends on the expression of the bacterial and of the cellular interactor. Indeed, SdrD expression is upregulated following contact with neutrophils (Sitkiewicz et al., 2011) and Hsc70 production is stimulated by S. aureus infection of EA.hy 926 cells (Hirschhausen et al., 2010). Moreover, Atl has also been shown to be involved in the secretion of several S. aureus proteins, among which SdrD (Pasztor et al., 2010). Atl could thus have an “active” role as a secondary mechanism of internalization in some cells and act as a regulator of those secondary mechanisms in other cell types. More generally, these observations might reflect the fact that the bacteria can adapt their internalization strategy to the environmental conditions, i.e., an absence or scarcity of Fn, by finding alternative binding partners and/or by upregulating one side or the other of the adhesion machinery (e.g., SdrD and Atl). For instance, ClfA binds to annexin 2 on the surface of bovine mammary epithelial cells (MAC-T cells) (Ashraf et al., 2017) but has two receptors on the surface of endothelial cells, namely αvβ3 integrin using fibrinogen as a bridge (McDonnell et al., 2016) and von Willebrand factor using self-secreted von Willebrand factor binding protein as a bridge (Table 1 and Figure 1C) (Claes et al., 2017). This might increase its capacity to adhere to and possibly enter endothelial cells, thereby supporting its role in the pathogenesis of endocarditis. Fibrinogen binding also primes platelets aggregation and abscess formation and can lead to thromboembolic lesions in the heart during sepsis (McAdow et al., 2011; Malachowa et al., 2016).

In other cases, the alternative mechanisms may support the FnBP-Fn-α5β1 integrin-mediated uptake of S. aureus instead. This is notably illustrated by S. aureus extracellular adherence protein (Eap) that plays a role in the adherence to fibroblasts and epithelial cells independently of any binding to Fn or fibrinogen (Figure 1C) (Hussain et al., 2002). Recently, its deletion has been shown not to affect the adhesion and internalization steps, contradicting the early idea that its role was to compensate for FnBP function loss (Palma et al., 1999; Grundmeier et al., 2004). Rather, Eap seems to promote the adhesion and internalization of S. aureus and other pathogenic bacteria encountered in the context of polymicrobial skin infection (Bur et al., 2013). As for all the previously cited mechanisms, the molecular and signaling pathways underlying this activity are not understood yet and more work is needed to clarify whether Eap (1) is secreted and binds to α5β1 integrin via Fn, triggering the internalization cascade pathway and enhancing the uptake of bacteria bound to molecules other than α5β1 integrin in cells that express this integrin poorly; or (2) triggers actin-dependent phagocytosis, as its effect is fully blocked by cytochalasin D.

The need to elucidate the molecular and signaling pathways of these alternative mechanisms is highlighted by their phenotypical relevance. ClfA and SraP are involved in the pathogenesis of endocarditis. SdrD-mediated binding appears to help the bacteria survive in vivo (Askarian et al., 2017)—whether this is because of an increased ability to kill neutrophils and/or to invade cells is an open question. Also, unraveling whether internalization occurs via the same FAK/Src/PI3K/Akt cascade as for the FnBP-Fn-β1 integrin pathway is essential as this may offer a potential avenue to treat chronic infections.

The minor mechanism involving GapC during the adhesion to and invasion of MAC-T cells is also noteworthy as it has been involved in the development of mastitis (Kerro-Dego et al., 2012). Finally, another point to bear in mind is that in addition to adhesins, other (probably less expected) molecules are involved in the adhesion to and invasion of NPPCs by S. aureus. In particular, the Tet38 efflux pump extrudes tetracycline and unsaturated free fatty acids but also interacts with CD36, a cellular transporter of long chain fatty acids, to trigger S. aureus adhesion and entry into A549 cells (Truong-Bolduc et al., 2017). Tet38 is also involved in phagosomal escape, facilitating the replication and persistence of S. aureus.