Candida albicans Cell Wall Proteins

Kre9p, along with the covalently attached Crh11p, Crh12p, and Utr2p (Table 2), is in the GH16 family. Kre9p is involved in β-1,6-glucan biosynthesis. The synthesis of this polysaccharide requires both cytoplasmic and cell wall activities, and its synthesis is not as well understood as are those of the other two structural polysaccharides (204, 322). The null (kre9Δ/Δ) strain is inviable when grown on glucose (224, 279). The mutant strain grows poorly on other carbon sources (224). Growth on sucrose, mannose, maltose, or fructose is similar to or less than that on galactose. When growth was on galactose, the amount of β-1,6-glucan in the heterozygote fell to about 20% of that seen for the wild type and was undetectable in the null strain. Hyphae failed to form in serum. The strain showed attenuated virulence in a murine systemic disease model. The absence of β-1,6-glucan would remove the normal attachment site for GPI-CWPs (Fig. 2) (177, 179, 322). Interestingly, in S. cerevisiae, the loss of Stt1p, which has a role in N glycosylation, also leads to a substantial reduction in β-1,6-glucan and links N glycosylation with the cell wall (52). Transmission electron microscopy of the Stt1p-deficient cells showed a diffuse cell wall with a loss of outer mannan. C. albicans STT1 is an ortholog of the S. cerevisiae gene, and there also may be a similar linkage between N glycosylation and β-1,6-glucan. KRE9 is required for the induction of a high-affinity fibronectin receptor presumably attached to glucan, which is discussed later (293). This observation would be consistent with the loss of attachment sites, although there may be other possibilities. Growth on galactose apparently partially relieves the loss of β-1,6-glucan (189). In a study some 20 years ago, McCourtie and Douglas (238) showed that growth on galactose enhanced adherence to acrylic compared to what was seen for glucose-grown cells. Galactose-grown cells were more resistant to spheroplast formation with zymolyase. In addition, transmission electron microscopy revealed the presence of an additional surface layer. Such an outer layer may be the source of a partial remediation of the β-1,6-glucan defect.

Glucan is synthesized as a linear polymer; therefore, the branching of polysaccharide found in the cell wall (shown schematically in Fig. 2) requires enzymes with the capability to form branches from the linear polymers. Bgl2p is in the same GH17 family as Mp65p and Scw4p, discussed below. Bgl2p catalyzes the splitting (removing a disaccharide) and linkage of β-1,3-glucan molecules, resulting in β-1,3-glucan chain elongation with a β-1,6 linkage at the transfer site. Bgl2p is the major β-1,3-glucosyltransferase (also sometimes called a glucan transferase); however, after deletion of both alleles, about 50% residual enzymatic activity is detected (reference 331; also reviewed previously in reference 49). The presence of residual activity suggests the presence of additional β-1,3-glucosyltransferase(s) that was associated with the formation of a product different from those formed by Bgl2p. There were no changes in β-1,3-glucan or β-1,6-glucan content in the mutant strain. However, the strain was more sensitive to the chitin synthesis inhibitor nikkomycin than was the parental strain. This suggests an increase in chitin and a compensatory role for chitin in cell walls with reduced branching. The expression of BGL2 alters in response to several conditions. Expression is reduced in low-iron conditions, as determined by microarray analysis (194). Several other cell wall-associated genes encoding GPI-CWPs (e.g., ALS2, RBT5, PHR2 [Table 2]) and PIR1 covalently attached proteins, and another putative glucanase, SCW11, responded to iron levels (194). In expression profiling, BGL2 expression is greater in the presence of fluconazole under conditions promoting SAP gene expression, as is GSC1, encoding a subunit of β-1,3-glucan synthase (62). Microarray analysis shows that BGL2 is also induced, not surprisingly, during cell wall regeneration along with other cell wall-associated genes, including genes for β-1,3-glucan synthase (46). In addition to this role in glucan metabolism, Bgl2p, also shows adhesin activity, which is discussed in a later section (160).

The exoglucanase Xog1p is the only C. albicans enzyme found in the GH5 family. Xog1p has a marked specificity for β-1,3-glucoside linkages compared to β-1,6-glucoside linkages (368). The enzyme catalyzes the successive removal of glucose residues from the nonreducing end of β-1,3-glucan, although at high acceptor concentration, transglucosylation with retention of the anomeric configuration can occur. The structure has been determined at 1.9 Å (67) and the insertion of serine at the CUG codon in the mature protein had a minor effect (66). Xog1p is the major exo-β-1,3-glucanase of the cell wall, as only residual activity is detected in the null strain (127). The loss of the enzyme did not impair yeast growth or morphogenesis, although there was a slight increase in sensitivity to chitin and glucan synthesis inhibitors. In addition, there was no apparent defect in the murine systemic infection model. Phr1p and Phr2p, which are glucan-attached proteins, may also have similar activities (Table 2). Another potential exoglucanase-encoding gene, EXG2, is induced during cell wall regeneration (46).

XOG1 is induced about 2.5- to 6.9-fold by morphogenesis in serum, as determined by semiquantitative RT-PCR and microarray analysis, and the level of enzyme activity is also increased (200). The expression of this gene in hyphae is dependent on SIT4, encoding a serine/threonine protein phosphatase that modulates morphogenesis. In the sit4Δ/Δ strain, enzymatic activity was repressed during hyphal formation. In the absence of Nrg1p and Tup1p, repressors of morphogenesis, the gene is expressed about fourfold more than in wild-type organisms (transcriptional profiling [259]). Exposure of opaque a cells to α pheromone results in a sixfold increase in expression (transcriptional profiling [22]). XOG1 is also activated by Rim101p in response to external alkaline pH (transcriptional profiling [221]). Constitutive expression of HGT4, encoding a glucose sensor required for growth on low glucose, elevated XOG1 expression compared to what was seen for the null strain (transcriptional profiling [33]). The hgt4Δ/Δ strain showed reduced the capacity to form filaments. XOG1 expression increased (1.6-fold) upon exposure to ketoconazole (transcriptional profiling [210]). The expression of a subunit of β-1,3-glucan synthase also increased. Ketoconazole does not directly affect the cell wall, so the mechanism by which the increased hydrolysis of β-1,3-glucan is involved in the response to fungistasis in unclear. Ketoconazole may exert its effect through changes in membrane sterols. Although XOG1 expression increases in hyphae and in strains lacking repressors of filamentation (259), the protein does not appear to have a critical role in morphogenesis, based on the ability to undergo this transition in the absence of Xog1p.

The cell wall enzyme Eng1p, along with cytoplasmic Acf2p, is in the GH81 family. Eng1p is a fungus-specific endo-1,3-β-glucanase involved in cell separation. It catalyzes the release of glucose from the nonreducing end. Eng1p had endoglucanase activity, as demonstrated by the release of reducing sugars from the substrate, and activity was specific for β-1,3-glucan (96). The eng1Δ/Δ strain showed normal growth and morphogenesis. However, microscopic examination showed clusters of yeast cells that had completed cytokinesis but failed to separate. ENG1 is a functional ortholog of S. cerevisiae DSE4 (ScDSE4). ScDse4p is a daughter cell protein involved in cell wall degradation from the daughter side during cell separation (61). The yeast septum is three layered, with the internal layer composed primarily of chitin and the layers on either side similar to general cell wall in composition (reviewed in references 35 and 409). ScDSE4 expression is cell cycle regulated and peaks in late M or at the M/G₁ boundary (359). Ace2p is a transcription factor regulating genes involved in cell separation, and ENG1 is downregulated in the ace2Δ/Δ strain (256). Ace2p is localized to the daughter cell nucleus (172). Although Eng1p has not been localized to the daughter side of the septum, it seems likely to be found here, as it is a functional ortholog of a protein that is so localized and as a regulator of expression is localized in daughter nuclei. Observations with CRL1 provided additional evidence for a role for Eng1p in cell separation (86). CRL1 encodes a predicted GTPase of the Rho family. The crl1Δ/Δ strain had elongated cells and a separation defect. The overexpression of ENG1 suppressed the separation defect but not the elongation phenotype.

Northern analysis of the ENG1 transcript showed a decrease in abundance 2.5 h into germination (96). Another study found that ENG1 and XOG1 expression increased during hyphal formation and that the increase was dependent on the serine/threonine protein phosphatase Sit4p (transcriptional profiling [200]). These two studies appear to be in conflict. Both studies were performed with similar serum-containing media. Additionally, ENG1 expression responds to other conditions that also impact the cell wall. ENG1 is repressed by treatment with caspofungin, an inhibitor of β-glucan synthesis (transcriptional profiling [210]) and in response to pheromone in Spider medium (transcriptional profiling [22]). In both conditions, cell budding decreases and presumably there is little need for septum degradation. The evidence supports the role of Eng1p in the catabolism of the secondary septum between the daughter and mother cells, probably from the daughter side.

Mp65p is possibly a β-glucanase and is also discussed elsewhere as an adhesin. La Valle et al. (197a) found by Northern analysis that MP65 expression increased upon germ tube formation in Lee medium at 37°C. On the other hand, Sohn et al. (356) reported that MP65 expression was constitutive in both yeast cell conditions and hypha-promoting conditions. This latter study used yeast cell growth in yeast extract-peptone-dextrose (YEPD) or minimal essential medium plus glucose at 30°C and hyphal growth in YEPD plus serum or in minimal essential medium plus glucose at 37°C. The difference between the two reports may represent a medium-dependent enhancement of expression, as noted for some genes in the two conditions used by Sohn et al. (356).

The deletion of MP65 had no effect on growth rate, cell size, or sensitivity to cell wall-perturbing agents (325). However, gene deletion severely impaired morphogenesis. Mp65p was not detected on the cell surface by an Mp65p-specific monoclonal antibody (MAb). Yeast cells had reduced ability to adhere to polystyrene. The deletion strain was attenuated in the murine model of disseminated infection. Although the protein sequence contains two conserved glutamate residues crucial for activity of the glucanase family, GH17, with which Mp65p has been associated, the enzymatic activity of the recombinant protein has not been tested. Additional evidence points to the involvement of MP65 in cell wall metabolism or structure. MP65 expression also increased transiently during cell wall regeneration (transcriptional profiling [46]) and increased in response to pheromone (transcriptional profiling [22]). Expression decreased in response to treatment with caspofungin (transcriptional profiling [210]) and increased in response to ciclopirox olamine (transcriptional profiling [201). MP65 expression also responds to other conditions. MP65 expression is greater in a strain with reduced PLC1 expression than with wild-type PLC1 expression (transcriptional profiling [189]) and at alkaline compared to acid pH (transcriptional profiling [23]). Other putative glucanase genes, e.g., SIM1, respond to pH. MP65 is the ortholog of S. cerevisiae SCW4. A recent study showed that Scw4p is incorporated into the S. cerevisiae cell wall by an unknown alkali-sensitive linkage (378). A similar linkage for C. albicans Mp65p has not yet been examined. Although this protein has been studied for other properties unrelated to a cell wall function, the effect of deletion on morphogenesis and other observations suggest that it does have a cell wall function, perhaps as a glucanase.

There are two other possible glucanases, Sun41p and Sim1p. The S. cerevisiae orthologs of CaSUN41 and CaSIM1 are ScSUN4 and ScUTH1, respectively. The S. cerevisiae genes are two of the four genes of the SUN family. ScSun4p and ScUth1p have been demonstrated to be in both the mitochondrion and the cell wall, where they are proposed to have glucanase activity (393). The Sim1p-invertase fusion was detected in export screening (249). In C. albicans, SUN41 and SIM1 both show an increase in expression during cell wall regeneration (transcriptional profiling [46]), during treatment with flucytosine (transcriptional profiling [210]), and under conditions of hypoxia (transcriptional profiling [346]). SIM1 is downregulated by Rim101p (transcriptional profiling [221]). SUN41 has decreased expression following infection of HEp2 cells (transcriptional profiling [326]) and treatment with caspofungin (transcriptional profiling [210]). It is also regulated by Efg1p and Cph1p (transcriptional profiling, Northern analysis [221]).

Four studies report defects associated with SUN41 deletion (103, 138, 279, 390). Deletion of both alleles resulted in defects in hyphal formation, cell separation, and biofilm formation (103, 138, 279). SUN41 expression in the parent strain did not change during entry into stationary phase and through 11 days of culture (390). However, the sun41Δ/Δ strain had a severe defect at day 8, with loss of viability affecting the maintenance of stationary phase. Mutant strains were more sensitive to caspofungin (279) and Congo red (138, 279) with slight or no change in calcofluor white sensitivity (103, 138). The mutant strain was unchanged compared to the parental strain in terms of sensitivity to other cell wall-perturbing agents or conditions such as nikkomycin Z treatment and oxidative stress (103). Mutant strains showed some reduction in adherence to a Caco-2 monolayer (138) but no difference in adherence or damage to FaDu oral epithelial cell line cells or human umbilical vein endothelial cells (HUVECs) (279). In murine models of hematogenous and oral infections, the mutant was attenuated (279). The deletion of SIM1 resulted in only minor phenotypic alterations (103). The failure to achieve the deletion of both SUN41 and SIM1 supported a synthetic lethal condition. When the control of either gene was by the methionine/cysteine-repressible MET3 promoter, there was no growth under repressing conditions, and there was growth with increased doubling time in the absence of repressors remediated the lysis phenotype. Under repressing conditions, mother and daughter cells remained attached, and after several generations, primarily mother cells lysed. Transmission electron microscopy showed defects at the septum. The conditional strains were more sensitive to several agents, such as nikkomycin, azole-class drugs, and calcofluor white. The latter reagent showed variable chitin distribution in the mutant. Under repressing conditions, sorbitol remediated the lysis condition in rich but not minimal medium. These studies provide strong evidence that these genes contribute to cell wall integrity and share an essential function.

Less information is available for several other putative glucanases. C. albicans Scw4p is another unattached protein that may have glucosidase activity. The best hit for S. cerevisiae is ScSCW4, although this gene is the ortholog for CaMP65, discussed above. CaSCW4 expression is greater in opaque than in white cells (transcriptional profiling [193]) and is downregulated in alkaline conditions (transcriptional profiling [23]) and during cell wall regeneration (transcriptional profiling [46]). The observation of greater expression in opaque cells is interesting in view of the observation that the ScSCW4 ScSCW10 double deletion mutant has defects in mating (43). The best hit for ScSCW10 in C. albicans is MP65, discussed above. Yet another gene, SCW11, encoding a putative C. albicans glucanase shows altered expression in response to different conditions. Caspofungin treatment reduced expression (transcriptional profiling [210]). Expression also decreased late in protoplast regeneration (transcriptional profiling [46]). On the other hand, expression increased in response to elevated iron concentration (transcriptional profiling [194]). Expression is less in hyphae than in yeast cells and is regulated by Ace2p in both morphologies (RT-PCR [172]; transcriptional profiling [256]). The gene appears to be essential, as homozygous mutant strains were not recovered (279). As Ace2p is a regulator of cell separation and morphogenesis and expression is greater in yeast cells than in hyphae, Scw11p may be a glucanase involved in cell separation.

There are four annotated genes encoding chitinases in C. albicans. Three of these are cell wall associated and all are members of the GH18 family. Cht4p lacks a signal for classical export and is an ortholog of S. cerevisiae Cts2p, a cytoplasmic protein implicated in sporulation. Like glucanase activity, both covalently attached and soluble proteins are predicted cell wall chitinases. Cht2p is a GPI-CWP (Table 2) that has been found in cell wall extracts (see Table 4). Deletion does not cause a defect in cytokinesis. Cht1p is a minor chitinase (85, 343). The null strain showed no growth differences in liquid medium from what was seen for the parental strain but had increased hyphal growth on solid medium (85). In some studies, the expression of CHT1 was not detected (172, 239). The gene was downregulated by Rim101p in alkaline medium (transcriptional profiling [23]). This observation is the likely reason for the failure of other studies to note expression, as in those studies cells were not grown at acidic pH. The mutant phenotype, if tested at acidic pH, may differ from that at alkaline pH. In addition to CHT1, PHR2 and SCW4 have a reduction in expression at alkaline pH (transcriptional profiling [23]). Changing the enzymes that contribute to cell wall polysaccharide structure and remodeling may reflect a pH optimum that is not broad enough to support adequate cell wall integrity under different conditions.

Cht3p is the major chitinase (85, 343). In a CHT3 null strain, cell-associated chitinase activity was greater than culture supernatant activity, and both decreased about 60% in yeast cells in the mutant strain (343). In the strain lacking Cht2p, cell-associated activity decreased about 20%, while there was no decrease in culture supernatant activity. This observation is consistent with the covalent attachment of Cht2p and the greater contribution of Cht3p to chitinase activity. In hyphae, the activity was severalfold greater. In hyphae, the loss of Cht2p had no effect, while the loss of Cht3p reduced both cell-associated and culture supernatant activity by about 80%. This observation suggested that most of the increase in hyphal chitinase activity derived from Cht3p. The greater abundance of chitinase activity in hyphal organisms was unexpected, as a previous study reported the preferential transcription of both genes in yeast cells (Northern analysis [239]). Whether the difference in hypha-inducing conditions is responsible for the difference or whether there is posttranscriptional regulation is unknown. Deletion of various chitin synthase genes had either a modest effect or no effect on chitinase activity (343). Thus, there was not a concomitant decrease in chitin synthase activity when chitinase decreased. The increase in the chitin content of yeast and hyphal cells in CHT2 and CHT3 null mutant strains was not significant. However, there was some hypersensitivity to calcofluor white, which can reflect an increase in chitin.

Cells in which Cht1p or Cht2p was absent had no defect in cell separation, and the budding was bipolar (85). In contrast, the loss of Cht3p resulted in the failure of cells to separate. The restoration of CHT3 expression remediated the defect, confirming the role of Cht3p in cytokinesis. A deletion of the only S. cerevisiae chitinase gene, CTS1, causes the same cell separation defect in that organism (190). C. albicans CHT3 but not CHT1 or CHT2, when expressed in S. cerevisiae, partially complemented the loss of ScCTS1 (85). This observation identifies C. albicans CHT3 as the functional homolog of S. cerevisiae CTS1. In S. cerevisiae, Cts1p localizes to the daughter cell (61). The expression of ScCTS1 is cell cycle regulated and peaks at M or the M/G₁ boundary (359). The functional homology of CaCht3p and ScCts1p suggests that CaCht3p also localizes to the daughter cell and hydrolyzes the primary chitin septum from the daughter side. In a C. albicans strain lacking the Ace2p transcription factor, which regulates cell separation, there was no effect on CHT2 expression. However, CHT3 expression was almost abolished in both yeast and hyphal cells (RT-PCR and transcriptional profiling [172, 256]). The expression of ENG1 and MP65, encoding glucanases, also decreased. This observation suggests that cell separation requires hydrolysis of both chitin and glucan. CHT2 and CHT3 expression increased in farnesol-treated biofilm (RT-PCR and transcriptional profiling [42]). Since this treatment favored yeast form growth, this observation would be consistent with the previously reported high transcript abundance in yeast cells compared to what was seen for hyphae (239) but not with the observations of enzyme abundance (343). Expression of CHT3 increased slightly during cell wall regeneration (transcriptional profiling [46]). This delay likely indicates that the regenerating cells do not make daughter cells during the early stages of cell wall replacement.

WSC1 encodes a CWP whose expression increases in the absence of cyclic AMP (134). Expression decreased during mating (transcription profiling [411]). A deletion strain was hypersensitive to caspofungin, suggesting a requirement for Wsc1p for normal cell wall structure (279). The mutant strain had no defect in biofilm formation. However, the mutant strain had a severe defect on day 4 of culture growth, affecting the entry into stationary phase on day 5 (390). Cells become smaller as growth slows and cells enter stationary phase at approximately day 5 when grown on rich medium. Electron microcopy shows that after the cell enters stationary phase, the cell wall is thicker (45). SUN41, discussed above, had a severe defect at day 8 of culture growth that affected the maintenance of stationary phase (390). The effect on stationary phase of two CWPs emphasizes the necessity of constant vigilance to maintain cell wall integrity and that contributions to this integrity in growing and nongrowing cells may differ.

Unlike the enzymes discussed above, acid trehalase (GH65 family) is not involved in cell wall remodeling. Trehalose is a nonreducing disaccharide. Atc1p, acid trehalase, which contains a possible signal sequence for classical secretion, is one of two enzymes hydrolyzing trehalose (290). This enzyme has been recovered from isolated cell walls. The ATC1 in vitro translation product hydrolyzed trehalose. After the deletion of ATC1, a 170-kDa protein was missing from the extract of the mutant cell wall. The atc1Δ/Δ strain could not be grown on trehalose. The strain was also attenuated in the murine model of disseminated infection. Growth on glucose repressed the expression of ATC1 detected by RT-PCR compared to the growth on trehalose, thereby indicating that ATC1 expression is subject to glucose repression. Trehalose is a protectant against oxidative stress (7). The loss of a neutral cytosolic trehalase, Ntc1p, had no effect on the cellular response to oxidative stress (296). On the other hand, the atc1Δ/Δ strain, lacking acid trehalase, was more resistant to heat, osmotic, and oxidative environmental stress (289). At 1 h of exposure to 42°C, both the mutant and the wild-type strains were similar in terms of survival; however, after 3 h the mutant strain had an advantage. Osmotic stress did not affect the mutant strain until 3 h, while the parental strain was sensitive at 1 h. Although affected by oxidative stress, the mutant strain was more resistant. The mutant strain had somewhat reduced formation of germ tubes under conditions inducing morphogenesis. During longer periods of hyphal development, the mutant strain accumulated more trehalose than the parental strain. Thus, in the ATC1 deletion strain, the organism has a reduction in hyphal formation that is a putative virulence factor and continues to accumulate trehalose as protection against oxidative stress.

There are 10 SAP genes. Two of them, SAP9 and SAP10, encode proteins attached to the cell wall matrix by covalent linkages and are summarized in Table 2. The products of SAP1 to SAP3 and SAP4 to SAP6 are approximately 67% and 89% identical, respectively, and cluster together (247, 248). There are three fairly recent reviews that cover the SAP gene family (261, 262, 336), and these should be consulted for more detail than will be presented here. The Saps are the only known proteinases excreted into the extracellular, non-cell-associated space. All Candida species secrete proteinases, but species other than C. albicans appear to do so at a lower level (321). The loss of SAP2 and SAP4 to SAP6 severely impairs growth with protein as the sole nitrogen source (149, 327). The Saps are synthesized as prepropeptides with a signal sequence and propeptide of about 60 amino acids (151, 268, 384). The signal peptide is lost in the endoplasmic reticulum after serving to target the protein to the endoplasmic reticulum, and the propeptide is lost in the Golgi. Once secreted, the Saps may hydrolyze host proteins. Sap2p is the major Sap protein produced in vitro and has been the most studied for its properties (148, 405). Sap2p has a broad substrate specificity and among various activities can degrade host ECM (extracellular matrix), keratin, mucin, stratum corneum, proteinase inhibitors α-macroglobulin and cystatin A, and immunoglobulins (Igs) and activate proinflammatory cytokine interleukin-1β (IL-1β) and clotting factor X (18, 57, 159-161, 245, 306, 382, 383). It is a classic aspartic proteinase like the prototype pepsin. Activity is inhibited by pepstatin A and as well by several clinically useful human immunodeficiency virus (HIV) proteinase inhibitors (27). Aspartic proteinases have an acidic optimum pH. This is a potential disadvantage for C. albicans, since host tissues are at neutral or less acidic pH. Candida is a commensal in the vagina, with an acidic pH, and in the oral cavity, with neutral pH. Normal blood pH is about 7.4. There are some differences among Saps in response to pH. Activity can span a pH range from 2 to 7. Sap1-3p have their highest activity at lower pH values, pH 3 to 5, while Sap4-6p have theirs at higher pH values, pH 5 to 7 (26, 354).

There are differences in expression, and gene expression generally correlates with protein abundance (405). SAP2 is the major SAP gene expressed during the early stage of growth in response to in vitro culture on protein as the nitrogen source (Northern analysis [148]). SAP8 production is temperature regulated, it is more strongly induced during early growth at 25°C than 37°C, and it is greater in opaque than in white cells (Northern analysis [247]). However, expression has been detected in infections, suggesting that additional regulatory mechanisms may operate in SAP8 expression (RT-PCR [263, 313]). For example, in humans SAP8 was expressed more during infection than in oral carriage and was preferentially expressed in vaginal compared to oral disease (263). SAP1 and SAP3, which have sequences related to that of SAP2, are regulated by phenotypic switching and are expressed in the opaque phase of strain WO-1 (Northern analysis [405]). SAP4 to SAP6 are almost exclusively expressed in hyphae at the neutral pH used for induction (Northern analysis [54, 148, 405]), and the corresponding proteins are found in the medium (54). Expression is not dependent on protein in the medium. The SAP genes are affected by exposure to α pheromone, with SAP1 and SAP3 expression repressed (Northern analysis, transcriptional profiling, and real-time RT-PCR [22, 211, 334]) and SAP4 to SAP7 expression enhanced (transcriptional profiling, real-time RT-PCR [22, 334]). There are conflicting reports as to whether SAP2 expression is repressed (transcriptional profiling [22]) or enhanced (real-time RT-PCR [334]). Mutant strains in which single SAP genes are deleted are viable, as are strains with deletions of multiple SAPs (69, 149, 186, 334, 375, 403). The proteinase activity in biofilm culture supernatant is greater than that in planktonic organisms (241).

There are numerous studies of SAP gene function and expression (69, 70, 98, 149, 156, 158, 186, 313, 335, 337-339, 341, 365, 375, 376). Various studies employed in vitro-reconstituted tissue and animal models infected with mutant strains as well as the detection of gene expression at various infection sites. There are differences among studies in the number of SAP genes expressed and the effects of mutants on the outcome of infection. Each of the genes from SAP1 to SAP8 is expressed in one or more models. In some studies, expression is detected for two SAP genes, while five or six are detected in other models (dot and Northern blot analysis, RT-PCR [70, 335]). The profiles of expressed genes also vary during infection. For example, during the progression of infection in a reconstituted vaginal epithelial model, the expression of SAP1, SAP2, SAP4, and SAP5 was detected prior to that of SAP6 and SAP7, and in this model strains lacking SAP1 or SAP2 were attenuated, while those lacking SAP3 or SAP4 to SAP6 were not (RT-PCR [335]). In a model of murine keratitis, the loss of SAP6 resulted in mild disease, while the loss of SAP1 to SAP3 or SAP4 and SAP5 resulted in infection similar to that seen for the parent strain (156). Collectively, the studies show the expression of SAP genes in vivo in human tissue, reconstituted tissue, and animal models and that the loss of one or more SAP genes may modify infection in model systems.

To reach the extracellular space, proteins such as Saps must pass through the cell wall and have been detected there. Immunoelectron microscopy with anti-Sap antibody shows protein in the cell wall. An anti-Sap MAb showed reactivity within the cell wall of in vitro-grown cells (367). In a rat vaginitis model, reactivity was detected in the fungal cell 1 day postinfection and, to a lesser extent, 5 days postinfection. Antibody to Sap1-3p was reactive with the walls of yeast cells above the epidermis, in contact with superficial cells of the reconstituted tissue, or within epidermal cells (338). In agreement with the predominant expression of SAP1 to SAP3 compared to SAP4 to SAP6 as detected by RT-PCR, antibody to Sap4-6p detected little reactivity. In a similar study with reconstituted vaginal epithelium, Schaller et al. (335) found more reactivity at 12 h postinfection in the cell wall with antibody to Sap1-3p than with antibody to Sap4-6. The expression of SAP2 was detected by RT-PCR at 6 h and additionally SAP1, SAP4, and SAP5 were detected at 12 h, while SAP6 and SAP7 were not detected until later. Saps may also contribute to adherence. Antibody that recognizes Sap1-3p bound to the cell surface and antibody to Sap4-6p to the surfaces of organisms several hours after phagocytosis (26). Strains lacking SAP1, SAP2, SAP3, or SAP4 to SAP6 showed altered adherence to poly-l-lysine, Matrigel (primarily laminin and other basement membrane components), or BECs, depending on the surface and organism growth in glucose or galactose (403). The strains with deletions of SAP1, SAP3, or SAP4 to SAP6 that were grown in glucose showed reduced adherence to poly-l-lysine; those strains lacking SAP3 and SAP4 to SAP6 showed reduced adherence to BECs; and only the strain lacking SAP3 showed reduced adherence to Matrigel. Some HIV protease inhibitors inhibit the adherence of C. albicans to a HeLa epithelial cell line (19). Adherence to Vero cells was inhibited by some HIV protease inhibitors, and this adherence was associated with Sap1-3p (27). On the other hand, the same drugs did not inhibit binding to endothelial cells (97), which is consistent with the observation that strains with deletions of SAP1, SAP2, or SAP3 are unaltered in their adherence to endothelial cells (152).

In S. cerevisiae, both pheromone secretion and extracellular acid proteinase (barrier activity) regulate mating. A search for a candidate for a similar barrier activity in C. albicans identified SAP30 (334). Sap30p was found in a large-scale fusion protein screening as one of the secreted proteins (249). SAP30 (alias BAR1) expression is associated with mating type a cells (transcriptional profiling [388]). Upon treatment of opaque a cells, SAP30 expression increased more than 200-fold (real-time RT-PCR [334]). The deletion of SAP30 results in opaque a cells having a hypersensitivity to α pheromone. Mutant a but not mutant α cells display a mating defect, with very few mating zygotes found.

Various extracellular phospholipase activities that include reactions associated with PLA, B, and C have been reported for C. albicans (see the reviews in references 49 and 123). However, currently only PLB-type proteins are thought to be secreted (336). There are five PLB genes with signals for secretion and one SPO1 gene encoding a protein with similarity to PLB, which has a possible signal sequence. Three of the PLB genes, PLB3, PLB4.5, and PLB5, have GPI anchors, and whether they are localized to the membrane or to the cell wall has not been clarified (379) (Table 2). Secreted phospholipase activity has been reviewed recently, and the reviews cover secreted phospholipases more extensively than here (123, 336). Extracellular enzymes transit the cell wall and may be found in the wall. Phospholipase activity (combined PLA and B) is found in the cell walls of both yeast cells and hyphae in an oral reconstituted human epithelium (RHE) infection model (159). Activity localized to the periphery of the yeast cell and at the tips of hyphae. Immunoelectron microscopy also detected the protein in the cell wall (203). Environmental factors influence the expression of PLB1. Expression was detected in rich medium at 30°C but not at 37°C, although the protein was detected at the higher temperature (Northern analysis [252, 254]). In defined medium, expression at the higher temperature was detected only with the addition of serum. Expression was found in both yeast cells and hyphae, although in early stages of yeast growth, expression may be greater in yeast cells than in hyphae (Northern analysis [142, 252]). Tup1p (transcriptional corepressor of filamentation) regulates the expression of PLB1, as in its absence expression increased (Northern analysis, transcriptional profiling [142, 258]). Expression was unregulated in the tup1Δ/Δ strain, as the PLB1 expression was similar in conditions that favored either yeast cell growth or hyphal formation in the wild type (Northern analysis [142, 258]). In another study, PLB1 expression determined by RT-PCR did not correlate with adherence or hemolysin production (324). This suggests that phospholipase activity is not responsible for hemolysis, and as noted above the mannan portion of a mannoprotein may have this activity.

A strain with deletion of PLB1 showed no obvious phenotype and was unaltered in its adherence to endothelial and epithelial cells, but its ability to penetrate the cells was diminished (203). The strain also showed attenuated virulence in a murine disseminated infection model. PLB2 has been cloned and expression was detected in rich medium (Northern analysis [370]). The gene has not been deleted. However, Plb1p accounts for most of the activity, as only about 1% of the phospholipase activity remained after the deletion of PLB1 (123). A more recent clinical isolate study that compared the expression levels of various lipase genes in phospholipase-positive and -negative clinical isolates supported a similar suggestion (323). The negative group expressed only PLB2 and PLD1, while the positive group expressed PLB1, PLC1, and PLD1.

PLB1 is one of the core genes repressed more than 1.5-fold in response to osmotic, oxidative, and heavy metal stress (transcriptional profiling [94]). Many of the genes in the repressed group are involved in transcription and metabolism, and the reduction may be related to slowed growth in the presence of the stress. Hog1p also contributes to the transcriptional regulation of some genes in the absence of stress. Compared to hog1Δ/Δ cells, PLB1 is expressed 3.8-fold higher in HOG1 cells. The stress associated with drug treatment appears not to affect the expression of PLB1. There was no effect of treatment with ciclopirox olamine or fluconazole (RT-PCR [272]), caspofungin (RT-PCR, transcriptional profiling [210, 272]), or amphotericin B, ketoconazole, and flucytosine (transcriptional profiling [210]). However, a bisquaternary ammonium salt whose likely target is phospholipase inhibits growth (269). PLB2 is induced sixfold in response to pheromone (transcription profiling [22]). Expression is not affected by exposure to ciclopirox olamine or fluconazole (Northern analysis [272]).

The expression of PLB1 and PLB2 has been detected in infection specimens. In saliva from humans with either oral carriage of C. albicans or infection, the expression of PLB1 was more frequently detected in specimens from infected individuals and correlated with infection (RT-PCR [263]). There was no difference in PLB2 expression among infected individuals and carriers. There was no difference between individuals with carriage and infection in the vagina. This suggests a site-dependent difference in expression. In an oral infection model in immunocompetent mice, PLB1 expression was sustained for several days early in infection (RT-PCF [313]). In a murine gastric infection model, PLB1 and PLB2 expression was observed in both immunocompetent and immunodeficient gnotobiotic mice (RT-PCR [341]). At 3 to 4 weeks after infection, PLB2 expression was noted in tongue, palate, esophagus, and stomach, while PLB1 was detected in stomach. This suggests a site-independent expression of PLB2 and a site-dependent expression of PLB1.

Enzymes that hydrolyze ester bonds either at the lipid aqueous interface (lipases) or in solution (esterases) have been reported for C. albicans but also have been rather neglected. They were included in a recent review (336), which is a source of more information than given here. The lipase family contains 10 genes, LIP1 to LIP10, with LIP7 lacking the signal for secretion. Molecular analysis of the LIP family began in 1997 with the cloning of LIP1 (112) and was followed in 2000 with the cloning of the other nine genes (150). The protein sequences are similar, with up to 80% amino acid identity. Expression analysis employed RT-PCR except for Northern blotting for LIP1. LIP3 to LIP6 are expressed in vitro in all the media and conditions tested. LIP1, LIP3 to LIP6, and LIP8 expression occurs in medium with Tween 40 as the carbon source, while LIP2 and LIP9 expression occurs only in medium without lipid. The expression of some lipase genes in the absence of lipid suggests that there may be additional functions for these enzymes. During hyphal formation in defined medium, the expression of LIP1 was not detectable by either Northern or RT-PCR analysis. LIP2 and LIP9 and LIP10 expression also was not detected. However, LIP3 to LIP5 and, to a lesser extent, LIP6 were expressed throughout the transition. LIP8 expression was detected at 1, 3, and 5 h but not subsequently. When serum was used for hyphal induction, the expression pattern was similar, except LIP1 was detected and LIP6 expression was enhanced. There are also differences in response to other in vitro conditions. LIP1 is repressed by α pheromone (transcription profiling [22]), LIP2 is repressed during mating (transcription profiling [411]), LIP4 is expressed more in opaque than in white cells (transcription profiling [193]), and LIP6 is induced upon adherence to polystyrene (transcription profiling [228]). Upon exposure to ciclopirox olamine, the responsivenesses of the genes were different, with LIP1, LIP2, and LIP5 showing strong, modest, and moderate repression, respectively (RT-PCR [272]). Organisms exposed to fluconazole showed a modest decrease in LIP4 expression and an increase in LIP8 expression. Another study found no effect of ketoconazole on LIP gene expression but observed a decrease in LIP6 expression upon exposure to caspofungin (transcription profiling [210]).

LIP4 has been expressed in S. cerevisiae and the recombinant protein analyzed (320). The enzyme is a true lipase hydrolyzing insoluble triglycerides and is able to use methyl, ethyl, and propyl esters. The enzyme had highest selectivity with unsaturated fatty acids. There was evidence that esterification was by acyltransfer. A screen for genes whose haploinsufficiency affected filamentation identified LIP5 through reduced filamentation on Spider medium (228). LIP8 has been deleted (115). Heterozygous lip8Δ and lip8Δ/Δ strains showed growth similar to that seen for the wild type in complete medium, but with Tween 40 as the carbon source, both strains showed less than 80% of the wild-type growth. Surface properties of the lip8Δ/Δ strain but not of the lip8Δ strain were altered, as the null strain showed a tendency to flocculate. There was no difference between mutant and wild-type strains in morphogenesis in serum-containing medium. There was no difference between mutants and wild type when they were grown on solid medium containing Tween 20 or Tween 80. However, in a specialized medium, the lip8Δ/Δ strain showed a moderate but significant reduction compared to other strains. The phenotypes on solid medium differed, as the mutant strains showed a rough phenotype at pH 7 and 10 compared to what was seen for the wild type or at pH 4. The introduction of one copy of the wild-type allele into the heterozygous but not the homozygous mutant reversed this appearance. Rough colonies primarily contained mycelia and smooth colonies of yeast cells. This difference could be due to a gene dosage effect or allelic differences, as seen in the ALS family discussed later. The lip8Δ/Δ strain showed attenuated virulence in the murine model of disseminated infection but not in peritoneal infection (115). As described below, LIP8 expression is frequently detected during infection.

As with SAP genes, there are differences in the expression of LIP genes in infection models. In a murine peritonitis model, the expression of lipase genes was determined in liver tissue (RT-PCR [150]). LIP5, LIP6, LIP8, and LIP9 were expressed in some or all livers analyzed. Subsequently, expression in this model was examined further with an analysis of livers obtained at 4 and 72 h postinfection (RT-PCR [366]). LIP10 was not detected. LIP4, 6, and 9 were expressed in most livers at the first time point, with LIP5 and LIP8 expression being found in all livers. LIP1 and LIP3 but not LIP2 expression was detected. At the later time, LIP5 and LIP8 expression was found in all livers, and LIP2 expression was then detected in some livers. The expression of other lipase genes, particularly LIP9, generally decreased. At 3 days, the expression pattern in kidney was similar to that in liver. In the RHE model, both constitutive and variable expression occurred. The expression of LIP1, LIP4, LIP5, LIP6, and LIP8 was observed at all three time points, i.e., 12, 36, and 48 h after inoculation (RT-PCR [366]). LIP2 and LIP9 expression was not detectable until 36 h and was still present at 48 h. LIP3 and LIP9 were just barely detectable at 48 h. In another murine model, expression was monitored in alimentary tract colonization in the cecum and mucosal infection in the stomach, hard palate, esophagus, and tongue (RT-PCR [340]). The expression was determined in immunodeficient gnotobiotic mice 3 to 4 weeks after inoculation. LIP4 to LIP8 were expressed in mucosal tissues and cecum contents, while LIP1, LIP3, and LIP9 and occasionally LIP10 were detected in gastric tissue but not oral tissue. LIP2 expression was detected only in cecum contents. Analysis of lipase expression in eight saliva specimens from patients with oral infection also showed selective expression (RT-PCR [366]). The most commonly expressed genes were LIP4, LIP5, and LIP8 in four or five specimens; LIP6 was detected in two specimens, while LIP1-3 were detected in a single specimen and LIP10 was not detected. These observations are similar to those discussed above for SAP genes, for which there is apparent condition-dependent use of genes of a family.

C. albicans has both internal and extracellular phosphatases. There are both constitutive and inducible acid phosphatases. The secreted acid phosphatase was one of the first mannoproteins to be characterized (reviewed in reference 49). These enzymes have received little recent attention as phosphatases. Pho100p is induced at acid pH and low phosphate (129). The protein is extracted from intact cells in buffer containing dithiothreitol (DTT). Pho100p is N glycosylated. PHO100 expression is induced in early biofilm formation (transcription profiling [260]). Pho112p and Pho113p, constitutive phosphatases, were also extracted from cells with DTT-containing buffer (129). Both phosphatases were N glycosylated. Pho112p was detected in the large-scale protein fusion screen for secreted proteins (249). PHO113 was negatively regulated by Rim101p (transcription profiling, RT-PCR [311]).

There are two potential glucoamylases. A 190-kDa protein was expressed more abundantly on the surfaces of galactose-grown cells (reviewed in reference 49). Peptide sequences from this protein were obtained and suggested a glucoamylase (369). Probing a genomic library with a sequence from a heterologous gene with protein homology to the peptides resulted in the isolation of the gene. GCA1 encoded a protein with potential N-glycosylation sites. Northern analysis showed that the gene was expressed at a high level in organisms grown on galactose compared to what was seen for those grown on glucose or sucrose. RT-PCR analysis showed that expression in cells grown on starch and cellobiose was the same as expression in cells grown on galactose. Expression was less for cells grown on sucrose and even less for cells grown on glucose. GCA1 was corepressed by Mig1p and Tup1p (transcription profiling [258]). GCA1 expression is downregulated at alkaline pH by Rim101p (transcription profiling [23]) and induced by ketoconazole (transcriptional profiling [210]). GCA1 is also expressed in a rat model of oral infection (RT-PCR [369]). Expression in cells grown in Lee medium was 3.5-fold greater than that in vivo. The second gene, SGA1, encodes a fungus-specific putative glucoamylase that has not been localized (31). Sga1p was identified as part of the large-scale genetic screen for exported proteins (249). Sga1p was one of proteins encoded by fusions of C. albicans DNA with the intracellular form of invertase that conferred growth when expressed in a heterologous S. cerevisiae suc⁻ strain. The protein has a predicted signal sequence. C. albicans SGA1 is the ortholog of S. cerevisiae SGA1, which encodes an intracellular sporulation-specific enzyme. C. albicans SGA1 expression decreased rapidly as protoplast regeneration was initiated (transcription profiling [46]).

The ALS gene family encoding GPI-CWPs (Table 2) includes eight genes and the family has recently been reviewed by Hoyer et al. (144) and should be consulted for a more extensive discussion. As expected, the Als proteins are found at the cell surface (Fig. 1B). Briefly, the encoded proteins have three domains (143). The central domain, consisting of tandem repeats, can vary in number between genes and alleles and can be used to classify three subgroups: based on cross-hybridization of the tandem repeats, ALS1, ALS2, ALS3, and ALS4 are in one group and ALS5, ALS6, and ALS7 are in a second group, while ALS9 is in neither of these two groups (143, 145, 146). The last two domains have multiple glycosylation sites that are predicted to lead to a stretched conformation that can extend the N-terminal probable binding domain from the attachment site. The presence of tandem repeats is a potential site for allelic variation. Allelic variation can occur within the same cell with differences in the two alleles borne by the diploid cell or between strains (144). For example, in the widely used strain SC5314, the larger alleles of ALS1, ALS5, and ALS9 are on one chromosome, while the smaller alleles are on the other (417). In one study with many isolates, the tandem repeat number within ALS1 was 16 for the most common allele, but alleles with 4 to 37 copies were also found (220). As noted below, alleles may differ in their contributions to the phenotype of the organism. Recombinant soluble Als5p fragments can aggregate in solution as amyloid-like fibers, and there are conserved sequences associated with this aggregation in other Als proteins (284). Such interactions may contribute to the adherence function of Als proteins, and this possibility undoubtedly will be explored.

The transcription of all family genes in vitro and during infection has been detected, but some genes (ALS6 and ALS7) have been observed with only low levels of expression (144). ALS gene expression may be affected by the stage of growth and by morphology (133). ALS1, ALS2, and ALS3 have a dynamic range of expression, with large increases and decreases. The expression of Als proteins was monitored by constructing strains that expressed a GFP fusion protein. Als1p-GFP expression (monitored through fluorescence flow cytometry) increased more than 20-fold upon transfer into fresh medium before slowly decreasing, and Als7p-GFP had a transient increase shortly after transfer, although at a low level. Als5p-GFP and Als6p-GFP were expressed at low levels, with a small, late increase in Als6p-GFP expression. Both Als3p-GFP and Als9p-GFP were at background levels. During germ tube formation, Als1p-GFP expression was the first to increase and attained the highest level; Als3p-GFP expression increased more slowly and rose abruptly coincident with microscopic germ tube emergence. The expression levels of Als5p-GFP, Als6p-GFP, Als7p-GFP, and Als9p-GFP were each comparable to the control level. The early expression of Als1p-GFP followed by a decrease in germ tube formation suggests that Als1p should be found in the initial growth of the germ tube. This was observed with a MAb to the N terminus of Als1p, where bright fluorescence was localized to the base of the germ tube (232). Transcript numbers determined by real-time RT-PCR generally agree with the relative abundance as assessed by GFP monitoring (132). Real-time RT-PCR and RT-PCR showed that when expression occurred ALS1, ALS2, and ALS3 were the most abundantly expressed group, while ALS4 was the most abundantly expressed of the low or weakly expressed group (130, 133). ALS3 expression was much greater in germ tubes than in a 16-h yeast culture. In biofilms, ALS1 was also a major differentially expressed gene compared to what was seen for planktonic gene expression by microarray analysis (119) and real-time RT-PCR (280). If protein abundance parallels gene expression, then some of the Als proteins may be present at very low levels and perhaps not detected, as in the studies by de Groot et al. (73), who detected only Als1p and Als4p, and Ebanks et al. (87), who detected either Als3p or Als6p (Table 4). The ALS genes as well as EAP1 are genes of the broader class of flocculation genes found in yeasts (see the review in reference 395). SFL1 (Orf19.454) is an ortholog of S. cerevisiae SFL1, a suppressor of flocculation (18). In the absence of Sfl1p, the expression of ALS1 and ALS3 increased. Recently, the regulation of ALS3 expression focused on the promoter region (8). Two regions were identified, one required for activation in hyphae and proximal to the coding region and a distal region that amplified the response (8).

In disease models and clinical samples, ALS genes show variation in expression. In a RHE model, the expression of ALS1, ALS2, ALS3, ALS4, ALS5, and ALS9 was present over time (12, 24, 36, and 48 h) for each of the three inoculum concentrations tested (RT-PCR [130]). For both ALS3 and ALS5, one replicate among all the analyses fell below the detection limit. However, for ALS6 and ALS7, this occurred more often and in the case of ALS6 both replicates were below the detection limit for two analyses. This suggests that ALS7 and particularly ALS6 are expressed at low levels. Two studies compared expression from human clinical specimens with expression in an animal model. Analysis of mRNA was more sensitive than total RNA for human vaginal fluid samples (RT-PCR [55]). ALS1-3 and ALS9 were detected in 70 to 87% of specimens; ALS4, ALS5, and ALS7 were detected in about 30% of samples; and ALS6 was detected in only 13% of samples. In a murine vaginitis model, ALS expression was determined at day 4 and day 7. The percentage of samples in which expression was detected was greater at day 7 than at day 4. At day 7, ALS, ALS2, and ALS3 expression was detected in all mice. ALS9 expression was detected in about one half of the animals on both days. ALS6 and ALS7 expression was detected on both days in one or two mice. ALS4 and ALS5 were detected on day 7 only in one of five mice. Expression in a vaginal RHE model was assessed at 12, 24, and 36 h. ALS, ALS2, ALS3, and ALS9 were consistently strongly expressed and ALS5 was consistently but somewhat more weakly expressed. ALS4, ALS6, and ALS7 expression showed negative or weak signals. Since ALS4 showed strong expression in inoculum cells, the reduction of its expression in samples recovered from the vaginal RHE suggested that expression may be downregulated.

In a hyposalivatory rat model of oral infection, ALS expression was greater at day 5 than at day 3 after inoculation (RT-PCR [131]). ALS1, ALS2, ALS3, and ALS4 expression was detected at both time points, while ALS5 and ALS9 expression was not detected until day 5. ALS6 was detected only once, and ALS7 expression was never detected. There were differences in the numbers of expressed ALS genes depending on the tissue. Generally, more genes were expressed in organisms from the tongue and mandible than in those from mucosae. Oral scrapings from HIV-positive individuals were analyzed for ALS expression. The expression of all ALS genes was detected in some specimens. ALS1, ALS2, ALS3, and ALS9 expression was found in all specimens. Expression of the other genes was more variable, with ALS4, ALS5, and ALS7 found in 66 to 83% of samples and ALS6 being detected least often, at 50%. In a murine model of disseminated infection, only GFP fusion constructs with ALS1 and ALS3 were detectable by protein expression, primarily in the kidney (132). The kidney was the most frequently infected organ, followed by the heart, with the liver and spleen occasionally infected. The more sensitive RT-PCR method was also used for detection. The other ALS genes were detected by more-sensitive RT-PCR. ALS1-3, ALS9, and, less frequently, ALS4 were expressed, while ALS5, ALS6, and ALS7 expression was not detected. Generally, both clinical samples and model samples show two classes of expressed genes. ALS1, ALS2, ALS3, and ALS9 are the most abundantly expressed genes, while the others are less frequently expressed, particularly ALS6. Generally, the genes most abundantly expressed in culture are also the most abundantly or frequently expressed in infection models, except for ALS9, which was often expressed at low levels in culture but was consistently found to be expressed in infection models.

The sequence- and homology-based prediction that Als proteins would have adherence properties was borne out by several studies (144). Some of these properties are summarized in Table 5. Since the Als proteins expressed on yeast cells can vary with the growth state, there should also be variation in the contribution of Als proteins to adhesive interactions. Adhesive properties of these proteins have also been examined by heterologous expression in S. cerevisiae. This approach offers an opportunity to study their adhesive properties in greater isolation than possible on the C. albicans cell but also has the risk that context modifies the properties. These studies are discussed in the review by Hoyer et al. (144) and will not be included here. Adherence to host cells or other ligands shows that strains with deletions of ALS genes have altered binding. The loss of Als1p reduced binding by 20% (strain 1) and 35% (strain 2) to HUVECs (113, 413), and overexpression increased binding (113). Antibody to the N terminus of Als1p reduced binding to endothelial cells (113) and again implicated the N terminus in binding. Using a third als1Δ/Δ strain and an assay for cell association, no reduction in cell association with HUVECs was noted (300). There are also conflicting reports on the effect of the deletion on binding to epithelial cells. Yet another strain showed a reduced binding to FaDu cells (pharyngeal epithelial cell line established from a squamous cell carcinoma) (1), while strain 3 showed no reduction in association with FaDu cells (300). A fourth als1Δ/Δ strain showed attenuation in an oral model of infection and reduced binding to murine tongue in an ex vivo assay (165). Strain 1 showed no loss of binding to FaDu cells, BECs, or RHE, although there was a reduction in the destruction of RHE compared to what was seen for the control strain (413). There was no effect on binding to fibronectin or laminin (144, 413). The apparent differences among strains may arise from one or more sources, such as unknown alteration(s) introduced into the mutant during construction, differences in the mammalian cells, differences in the growth levels, and differences in assays of the interaction between fungus and host cells. A side-by-side comparison of these strains should answer at least some of these questions.

The deletion of ALS2 could not be achieved, and the gene was placed under the MAL2 promoter (416). The deletion of ALS4 resulted in an increased expression of ALS2 and vice versa with the uninduced conditional ALS2 strain, as determined by real-time RT-PCR. This suggests that there may be some compensatory effects and perhaps some functional overlap. The loss of ALS4 reduced adherence to HUVECs but not to BECs or RHE and did not cause the destruction of RHE. The uninduced conditional ALS2 strain showed reduced adherence to HUVECs and RHE and reduced RHE destruction, compared to what was seen for the control, but loss of adherence to BECs.

Compared to the loss of Als1p, the absence of Als3p had a great effect on reducing adherence to HUVECs (413). While the absence of Als1p did not alter binding to BECs, the absence of Als3p reduced binding about 60%. The expression of the N-terminal region of Als3p in the null strain partially restored binding to BECs (412). Antibody recognizing the N-terminal region reduced binding. In the case of RHE, the strain lacking Als3p bound and there was little subsequent destruction of RHE (413). There was no effect on binding to fibronectin or laminin (144, 413). When a second strain was tested for cell association, there was a reduction in the interaction with HUVECs but no change with FaDu cells (300). Like the als1Δ/Δ strains, there may be several sources for differences between strains and assays. Indeed, as is noted below, the second strain showed a large reduction in endocytosis with both HUVECs and FaDu cells (300). In strain SC5314, the two ALS3 alleles have different numbers of tandem repeats (282). A heterozygous strain expressing the allele with more tandem repeats showed adherence to HUVECs and FaDu pharyngeal epithelial cells similar to that seen for the parent strain. The heterozygous strain expressing the allele with fewer repeats had diminished adherence compared to the parent, thus implicating this region in adherence. However, the role of this highly glycosylated repeat is likely that of extending the N-terminal binding domain further from the attachment site (144, 282).

The effect of deleting ALS5, ALS6, or ALS7 was surprising in that the strains with a deletion showed increased adherence to HUVECs and BECs (414). The loss of Als7p had no effect on cellular aggregation, while the loss of Als5p and Als6p increased aggregation slightly but not enough to account for the increased adherence of strains lacking Als5p, Als6p, or Als7p. The reintegration of one allele did not restore adherence comparable to that of the control strain, except for the strain with a reintegration of one ALS6 allele in terms of its adherence to HUVECs. Deletion had no effect on the destruction of RHE. Since expression of these genes is at very low levels, proteins may also be at very low levels. This potential low level of protein suggests that these proteins may have a role in cell surface conformation by exposing adhesins or by an alteration that is compensated by additional adhesins. The absence of Als9p altered adherence to endothelial cells but not that to epithelial cells or laminin (415). There was also no effect with the RHE model (414). In a strain with reintegration of the functional gene, one allele but not the other could restore function (415). This is another example of allelic difference.

In biofilms, organisms are in contact with other organisms and Als adhesins. BCR1 encodes a transcription factor required for biofilm formation that regulates expression of several adhesin genes, including ALS1 and ALS3 (273). Nobile et al. (273) found that an ALS1 null strain formed an in vitro biofilm that was not strongly attached to the surface. In addition, the strain with a loss of ALS3 had severely inhibited biofilm formation. Only a rudimentary biofilm that had mainly yeast cells formed in vitro on a silicone surface. A bcr1Δ/Δ strain did not form biofilm but was rescued by the overexpression of ALS3 and partially rescued by the overexpression of ALS1. Since ALS3 is not normally expressed in yeast cells, and cells of an als3Δ/Δ strain can adhere, the defect in the als3Δ/Δ strain is likely to be later in biofilm development. In an in vivo rat venous catheter model, the bcr1Δ/Δ strain also failed to form biofilm, but the als3Δ/Δ strain produced apparently normal biofilm. However, as in vitro, the overexpression of Als3p rescued the in vivo biofilm defect of the bcr1Δ/Δ strain. Hoyer and colleagues (412) found that throughout the biofilm Als3p-GFP was distributed diffusely on the germ tube surface. In this study, biofilms formed by als3Δ/Δ were disorganized and unstable, with hyphae that were parallel rather than exhibiting the usual entwined structure. Efg1p is required for the expression of ALS3, and a strain lacking EFG1 formed only a sparse structure. The overexpression of Als3p in an efg1Δ/Δ strain restored biofilm formation. Although both studies with ALS3 null strains reached similar conclusions that Als3p is important for biofilm formation, there were differences in the reduced biofilm formed. While this may reflect strain differences, it may also reflect the influence of environmental conditions, e.g., medium, on biofilm formation, as suggested by the differences in in vitro and in vivo biofilm formation described above. The loss of Als2p but not of Als4p reduced biofilm mass about 15% (416). This suggests that although there is some compensatory expression of ALS4 when ALS2 is deleted and vice versa (416), the compensation of Als4p is insufficient or does not extend to this function.

A recent report identified cadherin as another ligand for Als3p (300). C. albicans invades endothelial and oral epithelial cells by the hyphal induction of host cell endocytosis (102, 288, 298). Host cell N-cadherin is the surface receptor engaging hyphae (299). The deletion strain lacking Als3p but not Als1p failed to induce endocytosis by both HUVECs and epithelial cells of the FaDu cell line (300). The restoration of functional Als3p also restored endocytosis. An affinity purification scheme found that N-cadherin and E-cadherin from endothelial and epithelial cells, respectively, were missing from host proteins bound to the Als3p-deficient strain compared to what was seen for parental and Als1p-deficient strains. In the absence of endocytosis, little damage occurred to the mammalian cells. The binding site resides in the N-terminal portion of Als3p. Latex beads were coated with recombinant N-terminal protein fragment (∼400 amino acids) of Als1p or Als3p or control bovine serum albumin (BSA). The rAls3p-coated beads were endocytosed by CHO (Chinese hamster ovary) cells expressing either N-cadherin or E-cadherin, while control beads coated with BSA were not. Although Als1p and Als3p share considerable similarities, molecular modeling revealed differences that predicted that the Als3p N-terminal domain would interact with either cadherin, similar to what is seen for cadherin-cadherin interactions.

Als proteins may have an additional role in cell growth. Despite the expression of the corresponding proteins at very low levels in wild-type cells, the deletion of ALS5 or ALS6 resulted in strains with slowed growth (414). The growth rate did not change with the addition of uridine, suggesting that the defect was not associated with the URA3 marker. Only in the case of the ALS5 strain did the restoration of one allele restore a normal growth rate. Yeast cells of the als1Δ/Δ strain were smaller in size than the control strain when grown in the same conditions (144). Germ tube formation was also delayed in deletion strains (113, 413). The deletion strain also had delayed pathogenesis in several models (1, 113, 165). An earlier report by Chaffin and Sogin (50) associated decreasing cell size with delayed germ tube emergence. The delay in pathogenesis may be associated with the delay in hyphal formation. These observations with Als1p, Als5p, and Als6p suggest, at least with these proteins, that proteins may have more than a single role.

Eap1p was originally identified in a screen of a genomic library expressed for sequences that conferred adhesive properties on S. cerevisiae (205). In C. albicans, EAP1 is regulated by Efg1p, a regulator of cell morphology. Eap1p, like Als proteins, is a GPI-CWP (Table 2) and corresponds to one of the yeast flocculation genes reviewed by Verstrepen and Klis (395) and to yeast adhesins reviewed by Dranginis et al. (84). An indirect immunofluorescence assay demonstrated that Eap1p tagged with a hemagglutinin epitope is found at the cell surface (206). Glucanase digestion released a tagged protein construct from the cell wall, and the loss of the GPI anchor site also resulted in finding the protein in the medium. To assess the adherence of wild-type and eap1Δ/Δ strains to polystyrene surfaces, the shear force required for detachment was determined. Fewer mutant cells than wild-type cells adhered. The heterozygous EAP1/eap1Δ cells adhered with intermediate frequency. This observation suggested that Eap1p contributes to adherence to polystyrene but is not the only adhesive surface component. However, once adhered, both wild-type and mutant cells were fairly resistant to removal. Comparison of the adherence of wild-type cells and that of mutant cells to a human embryonic kidney cell line showed that mutant cells were reduced 37% compared to what was seen for the wild type (205). Adherence was dependent on Efg1p, as the efg1Δ/Δ strain had reduced adherence compared to the wild-type strain and adherence was restored by the expression of EAP1 under the actin promoter in the efg1Δ/Δ strain (205). Since the deletion of EAP1 did not abrogate adherence, this observation suggests that other factors also contribute to adherence. Northern analysis showed that EAP1 was expressed in both yeast cells and hyphae and that expression was dependent on Efg1p (205). An earlier transcriptional profiling study of the yeast-to-hypha transition reported a twofold increase in EAP1 at 6 hours (264). However, in this study no morphological difference in expression was noted at 6 h with Northern analysis (205). EFG1 is a regulator of morphogenesis and cell wall remodeling. Although EAP1 expression is not induced in germ tube formation, its expression is still controlled by Efg1p.

Biofilms for both the wild-type and null strains were formed under constant medium flow with defined shear force (206). The wild type formed biofilm as described in other biofilm studies. Biofilm developed from adhered yeast cells to a multilayered structure with yeast cells and hyphae and ECM. Yeast cells of the mutant strain adhered and formed pseudohyphae and hyphae at 8 h in a biofilm several layers thick. However, by 20 h most of the cells washed out. Thus, the eap1Δ/Δ strain was able neither to form a thick biofilm nor to maintain a reduced thickness biofilm in vitro. The strains were also tested in vivo in the rat venous catheter model and, as described elsewhere for ALS3 and HWP1 strains, the wild type formed a biofilm in the catheter lumen. The null strain did not. EAP1 gene expression in sessile cells from both the in vitro and in vivo models was about twice that of planktonic organisms (real-time RT-PCR). The planktonic control contained a mixture of morphological forms, as did the biofilm. This difference is the same for morphological forms found by transcriptional profiling (264) but differs from the lack of difference determined by Northern analysis (205). This difference may be associated with biofilm formation or may be influenced by the different growth media used in the determinations.

Hw1p is a unique adhesin expressed on the hyphal surface that becomes covalently attached to host cells (see the reviews in references 49 and 371). An initial description of HWP1 expression found the transcript only in hyphal organisms by Northern analysis (363), and a subsequent analysis found very faint levels in yeast cells by Northern analysis (68). Antibody to the protein showed expression only on hyphal surfaces both from in vitro-produced hyphae (68, 363) and hyphae in the superficial layer of the stomach of a highly colonized beige mouse (363). HWP1 encodes a GPI-CWP whose expression is regulated by a number of transcription factors that control morphogenesis and mating (Table 2). Exposure of opaque a/a cells but not α/α cells to α pheromone increased HWP1 expression in Lee medium (Northern analysis [68]) and also on Spider medium (transcription profiling [22]). While opaque cells do not form germ tubes in response to α pheromone, they do form conjugation tubes in preparation for mating, and these conjugation tubes are elongated structures. Treatment with anti-Hwp1p antibody specifically stained the emerging and elongating conjugation tubes of the a/a cells (68). When the growth reverted to yeast growth at the apex, expression terminated and the yeast cell was unstained. When α/α cells along with a/a cells were exposed to α pheromone in a mating culture, only the conjugation tubes of a/a cells expressed Hwp1p. The differential Hwp1p expression on parent cells facilitated the determination that the first daughter cell of a mating emerges from the conjugation tube contributed by the a/a parent. Adherence to bring the mating cells together is involved in mating, but whether Hwp1p functions as an adhesin in this process is unknown.

Recently, an analysis of a segment of the promoter region was found to bind several proteins that differed in yeast cells and during hyphal induction (175). Like the promoter analysis of ALS3 above (8), two regions were found. The model proposed for yeast cells was that the promoter was occupied by DNA-histone octamers and Gfc1p and Hbp3p that inhibited the binding of activators (175). During hyphal formation, the removal of this blockage allowed activation that was amplified by the interaction of the distal region with transcription factors previously described, e.g., Efg1 and Nrg1. HWP1 is also regulated by the recently described Sfl1p, with increased expression in the absence of Sfl1p (RT-PCR [18]).

This GPI-CWP protein is a substrate for mammalian transglutaminases that cross-link the C. albicans protein to the surface of oral mucosa (362). The cross-linking occurs in the N-terminal domain of Hwp1p. There is similarity in this domain to small proline-rich proteins (PRPs) expressed on the surface of stratified squamous epithelium. Like the adhesive interactions between Als3p and cadherins described above, the C. albicans protein may mimic a host protein. Not all BECs are capable of cross-linking with C. albicans via Hwp1p (362). Recombinant Hwp1p attached to only the more terminally differentiated BECs displaying SPR3 (a small PRP) and keratin 13 and not to the less terminally differentiated cells displaying involucrin (305).

Recent reports have suggested the possibility of a new adhesive role for the protein in biofilms (273, 277). The expression of the transcription factor Bcr1p was required for biofilm formation (276). Targets for control by this factor include several genes encoding surface proteins: ALS3, ALS1, and HWP1. A strain lacking Hwp1p formed a biofilm which was prone to detaching from the abiotic substrate (273). The hwp1Δ/Δ strain produced a biofilm with reduced biomass compared to that seen for the wild type (277). The biofilm had few hyphae, while both planktonic yeast cells and hyphae were in the surrounding medium. Several years ago, a study showed that the deletion of HWP1 does not cause a filamentation defect (347). Biofilms formed when bcr1Δ/Δ mutant cells were mixed in various ratios with cells of strains with wild-type HWP1 or with heterozygous or homozygous null mutations (277). The surrounding medium of a biofilm formed with the bcr1Δ/Δ strain had many planktonic organisms. The mixture with hwp1Δ/Δ mutant cells also had many planktonic organisms but fewer than the bcr1Δ/Δ mutant alone. When mixed with a strain expressing Hwp1p, the planktonic population was substantially lower, even when 87.5% of cells were from the BCR1 null strain. The authors suggested that Hwp1p acts as an adhesin to retain organisms in the biofilm. In the in vivo model for biofilm formation in a rat venous catheter, the HWP1 null strain was also defective in in vivo biofilm production. This is unlike what is the case for the als3Δ/Δ mutant discussed previously, which formed biofilm in vivo but not in vitro. Also, unlike what was seen for Als3p, the overexpression of Hwp1p in the BCR1 null strain did not restore the biofilm formation defect. However, in view of the differences with als3Δ/Δ mutants in vitro and in vivo, environmental conditions may be a factor.

Ywp1p is a GPI-CWP (Table 2) whose expression is greatest on yeast cell surfaces just after the end of the most rapid exponential growth and thereafter declines (129). In low-phosphate medium, GFP-tagged protein persisted longer on growing cells and cells from several-day-old cultures. Expression was not observed on hyphae and pseudohyphae. The protein in the medium increased as the culture approached stationary phase through either shedding of attached protein or secretion without attachment or both. The ywp1Δ/Δ strain adhered more to polystyrene and other surfaces than did the wild type and formed yeast cell biofilm with several layers, in contrast to wild-type cells. Wild-type cells showed less adherence and monolayer formation under biofilm conditions. At conditions conducive to hyphal formation, there was less effect, as Ywp1p is not expressed on hyphal surfaces. Thus, this protein appears to block adherence of wild-type cells, a property that it shares with Als5p, Als6p, and Als7p (Table 5). The suggestion put forth was that the protein promoted dispersal through limited adherence or perhaps inhibition by soluble protein, which would allow the organism to seek new sites for colonization. This suggestion was supported by the observation that the addition of culture medium containing the protein was more inhibitory of adherence and biofilm formation than that of culture medium without the protein. To colonize new sites, cells from biofilm must disperse and, as most wild-type biofilms have abundant hyphae, this protein on yeast cells might promote this. YWP1 expression is upregulated in white cells of the WO-1 strain, where there are distinct microscopic morphological differences between the white and opaque yeast cells (transcriptional profiling [193]). This suggests that the role of Ywp1p is more important in white cells. The expression of this gene is also regulated under other conditions (Table 2). One transcription regulator displays a complex regulation. YWP1 expression is dependent on EFG1 as well as downregulated in both an EFG1 null strain and an EFG1-overexpressing strain, perhaps through both the activation and the repression of YWP1 (83, 356).

Ecm33p is a GPI-CWP (Table 2). ECM33 is a member of a two-gene family that includes ECM331. ECM33 contains an intron (231). Genes involved in splicing, translation, and mitochondrial respiration are overrepresented among genes with introns (244). ECM33 is not part of this bias representation, as it encodes a CWP. The deletion of this gene affects both yeast cell and hyphal morphology. On solid medium, ecm33Δ/Δ cells seemed to be rounder and larger than in liquid media, with this appearance being even more pronounced in stationary phase. However, the period of growth appeared to be 24 h, which is less time than in a more rigorous description of stationary phase (390). As normal cells progress into stationary phase, they become smaller. This ecm33Δ/Δ strain may also have abnormalities after an additional growth period to reach and maintain stationary-phase status. The doubling time of the mutant strain was greater than that of the parental strain (232). In the ecm33Δ/Δ cells, filamentation was delayed in YEPD with serum, and fewer hyphal organisms were produced, with the width of those filaments being greater than that seen for the parental strain (231). In wild-type strains, the more rapidly growing cells are larger than the slower-growing cells (48). In addition, germ tubes are initiated more rapidly from large cells than from small cells (50). Comparison of the ecm33Δ/Δ cells with wild-type cells suggests that there is a disregulation in cell size. Cells of the als1Δ/Δ strain also have delayed filamentation, but the yeast cells are also smaller (144, 413), and the size relationships with wild-type cells appear to be retained in this mutant. Calcofluor white staining of ecm33Δ/Δ cells showed an alteration, with much of the reactive material appearing as an aggregate (231). Both ECM33/ecm33Δ and ecm33Δ/Δ cells had a tendency to flocculate in a medium-dependent manner, as this flocculation was observed in YEPD and not in yeast nitrogen base. This may represent a protein interaction or a loss of interaction with another medium-dependent CWP. For example, PIR1 is not expressed in YEPD but is in yeast nitrogen base (166, 229a). The gene dosage effect was also seen for the heterozygous and homozygous deletion strains in sensitivities to Calcofluor white, Congo red, and hygromycin B. These observations implicate Ecm33p in cell wall integrity and shape.

Electron microscopic imaging of the ecm33Δ/Δ and ECM33/ecm33Δ strains and comparison to wild-type and complemented strains showed differences (232). Both the heterozygous, complemented strain with one allele and the null strain showed differences. Strains with one wild-type allele showed an abnormally wide and electron-dense outer layer in comparison to the wild type (Fig. 1 shows examples of the wild type). In the ecm33Δ/Δ strain, not only was the outer layer unusually electron dense but the inner electron translucent layer was wider and less electron dense in comparison to the wild type. Additional evidence of aberrant wall structure was obtained with the reaction of a MAb to the N terminus of Als1p. In the wild-type hyphae, Als1p was localized to the region adjacent to the mother cell. In a heterozygous strain, the immunofluorescence associated with the binding of the MAb extended further up the developing hyphae, and in the null strain the fluorescence was diffusely observed everywhere on the hyphae except the tip.

The slow growth of the ecm33Δ/Δ yeast cells makes it difficult to interpret the reduced virulence of these cells in the murine systemic infection model (231). In vitro models where growth is not expected to be as critical were examined. When added to HUVECs and FaDu cells, the null strain rapidly germinated and produced hyphae of normal length. The conditions appear to affect the germination of the null strain, since in liquid and solid media, as discussed above, filamentation was delayed. The number of organisms adhered to the two cell surfaces decreased as a wild-type allele was deleted such that deletion strains had the least adherence and heterozygous strains had intermediate adherence. The same gene dosage effect was noticed when endocytosis and HUVEC and FaDu cell damage were examined. These observations support a contribution of Ecm33p to adherence and host cell interactions. However, the aberrant architecture of the cell wall raises the questions of whether the effect is a direct one.

The binding of fibrinogen was particularly well characterized when studies began on host proteins that could serve as ligands for C. albicans surface receptors (see the review in reference 49). Briefly, as a fibrinogen binding protein, Pra1p was found on both yeast cell and hyphal surfaces (44) and in vivo and in vitro (218). The protein was ubiquitinated (345). The binding of fibrinogen and antibody to Pra1p was heterogeneous, with a patchy distribution on the surface (229b). This suggests that at least Pra1p and perhaps others are asymmetrically exposed at the cell surface. Whether the pattern of surface exposure reflects the distribution of any nonexposed protein is unknown. The gene encoding Pra1p was identified both through its fibrinogen binding property (alias FBP1 [4]) and by its being a pH-regulated antigen (344). PRA1 expression was regulated in response not only to pH but also to medium (Northern analysis [4, 344]). Expression was found in yeast cells and germ tubes in Lee medium and tissue culture medium 199 at 25°C and 37°C. However, expression was not detected at either temperature when organisms were grown in YEPD medium. Also, transcript was not abundant immediately following inoculation into fresh medium, despite cell growth (Northern analysis [344]). Transcriptional profiling subsequently showed that the pH regulation is mediated through Rim101p (transcription profiling, RT-PCR [311]). Recently, Pra1p was found among proteins biotinylated at the hyphal surface but not at the yeast cell surface (391). However, yeast cells were grown in YEPD, where previous studies would predict that PRA1 would not be expressed. Pra1p contains a signal sequence for classical secretion.

The strain with both alleles deleted is viable (344). The general growth characteristics of the strain were normal. The pra1Δ/Δ strain formed germ tubes at 37°C but not at 42°C. At the higher temperature, the chitin distribution was altered. These latter two characteristics suggest a contribution to the cell wall but are clearly milder than the effects of other genes when deleted. PRA1 has not been reported among the genes regulated by morphogenesis in transcription profiling studies, e.g., general morphogenesis (264) and regulation by Nrg1p (259). However, in addition to the regulation of expression by pH and medium noted above, PRA1 expression rapidly decreases when cells are transferred into blood (transcription profiling, RT-PCR [111]). Response to estrogen compounds supports additional regulation (transcriptional profiling [56]). The abundance of PRA1 transcript decreased in response to exposure to 17-β-estradiol by over twofold in a strain lacking CDR1 and CDR2, encoding drug efflux pumps. Only a small number of genes responded to estrogen compound exposure in the three strains tested. The changes were confirmed by RT-PCR.

In a proteomic study, three of six spots that were more abundant in a silver-stained gel of hyphal whole-cell lysates were identified as Pra1p (58). Two of the spots were identified as Phr1p. A similar comparison between the wild type and the pra1Δ/Δ strain detected the absence of Pra1p only in the mutant strain. Surprisingly, when similar profiling was performed on the phr1Δ/Δ strain, not only was Phr1p missing but Pra1p was decreased in abundance relative to what was seen for the wild type. Northern analysis of PRA1 showed no difference in expression in yeast and hyphal cells and the absence of the transcript in the pra1Δ/Δ strain but not the phr1Δ/Δ strain. These observations support a possible posttranscriptional regulation in hyphae and phr1Δ/Δ cells. The extent to which extracts and separations are replicated may affect the detection of differential protein abundance, as may the presence of multiple species, not all of which may be identified and therefore considered. The more recent studies of Pra1p have focused not on adherence properties but on antigenic properties and potential as a vaccine candidate rather than as an adhesin (see references 397 and 398, for example).

Recently, Pra1p has been identified as a ligand for a host integrin receptor, α_Mβ₂ (CD11b/CD18, Mac-1) (357). α_Mβ₂ is the major receptor on PMNs. This receptor is the PMN partner for binding with C. albicans germ tubes, and it mediates the migration of leukocytes to infection sites and the subsequent PMN interaction with the fungal cell (106-109). Supernatants of 2- to 3-day cultures were the starting material to identify the C. albicans ligand. Supernatant immobilized in wells of a tissue culture plate supported the adherence of THP-1 cells and their migration through filters in a transwell migration assay. The supernatant activity increased over several days of fungal growth. The addition of antibodies to either the integrin subunit or a high-affinity ligand of the integrin blocked activity. The activity of the supernatant was sensitive to proteases but not to other hydrolytic enzymes. Hybridomas prepared from mice immunized with culture supernatant were tested for the ability of secreted MAb to block the activity of the supernatant. The antigen recognized by the most potent MAb was obtained by affinity chromatography with immobilized MAb. The isolated material was a large complex that dissociated after heating at or above 65°C to give several bands detected by staining a SDS-polyacrylamide gel electrophoresis gel. Western blotting with the MAb identified a single band of 52 kDa in the heated sample and with the large complex (∼250 kDa) in the unheated sample. The protein in the reactive bands was digested with trypsin and the peptides were identified by mass spectrometry. The smaller, 52-kDa band yielded peptides only from the Pra1p sequence. The 250-kDa band yielded peptides from the Pra1p sequence as well as Mp65p and Hyr1p. A second method to identify the fungal ligand utilized affinity chromatography with recombinant purified α_Mβ₂. Conventional N-terminal sequencing as well as mass spectrometry of the bound protein again identified Pra1p. In the affinity purifications, the eluted material supported the adherence of THP-1, while the column pass-through material was depleted compared to the starting material. The supernatant obtained from the PRA1 null strain failed to support the adherence or migration of THP-1 cells. The role of Pra1p in PMN killing and phagocytosis was further investigated by testing the ability of human PMNs to kill and phagocytose cells of either the parental strain or the mutant strain. After 5 h of coincubation, 52% of the wild-type C. albicans cells were viable, compared to 90% viability for the mutant. The addition of antibodies to the α_Mβ₂ receptor blocked the killing of wild-type cells. Purified Pra1p also protected wild-type cells from killing. The release of Pra1p from fungal cells provides an evasion from PMN action by blocking adherence to the hyphae, which is required for killing.

In addition to the above-described study, an earlier study also found Pra1p to be part of a larger complex, but the other components were not identified (217). The identities of the complex components are interesting with respect to the nature of their association when in the cell wall. As noted above, Pra1p has a signal sequence for secretion and Hyr1p is a GPI-CWP (Table 2). The other component, Mp65p, may be held in the cell wall by as-yet-uncharacterized alkali-labile linkage, as described earlier for this protein. Along with the fibrinogen binding activity, Pra1p contributes in several ways to the interaction between fungus and host.

Cell surface hydrophobicity has been implicated as a contributor to adherence to host ECM and cells and resistance to macrophage killing (reviewed in references 49 and 135). This could be a factor in the high virulence of hydrophobic cells compared to that of hydrophilic cells in mice. MAbs to cell surface hydrophobic protein inhibited the adherence of cells to fibronectin and laminin (237). Pretreatment of hydrophobic cells with MAb also reduced adherence to endothelial cell monolayers (125). These studies implicated the protein recognized by MAb and perhaps the hydrophobic properties in the adherence of ligands and cells. The acid-released fraction of mannan from hydrophobic cells was longer and more abundant than that from hydrophilic cells (236), and hydrophobicity was independent of serotype A or B (235).

A scheme to identify the protein and encoding gene employed antibody that recognized the hydrophobic protein and peptide sequences of the isolated protein (352). The identification of CSH1, which encodes a predicted protein of 38 kDa, was the result. The predicted protein sequence lacked an N-terminal secretion signal suggesting cytoplasmic localization and was a protein of the aldo-keto reductase family. Several other sequences in the genome had similarity to CSH1. The deletion of CSH1 decreased cell surface hydrophobicity about 75% and decreased adherence to fibronectin. By Western blotting, a 38-kDa protein was missing from the whole-cell lysate of the mutant strain separated in one dimension. The protein is apparently not glycosylated but must influence the acid-labile mannan composition, as there are differences between hydrophobic and hydrophilic cells, as noted above. Western blot analysis of a two-dimensional separation showed that the major activity observed in the parental strain was lost in the mutant strain (351). There was a minor cross-reactivity unaffected by the deletion of CSH1. In a strain in which one copy of CSH1 was reintegrated, the expression of Csh1p increased with increasing temperature between 23°C and 37°C. Interestingly, by Northern analysis CSH1 expression was greater at the lower temperature. Thus, there appeared to be an inverse relationship between transcript and protein abundance. Maximum CSH1 expression occurred at 11 h in actively growing cells at 37°C but later (26 h) in a culture growing at 23°C. Western blotting of extracts of proteins biotinylated on the intact cell, proteins released by limited glucan digestion of intact cells, and cytoplasmic lysate showed that most of the 38-kDa protein was in the cytoplasm. However, it was present in the cell wall extracts and the amount of protein in the cell surface extracts was increased in extracts of hydrophobic cells compared to what was seen for hydrophilic cells.

During frozen storage, the null strain reacquired cell surface hydrophobicity levels similar to those of the parent strain (350). However, the 38-kDa protein was absent from the cell extract, and thus the increase in hydrophobicity is associated with an unknown mechanism. The strain also showed an increase in binding to fibronectin that supports multiple adhesins for this ligand. The strain lacking CSH1 and strain in which one copy of CSH1 was reintegrated showed similar hydrophobicities. In a murine model of disseminated infection, both strains showed similar levels of virulence, although the deleted strain had a small increase in mean survival. In a murine vaginitis model, the strains were similar.

A study undertaken to study the variation in biofilm formation with multilocus genotypes isolated from different sources found little correlation (207). However, there was a positive correlation between the cell surface hydrophobicity of the inoculum and biofilm formation. CSH1 expression decreased in biofilm treated with farnesol (transcription profiling, RT-PCR [42]). Cell surface hydrophobicity decreased with increasing farnesol concentration added for biofilm formation. These two studies suggest that cell surface hydrophobicity contributes to biofilm formation and further links CSH1 expression with that phenotype. Under oxidative stress conditions, expression increased compared to what was seen for unstressed organisms (transcription profiling [93]). Three studies find increased expression in response to inhibitors or drug resistance. Expression was greater in ketoconazole-treated cells than in untreated cells (transcription profiling [210]). In benomyl-treated cells in which MDR1 (encoding a multidrug efflux pump) expression increases, CSH1 expression increased. Expression was similarly increased more than 100-fold in a strain with upregulated MDR1 (transcript profiling [171]). In a series of clinical isolates with stepwise acquisition of fluconazole resistance, CSH1 expression was increased in expression coordinately with MDR1 (transcription profiling, RT-PCR [318]). These changes may be related to the putative cytoplasmic function as an aldo-keto reductase. Whether these challenges also result in changes to surface hydrophobicity is untested.

In addition to their role as proteinases discussed previously, an adherence role has been proposed for Saps (see the review in reference 262). Adherence to three substrates was examined (403). Poly-l-lysine and ECM obtained from Engelbreth-Holm-Swarm cells were immobilized on glass slides. Adherence to BECs was also examined. The adherence of strains defective in one or more SAP genes to these substrates was tested. The observations supported a generally modest contribution to adherence. Galactose-grown cells adhered more than glucose-grown cells. Preincubation of cells with pepstatin A inhibited adherence to poly-l-lysine more than that to ECM. The differences with what was seen for mutant strains were generally small: strains lacking SAP1, SAP3, or SAP4 to SAP6 grown on glucose but not galactose adhered less to poly-l-lysine than did parental cells. The loss of SAP2 had no effect for growth on either sugar. On ECM, only the sap3Δ/Δ strain showed a small reduction in adherence. Adherence to BECs was less for the glucose-grown strains lacking SAP3 or SAP4 to SAP6 and for galactose-grown cells for all mutants.

A 65-kDa mannoprotein present in the cell wall and in culture supernatant has been studied extensively as an important target for host defense. It is a putative β-glucanase that was described in an earlier section (“Cell Wall-Localized Proteins”). The sequence contains an RGD motif that has been implicated in a variety of adhesins (197a). Recently, the adherence properties of this protein have been examined (325). Although yeast cells of a null and wild-type strain did not differ in generation time or size, cells of the null mutant did not form germ tubes. Cells of the mutant strain adhered less well to polystyrene than did the wild-type cells. Cells of heterozygous strains (including a null strain with one allele replaced) had an intermediate adherence level. Antibody to recombinant Mp65p inhibited the binding of wild-type yeast cells to plastic. This observation suggested that the loss of Mp65p and not failure to make germ tubes contributed to reduced plastic adherence. Human variable heavy- and light-chain domains lacking the Fc region that recognized Mp65p were generated from a phage library (71). This antibody construct inhibited the binding of C. albicans wild-type cells to epithelial cells of rat vagina. These recent studies support that Mp65p not only is a major target for cell-mediated immune response but also has adhesive properties. Whether the protein has the putative β-glucanase enzymatic function is unknown.

Histatins are small histidine-rich salivary peptides with potent antifungal activity. Over the last decade, a clear picture has been emerging of how these peptides interact with and kill C. albicans. Histatin 5 binding to C. albicans is saturable (91). A binding protein of about 70 kDa was identified from surface extracts of C. albicans as well as sensitive S. cerevisiae. The binding protein was isolated and identified as Hsp70 (208). Hsp70 proteins have been demonstrated on the cell surface (Table 4) (reviewed in reference 49). However, unlike the adhesins discussed above with the exception of Csh1p, Hsp70 proteins lack a signal for classical secretion. The S. cerevisiae parental strain and the ssa1 ssa2 mutant (missing two genes of stress-70 subfamily A) strains provided further confirmation, as the mutant strain was much less sensitive to histatin 5. Recently, both pulldown assays and yeast two-hybrid assays showed greater interaction between histatin 5 and C. albicans Ssa2p than between histatin 5 and Hsp70p (alias Ssa1p) (209). C. albicans mutants with deletion of either SSA2 or HSP70 showed normal growth and filamentation. However, the SSA2 null strain was highly resistant to killing, while the HSP70 null strain was not, suggesting that Ssa2p has the major receptor role. P-113 (amino acids 4 to 16), a peptide fragment that was as active as histatin 5, was inhibited when the salt concentration was increased (319). P-113 lost activity when two lysine residues were replaced by glutamine. This observation suggested that binding involved ionic interaction. The glutamine-substituted peptide binds to the surfaces of both wild-type and SSA2 null cells at levels higher than those seen for the unsubstituted peptide but is not translocated into the cytoplasm (157). A peptide array of overlapping peptides of Ssa2p showed that P-113 had strong binding to two peptides, while the glutamine-substituted peptide did not bind to the array peptides. Wild-type and mutant cells failed to import the substituted peptide into the cytoplasm, while the wild-type cells imported the unsubstituted peptide more rapidly and more extensively than mutant cells. These observations support an ionic interaction between surface protein and histatin 5.

Histatin 5 interaction with C. albicans leads to ATP release in a nonlytic manner as a prelude to fungal cell death (184). The ATP itself has a cytotoxic role because the removal of the released ATP can inhibit the killing. When histatin 5 was directly introduced into the fungal cell, bypassing the entry step, it was still capable of inducing the release of ATP and killing C. albicans (11). This observation suggested that the signal for killing did not come externally. The loss in cell viability was concomitant with a decrease in cell volume and in the number of G₁-arrested cells (12). As the cell volume decreased, the proportion of unbudded cells increased. RT-PCR analysis of histatin 5-treated cells showed a drop in G₁ cyclin, CLN1, and CLN3 expression levels. Anion channels and ATP release have been implicated in cell volume regulation, and DIDS (4,4′-dithiocyanatostilbene-2,2′-disulfonic acid), a chloride anion-channel-blocking drug, inhibited the histatin 5-associated loss in viability. Potassium and magnesium ions are lost along with ATP, and one route for nonlytic loss may be through potassium channel Tok1p. The deletion of TOK1 reduced but did not abolish ATP efflux (13). Histatin 5 treatment of tok1Δ/Δ cells still produced substantial killing. These observations suggested that Tok1p contributes to but is not the main site for action. TRK1 encodes the potassium transporter. Attempts to delete the second allele were unsuccessful, suggesting that TRK1 is an essential gene (14). However, the deletion of one allele resulted in reduced histatin 5 killing, histatin 5 induced ATP release, and the influx of K⁺ (measured as ⁸⁶R⁺) was reduced five- to sevenfold. These observations suggest that Trk1p is the mediator of ATP loss.

Very recently, the Edgerton laboratory reported that transcriptional profiling of histatin 5-treated C. albicans showed that the major response was in genes involved in adaptation to osmotic stress (400). Activation of the HOG pathway was shown. Prestressed (sorbitol-treated) cells were hypersensitive to histatin 5, while concurrent stress did not change sensitivity. Oxidative prestress did not alter histatin 5 sensitivity. This suggests a common response to osmotic stress and to histatin 5.

Other small peptides of the oral cavity may use the same pathway as histatin 5 in their antimicrobial action. Human neutrophil defensin 1 is active at a concentration similar to the active concentration of histatin 5 and also induces nonlytic efflux of ATP (90). The binding protein for histatin 5 was identified by a blot overlay method. C. albicans proteins were electrophoretically separated and transferred to a membrane for blotting with labeled histatin 5. Preincubation of the blot with neutrophil defensin 1 or unlabeled histatin 5 abrogated the binding of labeled histatin 5. This observation suggested that the two peptides shared a binding protein on the fungal surface. When neutrophil defensin 1 was incubated with wild-type, hsp70Δ/Δ, and ssa2Δ/Δ strains, the mutant strains remained susceptible to killing (401). This observation is at odds with the first observation of competition for the binding site. The binding protein for neutrophil defensin 1 remains unresolved. When human β-defensin 2 was incubated with the three strains, at low peptide concentrations (2 to 10 μM) there was reduced susceptibility of both mutant strains, but at 12 μM the killing was about 90% of that observed with the wild-type strain. When β-defensin-3 was incubated with the strains, there was again reduced susceptibility with both mutant strains, which was particularly marked in the case of the hsp70Δ/Δ strain. When these three peptides were tested with the parental and TRK1/trk1Δ strains, there was no difference. These results suggest that β-defensin-2 and β-defensin-3 may share binding proteins with histatin 5 but that the subsequent mechanism of killing follows a different path(s).

C. albicans adheres to type I, denatured type I (gelatin), and type IV collagen to different extents (reviewed in reference 49). Competition with fibronectin, various peptides, heparin, and dextran sulfate reduced cell adherence. However, dextran sulfate may bind to collagen rather than to the fungal cell. Collagen affinity chromatography yielded two proteins as potential adhesins. Collagen IV monomers have three domains; an N-terminal 7S domain (7S), a central collagenous triple alpha helix domain (CC), and a noncollagenous C-terminal domain (NC1). Binding to the three domains of collagen IV was examined further (6). Binding to immobilized collagen IV was greater than that to laminin or fibronectin, and the presence of divalent cations enhanced binding. Binding to the 7S domain, NC1, or CC was cation dependent, partially dependent, or independent, respectively. Levels of adherence to the 7S and CC domains were similar and somewhat greater than that to NC1. Enzymatic removal of N-linked oligosaccharides from the 7S domain reduced adherence, and this along with cation dependence suggested a lectin-like C. albicans adhesin. Addition of glucose, galactose, lactose, or heparin sulfate resulted in no reduction of binding. However, N-acetyllactosamine followed by N-acetylglucosamine, methylmannoside, and l-fucose, sugars known to be in the N-linked oligosaccharide, inhibited binding up to 50%. Trypsin treatment of yeast cells reduced binding. The putative protein(s) mediating adherence was not identified.

A second study examined adherence to collagen type I and gelatin immobilized on nitrocellulose filters and placed in polystyrene wells for assay (226). C. albicans adhered to both ligands. Preincubation with either of two peptides containing RGD sequences but not the peptide in which E was substituted for D inhibited binding to gelatin. There was no effect on binding to collagen I. Thus, it appeared that C. albicans recognized RGD in the denatured but not in the native collagen. Further, either one or more adhesins recognize more than one ligand sequence, or else there are multiple collagen adhesins.

Adherence to vitronectin was reviewed previously (49). Briefly, both carbohydrate β-glucan and protein were potential adhesins for this constituent of vascular wall and serum. Far-Western (ligand affinity) blotting identified a 30-kDa protein. Since integrins (αβ heterodimers) recognize host ECM ligands, frequently involving an RGD binding motif, there was considerable speculation that C. albicans adhesins would share some integrin characteristics (see the review of integrins in reference 222). Since the last review (49), that notion guided subsequent studies of adherence to vitronectin. MAbs and polyclonal antibodies recognizing human α_v, β₃, β₅, α_vβ₃, or α_vβ₅ were used (361). MAbs to both subunits and complexes bound to C. albicans, as determined by flow cytometry. Immunoprecipitation of yeast cell lysate with antisera to either α or β₃ subunits yielded the same 130-, 110-, and 100-kDa proteins. This is the same size range as for human integrin subunits. The antibody to β₃ precipitated a dominant 84-kDa protein along with 130- and 100-kDa proteins. Antibodies, particularly the combination of antibodies to α_vβ₃ or α_vβ₅, blocked fungal adherence to immobilized vitronectin, as did RGD-containing peptides. Antibodies also blocked adherence to an endothelial cell line expressing surface vitronectin. Similar experiments were performed on germ tubes (329). MAbs to the α subunit or complexes with either β subunit stained organisms. Organisms reactive with antibody to α_vβ₃ increased, while organisms reactive with antibody to α_vβ₅ decreased. About 65% of the germ tubes adhered to vitronectin, with the extent of adherence affected by the presence of divalent cations. Antibody to α_vβ₃ and its subunits but not antibody to α_vβ₅ inhibited binding. Antibodies, heparin, or RGD-containing peptide reduced adherence to an endothelial cell line, and their combination abolished binding. The adherence of germ tubes differed from what was seen for yeast cells by the apparent absence of a contribution to binding of moiety(ies) reactive with antibody to α_vβ₅. Such observations continue to support the notion of shared function or structure between mammalian cells and fungus. However, the binding proteins have not been identified to permit more than a superficial understanding of the relationship.

When mammalian integrins bind ligands, outside-in signaling can be initiated by receptor clustering and with involvement of kinases, prominently including FAK (focal adhesion kinase) (reviewed in references 222 and 348). Tyrosine kinase inhibitors, e.g., genistein, inhibited yeast cell binding to vitronectin and an endothelial cell line (328). Immunoprecipitation and Western blotting with antibody to mammalian FAK revealed a 105-kDa cytoplasmic protein band that is similar in size to mammalian FAK. Confocal microscopy of permeabilized yeast cells adhered to vitronectin showed that protein reacting with anti-FAK antibody was in the membrane and colocalized with protein reactive with anti-α_v, -β₃, or -β₅ antiserum. The protein immunoprecipitated with anti-FAK antibody from cells bound to vitronectin or endothelial cells, compared to unbound cells, showed a high level of reactivity with anti-phosphotyrosine, suggesting phosphorylation and putative activation upon binding to vitronectin. Binding to fibrinogen, BSA, or poly-l-lysine did not result in a similar change. None of the reactive proteins was identified. In this area, the studies have not progressed to the identification of putative proteins participating in the binding.

In another study, antibody to the human fibronectin integrin receptor, α₅β₁, was used to screen yeast cell and germ tube cDNA libraries (182). The reactive clones had greater activity with the anti-vitronectin receptor (α_vβ₃) antibody but did not react with antibodies to the human iC3b receptor (α_Mβ₂) and did not hybridize in Southern analysis with the fibronectin receptor probe. In vitro protein expression from some clones yielded a 37-kDa protein that could be precipitated with appropriate antisera. The nucleotide sequence was that of ADH1 (Orf19.3997). Western blot analysis of a detergent-based cell wall extract with anti-fibronectin receptor and anti-vitronectin receptor antibody showed a reactive 37-kDa protein similar to S. cerevisiae Adh1p, which was used as a control. However, there was reactivity with multiple bands, with prominent bands being larger or smaller depending on the antibody. In particular, anti-fibronectin receptor antibody detected larger species near the top of the gel. A 37-kDa reactive band was also detected in culture supernatant. There are differences in the sources of extracts and proteins for Western blotting between this and a previous study of the vitronectin receptor (361). These and previous studies reviewed elsewhere (49) consistently show antigenic relatedness between C. albicans proteins and human integrin proteins.

Fibronectin is a large dimeric glycoprotein that circulates in plasma as well as being part of the host ECM. Fibronectin as a ligand for C. albicans adhesins was the focus of many studies in the 1990s (reviewed in reference 49), in contrast to what has been the case more recently. Multiple protein species were identified by various means of monitoring interactions with fibronectin. Some studies reported that RGD peptides inhibited fibronectin binding to C. albicans, while other studies did not. As discussed for vitronectin, the notion that C. albicans adhesins for ECM may be related to human integrins underlay a strategy to clone a gene encoding a protein that reacted with anti-fibronectin integrin receptor antibody (182). As noted above, the cloned sequence was homologous with ADH1. The encoded protein as well as authentic alcohol dehydrogenase reacted with antibodies to human fibronectin and vitronectin receptors.

Among the receptors for fibronectin previously reviewed was a low-affinity receptor induced by the addition of hemoglobin to a defined medium (49). The induction of this activity greatly increased the binding of soluble fibronectin. The binding was saturable, with a dissociation constant (K_d) of 2.7 × 10⁻⁸ M. The yeast cells grown in hemoglobin also had increased binding to immobilized fibronectin and bovine corneal endothelial cells. The researchers went on to demonstrate that the receptor induced by growth in the presence of hemoglobin was promiscuous (407). It was also a receptor for soluble laminin, fibrinogen, and collagen type IV but not collagen type I. A 55-kDa protein present in DTT and lyticase extracts of the cell surface was identified as the binding protein. The binding to fibronectin was through the fibronectin cell binding domain (408). A C-terminal portion of the cell binding domain containing an RGD sequence was an effective inhibitor with or without the RGD sequence. The binding of the cells grown in hemoglobin was enhanced to immobilized fibronectin and fragments containing the cell binding domain. The 55-kDa protein has not been identified.

As reviewed previously (49), Negre et al. (265) reported a high-affinity fibronectin receptor (K_d = 1.3 × 10⁻⁹ M) detected on the surfaces of organisms grown in complex media as well as a low-affinity receptor. The ligand for the high-affinity receptor was in the collagen binding domain of fibronectin. Although other peptides inhibited the binding of soluble fibronectin, they were much less effective. Maximal binding to immobilized fragments was observed with the gelatin/collagen binding domain. Pendrak et al. (293) examined the effect of caspofungin on the induction of the high-affinity fibronectin receptor. Yeast cells were grown in defined medium with or without subinhibitory levels of caspofungin. Cells grown in caspofungin alone showed high binding of soluble fibronectin compared to control cells, although binding was less than that of cells grown with hemoglobin. The maximum caspofungin effect was at doses at which there was also development of misshapen cells. A lower dose of 10 ng/ml, at which there were no obvious morphological or growth defects, was used for subsequent experiments. Nikkomycin, a chitin biosynthesis inhibitor, also induced enhanced fibronectin binding but to a lesser extent than caspofungin. Fluconazole had no effect. The binding constant for soluble fibronectin to caspofungin-treated cells was about 1 nM, the same as previously reported for the high-affinity receptor. The binding constant was similar for the nikkomycin-treated cells, but the receptor surface density was less. Competition with the soluble fibronectin for binding occurred with the collagen binding domain fragment but not with the cell binding domain fragment. Cells from which proteins had been sheared in a blender had reduced fibronectin binding capacity. Proteins sheared from the cell were bound to polystyrene wells, and the binding of fibronectin to these immobilized proteins was greater for the fraction recovered from caspofungin-treated cells than for what was seen for untreated controls. The binding activity was concentrated in a fraction of a molecular weight greater than 500,000. Treatment with trypsin and DTT and to a lesser extent with β-1,3-glucanase reduced binding to the complex. In a KRE9 null strain lacking β-1,6-glucan, there was no effect of caspofungin on the induction of fibronectin binding. Also, in a heterozygous strain in which one allele of the essential HBR1 (hemoglobin response) gene was deleted, the induction of fibronectin binding by caspofungin was absent. However, the heterozygous deletion had no effect on the induction of the low-affinity receptor by hemoglobin. This study and that of Soustre et al. (358), also discussed elsewhere, report opposite effects of caspofungin treatment on binding to fibronectin. In both studies, organisms were grown in yeast nitrogen base with glucose (2% [111 mM] in one case and 50 mM in the other). There is a large difference in the caspofungin MICs of 625 ng/ml (293) and 8 ng/ml (358) but less difference in the amounts of caspofungin used, as the former used 10 ng/ml and the latter used MIC/2. One study used soluble fibronectin (293) and the other immobilized fibronectin (358); however, with the high-affinity receptor, maximal activity was found with the gelatin/collagen binding domain fragment in both soluble and immobilized assays.

Adherence to laminin received considerable attention in the previous decade (see the review in reference 49) but has received less attention more recently. Laminin is a large (900-kDa) protein composed of A, B1, and B2 chains. Previously, GAPDH encoded by TDH3 was identified as an adhesin for laminin, and the surface enzymatic activity was observed. This was confirmed by use of S. cerevisiae, where strains lacking either Tdh1p, Tdh2p, or Tdh3p or double mutants lacking Tdh1p and Tdh2p or Tdh2p and Tdh3p had cell wall-associated enzymatic activity (78). The recombinant Candida Tdh3p protein was purified and was enzymatically active (396). The purified protein also bound to laminin and fibronectin. The expression of the C. albicans Tdh3p on the surface was examined further using a Tdh3p fusion with the internal form of S. cerevisiae invertase (76). When the plasmid containing this construct was transformed into a suc⁻ strain of S. cerevisiae, the transformant was able to grow on sucrose. Further constructs localized the region of Tdh3p required to support the growth of transformants on sucrose to the N-terminal half of the protein. The cell wall-associated ScGAPDH activity increased in response to starvation and temperature increase (77). When a transformant with a C. albicans Tdh3-invertase fusion was examined, there was a similar response in terms of an increase in both GAPDH and invertase activity. This increase was not dependent upon either protein synthesis or the ubiquitin stress response. These observations provide additional support for Tdh3p as an adhesin and for its presence on the cell surface.

Tenascin is a hexameric protein with subunits generally of around 200 kDa (see the review in reference 147). The protein has repeated structural motifs, typically including 14.5 epidermal growth factor-like repeats, fibronectin type III repeats, and a globular fibrinogen-like domain. Tenascins have tightly regulated expression during the development and lifetime of vertebrates. Tenascins modulate cell-matrix interactions. With a sharing of the various structural motifs present in other host ligands to which C. albicans binds, tenascin-C would be predicted to bind to the fungus. Soluble tenascin-C bound to germ tubes but not to yeast cells, as determined by indirect immunofluorescence (214). Cell surface extracts from both yeast cells and germ tubes bound to immobilized tenascin-C. Antibody to either C. albicans or tenascin but not to laminin inhibited binding. This observation supported a specific interaction. Inhibition of the binding of fungal proteins to tenascin-C by fibronectin but not fibrinogen implicated the fibronectin type III repeats in binding. The binding of fungal protein was RGD peptide and divalent cation independent. The failure of soluble tenascin-C to bind to yeast cells coupled with the binding of a component(s) of a yeast cell surface extract to immobilized tenascin-C suggests that either the adhesin or the adhesive motif is blocked on the yeast cell surface but exposed at the germ tube surface.

Several studies in recent years examined the effect of pretreatment with subinhibitory drug concentrations on adherence to various host ligands. Imbert and colleagues (47, 153, 358) examined adherence to uncoated polystyrene, fibrinogen, fibronectin, laminin, gelatin, collagen type IV, and ECM (a mixture of laminin, collagen type IV, entactin, and heparan sulfate proteoglycan) for yeast cells treated with one-fourth the MIC. Treatment with flucytosine had no effect. Amphotericin B, itraconazole, fluconazole, and terbinafine decreased adherence to polystyrene, fibrinogen, and fibronectin. All except amphotericin B also decreased adherence to laminin, gelatin, collagen type IV, and ECM (153). Caspofungin at one-half the MIC decreased adherence to ECM for all six fluconazole-sensitive strains tested (358). However, when strains were fluconazole resistant in vitro, three of five strains showed reduced adherence when treated with the low concentration of caspofungin. Another echinocandin, aminocandin, was studied at one-half the MIC with fibronectin as well as the ECM complex (47). Five of six fluconazole-sensitive strains showed reduced adherence to ECM, and three were less adherent to fibronectin. Five fluconazole-resistant strains all showed reduced adherence to ECM and four strains reduced adherence to fibronectin. The outcome differed when the same fluconazole-resistant strains were tested with caspofungin, as two strains were unaffected. Although the number of strains is small, this suggests a difference between two drugs of the same class on fluconazole-resistant strains. As the mechanism of fluconazole resistance for these strains is unknown, there is a possibility that the fluconazole resistance mechanism may underlie differences in the echinocandin drug effect. The effect of caspofungin treatment of reducing the adherence to fibronectin differed from what was found in the previously discussed study of Pendrak et al. (293), who observed increased binding to fibronectin.

Klotz and colleagues have examined the adherence of C. albicans and S. cerevisiae expressing Als1p and Als5p to beads coated with 7-mer peptides from a peptide library (180). The sequence was determined for five beads that bound C. albicans as well as the Als1p- and Als5p-expressing S. cerevisiae. All peptides shared common characteristics in which the first residue had a high turn property, e.g., A, D, or P; the second was a bulky hydrophobic or aromatic amino acid, e.g., F, T, or W; and the third was an R or a K. This motif was termed “τφ⁺.” Scrambling sequences had little effect on C. albicans cells but reduced binding of the S. cerevisiae strains. Perhaps this difference was due to the redundancy of adhesins on the C. albicans surface. When BSA was denatured, exposing several of these motifs, and mixed with C. albicans yeast cells, the cells aggregated, while they did not with native BSA. The addition of a 23-mer peptide derived from the cell binding domain of fibronectin inhibited binding to a bead coated with a 7-mer peptide (181). Smaller peptides derived from the sequence were ineffective. This peptide also killed C. albicans in a manner independent of the adhesive interaction. The fungicidal activity was inhibited by physiological salt concentrations, as are host microbicidal peptides.

Beads coated with different ECM proteins, fibronectin, type IV collagen, laminin, BSA, or casein supported the binding of C. albicans yeast cells, germ tubes, and pseudohyphae (121). Homopolymers composed of 10 serine, threonine, or alanine residues attached to beads supported the binding of the various C. albicans forms, while the single amino acids did not. After a few hours, the cells separated from the polyalanine-coated beads but not from other two homopolymers. Urea, formamide, and a pH of 12 inhibited binding to fibronectin- or polythreonine-coated beads. The addition of glucose, galactose, mannose, fucose, 1 M NaCl, EDTA, or Tween 20 had no effect. This suggested that the adhesive interactions with the two ligands were chemically similar. The addition of accessible threonine residues to the C terminus of a nonadhesive peptide stimulated binding.

In addition to C3d and iC3b, three other complement pathway proteins, C4BP (from the classical and lectin pathways) and factors H and FHL-1 (regulatory components from the alternative pathway), bind to the cell surface (242, 243). The binding of these regulatory factors is a possible mechanism of complement escape by microbes (185, 242, 243, 304). These factors affect the decay of the C3 convertase and factor I-mediated degradation of C3b, thus diminishing complement activation. Factors H and FHL-1 bind to the yeast surface in a patchy pattern (243). A heterogeneous distribution of binding is reported for other ligands such as fibrinogen and laminin (49). Binding of the two factors from EDTA-treated normal human serum was dose dependent and specific, as purified recombinant proteins competed with radiolabeled protein (243). Factors H and FHL-1 contain multiple short complement regulator domains (SCRs), and binding of various recombinant proteins with deleted domains identified a common C. albicans binding region within SCRs 6 and 7. A second binding site unique for factor H mapped to SCRs 19 and 20. Factor H has heparin interaction sites localized to SCRs 7, 13, and 20. Heparin inhibited the binding of factor H. The attached factors bound to the yeast cells retained their activity and mediated the factor I-dependent cleavage of C3b.

C4b, a part of the C3 convertase, forms during the activation of the classical and lectin complement pathways. The multimeric C4BP can bind to C4b and accelerate the decay of the enzyme and the factor I-dependent degradation of C4b. C4BP bound to yeast cells and the tips of hyphae in a patchy pattern, as did factors H and FHL-1 (242). The binding was dose dependent and observed with recombinant protein as well as EDTA-treated normal human serum. The observation that unlabeled C4BP, factor FHL-1 at a high concentration, and to a lesser extent factor H inhibited binding of radiolabeled C4BP suggested a common binding site for C4BP and factor FHL-1. C4BP has seven α chains and a single β chain. Binding was mediated by α chains, and the binding was due primarily to ionic interaction. The chains are composed of eight complement control protein (CCP) domains. Binding of recombinant proteins lacking one of the domains to hyphae showed that binding was localized to CCP1 and CCP2 and possibly to CCP6. Like factors H and FHL-1 of the alternative pathway, C4BP retained cofactor activity. Bound C4BP mediated the factor I-dependent cleavage of C4b. Compared to yeast cells treated with buffer alone, yeast cells with bound C4BP were adherent to HUVECs. Thus, the bound C4BP may contribute to host interaction by the inhibition of complement activation and enhanced adherence.

To identify the fungal adhesin for these ligands, the binding of factors H and FHL-1 to a S. cerevisiae protein array was determined (304). Three proteins bound both factors and one protein bound only factor H. Phosphoglycerate mutase bound both factors and was selected for further characterization with the C. albicans ortholog, Gpm1p. Recombinant Gpm1p (rGmp1p) bound both factors H and FHL-1 but not C4BP. The binding of recombinant factors H and FHL-1, described above, to rGpm1p was determined, and this confirmed that SCRs 6 and 7 are a common region for the two factors and a second and lower-intensity binding region for factor H in SCRs 19 and 20. This confirmed the observations of binding to yeast cells. The factors bound to rGmp1p also had factor I-dependent cleavage of C3b, as they did when bound to intact cells. The presence of Gpm1p on the surfaces of intact yeast cells and hyphae was confirmed by antibody prepared to rGmp1p. Antibody reactivity showed the same pattern of surface distribution as noted for bound factors, with a heterogeneous distribution on yeast cells and at the hyphal tip. A deletion strain of GPM1 lacked protein reactive with the antibody in both cytoplasmic and cell wall fractions. Compared to cells of the wild-type strain, cells of the deletion strain bound somewhat less factor H or FHL-1. As suggested by the array studies, the presence of multiple adhesins on the surface is the likely explanation for the partial reduction.

Initial studies of hemoglobin on the induction of surface changes affecting binding to fibronectin were reviewed previously (49), and a more recent review focused on hemoglobin sensing and fungal response in greater detail than presented here (295). There are two reports describing receptors for hemoglobin (291, 404). Both reports provide evidence for a low-affinity receptor. However, there are some differences between conditions and parameters tested that complicate determining whether the same receptor is described. Pendrak et al. (291) examined receptor characteristics for the induction of fibronectin binding, while Weissman and Kornitzer (404) focused on Fe utilization from hemoglobin and hemin and specific receptors. Pendrak and colleagues went on to follow up on the signaling function (292-295).

As discussed above, cells grown in defined medium with hemoglobin have enhanced binding to fibronectin, laminin, collagen type IV, and fibrinogen mediated by a promiscuous receptor. Fibronectin receptor expression was induced by ferric, ferrous, and cobalt-protoporphyrin derivatives of hemoglobin (291). Since the cobalt-containing molecule was active, the requirement was not for iron. In addition, the internalization of iron from hemoglobin occurred after the detection of enhanced binding capacity. The hemoglobin binding protein haptoglobin inhibited binding, and globin was inactive, thus suggesting that some structural aspect of the globin-porphyrin complex is required. Yeast cells were able to bind to immobilized hemoglobin. Since the induction of fibronectin binding was saturable and dose dependent, the presence of a receptor was suspected. In solution, the binding of hemoglobin, at a K_d of 1.2 × 10⁻⁶ M, suggested the presence of a low-affinity receptor. This receptor has not been further identified. Prior to the demonstration of a low-affinity receptor on yeast cells, an earlier report found that hemoglobin bound to hyphae at levels higher than those to yeast cells (374). Hyphae can use hemoglobin as a source of iron. Whether the binding to hyphal organisms is through the same receptor that is more abundantly expressed on hyphae or via a different receptor is not known. The hemoglobin receptor also appears to have a signaling function. Pendrak et al. (295) reported that hemoglobin induces the expression of other genes, e.g., HBR1 and HMX1. Hbr1p is a suppressor of white-opaque switching (294). As opaque cells are more susceptible to host defenses than are white cells (183), the suppression of switching while in the vascular system may be an advantage for the microbe. More recently, hemoglobin has been identified as an inducer of morphogenesis under conditions of neutral pH and low glucose and ammonium ion (292).

C. albicans has the components for high-affinity iron import but can also use hemoglobin and hemin (oxidized heme). PGA10 (alias RBT51), with homology to RBT5, was identified by complementation in S. cerevisiae of a strain missing the high-affinity uptake receptor and in the presence of an iron chelator, ferroxine (404) (Table 2). PGA10 belongs to a family with RBT5 and CSA1, which encode predicted GPI-CWPs with the conserved CFEM domain (312). Rbt5p has been recovered from the cell wall (73), which conflicts with the suggestion of membrane localization (404). Each gene of the family was deleted in a strain missing the C. albicans high-affinity iron transporter (Ccc2p) (404). In the mutants, the RBT5 null strain was the one reduced in growth on hemoglobin in the presence of ferroxine. As monitored by Western blotting, Rbt5p increased in iron-starved cells. It is glycosylated, as indicated by its size, which was much larger than predicted, and a polydisperse band. Glycosylation was confirmed by other methods. Using antibody recognizing Rbt5p and zymolyase-treated cells, reactivity was associated with the cells. The authors suggested a membrane location. ⁵⁵Fe-hemin bound to C. albicans in a saturable manner that was susceptible to competition by unlabeled hemin and hemoglobin. The competition suggested a single receptor class. The receptor has high affinity for hemoglobin and requires about 0.2 μM to compete at a labeled ligand concentration of 1 μM. In iron-starved cells, there are an estimated 3 × 10⁶ receptors, compared to 1.2 × 10⁵ receptors in unstarved cells. Although strains with single deletions of PGA10 or CSA1 were not impaired in binding, there was some contribution, as the strain with deletions of all three genes (RBT5, RBT51, and CSA1) bound less hemin than the single-deletion strains.

Although there is little free hemoglobin in serum, C. albicans possesses a hemolytic factor (reviewed in reference 49) that has been partially characterized (402). There are also some pathological conditions that can increase availability (reviewed in reference 295). As an iron source, once hemin has entered the cell, the release of Fe from heme is catalyzed by Hmx1p, a cytoplasmic protein (174) whose expression increases upon hemoglobin binding. In addition to the question of whether the CFEM family proteins Rbt5p, Rbt51p, and Csa1p are the receptors for signaling, there are many unanswered questions. Among these questions are how the signal transmits from the adhesin-ligand binding and how heme is released from hemoglobin and imported into the cell.

C. albicans produces extracellular Saps that can hydrolyze host proteins, as discussed briefly above. Some microbes can hijack host proteolytic proteins by binding plasminogen and use the host fibrinolytic system to enhance invasion. Two contemporaneous studies showed that C. albicans can bind human plasminogen (65, 163). Yeast cells in rich medium in exponential growth (163) or from a 24-h culture (65) bound soluble plasminogen. Plasminogen bound to immobilized cells in a dose-dependent manner with a K_d of 70 nM, which was in the same range as for other microbes (65). Cells can also bind the activated plasmin from solution (163).

To identify proteins binding plasminogen, Crowe et al. (65) extracted CWPs by treatment of cells with βME and Quantazyme (β1-3 glucanase). Plasminogen bound to the extracted immobilized proteins with a K_d of 112 nM. Proteins extracted from the S. cerevisiae and C. albicans glycosylation-deficient och1Δ/Δ, mnn4Δ/Δ, mnt4Δ/Δ, mnt2Δ/Δ, mnt3Δ/Δ, mnt4Δ/Δ, and mnt5Δ/Δ strains gave similar levels of binding. A lysine analog, ε-aminocaproic acid (εACA), almost completely inhibited binding to proteins. The observation suggested that lysine residues were critical for binding. Far-Western blotting of SDS-polyacrylamide gel electrophoresis-separated proteins indicated five major and several less intensely reacting spots. The addition of εACA during the plasminogen binding step again blocked binding. To identify the reactive moieties, spots identified by far-Western blotting after two-dimensional separation were cut from the gel for sequence analysis. The major plasminogen binding spots were encoded by the genes GPM1, ADH1, TSA1, and CTA1. Each of the protein products contain C-terminal lysine residues. Tef1p was a minor binding protein although it has terminal lysine residues, suggesting that the lysine may not be readily exposed. Three other minor binding proteins were Pgk1p, Tdh3p, and Fba1p. These proteins lack signals for classical secretion, and such proteins are the topic of discussion elsewhere in this review (Table 4). Some of the proteins, in addition to having cytoplasmic functions, are associated with other surface functions, such as laminin binding by Tdh3p.

In addition to binding factors H and FHL-1, rGPm1p also bound plasminogen (304), as predicted by the study discussed above. The inhibition of plasminogen binding to rGmp1p by εACA again implicated lysine residues as the binding motif of Gmp1p. Consistent with the observations of Crowe et al. (65) of multiple plasminogen adhesins, the strain with a deletion of GPM1 showed a reduction but not a complete loss of plasminogen binding capacity (304).

In other microbes, enolase is a surface protein that binds plasminogen. Jong et al. (163) studied a recombinant C. albicans enolase expressed in Escherichia coli. The recombinant protein was immobilized by a six-histidine tag-bound plasmin. The plasmin was enzymatically active and inhibited by known plasmin inhibitors. The binding was lysine dependent. Since the predicted sequence for Eno1p lacks C-terminal lysine but does have two residues within the last nine positions, it may be a relatively weak binder and not detected among the minor proteins by Crowe et al. (65). Plasminogen bound to a receptor undergoes a conformational change that alters the affinity for activators. The K_m for streptokinase activation of plasminogen bound to immobilized enolase was about 10% of that of unbound plasminogen (163). This suggests that plasminogen bound to enolase undergoes a conformational change.

The three studies examined the effect of bound plasmin on C. albicans interaction with various substrates. In one study, C. albicans cells with bound plasminogen showed hydrolysis of a fibrin clot when an activator of plasminogen was present (65). There was no activity from fungal cells in the absence of bound plasminogen. The requirement for an exogenous activator suggested that C. albicans does not have an appropriate proteolytic activity at the cell surface. Plasminogen bound to rGpm1p was converted to proteolytically active plasmin in the presence of an activator (304). The hydrolytic activity increased with increasing rGpm1p available to bind plasminogen. A third study used a fibrin matrix gel. C. albicans cells with bound plasminogen or plasmin showed hydrolysis, and the activity was abolished by inhibitors (163). There was no activity in the absence of bound plasminogen or plasmin. Thrombin in the gel may have been responsible for activating the bound plasminogen. The effect of bound plasminogen in the presence of an activator to facilitate the transversal of an ECM barrier or to damage an endothelial monolayer was examined (65). No advantage was observed with the plasminogen-coated cells. In contrast, there was an advantage when an in vitro blood-brain barrier system of human brain microvascular endothelial cells was tested (163). C. albicans cells with or without bound plasmin were added to the upper chamber containing the endothelial cell monolayer, in which tight junctions had been formed. Of two strains tested, the plasmin-coated cells of a rapidly growing strain had a small advantage (about 20%) in transversal into the bottom chamber over what was seen for uncoated cells. The advantage was larger for plasmin-coated cells of a slower-growing strain that still formed germ tubes, although the overall number of cells that traversed the monolayer was less than half of those for the faster growing strain.

Along with histatin and statherin, PRPs are the major salivary polypeptides. bPRPs (basic PRPs) can be either glycosylated or not. C. albicans can adhere to saliva-coated hydroxyapatite and to immobilized salivary proteins (39, 283a). bPRPs are major ligands for C. albicans adherence (283a). A recent study implicates Bgl2p, discussed previously for its enzymatic activity in the cell wall, as an adhesin for salivary proteins (160). An extract obtained from yeast cells by mild glucanase treatment inhibited binding to salivary proteins in two assays: binding to saliva-coated hydroxyapatite beads and binding to a blot overlay of separated saliva. In the latter assay, bPRPs were the major saliva binding components. Mannoprotein purified by chromatography on concanavalin A-Sepharose was enriched for inhibitory activity and contained two major components of 35 kDa and 97.4 kDa. The inhibitory activity of the proteins eluted from a gel was about 30- to 100-fold more than that of the crude extract. The 97.4-kDa material yielded no identifiable sequence. However, the 35-kDa sequence corresponded with Bgl2p. Bgl2p has the signal sequence for secretion by the classical pathway. In the cell wall, it appears to have two functions, one enzymatic within the C. albicans cell wall and one as an adhesin for an abundant protein in the oral environment.

Polymicrobial biofilms form on silicone surfaces of voice prostheses. The components of saliva from oral rinses of patients with voice prostheses that promoted C. albicans adherence were examined (141). Salivary components that bound to silicone were eluted, separated by electrophoresis, stained for protein, and transferred onto nitrocellulose filters. Two major salivary proteins of about 40 and 53 kDa were removed from the silicone. A blot overlay assay with radiolabeled yeast cells was employed to identify ligands. In both the initial saliva and eluted saliva, PRPs were ligands. However, in the eluted material, only one PRP bound yeast cells. The major binding was to the 40- and 53-kDa bands. This suggests that bPRPs are not major silicone-bound components and that the two unidentified proteins promote adherence to silicone voice prostheses.

The major saliva component statherin is a 43-amino-acid phosphorylated polypeptide. Protein peaks from a chromatographic separation of saliva were bound to hydroxyapatite beads, and the ability of the bound protein to promote adherence was determined. The binding occurred with several fractions identified as glycosylated PRPs and some uncharacterized peaks and primarily to statherin but not to acidic PRPs. Fuc-α-1,2-Gal-β-1,4-Glc partially inhibited binding to material other than statherin. Purified statherin mediated the dose-dependent binding of C. albicans to hydroxyapatite beads (161). Antistatherin antibody inhibited binding to saliva-coated beads and to BECs by 93% and 43%, respectively. The addition of Fuc-α-1,2-Gal-β-1,4-Glc reduced adherence to BECs by 79%. There is some difference between this study and prior studies, including one discussed above, which found that the major ligands of saliva after electrophoretic separation and blot overlay assay of fractionated or unfractionated saliva was to bPRPs (160, 283a). Collectively, however, these studies suggest that multiple ligands and adhesins mediate adherence in the oral cavity. Interestingly, in saliva, PRPs and statherin share the property of inhibiting the spontaneous precipitation of calcium phosphate salts.