Intracellular Invasion by Streptococcus pyogenes: Invasins, Host Receptors, and Relevance to Human Disease

ABSTRACT

The human oral-nasal mucosa is the primary reservoir for Streptococcus pyogenes infections. Although the most common infection of consequence in temperate climates is pharyngitis, the past 25 years have witnessed a dramatic increase in invasive disease in many regions of the world. Historically, S. pyogenes has been associated with sepsis and fulminate systemic infections, but the mechanism by which these streptococci traverse mucosal or epidermal barriers is not understood. The discovery that S. pyogenes can be internalized by mammalian epithelial cells at high frequencies (1–3) and/or open tight junctions to pass between cells (4) provides potential explanations for changes in epidemiology and the ability of this species to breach such barriers. In this article, the invasins and pathways that S. pyogenes uses to reach the intracellular state are reviewed, and the relationship between intracellular invasion and human disease is discussed.

Streptococcus pyogenes has evolved a variety of both surface-bound and extracellular factors that alter the inflammatory response and impair phagocytic clearance of the bacteria. The more than 150 genotypes use both similar and different strategies at the biochemical level to colonize their host and avoid protective defenses. These are reviewed elsewhere (5). Intracellular invasion is dependent on at least two classes of surface proteins, the M proteins (6, 7) and fibronectin (Fn)-binding proteins (3, 8). The M proteins serve many functions in the pathogenesis of S. pyogenes, including resistance to phagocytosis, adherence, and intracellular invasion (5, 9). The Fn-binding protein F (PrtF) and the allelic variant SfbI are also adhesins and invasins, produced by 50 to 60% of the M genotypes. The function of these proteins in the context of intracellular invasion is described below.

IMPACT OF INTRACELLULAR INVASION ON S. PYOGENES INFECTION

Many geographic and temporal clusters of sepsis and toxic shock, caused by a few serotypes of S. pyogenes, have been reported (10, 11). An M1 subclone was associated with many such clusters (10, 12). Surprisingly, devastating soft tissue infections were often reported in patients who had not experienced previous trauma or wounds that might initiate systemic spread of the organism (11). Cleary and coworkers (2, 13) considered the possibility that intracellular invasion of mucosal surfaces provides a window for streptococci to reach deeper tissue, even the bloodstream. Indeed, a globally disseminated M1 subclone invaded A549 epithelial cells at significantly higher frequency than did other subclones of M1 streptococci. The source of M1 subclones, i.e., uncomplicated disease or more invasive disease (blood isolated), did not correlate with efficiency of ingestion by epithelial cells (13). Only the M1 subclone that had acquired 70 kb of DNA sequence from two distinct prophages (13) was associated with high-frequency invasion of cultured cells. To date, no relationship to prophages and intracellular invasion has been discovered.

Although an association between high-frequency intracellular invasion and systemic streptococcal disease is still uncertain, the intracellular state may significantly increase the capacity of this bacterium to disseminate in human populations. When strains from carriers and patients with uncomplicated pharyngitis and sepsis were compared, those from carriers were observed to be internalized by HEp2 cells at the highest frequency (14). A highly variable M1 culture could be enriched for more invasive streptococci by in vitro serial passage through human epithelial cells (15); therefore, in vivo cycling of streptococci between the interior and exterior of the mucosal epithelium may select for variants that are more efficiently internalized. From 10 to 40% of children continue to shed streptococci after treatment with penicillin (16). Brandt et al. reported 80% relapse, by the same initial strain following vigorous penicillin therapy, when 40 consecutive isolates from 18 patients were characterized (17). Treatment failures and subsequent carriage of streptococci were reported to be more common when the infecting strain produced PrtF (also termed SfbI), an invasin that promotes high-frequency ingestion of streptococci by epithelial cells (18). This correlation was not, however, confirmed by a smaller study (17). In vitro, intracellular S. pyogenes can resist at least 100 μg/ml of penicillin (Cleary, P. unpublished data); therefore, penicillin treatment may further select for strains that can be efficiently internalized by epithelial cells. High-frequency intracellular invasion may increase the rate of antibiotic therapy failure and therefore increase the size of the human reservoir that can disseminate the organism to others in the population. As the reservoir enlarges, the probability of serious, systemic infection also increases. Thus, strains or serotypes that are less able to acquire an antibiotic-free niche may be less often associated with severe disease. Facinelli and colleagues observed an association between genetic resistance to erythromycin and more efficient uptake of S. pyogenes by epithelial cells (19). They suggested the frightening possibility that erythromycin resistance had become genetically linked to efficient invasion of human cells and thus coselected by antibiotic therapy. A genetic link was not established, however, nor did the authors exclude the possibility that a single subclone had disseminated in the study population. A thorough analysis of a cluster of toxic shock in southern Minnesota showed that nearly 40% of the school children in nearby communities were carriers of a serotype M3 clone that was responsible for systemic disease in adults (20). These investigators suggested that school children served as the reservoir for S. pyogenes that produced toxic shock in adults with underlying physical disabilities.

ARE TONSILS A RESERVOIR FOR S. PYOGENES?

The best and most direct evidence that intracellular bacteria are an important source for dissemination of streptococci and the cause of recurrent tonsillitis is based on microscopic studies of surgically excised tonsils (16). This study found that 13 of 14 tonsils from children plagued by recurrent tonsillitis harbored streptococci within epithelial cells in the tonsillar crypts. S. pyogenes was also observed in macrophage-like cells at high frequency in these specimens. Tonsils from control subjects who had tonsils removed for other reasons did not contain streptococci. Intracellular streptococci were also observed in cultured tonsils infected with S. pyogenes (21). The invasive M1 subclone was confirmed in vitro to efficiently invade primary keratinized tonsillar epithelial cells (22). In a murine intranasal infection model this M1 subclone and an M49 strain were shown to specifically invade nasal-associated lymphoid tissue, tissue known to be functionally homologous to human tonsils (23). Initially, streptococci were associated with M-like cells, sporadically located along the base of the nasal epithelium, and by 24 h postinoculation had formed microcolonies throughout the lymphoid tissue. From 1 to 10% of the streptococci, depending on the mouse, were intracellular. This study suggests that M-like cells are the primary portals for transport of S. pyogenes across the nasal mucosa into lymphoid tissue. Subsequent studies also demonstrated that nasal-associated lymphoid tissue is the initial, primary location for antigen-specific T cell priming and activation following intranasal infection by S. pyogenes (24, 25).

S. PYOGENES INVASINS AND HOST RECEPTORS

The intracellular state has many advantages for a bacterium’s survival in a host, such as evading phagocytic killing, escaping neutralization by antibodies, and avoiding antibiotics. In many respects S. pyogenes’ invasion of human cells is analogous to that of Yersinia pseudotuberculosis, which involves a bacterial outer membrane protein, InvA, binding to integrins (26). Integrins are a mammalian family of heterodimeric, transmembrane glycoproteins that bind to cell-secreted extracellular matrix (ECM) proteins, a collection of extracellular molecules, such as collagen, Fn and laminin (Ln), secreted by mammalian cells (27). Engagement of integrins by ECM proteins results in activation of host signal transduction pathways that lead to rearrangements of host cell cytoskeleton, cell membrane movement, and other cellular activities (28). Bacterial invasins are a subclass of bacterial adhesin molecule required for ingestion by host cells. Typically, invasins are proteins expressed on the surfaces of bacterial cells that directly or indirectly recognize specific host cell receptors (26). In general, invasins are capable of inducing reorganization of the host cell’s cytoskeleton by generating specific signals from receptors with which they interact. Integrins are often exploited for microbial entry into mammalian cells. This may be due to their ubiquitous expression and their ability to affect cytoskeletal arrangement (27).

S. pyogenes has evolved multiple surface molecules that bind to various ECM proteins, which tether them to and activate integrin signaling pathways required for invasion. In addition, this bacterium has also evolved multiple mechanisms for invasion of a wide variety of mammalian cells. The host cell receptors, bacterial ligands, and biochemical pathways vary, depending on the strain of streptococcus and host cell type, adding another layer of complexity (6).

Streptococcal adhesins are a variety of surface molecules which can be grouped into four families, depending on their association with their surfaces: (i) those that are covalently linked by their C-terminus to the cell wall peptidoglycan through an LPxTz motif, (ii) those that are tethered to the bacterial cell membrane through N-terminal modifications of lipids to form lipoproteins, (iii) those that are bound to the bacterial surface by noncovalent interactions, and (iv) those that are expressed and retained on the surface by an as yet unknown mechanism. The most prominent adhesins belong to the family of cell wall-anchored proteins that are covalently linked to the peptidoglycan by a membrane-associated transpeptidase called sortase A (29).

ZIPPER AND OTHER MECHANISMS OF STREPTOCOCCAL INGESTION

Cytoskeletal events accompanying the uptake of a variety of pathogenic bacteria have been studied. Ruffling of the membrane precedes the entry of Shigella and Salmonella spp. (26). In contrast, internalization by a zipper mechanism is mediated by interactions between surface invasins, ligands, and host cell receptors. These interactions lead to pseudopod formation, which requires the extension of actin filaments beneath the host cell membrane (7).

Scanning electron microscopy revealed that adherent M1⁺ bacteria are often in close contact with host cell microvilli, suggesting that this association may be an initial step in internalization (Fig. 1A). Microvilli frequently extend across bacterial surfaces and appear to entrap streptococci. Microvilli increase in number near adherent streptococci and appear to morphologically expand into pseudopodia-like structures, although these structures may arise independently (Fig. 1B). Membrane engulfment of streptococci suggests a zipper mechanism of uptake, as shown in Fig. 1C. The M1 strain of streptococcus, like Listeria and Yersinia spp., can be internalized by this zipper-like mechanism (7, 26). In contrast to M1⁺ streptococci, M1^– bacteria are rarely observed to interact with microvilli, and no pseudopodia-like structures were seen (7). De novo actin polymerization is essential for uptake of these streptococci, whether M⁺ or M^– strains, because invasion is blocked by cytochalasin D, an inhibitor of actin polymerization (2). Confocal immunofluorescent microscopy confirmed that vacuoles containing internalized streptococci are surrounded by polymerized actin. Actin is occasionally associated with membrane-spanning chains and seems to disappear soon after entry.

FIGURE 1.

Scanning electron micrographs showing different stages of streptococci-HeLa cell interactions. (A) A high density of microvilli surrounding and in contact with adherent streptococci. (B) A common morphological change that microvilli undergo. The cup-like structure appears to gradually engulf the bacteria and ultimately pull them into a vacuole. (C) A streptococcal chain has been partially ingested by a mechanism that morphologically resembles a receptor-ligand interaction. It is not known whether microvilli are involved in the latter or whether streptococci are ingested by two physically different cytoskeletal rearrangements.

Other microscopic studies demonstrated that physical events leading to ingestion of S. pyogenes vary dramatically between strains and serotypes (30, 31). Strain A40 induces the accumulation of small-membrane omega-shaped cavities in the host cell membrane near adherent streptococci. Shortly thereafter, membrane invaginations are formed into which streptococci enter. The integrity of the membrane is not disrupted, and streptococci end up in a cytoplasmic vacuole. These cell membrane changes are dependent on Fn and α₅β₁ integrins, but focal adhesion complexes were not observed to contain polymerized actin (30). Further studies showed that these membrane structures are caveolae, structures previously found to mediate entry of both viruses and parasites into cells. In contrast, invasion of HEp2 cells by strain A8 (M40 Sfb1^–) was not inhibited by an Fn peptide analogue, nor by anti-α₅β₁ antibody. This strain induced dramatic aggregation of microvilli near and around attached streptococci, and ingestion of this strain was independent of caveolae and lipid rafts. However, invasion of HEp2 and A549 epithelial cells by the M1 strain is dependent on Fn and α₅β₁ integrins (22, 32). Comparison of strains A40 (M12 Sfb1⁺) with A8 (M40 Sfb1^–) revealed striking differences. It is unclear whether these differences reflect yet another mechanism of streptococcal uptake or merely depend on different experimental conditions. Irrespective of differences, this study further implicated participation of microvilli in the uptake of streptococci by epithelial cells.

M PROTEINS AND OTHER Fn-BINDING PROTEINS ARE INVASINS

S. pyogenes expresses an array of cell surface proteins that facilitate bacterial colonization of a variety of human tissues (33). Some are adhesins that also function as invasins. The proteins SfbI and F1 (PrtFI) are closely related streptococcal adhesins that bind Fn with high affinity. Approximately 50% of S. pyogenes isolates carry the gene encoding SfbI/PrtFI, although many do not express the gene under laboratory conditions (3, 14, 18). Prebinding of soluble Fn to PrtFI-bearing streptococci can promote bacterial adherence to cultured cells and dermal tissue. This and experimental evidence from several laboratories are consistent with SfbI/PrtFI promoting intracellular invasion via a similar mechanism.

Molinari et al. (3) demonstrated that SfbI can mediate invasion of HEp2 cells. Streptococcal invasion can be ablated by antiserum raised against SfbI or by preincubation of host cells with recombinant SfbI. Latex beads coated with recombinant SfbI readily adhere to and are efficiently ingested by epithelial cells, demonstrating that the interaction of SfbI with host cells is sufficient for internalization. Fn binding by SfbI is required for invasion, because beads coated with a recombinant SfbI peptide lacking the Fn-binding domains do not efficiently adhere to HEp2 cells. Ozeri et al. (8) reported that invasion of HeLa cells by PrtFI⁺ bacteria is dependent on addition of serum or purified Fn. In both studies, antibodies against Fn inhibited invasion. Also, PrtFI peptides, containing at least one Fn-binding domain, were found to inhibit PrtFI-mediated invasion. PrtFI binds to a 70-kDa N-terminal fragment of Fn. This region of Fn differs from the region bound by epithelial cells. The 70-kDa fragment can inhibit Fn-mediated invasion by competing with intact Fn for PrtFI binding; however, only intact Fn molecules are capable of supporting bacterial invasion (8). Antibody directed against the integrin β1 subunit specifically blocks Fn-PrtF1-mediated invasion of HeLa cells (8). This result suggests that one or more β1-containing integrins are involved in bacterial internalization. In contrast, invasion of GD25 (embryonic mouse stem) cells appears to be mediated by integrin αvβ3 (the major Fn-binding integrin of this cell line), since invasion is inhibited by a peptide that specifically blocks Fn binding to this integrin (8). Collectively, these results demonstrate that Fn functions as a molecular bridge between SfbI/PrtFI and host cell Fn receptors. At least 11 Fb-binding proteins of S. pyogenes have been identified (34), indicating that S. pyogenes has evolved multiple invasins to utilize the Fn-integrin mechanism for internalization of host cells.

Expression of Sfb1/PrtF1 is not a prerequisite for high-frequency intracellular invasion. For example, serotype M1 strains typically lack the gene encoding SfbI/PrtFI, but some can invade human epithelial cells with high efficiency (13, 18). Invasion of cultured cells by the globally disseminated M1 strain 90-226 is primarily dependent on expression of M1 protein. Inactivation of the M1 protein gene, emm1, decreased invasion of HeLa, A549, and HEp2 cells by 50-fold (7, 32), and latex beads coated with M1 are readily ingested by HeLa cells (7). These results indicate that invasion of epithelial cells by strain 90-226 is mediated in large part by M1 protein.

Expression of M1 protein is not sufficient for invasion, however, because bacterial internalization is dependent on exogenous serum, the ECM proteins, Fn or Ln (32). M1protein appears to be the major Fn-binding protein expressed by strain 90-226, because inactivation of emm1 reduces binding by 88%. Also, purified M1 protein can bind Fn in vitro (32). This protein is also able to bind collagen VI, an ECM molecule that is distributed underneath the respiratory epithelium in the upper and lower airways (35). However, this is not a general property of all M proteins; only three, M1, M6, and M3 proteins are known to bind Fn (32, 36; Cleary P. unpublished data). Interestingly, M3 protein was also found to promote serum-dependent invasion of HEp2 cells (6).

In the presence of Fn, internalization of M1 streptococci is dependent on the interaction of M1, Fn, and the epithelial cell Fn receptor α5β1 (Fig. 2). A monoclonal antibody that specifically blocks Fn binding to this integrin can ablate invasion of A549, HeLa, and human tonsillar epithelial cells (22, 32). Low-molecular-weight nonpeptidyl α5β1 antagonists are also effective inhibitors of invasion of A549 and human tonsillar epithelial cells (22). The inhibitory effects of α5β1 antagonists are not due to generalized effects on either bacterial or host cells, because the inhibitors do not abrogate Ln-mediated invasion. Rather, the inhibitory effects of α5β1 antagonists are observed only when bacteria are exposed to either serum or purified Fn. Thus, as in the case of Sfb1/PrtF1⁺ strains, Fn appears to function as a bridging molecule in promoting invasion by M1⁺ bacteria. Ln can replace Fn as the agonist in invasion assays (Fig. 2). Failure of Ln to promote invasion of A549 cells by M^– mutants suggests that this ECM protein binds sufficiently to M1 protein to serve as a bridge between streptococci and α3β1 integrins.

FIGURE 2.

The various agonist and cellular receptors which streptococci commandeer to promote their own phagocytosis by epithelial cells. Both the M protein and high-affinity fibronectin proteins SfbI/PrtF1 are known invasins which depend on formation of a fibronectin bridge between the bacterial surface and integrin receptors α5β1 or αVβ3. Although less well studied, laminin can also serve as an agonist for ingestion of streptococci by linking them to α3β1 integrins. For some strains fibronectin-mediated invasion of epithelial cells also requires interaction between M protein and the complement regulatory protein CD46.

The invasion efficiencies of M1 strains are not dictated solely by the presence of invasion agonists. LaPenta et al. (2) and Cleary et al. (15) reported that M1 isolates exhibit widely varying invasion efficiencies, even when experiments are performed in the presence of serum. These variations are likely due, at least in part, to varying levels of M1 protein expression. Transcription of emm1 can vary considerably between isolates and even between different cultures of the same isolate (15). While this possibility has yet to be thoroughly investigated, it is consistent with the finding of Ozeri et al. (8) that the number of integrin-binding sites on bacteria can greatly affect invasion efficiency. Expression of different M proteins, M6, M3, or M18, in a common streptococcal background demonstrated that M proteins can differ widely in their capacity to promote ingestion of streptococci. Serum and the cellular receptor CD46 (Fig. 2) were required for uptake of streptococci that expressed M3 protein, whereas neither were required when the recombinant expressed M6 protein (6). Efficient invasion of A549 cells by strain 90-226 also requires that the M1 protein engage both CD46 and, indirectly, α5β1 integrins (37). Thus, it appears that M proteins of different serotypes often recognize different receptors on surfaces of mammalian cells. Berkower’s results suggest that the invasion potential is actually determined by the primary amino acid sequence of the respective M protein (6). While it is clear that several types of M protein can facilitate invasion by S. pyogenes, there is no single mechanism underlying M protein-mediated ingestion of streptococci.

STREPTOCOCCAL FOCAL ADHESION AND SIGNAL TRANSDUCTION

The complex networks of intracellular events that lead to ingestion of several microbial pathogens have begun to be unraveled. Studies have not only identified potential targets for therapeutics but in some cases revealed insight into basic cell biology questions. Uptake by host cells requires streptococci to manipulate normal outside-inside signals spawned from cellular receptors. Interactions of Fn-binding-proteins and integrins trigger a cascade of signals that cause cytoskeleton rearrangement leading to ingestion of streptococci by nonphagocytic epithelial cells. Analyses of streptococci-induced signaling pathways have focused on two major invasins and Fn-binding proteins, SfbI/PrtF1 and M1 protein. Ozeri et al. reported that infection of epithelial cells with an M6 SfbI/PrtF1⁺ strain caused integrins to be recruited to sites of bacterial adhesion and colocalization of phosphorylated focal adhesion kinase (FAK) and paxillin, common signaling and structural components of focal adhesion complexes (38). Invasion of epithelial cells by the M6 strain was significantly reduced when host cells were pretreated with genistein, a general inhibitor of protein tyrosine kinases, such as Src kinases and FAK. Invasion was also reduced by inhibitors of the small GTPases Rac and Cdc42. Moreover, internalization of this strain by FAK^–/^– and Src^–/^– cells and cells expressing dominant negative forms of Rac1 or Cdc42 was less efficient. From these findings, Ozeri et al. proposed a model for SfbI/PrtF1-mediated S. pyogenes invasion which is incorporated into Fig. 3: Fn bound to surface SfbI/PrtF1 protein associates with integrins to cause integrin clustering, followed by recruitment of paxillin, FAK, and other focal adhesion proteins at the site of entry. This initiates autophosphorylation of FAK, which creates a docking site for Src kinases. Paxillin present in this complex and additional sites on FAK is then phosphorylated by Src. Ozeri et al. proposed that Rac and Cdc42 are required to orchestrate signaling events initiated by Fn bound to SfbI/PrtF1 protein. Then the focal adhesion complex provides an anchor for actin polymerization and cytoskeleton rearrangements required for internalization of streptococci. To date, however, the relationship between integrin-generated signals, formation of focal adhesion complex, and Rac and Cdc42 is poorly understood and is unsupported experimentally. It is also possible that streptococci interact with additional receptors which in turn activate other mechanisms of signaling and pathways of ingestion.

FIGURE 3.

Postulated and confirmed components of streptococci-induced focal adhesion complexes and signaling molecules required for formation of those complexes and subsequent cytoskeletal changes that lead to ingestion of streptococci by epithelial cells. Fibronectin (Fn) bound to surface M1 or SfbI/PrtF1 proteins associates with integrins to cause integrin clustering leading to ILK activation in a PI3K-dependent manner. For some strains contact between M protein and CD46 is also required for high-frequency invasion. The function of CD46 in this process, however, is unknown. Following recruitment of paxillin, FAK, and other focal adhesion proteins, autophosphorylation of FAK creates a docking site for Src kinases to phosphorylate recruited paxillin and additional sites on FAK. Alternatively, Rac and Cdc42 are activated by ILK through an intermediate to participate in formation of focal adhesion complexes, which provide an anchor for actin polymerization and cytoskeleton rearrangement that ultimately results in uptake of S. pyogenes.

M1⁺ strains of S. pyogenes invade epithelial cells efficiently without expression of SfbI/PrtF1 protein. M1-mediated invasion of epithelial cells was shown to require phosphatidylinositol 3-kinase (PI3K) (39), a heterodimer consisting of the regulatory subunit p85 and the catalytic subunit p110. Activation of PI3K catalyzes phosphorylation of membrane-associated phosphatidylinositol, which binds to downstream targets, including integrin-linked kinase (ILK) (40). Experiments that disrupted activity found that ILK is also required for S. pyogenes invasion of HEp2 cells (41). ILK is known to be a crucial link between integrins and the cytoskeleton, is capable of phosphorylating the β₁ integrin cytoplasmic domain, and can be coimmunoprecipitated with β₁ integrin in lysates of mammalian cells (42). Invasion of epithelial cells by M1⁺ group A streptococci was impaired up to 80% by a specific ILK inhibitor. Inhibition of ILK expression with small interfering RNA or inhibition of activity by transfection with a kinase-inactive ILK gene also significantly reduced GAS invasion. Controls, assays of the non-integrin-dependent invasion of epithelial cells by Salmonella, confirmed that inhibitors of either PI3K or ILK did not have a general negative impact on the host cell’s ability to ingest bacteria but were specific for Fn-mediated uptake of S. pyogenes. ILK is known to be activated upon integrin engagement in a PI3K-dependent manner and to require that its pleckstrin homology-like domain be bound to phosphatidylinositol 3,4,5-triphosphate, a product of PI3K. Therefore, it is possible that PI3K indirectly provides the cytoplasmic membrane anchor for ILK.

Invasion of epithelial cells by M6 SfbI/PrtF1⁺ streptococcus was also reduced 70 to ∼80% by inhibitors of PI3K and ILK. In addition, interaction of M49 streptococci with plasminogen facilitates integrin-mediated and ILK-dependent internalization of bacteria into keratinocytes (43), also suggesting that ILK is a critical signal component involved in integrin-mediated invasion. This is consistent with the possibility that ILK indirectly activates Rac and Cdc42 (44, 45), which can regulate actin cytoskeleton rearrangement (46, 47), leading to various forms of cell adhesion complexes and stress fibers (41). Therefore, ILK may be an intermediate between integrins and Rac and/or Cdc42. Although M1-Fn and SfbI/PrtF1-Fn complexes may induce cellular changes by shared mechanisms, M1-mediated invasion is significantly less sensitive to genistein (41), again indicating that different mechanisms of signaling likely exist.

Paxillin, a cellular signaling molecule involved in cytoskeleton rearrangement downstream of ILK, is phosphorylated in response to Fn-bound streptococci that express either M1 or PrtF1/SfbI protein but is not phosphorylated in response to a mutant streptococcus deficient in both proteins. Inhibition of paxillin phosphorylation significantly reduced invasion by M1⁺ but did not affect ingestion of PrtF1/SfbI⁺ strains, indicating that paxillin phosphorylation is induced by both invasins but only required for M1-mediated invasion (48). M3 streptococci lack the major epithelial invasins SfbI/PrtF1 and M1 protein but have the potential to cause invasive disease. Experiments demonstrate that Src kinase and Rac1 but not PI3Ks are essential for internalization of these streptococci (49). There appears to be a bifurcation point downstream of ILK leading to PI3K or Src/Rac1, and it appears that S. pyogenes has evolved different invasins, which utilize various cellular signaling pathways to promote ingestion, pathways that lead to the distinct morphological changes observed in electron microscopic studies.

M1 protein can interact with multiple cellular receptors, either directly or indirectly (Fig. 2). For example, CD46, a cofactor in factor I-mediated inactivation of complement proteins C3b and C4b (50) and expressed on the surface of all nucleated human cells, enhances invasion of epithelial cells by M3⁺ and M1⁺ streptococci (6, 37). Overexpression of a mutant form of CD46 with a deletion in the cytoplasmic domain results in partial reduction of invasion. Feito et al. analyzed CD46 binding to clinical isolates of S. pyogenes using a soluble construct encompassing the extracellular domain and showed strong binding to M types 1, 3, 8, 18, 24, 28, 29, 31, and 78, weak binding to M6 and M29 strains, and no binding to M types 11, 12, 27, or 30. Although only a limited number of strains have been studied for the interactions between CD46 and streptococci, it is likely that the well-conserved C repeats of M proteins expressed by these different clinical isolates also contain the CD46-binding sites (51). CD46’s role in intracellular signaling and identification of its docking partners are poorly defined; however, these preliminary studies suggest that M1 protein-mediated ingestion of S. pyogenes requires costimulation from two cellular receptors (37).

Diverse invasins, agonists, and cellular receptors predict that the signaling networks which lead to ingestion of streptococci by epithelial cells may also be varied and multiple. Although our knowledge of the signaling mechanisms responsible for ingestion of S. pyogenes is rudimentary, Fig. 3 attempts to interpret what is known. Association of multimeric Fn-binding protein-Fn complexes with α5β1 integrins on the bacterial surface cause integrin clustering that leads to activation of ILK in a PI3K-dependent manner and recruitment of paxillin and FAK. Autophosphorylation of FAK creates a docking site for Src kinase resulting in phosphorylation of paxillin and additional sites of FAK. Rac and Cdc42 proteins are likely activated by ILK signaling, building focal adhesion complexes which ultimately regulate localized actin depolymerization and polymerization and the cytoskeletal changes that engulf the bacteria. Many gaps in the pathway exist, but the tools are now available to further dissect the mechanisms.

OTHER INVASINS

A few serotypes, including M1 strain 90-226, produce the Fn-binding protein Fba. Like M protein, the C5a peptidase (SCPA) and Fba expression are positively controlled by Mga, a multigene transcriptional activator. Mutation and complementation experiments demonstrated that Fba promotes adherence and uptake of streptococci by epithelial cells in the presence of Fn (52). More recent studies found that Fba binds the complement regulatory factor FHL-1 and that this agonist also promotes uptake of streptococci in M1^– mutants of strain 90-226 (53). Elsner et al. reported that mutants which lack the lipoprotein Lbp/Lsp showed reduced adherence and invasion of epithelial cells (54). However, others found that Lbp/Lsp functions as an adhesin but reported that mutations which eliminate expression of Lpb/Lsp had no impact on invasion of HEp2 cells (55). Our laboratory also found that mutation of Lpb/Lsp protein had no effect on invasion of A549 cells by strain 90-226. Expression of that protein on the surface of lactococcus imparted the organism with the potential to adhere to A549 cells but did not promote bacterial uptake. It is possible that this lipoprotein varies between strains or that differences in experimental conditions account for the different conclusions. Thus, binding Fn to the streptococcal surface is not in itself sufficient to promote invasion of epithelial cells. Fn density and/or conformation in the context of the streptococcal surface more than likely influences its potential to promote integrin aggregation or the signals generated by the occupied integrin.

The surface-bound protease SCPA was demonstrated to be the primary, Fn-independent invasin for Streptococcus agalactiae (56). Although M1 protein is the primary invasin for strain 90-226, SCPA accounts for 15 to 20% of bacteria internalized by A549 cells (57). Complete deletion or amino acid substitutions that block protease activity also reduced the capacity of SCPA to promote ingestion of S. pyogenes. Cheng et al. postulated that autocatalytic removal of the propeptide during protease maturation might uncover the protein’s cell-binding site (56).

Serum opacity factor is known to mediate invasion through binding of Fn to its C-terminal domain (58). Experiments revealed that the N-terminal, opacification domain of serum opacity factor also contributes to epithelial cell invasion independently of the C-terminal Fn-binding domain (59). Although the host invasion agonist and cellular mechanisms have not been identified, it appears that like M1 protein, serum opacity factor can use different pathways for cellular internalization.

Invasins that are independent of Fn have also been identified. Human plasminogen that is present in plasma and extracellular fluids has been linked to streptococcal invasion of human cells. Several plasminogen-binding proteins have been identified; however, the plasminogen-binding M-like protein (PAM) and PAM-related protein (Prp), which have high affinity for human plasminogen, are the most studied and are associated with an S. pyogenes strain isolated from a severe invasive infection (60, 61). Plasminogen-mediated invasion is significantly reduced after preincubation of bacteria with either α₁β₁ or α₅β₁ integrins but not after preincubation with α_vβ₅ integrin. Uptake of streptococci via plasminogen was also shown to be dependent on ILK signaling (43), indicating that S. pyogenes can use a variety of agonists to instigate integrin-mediated internalization by human cells.

M-like proteins bind C4BP, an inhibitor of the classical pathway of complement activation that blocks opsonization by C3b and phagocytosis of S. pyogenes by polymorphonuclear leukocytes (PMNs). Recent work demonstrated that a highly virulent M1 strain expresses protein H, a member of the M protein family that binds C4BP and enhances invasion of endothelial cells, independent of M1 protein (62). In addition, the proline-rich protein in human saliva associates with GrpE, a chaperone protein primarily expressed at cell division sites on the surface of S. pyogenes and significantly promotes internalization of streptococci, independent of M1 and PrtF1/SfbI proteins (63). Whether ingestion of streptococci, mediated by C4BP or proline-rich protein requires integrin signaling, is dependent on cytoskeleton rearrangement should be investigated.

FATE OF INTRACELLULAR STREPTOCOCCI

Listeria spp. can escape from vacuoles, and subsequent formation of actin tails allows the bacteria to spread to neighboring cells (26). Immunofluorescent microscopy did not reveal actin tails associated with streptococci at 5 or 7 h postinfection (7). Transmission electron microscopy showed streptococci enclosed in vacuoles within host cells (1, 2). Both Mycobacterium tuberculosis and Listeria are able to block the killing mechanisms of macrophages, providing a safe cellular compartment for persistent infection. The ultimate fate of internalized streptococci has, however, been controversial. Several studies failed to observe growth of intracellular streptococci, and infected cell lines eventually clear most of the bacteria (1, 64). Consistent with these findings are transmission electron micrographs taken 8 h postinfection showing degraded streptococci inside phagolysosomes (1). An early study described a unique strain of S. pyogenes, strain A8, that survived and multiplied within HEp2 cells (14), but to date this observation has not been confirmed or expanded. Thulin et al. found live S. pyogenes within macrophages at the noninflamed edge of soft tissue infections sampled by biopsy (65) and suggested that phagocytes containing intracellular streptococci contribute to the expansion of these rapid and sometimes lethal wound infections. Microscopic analysis of PMNs from intravenously infected mice contained intracellular streptococci in vacuoles, and mice inoculated with these purified infected PMNs died within 5 days, suggesting that the intracellular bacteria were not only viable but more virulent than those grown in culture medium (66). Staali et al. showed that M protein and M-like protein H are required for S. pyogenes to survive ingestion by PMNs. They discovered an undefined mechanism that prevents fusion of azurophilic granules with the phagosome (67). Although Cywes et al. originally suggested that uptake of S. pyogenes by epithelial cells was a host defense mechanism (4), their more recent experiments seem to show the opposite, i.e., intracellular invasion provides a safe haven for S. pyogenes (68). They demonstrated that two extracellular toxins, NADase and streptolysin O, produced by the highly invasive M1 clone of streptococcus, work together to prevent trafficking of intracellular streptococci to lysosomes, allowing 15% of bacteria to survive within immortalized keratinocyte lines. We reported many years ago that M1T1 streptococci isolated before 1980 did not efficiently invade epithelial cells (2), whereas the highly invasive M1T1 clone isolated in the 1990s invaded A549 cells at high frequency. The highly invasive M1T1 strain that Sharma et al. (68) used harbored a 36-Kb genetic element that encoded the nga and slo genes, providing a mechanism for coexpression of NADase and SLO, which in turn promoted intracellular survival. One wonders whether this genetic element is related to the prophage which we reported to distinguish the invasive M1 clone from less invasive M1 strains (13).

Historically, a primary focus of S. pyogenes pathogenesis has been the M protein and its capacity to block phagocytic uptake by PMNs. Now the field faces the dichotomy that this pathogen can not only resist engulfment by professional phagocytes but also survive ingestion by PMNs, which may then become a Trojan horse for dissemination of bacteria through human tissues. Finally, the outcome of streptococcal uptake by epithelial cells or phagocytes depends on the strain and the spectrum of proteins expressed on their surface.

OTHER MECHANISMS

The levels and distribution of cellular agonists are important determinants of the frequency of ingestion of S. pyogenes by human cells. The growth factor TGF-β1 is known to regulate the expression of Fn and other ECM proteins and positively control the integrin signaling pathway. Wang et al. (25) reported that infection of HEp2 cells or mice with S. pyogenes induces TGF-β1 expression by host cells. Moreover, pretreatment of HEp-2 cells with TGF-β1 increases expression of α5 integrin and Fn by HEp2 cells and therefore increases the frequency of ingestion of streptococci by these cells. The increased ingestion is blocked by inhibition of TGF-β1 signaling or by antibodies directed against α5 integrin or Fn. More recently, Li et al. reported that the neuraminidase of influenza A virus (IAV) enhances expression of α5 integrin and Fn on the surface of lung epithelial cells through activation of TGF-β1 (Fig. 4). This promotes FnBP-dependent adherence to these cells in vitro and secondary infections of lungs following inoculation with S. pyogenes (69). They found that inhibition of TGF-β1 or the activity of viral neuraminidase impeded upregulation of these cellular adhesins and IAV-enhanced bacterial adherence. Further study with Smad^–/^– mice, which are TGF-β1 signaling deficient, revealed that significantly larger numbers of streptococci were recovered from the lungs of IAV-infected Smad3^–/^– mice than from Smad3^+/+ littermates. Flow cytometry demonstrated that IAV infection of Smad3^+/+ mice increased the number of lung cells with high levels α5 integrin/Fn expression, whereas the number of these cells in lungs from Smad3^–/^– mice did not change after IAV infection. These results indicate that loss of TGF-β1 signaling prevents the upregulation of α5 integrin/Fn expression in IAV-infected mice and reduces host susceptibility to S. pyogenes coinfection. These finding are consistent with the possibility that neuraminidase in the presence of S. pyogenes (70) could enhance TGF-β1 activity, leading to global upregulation of integrins and ECM proteins, and increased invasion of lung epithelium, leading to systemic streptococcal infections subsequent to or during IAV infections.

FIGURE 4.

IAV facilitates bacterial adherence through the activation of the TGF-β signaling pathway. IAV neuraminidase activates latent TGF-β, which launches Smad signals that locally upregulate integrin and Fn expression by infected cells, providing increased numbers of receptors for adherence of streptococci. This in turn could lead to increased susceptibility to more frequent and more serious streptococcal infections.

SUMMARY

S. pyogenes has evolved multiple mechanisms for invading epithelial cells. Strains of this species differ substantially in invasion frequency and their dependence on invasion agonists. High-frequency internalization is mediated by ECM proteins, such as Fn, Ln, and collagen proteins that form bridges between a surface invasin and the appropriate integrin receptor on the target mammalian cell. Because adherence and invasion can be independent events, mediated by distinct surface proteins and distinct integrins, a two-receptor process is proposed for high-efficiency internalization. The spectrum of target host cells, invasion efficiency, and the ECM-binding proteins displayed on the bacterial surface determine the requirement for serum agonists. No single invasin or surface protein accounts for high-efficiency invasion of epithelial cells by all strains of streptococci. These findings suggest that intracellular invasion may be triggered by different agonists in different tissues or at different stages of infection. Intracellular streptococci appear to account for the persistence of streptococci after penicillin therapy and to be one source of streptococci that are responsible for recurrent tonsillitis. Therefore, the capacity to invade epithelial cells may influence the prevalence of a given strain of S. pyogenes in human populations. Multiple invasins may be differentially expressed in response to varying environmental cues at a specific tissue site. Invasins can bind the same, different, or multiple host proteins and then trigger different cellular signaling pathways. These variations may permit the bacteria to colonize various sites of a host and help explain the diverse manifestation of streptococcal diseases. A better understanding of streptococcal invasins, host cell receptors, and downstream signals may reveal the underlying mechanisms, which could lead to new and improved treatment strategies.