The molecular interactions between commensal microorganisms and their host are basically different from those triggered by pathogens since they involve tolerance. When the commensal is genetically equipped to become an opportunistic pathogen, as is the case with Candida albicans, the picture becomes more complex. In this case, the balance between protection and invasion depends on host reactivity to altered microbial expression of ligands interacting with innate immune sensors. Based on experimental evidence obtained with C. albicans, we discuss the different molecular processes involved in the sensing of this important opportunistic human pathogen by a panel of pattern recognition receptors (PRRs) according to the numerous pathogen-associated molecular patterns (PAMPs) that can be exposed at its surface. Beneficial or deleterious immune responses that either maintain a commensal state or favour damage by the yeast result from this dynamic interplay.
To distinguish friend from foe, the immune system has evolved receptors that recognize molecules present on pathogenic microorganisms. These receptors, referred to as pattern recognition receptors (PRRs), function to promote the innate immune response, which is mainly described as a pro-inflammatory response. PRRs bind to conserved microbial structures called pathogen-associated molecular patterns (PAMPs). PRRs include membranous Toll-like receptors (TLRs) and lectins, which confer additional extracellular surveillance mechanisms promoting phagocytosis and signalling (Mukhopadhyay et al., 2004; Taylor et al., 2005). The expression of these receptors allows the immune system to recognize a wide variety of pathogens that individually express one or several of these PAMPs. Their engagement initiates an early immune response dedicated to the killing of the pathogen while enabling organization of the adaptive response.
Candida albicans as a prototype of commensal
In contrast to pathogens, commensal microorganisms persistently colonize the host without causing any symptoms. The yeast Candida albicans is one of the most common human commensals. It colonizes the mucocutaneous surfaces of the oral cavity, gastrointestinal tract and vagina. Under certain host circumstances, C. albicans can proliferate in a saprophytic state and can become pathogenic. Among the host conditions favouring C. albicans proliferation in the gut and subsequent systemic infections are those encountered in hospital inpatients. In such ‘at-risk patients’, medico-surgical treatment of severe underlying disease and prevention of bacterial infection by broad-spectrum antibacterial antibiotics induce C. albicans proliferation and saprophytic–pathogenic transition. Given the high levels of morbidity and mortality associated with nosocomial candidiasis, the pathogenic adaptation of C. albicans has been the topic of extensive investigations (Calderone, 2002).
Yeast PAMPs and their relation with host PRRs
It is now known that the protective response to fungal exposure is initiated by host recognition of specific PAMPs, which leads to a pro-inflammatory response (Netea et al., 2008). Through different approaches based on the use of either purified or synthetic individual molecules, purified complexes, or yeasts deleted for genes involved in PAMP synthesis, it has been possible to determine that recognition is mainly based on expression of components within the yeast cell wall, which is a complex structure composed of glycans, either as glucans and chitin – present in the deeper layers of the cell wall – or saccharidic moieties of mannan, mannoproteins and glycolipids – expressed at the cell wall surface (Jouault et al., 1995; 2000; Fradin et al., 1996; 2000; Tada et al., 2002; Netea et al., 2004a; 2006; Poulain and Jouault, 2004; Pietrella et al., 2006). Faced with these polysaccharides and glycoconjugates, several lectins have been shown to act either independently or in cooperation with TLRs or other signalling receptors, to recognize surface sugars of yeasts for induction of the cell response (Fig. 1). Notably, dectin-2, which recognizes high mannoses (McGreal et al., 2006), associates with Fc receptor gamma (FcRgamma) chain (Sato et al., 2006). Macrophage mannose receptor (MR), which recognizes fungal mannan and mediates recognition and phagocytosis of yeasts by macrophages (Stahl and Ezekowitz, 1998), has been shown to associate with TLR-2 during stimulation by fungi such as Pneumocystis (Tachado et al., 2007). DC-SIGN, another C-type lectin, is able to internalize C. albicans (Cambi et al., 2003) by recognizing complex mannoside structures exposed on yeast cells (Cambi et al., 2003; Taylor et al., 2004) and a regulatory role on TLR-4-dependent signalling has been demonstrated for SIGNR1, the DC-SIGN homologue in mice (Nagaoka et al., 2005). Although the nature of the mincle ligand has not yet been characterized, this C-lectin also binds to yeast cell wall components and participates in the cell response with TLR-2 (Wells et al., 2008). Galectin-3, a S-lectin originally described for its binding specificity for galactose, binds Candida-specific β-1,2 mannosides (Fradin et al., 2000; Kohatsu et al., 2006) and associates with TLR-2 when macrophages interact with yeasts (Jouault et al., 2006). Finally, dectin-1 acts either alone (Brown, 2006; Rosas et al., 2008) or with TLR-2 to confer cellular responsiveness to microbial β-glucans (Brown et al., 2002; 2003; Gantner et al., 2005).
A large number of yeast interactions with host cells have been characterized in different studies, and examination of host PRRs and C. albicans PAMPs has shed light on the interplay between these two organisms. All these studies reflect the complexity of this interaction, but it is now clear that the resultant signalling leads, in the majority of cases, to a pro-inflammatory response towards the yeast which has been described as necessary for acquiring resistance to infection. Notably, it is now accepted that both TLR-2 and TLR-4 are the main TLRs involved in the signalling induced by C. albicans, although, depending on the model used, conflicting results have been obtained (Netea et al., 2002; Villamon et al., 2004; Blasi et al., 2005; Gil et al., 2005; Gil and Gozalbo, 2006a,b). Nevertheless, this cell response does not seem to be specific to pathogenic microorganisms since, for example, C. albicans which is both commensal and pathogenic, and the non-pathogenic yeast Saccharomyces cerevisiae can both induce TLR-dependent pro-inflammatory cytokine production (Netea et al., 2004b; Roeder et al., 2004). The concept that, in contrast to S. cerevisiae, pathogenic Candida may interfere with regulation of the immune response at the cellular level has not been widely explored. It appears that an integrated immune response determines the lifelong commensalism of this yeast at the mucosal level, as well as transition from mucosal saprophyte to pathogen (Zelante et al., 2007; Romani, 2008). However, several studies have indicated that an anti-inflammatory response during the initial interaction of yeasts with immune cells is necessary for the yeast to establish infection, showing that balancing the pro-/anti-inflammatory response is an important requisite for pathogenicity of candidiasis.
The pro-/anti-inflammatory response, a balance between beneficial and deleterious effect of both microbe and host response
Candida albicans infection does not induce significant acute toxicity in IL-10-deficient mice (Del Sero et al., 1999). Instead, improved clearance of yeasts, reduced fungal load and fungal-associated inflammatory responses are observed in IL-10 knockout (KO) mice, leading to early resistance to acute systemic candidiasis (Vazquez-Torres et al., 1999). Conversely, in the gastrointestinal model of infection, exogenous IL-4 and IL-10 were found to exacerbate the course of infection (Tonnetti et al., 1995), and an overall suppressive role of IL-10 on human monocyte function against C. albicans was also demonstrated (Roilides et al., 1998). The absence of IL-10 therefore reinforces innate and acquired antifungal immunity. This allows upregulation of type 1 cytokine responses in particular, associated with greater production of IL-12 and TNF-α (Del Sero et al., 1999).
These observations revealed a fundamental role for IL-10 in the initial development of Candida infection. This suggests that the yeast is able to play a role in stimulating the production of this cytokine during the initial processes and influence the pro-/anti-inflammatory cytokine balance that may then switch over the yeast from the commensal to the pathogenic state (Fig. 2). Such induced immunosuppression could therefore be a powerful immunoevasion strategy for the yeast (Romani and Puccetti, 2006).
Initial description of the role of TLRs in the cell response to C. albicans revealed that TLR-2 and TLR-4 were involved in the induction of pro-inflammatory cytokine production allowing protection of the host (Villamon et al., 2004; Netea et al., 2008). However, some data suggest that TLRs offer escape mechanisms to certain pathogenic microorganisms, especially through TLR-2-driven induction of anti-inflammatory cytokines, rendering the host response feeble. Indeed, it is surprising that, like IL-10 KO mice (Vazquez-Torres et al., 1999), TLR-2 KO mice are more resistant to disseminated Candida infection (Netea et al., 2004c). TLR-2 (−/−) macrophages have increased candidacidal activity in vitro. In vivo, C. albicans induces immunosuppression through a TLR-2-derived signal that mediates increased IL-10 production and induction of regulatory T-cells (Netea et al., 2004c). Such a link between TLR-2 and production of IL-10 (Kang et al., 2004; Dillon et al., 2006), as well as induction of regulatory T-cells (Sutmuller et al., 2006), has been suggested. Macrophages are involved in regulation of the immune response through their ability to secrete large quantities of IL-10. They have also been shown to participate in suppression of the differentiation of Th1 and IL-17-producing T-helper cells, and promote the differentiation of Treg cells (Denning et al., 2007). Thus, C. albicans may be able to induce both pro- and anti-inflammatory responses when recognized by macrophages, both responses originating from an interaction that seems to depend on TLR-2.
Regulation of an adequate or inadequate response: the role of some PAMPs
The most potent stimuli of macrophages have been demonstrated to be glyco-conjugated components such as mannans (Tada et al., 2002), including β-1,2 oligomannoses (Jouault et al., 1995), phospholipomannan (PLM) (Jouault et al., 2003) and β-glucans (Brown et al., 2003). Mannans are found in part of the cell wall directly available at the surface of yeasts. However, this complex structure is not specific to C. albicans since it is a basic entity present on all yeasts (C. albicans and S. cerevisiae), whether they are pathogenic or not. However, on C. albicans, some unique mannosides, namely β-1,2-linked mannosides, have been described to act as stimuli for macrophages (Jouault et al., 1995; 2000). Unlike these mannans, β-glucans, which are part of the backbone of all yeasts, are not accessible directly to the immune cells. β-Glucans are available on blastoconidia, mainly after treatment that opens up the deepest cell wall components, such as heat treatment (Martinez-Esparza et al., 2006). Exposure of β-glucans, which may reflect the partial degradation of the yeast cell wall within phagosomes and the loss of the superficial cell wall layer (Gantner et al., 2005; Martinez-Esparza et al., 2006), may allow their presentation to specific PRRs, such as dectin-1. β-Glucans are also soluble and can be detected, like mannan components (either α- or β-1,2-linked mannose residues), in patients' sera, rendering them available as soluble stimuli for immune cells (Sendid et al., 2004; Ostrosky-Zeichner et al., 2005). β-1,2-Linked mannoses are specifically exposed at the surface of C. albicans (Kobayashi et al., 1992). However, the glycolipid PLM, which is located within the yeast cell wall and is accessible at the surface (Poulain et al., 2002), is shed by C. albicans, allowing direct contact with the host cells even in the absence of phagocytosis (Jouault et al., 1998). β-1,2-Linked mannoses (Jouault et al., 1995) and PLM (Jouault et al., 1994) induce a pro-inflammatory response that depends mainly on TLR-2 (Jouault et al., 2003). Structural analysis has shown that PLM is a mannose inositol phospho ceramide (MIPC), thereby presenting two moieties which can induce opposite effects on target cells (Trinel et al., 2002): a glycan component composed of up to 14 β-1,2-linked mannoses, identical to that present in the acid-labile part of O-linked mannan, acting as signals for TNF-α stimulation (Jouault et al., 1995) and recognized by galectin-3 (Fradin et al., 2000; Kohatsu et al., 2006) in association with TLR-2 (Jouault et al., 2006), and a lipid moiety whose role is less well defined. However, through this latter moiety, PLM may act as a negative regulator of the cell response through activation of MKP-1 (Ibata-Ombetta et al., 2003a), a phosphatase that is involved in the regulation of pro- and anti-inflammatory responses (Chi et al., 2006). This leads to inactivation of ERK-1 phosphorylation (Ibata-Ombetta et al., 2003b), which is necessary for the efficient killing of Candida (Ibata-Ombetta et al., 2001; Wozniok et al., 2008) and which is involved in the activation of several downstream targets including p90RSK and BAD, which are involved in mitochondrial apoptosis of cells (Ibata-Ombetta et al., 2003a,b). Interestingly, recent data show that inhibition of MNK, a Mapk downstream of ERK-1, blocked the pro-inflammatory response but permitted IL-10 production in response to TLR stimulation (Rowlett et al., 2008), an effect that has been observed when large amounts of yeasts expressing PLM are in contact with macrophages (A. Sarazin, personal communication). In addition, control of TNF-α mRNA nuclear export has been shown to operate specifically through ERK-1 (Skinner et al., 2008), TNF-α mRNA being trapped in the nucleus when ERK-1 is inactive (Dumitru et al., 2000). Together, these observations suggest a central role for ERK-1 activation in the regulation of TNF-α production, either positive or negative, which could depend on the yeast burden, and efficient killing of yeasts. However, the exact participation of Mapks and upstream adaptors in the direction of cytokines produced in response to yeasts has not yet been clearly established.
Role of Mapks and mediators in regulation
Members within one family of PRRs can trigger opposite signalling effects, indicating that the ultimate outcome of a pathogen-induced immune response depends on the pathogen signature and the PRRs involved. This may lead to the recruitment of different adaptors and Mapks according to the PRRs and PAMPs involved. Slack et al. (2007) showed that induction of IL-10 by zymosan, a yeast cell wall derivative, requires activation of ERK depending on recruitment of Syk but not involving MyD88, the well-characterized TLR-2 adaptor participating in TNF-α induction. Conversely, induction of IL-10 by Streptococcus pneumoniae cell wall fragments also requires TLR-2 and participation of both MyD88 and NOD2, but is independent of ERK activation (Moreira et al., 2008). More recently, evidence has been obtained for the role of Raf-1-dependent acetylation of NFκB in the modulation by DC-SIGN of TLR-dependent signalling, leading to both prolonged and increased IL-10 transcription (Gringhuis et al., 2007). Interestingly, a link between ligation of TLRs and subsequent activation of ERK and NFκB signalling pathways has been observed through engagement of Tpl2, a Mapk upstream of ERK, that is involved downstream of TLR-dependent signalling (Banerjee et al., 2006), and Syk-dependent activation (Eliopoulos et al., 2006). Alternatively, a signalling pathway can be initiated by dectin-1 independently of TLR but involving CARD9 to link dectin-1/Syk to NFκB activation through Bcl10 and Malt1 for innate antifungal immunity (Gross et al., 2006; LeibundGut-Landmann et al., 2007).
Relation between phagocytosis and signalling
Beside, or in association with, their role in signalling, PRRs are also involved in the recognition and phagocytosis of pathogens. However, few studies have concerned this part of their biological activity and their role in the antimicrobial activity of innate immune cells has not been clearly established. Lectins are engaged in phagocytosis and some have been demonstrated to facilitate this process (Cambi et al., 2003; Porcaro et al., 2003; Sano et al., 2003). This has been well studied for mannose-binding lectin (MBL) which eliminates potential pathogens by activating the lectin pathway of complement or by opsonization of invading microorganisms (Turner and Hamvas, 2000). However, with certain lectins the mechanism of phagocytosis differs from other more conventional processes. Dectin-1-mediated phagocytosis depends on its ITAM-like motif but is Syk-independent. This involves different mediators such as Rac-1 and Cdc42, but not Rho, making the mechanism different from that used by FcR (Herre et al., 2004). Similarly, peritoneal macrophages from galectin-3 KO mice showed attenuated phagocytic clearance of apoptotic thymocytes and reduced IgG-mediated phagocytosis of opsonized erythrocytes (Sano et al., 2003). However, galectin-3, which associates with C. albicans within the phagosomes and delivers specific signals, does not interfere with the efficiency of yeast phagocytosis (Jouault et al., 2006). Although controversial according to the ligand examined (Yates and Russell, 2005; Russell and Yates, 2007), at least for bacteria, TLRs have also been shown to be involved in activation of the TLR signalling pathway regulating phagocytosis at the level of internalization and phagosome maturation (Blander and Medzhitov, 2004; Underhill and Gantner, 2004; Blander, 2007), with subsequent enhancement of actin-cytoskeleton mobilization that facilitates antigen processing and presentation (Blander and Medzhitov, 2006). Interestingly, TLRs engaged during phagocytosis of microbial pathogens by dendritic cells, but not syngeneic apoptotic cells, control both phagosome kinetics and maturation, with individual phagosomes undergoing separate programmes of maturation (Blander, 2007).
The link between recognition of microbes through PAMPs, their phagocytosis, and subsequent induction of signalling is not well established. Indeed, it has not been determined whether efficient signalling induced by whole microbes requires phagocytosis or not. The reason for this lack of data may be due to the multiplicity of PAMPs available at the surface of microbes and the multiplicity of PRRs possibly involved in recognition and signalling. Thus, most studies have been conducted using soluble PAMPs or particulates. In a recent study (Rosas et al., 2008), it was demonstrated that depending on the nature of the β-glucans and the size of the particulates, either ligation of dectin-1 at the cell membrane was sufficient to induce a pro-inflammatory response through the Syk/CARD9 pathway, or, in the context of ‘frustrated’ phagocytosis, increased expression of TLR was induced, favouring a different signalling pathway also ending with a pro-inflammatory response.
Human response and PRR–PAMP interaction: assistance of SNPs
As a result of what is observed in KO mice, several PRRs have been explored for single-nucleotide polymorphisms (SNPs), with the aim of finding a link between observations in animal models and human reality. On the basis that SNPs in humans may reflect KO mice models, investigations have pointed to the role of several receptors in the cellular response to yeasts. Although these studies are relatively recent, some have indeed demonstrated a role of TLRs and other PRRs in human candidiasis, demonstrating the relevance of in vitro and in vivo studies. SNPs affecting TLR-2 responsiveness have revealed that these receptors contribute to the course of sepsis, which is associated with substantial morbidity and mortality during intensive care treatment. In this respect, the TLR-2 SNP Arg753Gln results in altered cytokine release in response to Candida but not to Gram-positive sepsis. In contrast, in Candida sepsis, Arg753Gln heterozygous patients showed biomarker patterns that differed from wild-type patients with elevated TNF-α plasma concentrations, but reduced IFN-γ and IL-8 levels (Woehrle et al., 2008). For TLR-4, Asp299Gly/Thr399Ile polymorphisms have been shown to be associated with an increased susceptibility to Candida bloodstream infections. This may lead to increased production of IL-10, probably initiating the observed effect (Van der Graaf et al., 2006). Consistent with evidence that pathogens can utilize DC-SIGN binding to suppress the pro-inflammatory response, SNP Ala336Gly has been shown to decrease DC-SIGN expression and to be associated with lower susceptibility of humans to different pathogens (Vannberg et al., 2008). Recent advances in galectin-3 SNP analysis have demonstrated a new alteration that affects the capability of individuals to develop a secondary humoral response specific to glycans, a tag for the initial sensing of Candida by the immune system (A. Sarazin, personal communication) which makes TLRs and thus, associated lectins, important for the secondary response to yeasts. Both PRRs and associated lectin receptors seem to be important for the definitive response against infection. Thus, a link can be made between laboratory observations and patient behaviour. However, much more information is needed to understand the complex interplay required for the development of an immune response.
This brief review highlights the complex signalling leading to antagonistic pro- or anti-inflammatory responses, based on the recruitment of different PRRs (either TLRs or lectins) able to recruit cytoplasmic transmitters, depending on multiple ligands exposed on microorganisms such as C. albicans. Better understanding of the pathophysiology of such ‘patho-commensal’ microorganisms lies in the study of the early steps of the interplay which conditions the nature of the recruited transmitter. This model clearly shows that the interplay among innate signals and shaping of the adaptive response is the consorted action of many components that influence each other. TLRs may interfere with antigen processing and presentation regulated by lectins, whereas, in turn, lectins as antigen uptake receptors influence signalling pathways initiated by TLRs. The ultimate outcome is dictated by the pathogen's PAMPs (glycans and TLR agonists) which determine the set of receptors involved in shaping the immune repertoire of cells. What makes pathogens like C. albicans successful is their ability to rapidly modulate their PAMPs. In this respect, much is still unknown about the impact of cell wall molecules exposed on the yeast as a function of host environment, on the numerous receptors and signalling pathways leading to either beneficial or deleterious responses. The use of live in vitro methods enabling us to follow the sequence of events actually induced, instead of time-point examination, would be of great value in increasing our understanding of this complex dynamic process.
This work was supported in part by the European Community-Feder fund, Région Nord-Pas de Calais, ANR and Inserm. Aurore Sarazin was supported by a grant from Inserm/Région Nord-Pas de Calais.