Previous Article | Next Article ![]()
Clinical Microbiology Reviews, April 2004, p. 281-310, Vol. 17, No. 2
0893-8512/04/$08.00+0 DOI: 10.1128/CMR.17.2.281-310.2004
Copyright © 2004, American Society for Microbiology. All Rights Reserved.
Department of Pathology, University of Virginia, Charlottesville, Virginia 22908-0904
SUMMARY INTRODUCTION GLYCANS IN THE CANDIDA CELL WALL Basic Components Glucan. Chitin. Mannan. Sialic acids. Combination To Form the Cell Wall FUNGAL GLYCANS IN PATHOGENESIS AND HOST RESPONSE Protein Folding and Tertiary Structure Adhesion Adhesion to host components. (i) ß-1,2-Mannan. (ii) Acid-stable mannan. (iii) O-linked glycans. Adhesion to other members of the microflora. Immune Cell Receptors and Interactions Glycan binding to and activation of immune cells. (i) Mannan. (ii) Chitin. (iii) Glucan. Glycan influence on cytokine production. (i) Mannan. (ii) Glucan. Recognition by antibodies. DIAGNOSTICS AND THERAPEUTICS Diagnostics Mannan assays. Glucan assays. Therapeutics ISSUES IN CARBOHYDRATE CHEMISTRY Detection and Separations Complexity and Characterization CONCLUDING REMARKS ACKNOWLEDGMENTS REFERENCES
|
|
|---|
|
|
|---|
Using the National Nosocomial Infection Survey (1980 to 1990) data, Jarvis and Martone (131) reported that Candida species accounted for 9.4% of nosocomial urinary tract infections, 10.1% of nosocomial infections in adult and pediatric intensive care units, and 7.8% of nosocomial bloodstream infections. In 1990, fungi as a whole accounted for 10% of all nosocomial bloodstream infections (17). Ten years later, the rates of bloodstream infections due to Candida are still rising, with half of these candidemias due to C. albicans (230). More recently, other Candida speciessuch as C. glabrata, C. tropicalis, and C. dubliniensishave been isolated with increasing frequency (220). A study carried out by five university hospitals in the Netherlands also reported increased fungemia rates. Between 1987 and 1995, episodes of fungal bloodstream infections rose from 0.37 to 0.76 per 10,000 patient days. Of these infections, 93% were due to Candida species while 7% were due to Cryptococcus species (307).
In addition to the observed increases in incidence, fungal infections continue to be a serious clinical problem with respect to increased morbidity and mortality (194, 314; see also reference 219 and references therein). An analysis of records from the National Center of Health Statistics indicated that mycoses ranked 10th among the underlying causes of death in the United States in 1980 but that by 1997 this rank had risen to 7th (193). Increased morbidity due to nosocomial fungal infections also takes a heavy toll, resulting in longer hospital stays (49, 178, 314) and higher patient care costs (91, 234, 285).
The increasing incidence of fungal infections has, to a certain degree, coincided with the incredible medical advances of the last few years. Life-prolonging technologies, although welcomed by medicine and those it cares for, have unfortunately created numerous opportunities for emerging fungal infections. The rising number of immunocompromised patients, whether through immunosuppressive therapy, AIDS, or other ailments, the prophylactic use of antimicrobics, and the increasing number of indwelling catheters and prosthetic devices seen among patients all unwittingly subject patients to the threat of invasive disease (68). Patients with central intravenous IV catheters were over three times more likely to have bloodstream infections than patients without such catheters (17). Patients undergoing orthotopic liver transplantation demonstrated an incidence of invasive fungal infection of 5 to 42%, with Candida species being the most commonly isolated organism, followed by Aspergillus species (54). A study of heart transplant patients reported that while fungal infections accounted for only 7% of all infections, they were associated with the highest (36%) mortality (194). Taken together, these clinical observations strongly reinforce the necessity for understanding the relationship between fungus and host.
The interaction between fungus and patient occurs first at the level of the cell wall (35). The cell wall of most fungi is composed of glycoproteins embedded within a polysaccharide matrix or scaffolding. Additionally, some fungal species produce a polysaccharide capsule that surrounds the cell wall (e.g. the glucuronoxylomannan capsule produced by Cryptococcus neoformans [221]). It is therefore possible, and even likely, that carbohydrate, not DNA or RNA and perhaps not even protein, is the first component to contact host tissue. Thus, characterization of these exterior carbohydrate groups could lead to (i) a better understanding of fungal adhesion and mechanisms by which fungi avoid the host immune system, (ii) proposals for countering this avoidance so as to enhance the immune response, and (iii) the development of diagnostic tests based on identification of carbohydrate components. The utility of cell wall and cell surface carbohydrate groups has been shown in other areas of clinical microbiology (e.g., Mycobacterium) (23). Fungal cell surface carbohydrates also play an important role in industry (132, 135) and in plant and nonhuman animal fungal pathogenesis.
The purpose of this review is to bridge the areas of fungal pathogenesis and glycobiology. To this end, the manuscript is divided into three major sections. First, a brief introduction to the carbohydrate components of the fungal cell wall and how they are assembled is provided. Second, relevant studies where carbohydrates have been shown to play a role in infection, function in diagnostic assays, and act as therapeutic agents are discussed. The final section is a discussion of the practical issues and difficulties that arise when working with microbial carbohydrates.
This review is limited in scope in two ways. First, although there is an extensive body of literature describing the various structures and pathways by which microorganisms interact with host species, the focus of this review is fungal glycobiology. Therefore, discussion is limited to the polysaccharide components of the fungal cell wall and does not include more general fungal cell fractions, fungal proteins and lipids, or elements of the host cells. In addition, examination of the C. albicans literature illustrates the need for both careful identification of fraction components and strategies for demonstrating that carbohydrate, rather than protein or lipid, is responsible for the observed effect. For example, while C. albicans mannan (defined below) is found only as a glycoconjugate with protein or lipid, many pathogenesis studies use the term "mannan" to refer globally to mannoprotein or similar glycoconjugate fractions.
Second, C. albicans will serve as the example organism for relating glycobiology to clinical microbiology. Studies describing the role of carbohydrate groups in pathogenesis have been carried out mainly with C. albicans; therefore, the review likewise focuses on this organism. A review by Nelson et al. (202) focused on Candida mannan and issues of mannan chemistry, immune suppressive effects, and mechanisms of action. This review expands on their discussion to include glucans and relevant mannan results reported since the publication of the previous review. Although the focus is C. albicans, it should be remembered that carbohydrates, as polysaccharides or components of glycoproteins, play a role in the pathogenesis and diagnosis of other fungal species such as Cryptococcus neoformans, Aspergillus fumigatus, Kluyveromyces lactis, and Paracoccidioides brasilienis. Where appropriate, selected results from studies of these species are included in the discussion.
|
|
|---|
![]() View larger version (14K): [in a new window] |
FIG. 1. Monosaccharides found in the C. albicans cell wall.
|
|
View this table: [in a new window] |
TABLE 1. Carbohydrate components of some of the organisms responsible for invasive mycoses
|
]D = 30°. Chemical analyses allowed them to conclude that C. albicans glucan has a higher percentage of ß-1,6-glucan cross-links than that seen in Saccharomyces cerevisiae (21). Rees and Scott (231), based on computational modeling, predicted that ß-1,3-glucan forms wide helices, like a wire spring or a spiral staircase (designated type B), while ß-1,6-glucan forms structures that are extended and flexible (type D). As discussed below, glucan forms the structural skeleton of the cell wall, although the exact molecular arrangement has not yet been determined (21).
Chitin.
Linear polymers of ß-1,4-D-GlcNAc, called chitin, provide cross-linking and strength to the glucan scaffolding. The amount of chitin found in the cell walls of C. albicans hyphae is three times that found in yeasts (50). In contrast, Kanetsuna et al. (149) found that the GlcNAc content of the mycelial forms of Paracoccidioides brasiliensis and Blastomyces dermatitidis was one-quarter to one-third that of the yeast form. Lipke and Ovalle, based on findings of Kapteyn et al., have suggested that chitin plays a role in a cellular integrity rescue mechanism (153, 181). That is, when the cell wall is weakened by decreased amounts or quality of glucan, the cells compensate by increasing chitin production. Bahmed et al. (12) noted an increase in cell wall chitin in Kluyveromyces lactis and K. bulgaricus strains resistant to amphotericin B. Their results indicate that the net increase in chitin is due to a decrease in chitinase activity rather than an increased chitin synthase activity. However, the specific link between antifungal resistance and chitin content remains to be identified. Chitin also forms the core of the septum during bud growth and separation (see reference 29 for a review of the process in S. cerevisiae). GlcNAc also appears in the fungal cell wall as a component of glycoproteins. A chitobiose [D-GlcNAcß-(1
4)-D-GlcNAc] forms the link between protein asparagine residue and the inner core of N-linked glycans (Fig. 2).
![]() View larger version (20K): [in a new window] |
FIG. 2. Glycan structures in C. albicans glycoproteins. The processed oligosaccharide core is extended from the -1,3-linked mannose at its C-6 position (R). R structures represent possible side branch structures that can be incorporated into this outer-chain elongation. X can be up to 60 to 70. The acid-labile phosphooligomannosides are added to a side branch mannotetraose. Circled mannose units are potentially phosphorylated. Inset: Map key for linkage points. All bonds are -1,n, unless otherwise indicated. Structure scheme adapted from reference 88 with permission. Structural information based on work in references 13, 167 to 169, 211, 264, 265, and 293 and R. C. Montijn (1998), Ph.D. thesis, University of Amsterdam. GlcNAc, N-acetylglucosamine; M, mannose.
|
-glycosidic bonds predominate. The authors concluded that yeast mannan comprised short
-1,2 linked oligomannosides joined by
-1,6 linkages (21). Subsequent work, mainly by Ballou and coworkers, resulted in a more complete structural characterization of mannan from S. cerevisiae and other species (reviewed in reference 13). Similarly, elegant experiments by Suzuki and colleagues (163-166, 261, 262, 264) described the structures of the N-linked glycans from Candida species (Fig. 2). These glycans are composed of an internal core, one branch of which is extended to form the outer chain consisting of an
-1,6-linked backbone from which side branches are linked. To some of these side branches, additional ß-1,2-oligomannosides are attached through a phosphodiester. Because this phosphodiester bond can be cleaved by 10 mM HCl and heat, the ß-1,2-mannan is referred to as acid labile while the remaining outer chain groups are described as being acid stable (Fig. 2).
As alluded to above, mannose can be linked through either
- or ß-anomeric bonds. O-linked glycans and the outer chain side branches of N-linked glycans are composed mainly of
-1,2- or
-1,3-oligomannosides. Three types of ß-1,2-mannose additions have been reported by Suzuki and coworkers (266). The first type forms the acid-labile mannan in strains of C. albicans (both serotype A and serotype B), C. glabrata, and C. tropicalis (164-167, 263) The second is found on the nonreducing terminal of
-1,2-oligomannoside outer chain side branches of serotype A C. albicans strains and of C. glabrata and C. tropicalis strains (164-166, 263). The third type is connected to an
-1,3-linked mannose in the acid-stable mannan of C. guilliermondii and C. saitoana (261, 266). In addition, the first mannose in the N-linked glycan core is linked ß-1,4 to the second GlcNAc (Fig. 2). The structural models of Rees and Scott (231) predicted that
-1,2-mannan forms type B helices like ß-1,3-glucan,
-1,3-mannan forms extended ribbon chains (type A), and
-1,6-mannan forms type D chains.
ß-1,2-Mannan was predicted to form crumpled or contorted chains (type C). Further, all 1,2-linked glycans, with the exception of
-1,2-mannan, were predicted to form these distorted chains (231). Because of steric clashes between nonadjacent residues, the authors expected such structures to be found only rarely in nature. This expectation is supported empirically by the determined structures of commonly occurring polysaccharides. For example, neither starch (
-1,4-D-Glc; type B), cellulose (ß-1,4-D-Glc; type A), nor glycogen (
-1,4-D-Glc with
-1,6-D-Glc branches; types B and D) forms type C chains. Likewise, none of the commonly occurring fungal polysaccharides (Table 1) are predicted to form these crumpled, contorted chains. Nonetheless, examples can be found of 1,2-linked sugars, in addition to C. albicans mannan. Some these examples are Cryptococcus neoformans capsule glucuronoxylomannan (ß-1,2-D-xylose and ß-1,2-D-glucuronic acid), C. neoformans capsule galactoxylomannan (ß-1,2-D-xylose), gum ghatti (ß-1,2-D-glucuronic acid), and gum tragacanth (
-1,2-L-glucose) (9, 10, 22, 52, 157). In these cases, however, the 1,2-linked sugars occur either as a terminal group of a side chain or discontinuously as a component of a heteropolymer. In either case, the potential steric conflicts resulting from chain formation are avoided. The two instances of 1,2-linked sugars forming oligomers are a ß-1,2-D-glucan from Agrobacterium species (94, 226) and the ß-1,2-D-oligomannosides from Candida species.
Sialic acids. With only a few exceptions, sialic acids have been found only in higher eukaryotes (i.e., vertebrates and some invertebrates such as Echinoderms) (300). Besides these organisms, sialic acids have been found in Drosophila embryos and the capsule material of some bacteria species. Benhamou and Ouellette (19) also found sialic acid in the cell wall of a fungal plant pathogen, Ascocalyx abietina. Where they are typically found, sialic acids are components of O-linked glycans or complex-type N-linked glycans (Fig. 3). As mentioned above, several groups have presented evidence for the presence of sialic acids in the cell wall of C. albicans (2, 140, 272, 308). Until these reports, there had been no indication that yeast (C. albicans or S. cerevisiae) produce sialic acid in any capacity.
![]() View larger version (16K): [in a new window] |
FIG. 3. Representative glycans from higher eukaryotic organisms. Inset: Key for linkage positions as in Fig. 2. Fuc, fucose; Gal, galactose; GalNAc, N-acetylgalactosamine; GlcNAc, N-acetylglucosamine; Glc, glucose; M, mannose; SA, sialic acid.
|
Several aspects of these results are puzzling. First, the C. albicans cells were disrupted by shearing with glass beads. The cell debris, containing the cell wall, was removed by centrifugation, and the supernatant fluid, containing mostly cytoplasmic proteins, was used as the source material for a C3 binding protein. Such a protein would be expected to be membrane or cell wall associated and would thus be more likely to be found with the cell pellet. Second, the removal of terminal sialic acids by neuraminidase seems insufficient to account for a 20% decrease in apparent molecular mass, observed as an increase in relative electrophoretic mobility. Relative electrophoretic mobility is also unlikely to be affected by the presence or absence of negative charge from sialic acids because such charge effects are overwhelmed by the sodium dodecyl sulfate present during electrophoresis. Thus, any change in relative electrophoretic mobility must be due to changes in apparent molecular mass. The neuraminidase preparation used (Sigma; type V) reports potential protease contamination. The presence of proteases could be an alternative explanation for the observed decrease in molecular mass. However, because the 42-kDa protein was unaffected by neuraminidase treatment, the presence of generally acting proteases seems unlikely. Third, Endo F treatment did not deglycosylate the C3d binding proteins. This enzyme cleaves N-linked glycans between the two GlcNAc residues of the oligosaccharide core. It is effective against high-mannose or hybrid glycans but not against complex glycans. While this is consistent with the types of glycans typically containing sialic acids, it is not consistent with the types of glycans made by C. albicans.
Similar evidence was presented by Wadsworth et al. (308) in their characterization of a type 2 complement receptor (CR2). Proteins were extracted from the C. albicans cell wall with dithiothreitol. Glycans were labeled with biotin-hydrazide either before or after protein separation by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (PAGE) and subsequent transfer to nitrocellulose. Sodium periodate oxidizes cis-diols (e.g., between C-2 and C-3 of mannose), cleaving the ring and producing two aldehydes. These aldehydes can then react with the hydrazide group, resulting in covalent linkage of the attached biotin to the glycan. Wadsworth et al. showed that a 50-kDa protein from C. albicans blastoconidia and a 60-kDa protein from hyphae were recognized by anti-CR2 antisera. Initial results indicated that both proteins were glycosylated by biotin-hydrazide labeling. A 50-kDa protein also bound concanavalin A (Con A) and, to a lesser extent, wheat germ agglutinin (which normally binds GlcNAc but has been reported to also bind sialic acids). However, treatment with N-glycanase (which cleaves between the first GlcNAc and the Asn) or Endo H (which acts in a similar manner to Endo F) did not significantly reduce lectin binding, as would be expected. Furthermore, N-glycanase, Endo F, or neuraminidase treatment had no effect in immunoblots using the anti-CR2 antisera as probes. Based on these results, and considering that the protein separations were one-dimensional PAGE, it seems more likely that the 50-kDa CR2 protein is the deglycosylated form and that the biotin-labeled protein was a comigrant.
The 60-kDa protein also appeared to bind Con A, although this binding also was unaffected by Endo H treatment. This protein gave a stronger signal with regard to wheat germ agglutinin binding, which was markedly reduced on digestion with N-linked glycanase. The remaining results, however, led to some conflicting conclusions. Neuraminidase treatment of the 60-kDa protein led to loss of signal in immunoblots, which might be expected if the epitope included the removed sialic acid. In addition, neuraminidase treatment, either before or after biotinylation, eliminated detection of the biotin signal. This result is consistent with the terminal sialic acid being linked to a galactose (Gal) residue, as is seen in complex-type N-linked glycans. Although galactose can be labeled with biotin-hydrazide, a preliminary reaction with galactose oxidase is required to generate the required hydrazide-reactive aldehydes (248). N-Glycanase and Endo F deglycosylated the 60-kDa protein, as indicated by a decrease in apparent molecular mass in immunoblots. N-Glycanase cleaves both complex and high-mannose N-linked glycans, but Endo F is effective only against high-mannose glycans. This result, in contrast to the neuraminidase result, suggests that the glycans on the 60-kDa protein are high mannose rather than complex.
Jones et al. (140) examined the electrostatic properties of the C. albicans cell surface by using a cationic fluorophore, 9-aminoacridone. The electronegative C. albicans surface attracts the fluorophore, leading to self-quenching and a decrease in detectable fluorescence. Treatment of the cells with neuraminidase resulted in a time-dependent decrease in fluorescence intensity, leading these authors to conclude that the C. albicans surface contained sialic acid groups. However, the neuraminidase used in this study was from the same source as that used by Alaei et al., so that protease contamination is a potential confounding factor for concluding that the observed loss of electronegativity was due specifically to loss of sialic acids.
More direct evidence was presented by Soares et al. (272), who used fluorophore-labeled lectins to detect sialic acids on the C. albicans cell surface. Fluorescein isothiocyanate-conjugated Sambucus niger agglutinin (SNA) strongly labeled the entire surface of blastoconidia but did not appear to label hyphae. This lectin is highly specific for the sialic acid N-acetylneuraminic acid linked
-2,6 to galactose or GalNAc. Labeling by SNA was completely eliminated by pretreatment of C. albicans cells with Clostridium perfringens sialidase. Fluorescein-isothiocyanate-conjugated agglutinins from Limax flavus and Limulus polyphemus also bound to the C. albicans cell surface. These lectins are not specific for sialic acid linkage position and do not require a specific subterminal sugar group. As with the SNA result, L. flavus and L. polyphemus agglutinin binding was significantly reduced by pretreatment of the Candida cells with sialidase. In addition, sialidase treatment allowed the binding of peanut agglutinin to occur. This lectin recognizes terminal ß-galactose units, groups that would be exposed from complex-type N-linked glycans following removal of the terminal sialic acids (Fig. 3).
Using similar methods, sialic acids have been found to be also associated with the cell surface or exogenous glycoproteins of Cryptococcus neoformans, Fonsecaea pedrosoi, Paracoccidioides brasiliensis, Pneumocystis carinii, and Sporothrix schenckii (237, 270, 271; reviewed in reference 6). In addition, Rodrigues et al. (236) have recently shown the presence of sialyltransferase activity in C. neoformans. Taken together, these results not only demonstrate the presence of sialic acids on pathogenic fungi but also show that these sugars are probably the termini for complex-type N-linked glycans. This observation is significant because these glycans had not previously been found on these organisms. In mammals, sialylated glycans are often a component of glycoconjugates that modulate intercellular interactions, such as the sialyl Lewis x selectin ligand (155, 252).
A similar role for sialylated glycans has been demonstrated for C. neoformans and S. schenckii, where treatment of fungal cells with neuraminidase significantly increased phagocytosis by murine peritineal macrophages (5, 6, 237). S. schenckii or C. neoformans yeast cells were treated with neuraminidase from Clostridium perfringens or Vibrio cholerae for 1.5 to 2 h and then washed. Treated cells and untreated control cells were then added to adherent peritoneal macrophages. Neuraminidase treatment of C. neoformans cells led to a twofold increase in the phagocytic index (the percentage of macrophages phagocytizing at least one yeast multiplied by the average number of phagocytized yeast cells) (237). For S. schenckii cells, the increase was almost eightfold (207). In the studies above, neuraminidase was specifically chosen to complement the other work demonstrating the presence of sialic acids. It is unknown what effect, if any, treatment with other glycosidases, such as mannosidase, would have on phagocytosis. However, the neuramindase results are consistent with a model, proposed by Schauer and coworkers, that terminal sialic acid groups mask ligands for macrophage-bounds receptors (155, 156, 177). This raises the possibility that sialylated glycans on fungi and other microorganisms can assist in adaptation to life in the host by avoiding recognition or killing by the immune system.
Previous studies of mannoprotein fractions from the cell walls of S. cerevisiae or C. albicans gave no indication of the presence of sialic acids or complex-type N-linked glycans. There are two potential explanations for this (until recently) lack of evidence. First, the usual methods of mannoprotein preparation (hot water or citrate extraction followed by precipitation with Cetavlon or Fehling's reagent [see below]) may be harsh enough to degrade the sialic acid groups, which are known to be sensitive to chemical conditions (301). Second, the N-linked glycans on mannoproteins are strictly high mannose, as described in all previous reports, but the sialic acids are components of other glycoproteins as complex-type glycans. Such glycoproteins would not be expected to precipitate with Cetavlon because they lack the long mannan chains needed to interact with the Cetavlon (discussed in reference 191) and the mannose cis-diols needed to interact with borate (in the Cetavlon precipitation) or copper (in Fehling's reagent).
Visualization of the cell wall structure began with electron microscopy, which showed layers of differing electrodensity (42, 43, 260). These electron micrographs, combined with biochemical data, led to the development of a cell wall schematic (Fig. 4) (35, 232, 260) that has been further refined as the cell wall components have been studied more closely (161, 213). Although these schematics are often depicted as having a layered structure, it has become increasingly apparent that the cell wall is actually a dynamic structure and that any given layer should be more properly referred to as a zone of enrichment (260).
![]() View larger version (33K): [in a new window] |
FIG. 4. Schematic of the C. albicans wall structure. The wall structural cartoon is based on information from references 150, 152, 213, and 269. Electron micrographs are reprinted from reference 108 with permission. GPI, glycosylphosphatidylinositol-linked glycoproteins; ManPr, mannoprotein; PIR, PIR-family glycoproteins; PM, plasma membrane.
|
The detailed structure of the fibrils is still unclear. One possibility is that the protein component is globular and lies close to the glucan scaffolding, with the observed fibril being entirely glycan. Alternatively, the fibril may contain a linear or tightly helical protein core decorated with glycan. Some cell wall proteins, which are anchored into the plasma membrane, possess a "lollipop on a stick" structure (134). A long stretch of polypeptide (the stick) extends from the membrane and passes through the wall. Short O-linked glycans are attached along the length of the stick, presumably to help maintain a linear structure. A globular domain (the lollipop) is then stuck out into the extracellular space. However, it is unknown if the fibrils are anchored in such a way into the membrane, covalently attached to glucan or chitin, or simply trapped within the glucan matrix. It is also not known if the fibril is composed of subunits, but some initial experiments suggest that a fibril subunit exists (319).
Some organisms produce extensive carbohydrate structures at the cell wall exterior. For example, C. parapsilosis has been reported to produce a slime layer similar to those produced by Staphylococcus epidermidis and other bacterial pathogens (186, 187). Although the composition of this slime layer has not been characterized in detail, Kuhn et al. (171) have shown that cell wall-like polysaccharides comprise the matrix. Their conclusions were based on staining with a fluorescent Con A conjugate, which binds to terminal mannose and glucose units, and staining with Calcofluor White, which binds to ß-linked polysaccharides such as chitin and ß-glucan. These authors also demonstrated that C. albicans, in addition to C. parapsilosis, was able to construct this slime layer biofilm. As mentioned above, Cryptococcus neoformans constructs a polysaccharide capsule composed mainly of glucuronoxylomannan (52, 297) on the cell wall exterior. Galactoxylomannan and mannoproteins are also elements comprising the capsule. The genetic and temporal regulation of capsule synthesis has recently been reviewed by Bose et al. (22).
|
|
|---|
Adhesion to host components. The presence of polysaccharide in the intercellular junction of C. albicans cells bound to epithelial or endothelial cells has been known for some time. Marrie and Costerton (188) described a ruthenium red-stainable matrix surrounding the C. albicans cell. Ruthenium red is a cationic dye with affinity for anionic polymers such as certain polysaccharides. The observed matrix was shown to exist between fungal cells and buccal tissue as well as between fungal cells and bacterial cells. Barnes et al. (16) demonstrated fibril connections between the yeast cell surface and the endothelium of renal peritubular and glomerular capillaries. Although the specific composition of the fibrils (i.e., carbohydrate versus protein) was not investigated, the authors speculated that they were the same as, or similar to, polysaccharide fibrils found on the S. cerevisiae cell surface as described by Johnston and Latta (136). Tronchin et al. (296) also reported that fibrillar structures and polysaccharide granules on the yeast cell surface appeared to mediate adhesion to epithelial cells. Some of this material was released from the cell surface into the surrounding medium under certain conditions (109, 183).
Other work focused on specific glycans that might be involved in adhesion events. The majority of evidence pointed to mannan as the active glycan. The studies discussed immediately below established the role of mannan in adhesion. Subsequent work extended these results to show which specific mannan components (ß-1,2-oligomannosides, acid-stable side branches, or O-linked glycans) were the critical ones.
Maisch and Calderone (184) found that the mannoprotein fraction from C. albicans, when conjugated to sheep red blood cells (SRBC), was sufficient to cause the conjugated SRBCs to adhere to a fibrin-platelet matrix. Mannosidase treatment eliminated the adherence, indicating that the mannan was the adhering component. Subsequent results indicated that the matrix component involved in binding to mannan was fibronectin (253).
Sandin et al. (245) showed that pretreatment of yeast cells with Con A inhibited the binding to buccal epithelial cells (BEC), from 100 to 18% adherence. Con A pretreatment of the BECs also reduced binding. Con A is a lectin that binds
-linked terminal mannose or glucose residues of an oligosaccharide (93). Of the carbohydrates in the Candida cell wall (mannan, glucan, and chitin) only mannan is
-linked. Furthermore, Con A-mediated inhibition of binding was partially reversible with increasing amounts of
-methylmannoside added as a mannan competitor. Lectins that bind N-acetylgalactosamine (GalNAc), GlcNAc, fucose, or galactose did not significantly alter percent adhesion. Sandin et al. also reported that
-methylmannoside itself inhibited the binding to BEC while GlcNAc, ribose,
-methylglucoside, galactose, and xylose did not. This is in partial disagreement with the results of Sobel et al. (274), who reported that treatment of C. albicans with fucose inhibited adhesion but treatment with mannose,
-methylmannoside, or galactose did not.
Centeno et al. (47) investigated the possible adhesion interactions between epithelial cells, bacteria, and fungi. Earlier observations had been made that bacterial binding to epithelial cells blocked similar binding by fungi. In addition, bacterial structures (e.g., pili) can bind to fungal cells, leading to fungal agglutination. Centeno et al. found that binding of yeast to BEC was enhanced by pretreatment of the BEC with piliated, but not nonpiliated, strains of Klebsiella pneumoniae or E. coli. The enhanced binding was reversed by the addition of mannose. Further, mannose, but not glucose, inhibited C. albicans binding to urothelial cells and BEC in the absence of bacterial cells. The authors hypothesized that the bacterial pili added mannose-recognizing lectin-like proteins that provided extra attachment sites for the fungal cells.
Sawyer et al. (250) examined a liver perfusion model for studying yeast adhesion. Mouse liver was isolated (in an ex vivo manner) into a perfusion system. C. albicans cells were then infused into the liver, and the system was switched back to perfusion medium. Clearance of cells from the perfusate exiting the system was due either to endothelial trapping or, to a lesser extent, to adherence to and subsequent phagocytosis and killing by Kupffer cells. Treatment of the yeast cells with
-mannosidase (from Jack Bean), subtilisin,
-chymotrypsin, and papain increased the hepatic killing of yeast cells but did not affect trapping. Mannoprotein, isolated from either C. albicans or S. cerevisiae, blocked fungal cell trapping in the liver and was also able to elute previously trapped cells. These results, combined with previous results from this group showing that mannose or
-methylmannoside (but no other sugar) was able to block liver trapping, suggest that it is the carbohydrate portion of the mannoprotein fraction that is involved in hepatic endothelium adhesion.
The use of monosaccharides (e.g., mannose or
-methylmannoside) to inhibit binding is commonly applied to demonstrate the recognition of glycans. In the case of competition for the Con A site, this practice seems reasonable since it is known that Con A binds to terminal saccharides. However, other lectins or receptors may recognize oligosaccharides or even oligosaccharides in the context of the polypeptide or lipid to which they are conjugated. Further, these binding interactions can be specific for not only chain length and saccharide composition but also anomericity and linkage. Some of these instances are mentioned in the discussion below. Thus, it seems that the use of bioses or longer oligosaccharides as competitors would be a more physiologically relevant choice. There are a couple of issues, however, that make this choice problematic. First, the specific configuration (i.e., saccharide, anomericity, linkage) needed for a competitor often cannot be known a priori. Therefore, several oligosaccharides may have to be tried before a suitable one is identified. Second, in cases such as the ß-1,2-oligomannoside adhesin described below, the oligosaccharides are not commercially available and must be fractionated and purified in-house, which can be time-consuming. The problem of oligosaccharide availability is discussed in more detail in the last section of this review.
(i) ß-1,2-Mannan. Miyakawa et al. (195) developed C. albicans mutants that lack a ß-1,2-oligomannoside sequence found in the acid-stable mannan of serotype A strains. They observed a decrease in adhesion to BEC and to a mouth squamous carcinoma cell line by the mutant compared to the parent. Adherence of the mutant was comparable to that seen for wild-type serotype B cells that lack the ß-1,2-oligomannosides in the acid-stable region. Pretreatment of the epithelial cells with mannoprotein from the parent strain decreased adhesion of parent cells, but mannoprotein from the mutant did not.
A series of papers from Cutler and coworkers identified specific mannan groups involved in C. albicans binding to macrophages in the splenic marginal zone. Kanbe et al. (147) first demonstrated this binding in an ex vivo assay. Subsequent studies (146, 179) investigated the possible cell wall groups that might be involved in the observed adhesion. C. albicans cells were extracted with ß-mercaptoethanol, and the released material was fractionated using Con A affinity and ion-exchange chromatography (146). The crude ß-mercaptoethanol extract inhibited yeast cell binding to the splenic marginal zone and lymph node epithelial cells in a dose-related manner, and this inhibition activity could be isolated to one particular fraction. Heat and protease treatment of the fraction did not affect inhibition, but periodate and mannosidase treatment did, indicating that the active moiety was carbohydrate. This conclusion was supported by experiments using a monoclonal antibody (10G) that recognizes a ß-1,2-mannotetraose (179). The antibody was able to inhibit C. albicans attachment to murine splenic marginal-zone macrophages in the ex vivo assay. Recently, Fradin et al. (83) identified the receptor molecule on the macrophage for ß-1,2-oligomannosides to be a 32-kDa protein with homology to galectin 3. Since the strain that was used for these studies, A9, is serotype B, the tetraose epitope must be part of the acid-labile fraction of N-linked glycans (Fig. 2).
In a recent study using an in vivo model, Dromer et al. (71) supported the importance of ß-1,2-oligomannosides in adhesion. Intestinal colonization in mice was assayed by both survival and clearance (measuring CFU in fecal pellets). Expression of, or presence of, ß-1,2-oligomannosides, as measured by agglutination with an anti-ß-1,2-oligomannoside monoclonal antibody, positively correlated with virulence as assayed in mouse and rat systemic models. In the developed infant-mouse gastrointestinal (GI) colonization model, the investigators tested virulent (10261, high ß-1,2) and avirulent (10231, low ß-1,2) strains and measured the CFU/fecal pellet. The median score was always lower for 10231, and colonization was more protracted/prolonged with 10261 (which contained more ß-1,2). Synthetic ß-1,2-oligomannosides, but not
-1,2-oligomannosides, administered by gavage 1 h preinoculation, decreased or eliminated the CFU/fecal pellet (day 7 postinfection) in a dose-related manner. The authors therefore concluded that ß-1,2-oligomannosides play a role in colonization, presumably at the adhesion step.
(ii) Acid-stable mannan.
Other work by Cutler's group (144) showed that the acid-stable mannan inhibited C. albicans binding to splenic marginal zone epithelial cells, demonstrating that this mannan is also involved in attachment. Inhibition was not affected by treatment of mannoprotein with pronase, confirming the importance of the carbohydrate portion. Although the exact structure of the mannan group involved is not yet known, certain characteristics have been determined. Evidence suggests that both the
-1,2-linked side branches and the
-1,6-linked backbone are involved (145). Mannoproteins from S. cerevisiae MNN2 deletion mutants, which do not attach side branches to the outer chain backbone, did not inhibit binding, showing that the
-1,6-linked backbone alone is not sufficient. Acetolysis of the acid-stable mannan cleaves the
-1,6-linked backbone bonds and releases the side branches. Acetolysates of acid stable mannan also had no effect on binding, indicating that the side branches alone are not sufficient to produce the observed acid stable mannan-dependent adhesion. Partial sulfuric acid digestion, which sequentially releases the terminal sugar groups from the nonreducing ends, was carried out on acid-stable mannan. As the digestion progressed, producing shorter side branch chains, the inhibition of C. albicans binding decreased. Taken together, these results indicate that the acid-stable mannan mediating C. albicans binding to splenic marginal zone macrophages comprises two or more side branches, joined through the outer chain backbone and of a certain minimum length. Based on their results, Kanbe and Cutler (145) further concluded that chain length was more important than a specific side branch conformation. Other specifics of the macrophage mannose receptor and mannose-binding protein have been recently reviewed (84).
(iii) O-linked glycans. Buurman et al. (27) suggested that O-linked glycosylation plays a role in adhesion. Mutant strains with defects in O-linked glycosylation (leading to truncated glycans) showed decreased adhesion to human buccal epithelial cells in vitro and decreased adhesion to rat vaginal epithelial cells in vivo. They also observed that guinea pigs and mice had increased survival when challenged with the mutant strains. This increased survival was coincident with a decreased organ burden. These results, combined with the in vitro adhesion experiments, led the authors to conclude that decreased adhesion resulted in decreased organ burden and increased survival.
Herrero et al. (110) looked at the issue of glycosylation from the supply side. GDP-mannose is brought in from the cytosol to the Golgi lumen, where mannose is transferred to the growing mannan chain. The other product, GDP, is then hydrolyzed by a GDPase to give GMP and inorganic phosphate. GMP is then transported to the cytosol through an antiporter that brings in the GDP-mannose. Thus, a defect in the GDPase should have an effect on protein glycosylation. Previously, deletion of the S. cerevisiae GDPase gene (20) decreased glycosylation in the Golgi. The authors cloned the C. albicans GDPase gene, designated CaGDA1, and used the URA-blaster (80) procedure to knock out both alleles in C. albicans. Unexpectedly, they observed no change in N-linked glycans. However, a significant change in O-linked mannan was observed. There was a loss of the mannobiose and mannotriose groups and a corresponding increase in the monomer. This suggested a loss of ability to extend the O-linked chain. However, this loss of O-mannosylation did not affect measured adherence to HeLa cells.
These O-linked glycan results present a few possibilities. (i) The defect in the Herrero strain was not severe enough to alter adhesion. The authors themselves speculate that there might be backup GDPases in C. albicans. (ii) the residual O-linked M1 glycans (from Buurman et al. [27]) or the residual M4-M5 O-linked chains observed by Herrero et al. (110) were sufficient to maintain whatever structural requirement was necessary for glycoprotein stability and adhesin function. (iii) Some of the results observed by Buurman et al. (e.g., increased survival, increased vaginal clearing, and decreased organ load) were due to processes other than adhesion per se, although this later case is harder to reconcile with the observed decrease in adhesion to BEC in vitro. (iv) If O-linked glycans play a structural role (as in the lollipop-on-a-stick model), the effect of disrupting O-linked glycosylation may have more to do with protein structure than the glycans acting directly as an adhesin.
There is some evidence for the involvement of glycans other than mannan in adhesion. Reversing the typical manner in which adhesion studies are done, Forsyth and Matthews (81) investigated carbohydrate inhibition of lymphocyte adhesion to C. albicans hyphal cells. While mannose and
-methlymannoside both inhibited interleukin-2 (IL-2)-activated mouse lymphocyte adhesion to C. albicans hyphal cells by approximately 40%, GlcNAc inhibited adhesion by activated mouse and human lymphocytes by 80%. No effect on mouse lymphocytes was observed when glucose or sucrose was used. As with the mouse lymphocytes, glucose had no effect on adhesion of IL-2-activated human lymphocytes, but small amounts of ß-glucan (4 mg ml1, or approximately 40 µM) inhibited adhesion by 70%. From these results, the authors concluded that lymphocyte binding occurs through Mac-1, which has a lectin-like domain, known ligands for which include ß-glucan and chitin. Placing their observations in context, Forsyth and Matthews concluded that interactions between fungal cells (C. albicans) and lymphocytes are multidimensional but involve lectin-like binding sites on the lymphocytes and glycans on the fungal cell surface.
The inhibition of adhesion by GlcNAc and the chitin binding domain of Mac-1 echoed an earlier result of Segal et al. (255), who found that, of the cell wall polymers, chitin (although not mannoprotein or glucan) inhibited the binding of C. albicans to vaginal epithelial cells. Chitin hydrolysate, GlcNAc, glucosamine, and mannosamine inhibited adhesion to a similar extent (255). On first inspection, these results would seem to be in conflict with the current model of the cell wall that has the chitin component in the inner layers of the cell wall, near the plasmalemma (Fig. 4). It has been shown, however, that the hyphal cell wall is thinner than that of the blastoconidia (43). Further, there is evidence for a small amount of chitin being present on the cell surface (295). These observations serve as illustrations of the dynamic nature of the cell wall and support the evidence by Forsyth and Matthews and Segal et al. for a role of chitin in adhesion, but they do not explain the ineffectiveness of mannose or mannoprotein as an adhesion inhibitor observed by Segal et al.
Another possible explanation is that the experimental conditions are playing a role. Researchers have used various strains, growth conditions, and substrates in adhesion experiments (Table 2). A comparison of additional experimental conditions used in adhesion studies has been done recently by Calderone and Gow (32). The multidimensional aspect of adhesion noted by Forsyth and Matthews may indicate that the relative importance of a specific adhesin depends on the combination of fungal strain and tissue substrate, as well as experimental design.
|
View this table: [in a new window] |
TABLE 2. C. albicans glycans and adhesion
|
|
View this table: [in a new window] |
TABLE 3. Coaggregation of C. albicans with other members of the microflora
|
Hsu et al. (117) looked at coaggregation of C. albicans and oral bacteria isolated from bone marrow transplant patients. They reported that in contrast to previous results (11), sugars did not inhibit coaggregation of C. albicans and A. viscosus. Neither did the sugars inhibit coaggregation with S. sanguis or S. epidermidis, in agreement with the other results. Coaggregation of C. albicans with S. mitis and Lactobacillus salivarius was inhibited by
-methylglucoside. Glucosamine inhibited coaggregation with S. salivarius, while glucose inhibited coaggregation with Bacteroides gingivalis. Mannose,
-methylmannoside, glucose, and
-methylglucoside all inhibited coaggregation with Lactobacillus amylovorus. Coaggregation with F. nucleatum was inhibited by the addition of mannose or
-methlymannoside. More recently, Jabra-Rizk et al. (121) showed that mannose inhibited the coaggregation of F. nucleatum with C. dubliniensis as well as C. albicans.
The summary of these results seems to be that the strains of A. viscosus (one strain, tested twice) and F. nucleatum (three strains) tested were positive for coaggregation and that the evidence is that a glycan on the fungal cell surface is involved in that coaggregation. Because the fungal cells were cultured and prepared for assay under different conditions, it is difficult to predict how other fungal phenotypes are involved in coaggregation. For example, results by Jabra-Rizk et al. (121, 122) suggest that cell surface hydrophobicity plays a role in coaggregation of C. albicans and C. dubliniensis with F. nucleatum. This conclusion is supported, to a certain extent, by the observation of Jenkinson et al. (133) that Streptococcus gordonii, S. sanguis, and S. anginosis adhered best to C. albicans germ tubes, structures that are highly hydrophobic, especially at the distal end (106, 107). These authors, however, concluded that cell surface hydrophobicity was not involved since starvation of fungal cells increased coaggregation but had no effect on cell surface hydrophobicity.
In contrast to F. nucleatum and A. viscosus, no strain of Klebsiella pneumonia, Lactobacillus casei, Peptostreptococcus anaerobius, P. magnus, P. micros, Porphyromonas gingivalis, Prevotella intermedia, Pseudomonas maltophila, P. stutzeri (2 strains), P. aeruginosa, or Staphylococcus aureus formed coaggregates. The Streptococcus species present more variable results that seem to be species and strain dependent. For example, S. sanguis appears to form some kind of coaggregate, but the evidence points to the relevant glycan being on the bacterial, not the fungal, cell surface.
|
View this table: [in a new window] |
TABLE 4. Glycan interaction with immune cells and immune system components
|
Because zymosan phagocytosis was inhibited by a mannoprotein fraction, Speert and Silverstein (275) concluded that phagocytosis occurs after recognition by a mannose receptor. Similarly, Cross and Bancroft (56) reported that phagocytosis of an acapsular Cryptococcus neoformans mutant was inhibited by the addition of mannoprotein from S. cerevisiae. Scaringi et al. (251) reported that injection of mice with C. albicans mannoprotein stimulated the generation of natural killer cells and activated macrophages. Podzorski et al. (224) reported an increase in lymphoproliferation due to mannoprotein. The preceding results, however, lack the evidence necessary to prove that the effect was due to the carbohydrate portion of the mannoprotein fraction rather than the protein.
Several recent studies have linked glycan specifically to effects on the immune system. Podzorski et al. (224) refined their results mentioned above and examined the effect of O-linked mannan on lymphoproliferation. These oligosaccharides, separated from the mannoprotein fraction by beta elimination, inhibited lymphoproliferation in response to Candida antigen, tetanus toxoid, and herpes simplex virus type 1 (224). Fradin et al. (82) showed that pretreatment of macrophages (either peritoneal macrophages or the J774 macrophage cell line) with ß-1,2-linked, acid-labile oligomannosides decreased the macrophage binding of both live and heat-killed C. albicans cells. The extent of inhibition was the same when a synthetic ß-1,2-mannotetraose was used as the inhibitor. Another study from the same group showed that the ß-1,2-oligomannosides can also be released from C. albicans cells during C. albicans-macrophage coculture as a component of a glycolipid fraction (141). A recent paper by Cambi et al. (39) reported that along with the mannose receptor, dendritic cells (DC) bind C. albicans cells through a DC-specific intercellular cell adhesion molecule (ICAM)-grabbing nonintegrin (DC-SIGN) protein. Binding of fungal cells was inhibited in the presence of mannose, but only by 40%. Even with the addition of an anti-DC-SIGN antibody, binding was still 20% of control levels. In addition, when DC-SIGN was transfected into K562 cells (a moderately undifferentiated leukemic derivative cell line), mannose had no effect on C. albicans binding. These results suggest that the interaction of C. albicans cells and DC-SIGN is not mediated solely by lectin-like activity.
Watanabe et al. demonstrated that the glycan component of C. albicans mannoprotein acts as a hemolytic factor (311). RBC were mixed 1:1 (vol/vol) with cell fraction samples, and hemolysis was monitored by measuring the absorbance of the suspension at 405 nm. Hemolysis was associated with a sugar-rich, high-molecular-weight chromatographic fraction characterized as having low protein and high sugar content. Treatment of this fraction with periodate, which cleaves the glycan chains, eliminated hemolytic activity, demonstrating that carbohydrate was the active group.
Cross and Bancroft (56) extended their inhibition experiment to show that mannose binding protein bound to the acapsular Cryptococcus neoformans mutant, but not to encapsulated wild-type cells. These results indicated that the glucuronoxylomannan capsule produced by C. neoformans could mask potential glycan ligands recognized by immune cell receptors. Subsequent work by Mansour et al. (185) showed that mannoprotein released by C. neoformans into the culture supernatant fluid inhibited, in a dose-related manner, the binding of the macrophage mannose receptor to one of its natural ligands, horseradish peroxidase type II. They further showed that deglycosylation drastically reduced the ability of C. neoformans mannoprotein to activate primary T cells. In addition, binding of C. neoformans to DC is mediated, at least in part, by a mannose receptor (283). There is also evidence that peritoneal macrophages bind and phagocytose Paracoccidioides brasiliensis cells via a mannose receptor. Phagocytosis of P. brasiliensis cells was significantly inhibited by the addition of mannose or, to a lesser extent, fucose (3). Mannose reduced the phagocytic index to between one-third and one-half that of the control. This result again raises the issue of using monosaccharides as substitutes for oligomers or polymers. An oligomannoside, used as a competitor, may have completely inhibited phagocytosis because of more complete blockage of the macrophage receptor.
(ii) Chitin. Suzuki et al. (279) first reported the immunological effects of chitin in a murine model of candidiasis. Mice receiving pretreatment with chitin (50 mg kg1) were better able to survive a lethal challenge with C. albicans than were mice receiving no treatment. In addition, polymorphonuclear leukocytes (PMN) from circulating blood and peritoneal exudate cells from treated animals showed a higher level of reactive oxygen species (O2 and H2O2) generation than did those isolated from control animals. However, when macrophages were separated from peritoneal exudates, candidacidal activity was not dependent on reactive oxygen species (280) but was due to a serine protease released by the chitin-treated macrophages. Candidacidal activity was present in macrophage culture supernatant fluid and was inhibited by actinomycin D and diisopropylfluorophosphate, which inhibit protein synthesis and serine protease activity, respectively. One potential confounding factor of these studies is that the authors imply that the control animals were simply untreated rather than receiving sham injections. Thus, it is possible that some of the response is due to the act of injection rather than to receiving chitin specifically. However, chitosan, a deacetylated chitin derivative, was also tested for immunological effects (279, 280), and although chitosan-treated mice also had increased levels of PMN and macrophages, the distribution of these cells (circulation versus peritoneal exudate) was different. Thus, it seems less likely that the immunopotentiating effects of chitin were due entirely to the injection itself.
The effect of fungal chitin (as opposed to chitin extracted from crustacean shells) on immune system components is described in a single study. Rementería et al. (233) extended the earlier studies to examine chitin purified from the C. albicans cell wall. They reported that, as observed by Suzuki et al. (279), injection of mice with purified C. albicans chitin (30 mg kg1) increased the survival time in response to a C. albicans challenge of 105 cells. In these experiments, it was explicitly stated that control mice were given an injection of saline to control for the stress of the injection itself. As a further refinement, the authors quantified the chitin present in the organs of experimentally infected animals. Chitin treatments were then based on these measured values.
To begin to look at more specific reasons for this increased survival, peripheral macrophages were isolated over the course of a 20-day study period (233). Macrophages isolated from mice treated with chitin and macrophages isolated from mice treated and then infected with C. albicans both showed increases in phagocytosis and candidacidal activity over those for macrophages isolated from control animals. However, the peak of phagocytosis, measured as (number of macrophages ingesting at least one yeast cell/total number of macrophages) x 100, did not coincide with the peak of cytotoxicity, measured as percentage of killed yeast cells in samples percentage of yeast cells killed in controls lacking macrophages. Nitric oxide production was increased only in macrophages from mice receiving both chitin treatment and C. albicans infection. Because peak nitric oxide production preceded peak cytotoxicity, the authors concluded, as did Suzuki et al. above (280), that macrophage cytotoxicity in these animals was not occurring through an oxidative burst. In their study, Rementería et al. suggested that a lower dose of chitin (10 mg kg1) also produced immunosuppressive effects. However, these data, while suggestive, are not as convincing as those from the experiment with the higher dose.
(iii) Glucan. Studies that examine the interaction between fungal glucan and immune cells can be separated into two groups: those that use soluble glucan and those that use insoluble or particulate glucan. Szabó et al. used soluble glucan (prepared by sonication of S. cerevisiae glucan particles) to characterize the ligand binding domain of a previously identified ß-glucan receptor on mononuclear phagocytes (61, 130, 284). These receptors were shown to also bind C. neoformans glucan (56). In a study mentioned previously, Fradin et al. (82) found that laminarin, a polymer of ß-1,3-linked glucose units with ß-1,6-glucan branches, inhibited the binding of heat-killed, but not live, C. albicans cells to macrophage cells. The authors speculated that exposing the cells to heat released cell surface components and exposed the deeper glucan layers. This supports the evidence that macrophages contain proteins that bind glucan in addition to the mannose receptor. It further suggests that the mannose receptor is not sterically blocked when glucan is bound or that binding of mannan is able to displace bound glucan.
Tokunaka et al. (290) examined the immunopharmacologic activity of a soluble glucan preparation from the C. albicans cell wall (Candida soluble beta glucan [CSBG]). They observed that CSBG increased the hematopoietic response to cyclophosphamide-induced leukopenia, increased vascular permeability, and acted as an adjuvant for the purposes of antibody production. They also observed several effects with respect to cytokine production (discussed below). Kataoka et al. (154) looked at macrophage activation by several types of soluble ß-1,3-glucans by using an NF-
B-luciferase reporter assay. Activation culminated in the induction of proinflammatory mediators, inducible nitric oxide synthase (iNOS), tumor necrosis factor alpha (TNF-
), and macrophage inflammatory protein 2 (MIP-2). The authors found that a linear ß-1,3-glucan, such as curdlan, gave a stronger response than branched glucans, such as laminarin. They also reported that ß-1,3-oligoglucosides were ineffective at macrophage activational, although a ß-1,3-glucoheptaose was able to inhibit curdlan activation of macrophages. Kikuchi et al. (159), also using the CSBG preparation, showed that ß-glucan augmented or induced maturation of DC. Xia et al. (315) demonstrated that, as with human leukocytes, soluble and particulate ß-glucan binds to mouse leukocytes via CR3 (CD11b/CD18,
Mß2-integrin). Binding of ß-glucan to natural killer (NK) cells, macrophages and the monocytic cell line P388D1 was inhibited by other ß-glucans but not yeast
-mannan. Further, glucan binding was inhibited by anti-CR3 antibody and did not occur on cells from CR3-deficient mice. Earlier work by this group demonstrated that binding of soluble glucan to CR3 enabled NK cells and neutrophils to lyse iC3b-opsonized SRBC (303). The authors proposed that binding of soluble ß-glucan to these cells primed them for cytotoxicity against iC3b-opsonized cells, even when the opsonized cells are normally resistant to CR3-dependent killing.
Particulate glucan (e.g., in the form of zymosan or unopsonized yeast cells) binds to and is phagocytosed by neutrophils and monocytes in the absence of opsonins (60, 238). Ross and coworkers further demonstrated that binding of particulate, although not soluble, ß-glucan induced superoxide bursts from neutrophils (239, 303). Binding of particulate ß-glucan also led to upregulation of CR3 expression on the surface of NK and cytotoxic T cells (198). Czop and Austen (60) showed that binding of particulate ß-glucan to monocytes induced leukotriene production. This action is inhibited by pretreatment of the monocytes with soluble ß-glucan but not
-mannan. In an earlier paper, Glovsky et al. (90) reported that particulate glucan affected human serum complement activity, mainly by decreasing the C2 and C3 activity. Lesser effects on C1 and C5 activity were also seen. This coincided with a release of C3a from normal human serum. In vivo experiments showed that particulate glucan activates guinea pig complement via the alternative pathway. Ishibashi et al. (120) compared soluble and particulate forms of ß-glucan having the same primary structure. The particulate glucan had a significantly greater effect than did the solubilized glucan on the activation of various cells, as measured by production of IL-8 by human peripheral blood mononuclear cells (PBMC), TNF-
production by a murine macrophage cell line (RAW 264.7), and peroxide production in ICR mouse peritoneal exudate. These results indicate that immune cell activation is a function of both solubility and particle size (termed degree of assembly) (120). Vassallo et al. (302), working with particulate ß-glucan from Pneumocystis carinii, demonstrated that dose is also a factor in the immune cell response (see below).
Alkali-insoluble polysaccharide from Paracoccidioides brasiliensis, containing ß-glucan, also activated PBMC isolated from healthy volunteer donors (7). Incubation of PBMC in the presence of this cell wall fraction (F1) induced production of peroxide similar to that induced by gamma interferon (IFN-
). Incubation of PBMC with IFN-
and F1 increased fungicidal activity and TNF-
production over the levels produced by either component alone.
Glycan influence on cytokine production. (i) Mannan.
Several studies implicated mannan in the modulation of cytokines. Castro et al. (45) reported that mannoprotein from C. albicans, but not S. cerevisiae, inhibited arachidonic acid release by alveolar macrophages. In addition, pretreatment with mannoprotein inhibited the phagocytosis of C. albicans cells. Garner and Hudson (86) reported that injection of mice with mannoprotein induced TNF-
production. Wang et al. (310) injected mannoprotein into mice that were sacrificed over time to determine the mRNA levels of various cytokines. They observed no apparent induction of IL-2 but did find some increase in IL-4 and IFN-
levels between 48 and 96 h. The IL-10 signal was strong throughout the assay period (2 to 96 h), while IL-12p40 levels were reduced in the first 12 h but returned to normal levels by 48 h.
All of these studies were carried out using preparations or extracts that release mannoproteins, not isolated polysaccharide. It is not clear, therefore, if the active portion is the glycan or the protein. More recent studies have identified specific glycan components that are involved in cytokine interaction and induction. Garner et al. (87) observed that mannan induced TNF-
production by mouse alveolar macrophages. They implicated the carbohydrate portion specifically based on two results. First, pretreatment of the mannoprotein with Con A (a lectin that binds terminal mannose, as well as glucose, units) inhibited TNF-
induction. Second, pretreatment of the macrophages with mannose or
-methlymannoside also inhibited induction. These two results indicated the presence of a mannose receptor or mannose binding protein on macrophages. They further showed that there was some species specificity involved with regard to the ability of mannoprotein to induce TNF-
secretion. Mannoprotein from S. cerevisiae had no effect on TNF-
secretion. Mannoprotein from C. tropicalis induced TNF-
secretion, but the effect was about one-fifth the effect produced by C. albicans mannoprotein. N-linked glycans from C. tropicalis mannoproteins contain ß-1,2-oligomannosides both in the acid-stable portion and as acid-labile groups, although the latter do not get as long as those from C. albicans (164). The comparison of the three fungal species suggests that the active mannan is a ß-1,2-oligomannoside because these groups are absent from S. cerevisiae glycans.
Jouault et al. (143) extended the results of Garner et al. and showed that individual oligosaccharides in the acid-labile N-linked glycan groups (specifically ß-1,2-Man4, Man5, and Man8) induced TNF-
production by murine alveolar macrophages. Furthermore, TNF-
production following macrophage stimulation by the octaose was nearly fivefold higher than that due to stimulation by the tetraose or pentaose. TNF-
production in response to the heptaose, although the heptaose differs from the octaose by only a single mannose, was virtually undetectable. As the authors note in their discussion, this suggests a structural specificity to mannan binding (143). The relative activity of the tetra- and pentaose also suggests that the ß-1,2-oligomannoside has a repetitive characteristic. Four or five mannose units may form the base of the structure recognized by the macrophage receptor, but the additional four units, perhaps completing another helical turn, are needed to form the complete ligand.
A subsequent study showed that this stimulation had the effect of desensitizing the macrophages, causing an inhibition of TNF-
release after a second stimulation (142). The results from these two groups support, and are supported by, those from Cutler's group demonstrating that the C. albicans macrophage adhesin is a ß-1,2-mannotetraose (see the discussion above). This does not yet, however, provide a mechanism for the observation of Garner et al. (87) that TNF-
induction by Candida mannan is not a simple on/off switch. One possibility is that cross-linking of the mannose receptors is required for full induction and that C. tropicalis glycosylation is not extensive enough to provide maximal cross-linking. Another possibility is that some other element required for full induction is present in C. albicans but absent in C. tropicalis (87).
Studies by Zanetta and coworkers (46, 323) demonstrated that IL-2 directly interacts with C. albicans mannoprotein in a lectin-like manner. Binding was inhibited by the Man5 and Man6 cores, but not the Man9 glycan core (Fig. 5), leading the authors to speculate that the IL-2 carbohydrate recognition domain requires an unsubstituted Man(
1-6). Based on these and previous results, this group proposed a model that states that mannan on the Candida cell wall, or larger mannan pieces released from the cells into the surrounding-medium, binds IL-2 and prevents IL-2 signaling in the host. Additionally, IL-2 may thus act as a bridging unit, allowing Candida cells to bind T cells, possibly acting as the mannan binding receptor described by Durandy et al. (73).
![]() View larger version (13K): [in a new window] |
FIG. 5. N-linked glycan core structures used as inhibitors of IL-2 binding to C. albicans mannan (46, 323). Inset: Key for linkage positions as in Fig. 2.
|
B. They further showed that activation does not go through the classical lipopolysaccharide (LPS) receptor, TLR4. However, some Toll receptor is likely to be involved since depletion of the Toll adapter protein, MyD88, severely attenuated macrophage response.
Tokunaka et al. (290) published the results of immunological studies performed with a preparation of CSBG. Their results led to several conclusions: (i) ß-Glucan, in a dose-related manner, augmented IL-6 synthesis in IFN-
-stimulated peritoneal macrophages. (ii) CSBG competes with zymosan for sites on macrophages. This conclusion was based on the observation that addition of CSBG to the reaction decreased TNF-
production in response to zymosan. However, since CSBG did not itself stimulate TNF-
production, it is difficult to say if CSBG is truly acting as a competitive inhibitor of zymosan or acting through some other mechanism. (iii) LPS-induced TNF-
and nitric oxide production in RAW264.7 cells was enhanced by the presence of glucan. The authors conclude here that glucan was working thorough an additional or independent signal mechanism. However, again, since CSBG alone had no effect, it is difficult to see what CSBG-specific signal mechanism this could be.
Olson et al. (212) reported that while low and intermediate concentrations of glucan induce TNF-
production, higher concentrations suppress it. In a similar fashion, P. carinii ß-glucan induced TNF-
release by rat alveolar macrophages. This induction peaked at 2.5 x 107 particles ml1 (302). Higher concentrations of glucan resulted in diminished TNF-
release until the baseline was reached at 1x 108 particles ml1. The paper by Olson et al. was an extension of an earlier paper by Hoffman et al. (111), who showed that ß-glucan suppressed TNF-
release by rat alveolar macrophages stimulated with LPS. Olson et al. suggested that these observations might help explain some of the earlier, seemingly contradictory results reported in the literature where, under certain experimental conditions, C. albicans cell fractions produced a stimulatory effect in the immune system while others produced immune system suppression. This might also explain how the organism produces a seemingly self-harming interaction with the immune system. Another observation from this work is that TNF-
seems to bind glucan directly. This binding was inhibited by diacetylchitobiose (ß-1,4-GlcNAc2) and glucose but not cellobiose (ß-1,4-Glc2) or fucose. Olson et al. suggested that at low concentrations all the glucan was phagocytosed but at the higher concentrations glucan saturated the system, so that the remaining glucan could bind up circulating TNF-
, thus preventing its activity. TNF-
release from murine macrophages was also induced by ß-glucan from Paracoccidioides brasiliensis but not by
-glucan, chitin, or galactomannan fractions (79). The level of induction did not reach that due to LPS but was at least fourfold higher than the control level. Czop and Austen (60) reported that particulate, as well as soluble, C. albicans ß-glucan was phagocytosed by monocytes. They observed that this action stimulated oxidative generation of leukotrienes (LTB4 and LTC4). LTB4 acts as a chemotactic factor, while LTC4 is a permeability-enhancing factor.
Recognition by antibodies. The first separation of C. albicans strains into serotypes A and B was subsequently attributed to recognition by different antisera to specific mannosyl side chains on the mannoproteins (44, 105). Another illustration of mannan antigenicity is that an endotoxin-free C. albicans mannan preparation, when used to treat SRBC, significantly activates a primary antibody response to those SRBC (44). Faux et al. (78) described an enzyme-linked immunosorbent assay (ELISA) using specific antibodies (immunoglobulin G [IgG]) to Candida mannan. Although the epitope was not specified, the presumption by the authors is that a glycan group is recognized. The authors used this assay to determine the normal ranges of serum mannan in a population. Mannan/mannoprotein-specific IgG levels increased with age, and patients with primary antibody deficiency had uniformly low responses.
Anti-mannoprotein antibodies have been of particular interest in the field of atopic dermatitis. IgE-mediated hypersensitivity to mannoprotein (presumed to be due to the polysaccharide part) was thought to be associated with atopic dermatitis and respiratory allergies. Doekes et al. (69) reported that a large proportion of atopic dermatitis patients had IgE specific for a glycan epitope on Pityrosporon ovale (now a Malassezia species), another commensal yeast. They concluded that the epitope was a glycan because the IgE bound to cell extract fractions that were both high molecular weight by size exclusion chromatography and bound to the lectin Con A by affinity chromatography. Further, pretreatment of the extract with periodate eliminated its ability to prevent IgE from binding to extract adhered to the bottom of an ELISA well. Nermes et al. (203) looked at an ELISA for anti-mannoprotein IgA, IgG, and IgM, and compared it to a radioallergosorbant test for anti-mannoprotein IgE. Savolainen et al. (249) reported higher specific IgE responses to C. albicans mannoproteins and cell lysate proteins in atopic dermatitis patients than in healthy controls. No difference existed between the two groups in IgG response, although IgE and IgG were measured using different protocols. They also reported that both mannoprotein and cell lysates from Candida strains isolated from atopic dermatitis patients produced higher levels of lymphoproliferation induction than did those from healthy controls. However, as noted with other studies, the "mannan" used in the last two studies was a mannoprotein cell wall preparation, with no indication of separation of glycan from protein.
Although glucan and chitin in the cell wall of C. albicans act as immunomodulators, they have low antigenicity (44, 189). Uchiyama et al. also recently noted that there have been few reports of anti-glucan monoclonal antibodies, though they themselves describe an antibody raised against the CSBG (299). The lack of anti-ß-1,3-glucan antibodies was taken by Ballou to indicate that glucan is not exposed on the cell surface (13). Thus, immunological studies have tended to focus on the mannoprotein component.
|
|
|---|
Subsequent analyses of mannan epitopes have led to production of more specific antisera. A collection of such antisera (recognizing epitopes or factors designated 1, 4, 5, 6, 8, 9, 11, 13, 13b, and 34) is commercially available as the Candida Check kit (Iatron), which allows species identification of clinical isolates based on the combination of positive agglutination reactions (268). Shibata et al. (262), while defining antigenic factors 5 and 6, identified the specific structural differences between serotypes A and B (reviewed in reference 243). They also provided the first evidence for ß-1,2-linked mannose in the acid-stable portion of serotype A strain mannan. Lectins have been proposed as an alternative tool to these factor antisera for species identification. For example, Muñoz et al. (197) recently described using a panel of 14 commercially available lectins to produce agglutination profiles for 93 clinical Candida isolates. Only in the case of C. glabrata and C. tropicalis did all isolates (8 and 12, respectively) fall into a single pattern, although 90% of the C. albicans isolates gave the same profile. Further, 25% of the C. parapsilosis isolates gave the same profile as the C. glabrata isolates and another 37.5% produced the same pattern as the one most commonly produced by C. albicans isolates. Nevertheless, the results from this lectin panel suggest that it would be possible to discriminate C. albicans, C. glabrata, C. tropicalis, and C. guilliermondii. Thus, while this type of assay may not stand on its own, the use of lectins as agglutinating reagents can be useful as a confirmatory test or in conjunction with other types of diagnostic assays.
Diagnosis of disseminated candidiasis has proved particularly problematic because its clinical signs and symptoms are nonspecific. Also, blood culture isolation of the organism as an assay is low (50%) in specificity (208). Therefore, there has long been interest in a diagnostic assay for disseminated candidiasis. Many of the assays developed, especially early systems, were designed to detect C. albicans antigens in patient serum. In 1976, Weiner and Yount (313) developed a hemagglutination inhibition assay for Candida antigen detection (antigenemia), using mannan as the detected group. Antisera specific for mannan were first identified by passive hemagglutination of SRBC coated with a C. albicans mannoprotein fraction prepared from cell wall extracts. Because the protein level was low (3% [dry weight] protein), the authors concluded that the identified sera recognized polysaccharide. The presence of mannan in test serum was determined by inhibition of the agglutination reaction, as the serum mannan would compete with mannan coating the RBCs for antiserum binding sites. The authors note, however, that the success of this assay depends on the absence of any antimannan antibodies in the test serum. Weiner and Coats-Stephen (312) later developed a radioimmunoassay for mannan antigenemia detection, which they reported as being more sensitive. Serum mannan would compete with iodinated reagent mannan for antiserum binding sites, leading to lower counts on precipitation of the antigen-antibody complex. The method of iodination utilized chloramine-T, which incorporates 125I into protein through a covalent linkage of the iodinated label to amine groups. Although both assays effectively detected Candida-specific antigen, it can be seen that recognition of a protein, rather than a glycan, epitope cannot be completely ruled out.
Since they were originally developed, improvements have been made to antigenemia detection assays to make them both standardized and highly specific. There remains, however, a lack of the sensitivity necessary for early detection of systemic candidiasis. Sendid et al. (257) have noted that this lack of sensitivity is probably due to rapid clearance of antigen as well as the latex agglutination format typically used. Detection of antimannan antibodies in patient sera has also been studied as a diagnostic strategy and, in fact, was proposed before the antigenemia assays. However, in a 1992 review, Buckley et al. (25) pointed out that an inherent deficiency of immunodiagnostics designed to detect antibodies is that these assays will be positive only if the patient/host recognizes the invader and mounts a response. In addition, the specificity of these tests is diminished because false-positive results are likely to arise in patients who are colonized but do not have clinical candidiasis (34, 208, 257).
Poulain and coworkers sought to overcome the individual problems of antigen- and antibody-based tests and to develop a diagnostic test based on the detection of both mannan antigen and patient-derived antimannan antibodies. The test that they developed used a monoclonal antibody that recognizes an
-linked mannopentaose from the acid-stable mannan portion of N-linked glycans (PLATELIA CANDIDA Ag EIA; Bio-Rad) (128, 257, 318). The assay was subsequently tested for diagnosis of candidemia caused by species other than C. albicans (256). This strategy worked for C. albicans, C. glabrata, and C. tropicalis but did not work as well with C. parapsilosis, C. kefyr, or C. krusei. The differences in sensitivity were attributed to the particular antibody (EBCA1) used in the mannan detection test, which recognizes a mannose epitope present to a much lesser extent in mannan from the last three species (128). Because the incidence of invasive candidiasis due to non-albicans Candida species is increasing, Rimek et al. (235) recently tested the PLATELIA CANDIDA assay for cross-reactivity to antigens from 54 fungal species. They found that only Candida species gave positive results, although one strain of Geotrichum candidum and one strain of Fusarium verticilliodes gave borderline results. However, as was seen by Sendid et al. (256), C. kefyr, C. krusei, and C. parapsilosis tested negative in this assay.
Antigen detection has also been the focus of diagnostic tests for aspergillosis. The Aspergillus cell wall contains a neutral polysaccharide antigen, galactomannan (72). This polysaccharide comprises a branched core of
-1,2- and
-1,6-mannan with ß-1,4-oligogalactosyl side chains. One characteristic worth noting is that the side chain termini, as well as some of the internal units, are in the galactofuranosyl (five-member ring) rather than the galactopyranosyl (six-member ring [Fig. 1]) form (175). A study of experimentally infected rabbits and patients diagnosed with invasive aspergillosis suggested that antigen detection in urine produces higher sensitivity than detection in serum (72).
A different approach to using C. albicans mannan as a diagnostic was presented by Goins and Cutler (92), who released acid-labile and O-linked oligomannosides from intact cells. The released oligomannosides were then labeled with a fluorophore and electrophoretically separated. This treatment of intact cells released oligomannosides other than those from purified mannan. The authors suggest that these groups may come from other sources in the cell wall rather than being part of mannan purified by Fehling's reagent (92). Nevertheless, electrophoretic profiles of oligosaccharides released by either acid-hydrolysis or beta-elimination demonstrated species and strain differences, suggesting that this might prove to be a useful approach to species and strain identification.
Glucan assays. Candida glucan is immunologically less active than mannan (44, 170, 189). Nevertheless, direct measurement of glucan has been described in an assay for candidemia. Miyazaki et al. (196) reported the development of a diagnostic test, called the G Test, which detects ß-1,3-glucan. This test is based on the Limulus test originally developed for the detection of endotoxin (LPS). Comparison of this new G Test to other Limulus-based tests and mannan detection tests indicated that the G Test was more sensitive than the other assays tested. All rabbits infected in a systemic candidiasis model tested positive by day 2 after inoculation. A later study by Nakao et al. (201) determined that the amount of detected glucan was directly proportional to organism load. One complicating factor noted by Miyazaki et al. was that patients treated with anti-tumor polysaccharides could give false G Test positives. Thus, they recommend the combined use of the G Test and a mannan test for maximum sensitivity and specificity.
Several papers, primarily from Cutler and coworkers, describe Candida oligomannosides or antibodies raised against them as being useful therapeutic agents. This group reported the production and testing of monoclonal antibodies against fractions of C. albicans mannoprotein. One in particular, an IgM designated B6.1, protected mice from disseminated candidiasis, while another IgM, B6, did not (100, 101). B6.1 recognizes a ß-1,2-mannotriose (101). B6 recognizes a portion of the acid-stable mannan whose exact structure has not been completely characterized. B6.1, administered either intravaginally or intraperitoneally, also protected mice against vaginal candidiasis. This effect occurred whether the antibody was given prior to or after infection. B6 was protective, although to a lesser degree than B6.1, when administered before infection but had no protective effects when given after infection. The authors speculate that the more acidic environment in the vagina results in changes in glycan structure (162, 169) that allow greater exposure of the B6 epitope in the vaginal mucosa than in the bloodstream. B6.1, but not B6, was also protective against vaginal infection with C. tropicalis. Protection by B6.1 was expected because C. tropicalis N-linked glycans also contain ß-1,2-mannotriose groups in both the acid-labile and acid-stable mannan fractions (164). However, B6 does react with C. tropicalis, and so the reason why this antibody was not protective against vaginal candidiasis remains to be discovered.
A paper by Caesar-TonThat and Cutler (30) suggested that the protective nature of B6.1 in the case of disseminated candidiasis was due to its enhancement of phagocytosis and cytotoxicity of PMN in the presence of complement. Subsequent studies (324) showed that mannoprotein adsorbed with normal human serum delayed the accumulation of C3. Since this effect could be reversed by B6 or B6.1, they concluded that the glycan portions of mannoprotein fractions were involved in enhancement of the classical complement pathway. Han et al. (102) later presented data showing that an IgG3, recognizing the same epitope as B6.1, was also a protective antibody. Antibody isotype, in addition to epitope specificity, was first shown to be important in the protective efficacy of anti-Cryptococcus antibodies by Sanford et al. (246). In other studies of experimental cryptococcosis, a nonprotective IgG3 antibody became protective after a switch to IgG1 (320, 321). The protective nature of the IgG3 antibody is, however, consistent with the earlier findings and hypothesis that complement activation is involved in resistance to disseminated candidiasis (102). Han et al. (103) further showed that mannan, conjugated to bovine serum albumin, was effective as a vaccine against subsequent Candida infections. More recently, Bystricky et al. (28) reported similar results using a mannan-human serum albumin conjugate.
An extension of this work was presented by Glee et al. (89). They identified a series of peptides from a peptide display library that bound to IgM antibodies such as B6.1. It was thought that these peptides were acting as mimetics to the carbohydrate antigen. The results of this study, however, led the authors to conclude that these particular peptides were binding at a site other than the traditional antigen binding site. Each peptide was able to bind only IgM (no other class) but could bind all IgM antibodies tested, regardless of species or epitope. Thus, the binding site conformation and specificity must have been conserved among the antibodies. Nevertheless, the authors point out that these peptides may have therapeutic value as affinity matrices for IgM purification, as B cell activators, or to prevent unwanted or disadvantageous antigen-antibody interactions.
As described above, Suzuki et al. (279) demonstrated that mice treated with chitin from a commercial source (unspecified, but probably extracted from crustacean shells) had increased survival in response to a lethal challenge with C. albicans than did untreated mice. Rementería et al. (233) further showed that inoculation of mice with purified C. albicans chitin led to increased survival in response to a subsequent fungal challenge. Glucans extracted from mushrooms have been used as tumor immunotherapeutic agents for some time (see references in reference 303). Two recent papers discuss the use of C. albicans glucan as a therapeutic agent. Tokunaka et al. (289) demonstrated the use of glucan itself as an antitumor therapy. Soluble glucan (CSBG) inhibited the growth of P815 mastocytoma tumors and increased the survival of mice with implanted tumors. Ross and coworkers have investigated the mechanism by which these protective effects might occur. As mentioned above, glucan binds to CR3, which results in a magnesium- and protein tyrosine kinase-dependent conformation change in the receptor (303). This, then, without directly causing cellular activation, primes the cells to be cytotoxic toward C3-opsonized cells. Subsequent work showed that ß-glucan-primed cells were effective against opsonized mammary carcinoma cell lines (304). Furthermore, primed cells showed 32 to 54% cytotoxicity against freshly excised mammary tumor cells, indicating that at least a subpopulation of tumor cells are opsonized in vivo but that ß-glucan binding is required for NK cell recognition and killing of the opsonized cells (304). In the second study, Bromuro et al. (24) reported the presence of antiglucan antibodies. They also showed that some components of these sera aggravate infection while others inhibit infection. This led the authors to speculate that humoral protection requires not only the presence of the right antibodies but also the absence of other antibodies.
|
|
|---|
One early report describing improvements to glycan separations was presented by Honda et al. (114), who took advantage of the ability of borate to form complexes with the cis-diols on sugars mentioned above. The resulting anionic borate complexes not only have increased UV absorbance but also have mobility in an electric field, allowing separation by capillary zone electrophoresis or other electrophoretic methods. Other complexes were later formed by direct conjugation of groups to the oligosaccharides (Table 5). These conjugation reactions are typically carried out using reductive amination, in which an amine group on the molecule to be added reacts with the aldehyde at the oligosaccharide reducing end (C-1). There are two main advantages to this approach. First, reductive amination reactions can be done under relatively mild conditions (e.g., dilute acetic or citric acid at 37°C for 18 h), minimizing potential hydrolysis or other unwanted reactions. Second, because each oligosaccharide contains only one reducing end, the labeling reaction is stoichiometric at one label per oligosaccharide. As a result, depending on the specific detecting instrument, labeled oligosaccharides can be detected quantitatively.
|
View this table: [in a new window] |
TABLE 5. Selected reports of glycan labeling, detection and separation
|
Detection sensitivity was increased severalfold by the introduction of fluorescent labels. Several compounds with the ability to add fluorescence have been used to conjugate oligosaccharides. These include 2-aminopyridine (2-AP) (4, 77), 2-aminobenzamide (223, 267), anthranilic acid (8), 2-aminoacridone (AMAC) (126), 8-aminonaphthalene-1,3,6-trisulfonic acid (ANTS) (53, 125), and 1-aminopyrene-3,6,8-trisulfonic acid (APTS) (51, 97). ANTS, APTS, and similar compounds also contain strongly acidic functional groups that dissociate at low pH, thus intrinsically providing the capability of electrophoretic separation. One advantageous characteristic of APTS is that the unconjugated molecule has a much lower fluorescence than do APTS conjugates. While not affecting UV detection, the noise from the underivatized label is greatly reduced when using fluorescent detection (228).
In 1990, Jackson (125) was interested in finding a highly sensitive (subpicomolar range) alternative to radiolabeling. He labeled oligosaccharides with ANTS and carried out electrophoretic separations using high-percentage (20 to 40% gradient) polyacrylamide gels. This system is referred to as Fluorophore-Assisted Carbohydrate Electrophoresis (FACE). The method showed higher sensitivity than the paper chromatography, thin-layer chromatography, and noncapillary electrophoretic methods used previously. The basic FACE method was later combined with enzymatic digestions for more complete oligosaccharide characterization (127). The method was subsequently modified to use AMAC as the label. Since AMAC is not charged, only charged oligosaccharides maintain electrophoretic mobility in the absence of a borate buffer. This modification can be used as an additional tool for characterizing glycan structure, especially if there is evidence that some of the sugar residues are charged (e.g., sialic acids). The labeling reagents, precast gels, and gel buffers for carrying out FACE had, until recently, been commercially available. Gao and Lehrman (85) recently published a methods paper directly addressing this issue. Although perhaps less convenient, it is still possible to obtain all necessary reagents from other sources. Furthermore, FACE gels poured in-house are just as effective as precast gels, although more difficult technically to make than traditional protein PAGE gels.
Chiesa and Horváth (53) used capillary zone electrophoresis (CZE) for separation of ANTS-derivatized maltooligosaccharides (
-1,4-linked oligoglucosides). Detection was by either UV or laser-induced fluorescence. Alpert et al. (4), in a study of complex-type glycans, labeled them with 2-aminopyridine (2-AP) and carried out separations using hydrophilic interaction liquid chromatography (HILIC). These authors concluded that HILIC was the better chromatographic choice for combination with mass spectrometry (than, for example, high-pH anion-exchange chromatography) because volatile mobile phases can be used. They also showed that HILIC had better selectivity than reverse-phase high-performance liquid chromatography (HPLC). Chen and Evangelista (51) were able to resolve structural isomers by labeling with APTS and separating by capillary electrophoresis (CE). Based on their analyses, they concluded that the mechanism of separation they observed was different from that for separation in borate or high-pH buffers and was based on apparent hydrodynamic volumes. Guttman and Pritchett (97) used capillary gel electrophoresis to separate APTS-labeled high-mannose glycans isolated from glycoproteins. This method is distinguished from CZE in that the capillary is filled with a solid-phase gel, much like a chromatography column.
Plocek and Novotny (223) addressed some practical aspects of glycan separations. They noted that UV-absorbing tags are less expensive than fluorescent tags and that CE instruments with UV detectors are readily available. This led them to develop a method for labeling oligosaccharides with aminobenzoyl-glutamic acid in preparation for separation by CE. Most recently, Shimizu et al. (267) described the derivatization of glycoprotein hydrazinolysates with p-aminobenzamide and fractionation by cellulose chromatography. Raju recently reviewed electrophoretic separations of carbohydrates in more detail (228). El Rassi and Mechref have reviewed specific capillary electrophoretic applications (75).
Glycan labeling overcomes the limited UV absorbance and lack of color inherent to carbohydrates and thus provides higher sensitivity for detection. The labeling methods described above have the further advantage of using mild reaction conditions and do not require high temperature or pressure. The choice of label can also provide insight into glycan structure. Labels containing ionizable groups (e.g., ANTS) are often used to provide charge for mobility during electrophoretic separations. However, the presence of a charged group (e.g., sialic acids) can be confirmed by maintaining mobility using a neutral label (e.g., AMAC and 2-AP).
or ß conformation (Fig. 1 and Fig. 2 inset). This results in a much higher complexity per number of residues for oligosaccharides than for peptides or oligonucleotides (182, 258). A collection of four different nucleotides can combine to form 24 unique oligomers, but four different monosaccharides can combine to form 35,560 unique oligosaccharides (258). Additional complexity arises when describing the glycosylation of proteins. As glycans are added and modified during procession through the endoplasmic reticulum and Golgi, the glycan chains are elongated and terminated based on a number of factors, including transferase accessibility to the growing oligosaccharide chain and competition for substrate (14, 64). As a result, a single glycoprotein is present as several, or many, glycoforms (182). Even proteins with a single glycosylation site, isolated from a homogeneous cell population, contain glycans with several structures (95, 287). This increases the amount of structural information available to the cell in terms of sequence variability and ligand specificity (e.g., for cell recognition sites) but complicates structural studies and protein identification.
Many tools are available for the characterization of glycans, beginning with separation of the glycans from glycoproteins by chemical and enzymatic means. N-linked glycans have been chemically released by using hydrazinolysis (215) and TFMS (trifluoromethanesulfonic acid) (74). Until recently, there was at least one instrument designed to automate the deglycosylation of protein fractions by hydrazinolysis (GlycoPrep), but it is no longer commercially available or supported. One disadvantage of hydrazinolysis is that the protein component is destroyed during the glycan release reaction because hydrazine cleaves peptide bonds. On the other hand, deglycosylation with TFMS maintains the integrity of the aglycone (deglycosylated glycoconjugate) but not the integrity of the glycan (18, 74). O-linked glycans are typically released by treatment with mild alkali (beta-elimination), incubating the glycoprotein in 0.05 M NaOH at 23°C for 18 h (92, 116).
Several endoglycosidases are also available that will release glycans from their protein backbones. N-linked glycans can be released with peptide-N4-(N-acetylglucosaminyl)asparagine amidase (PNGase F), Endo H, or N-glycosidase F (Endo F), among others (see reference 286 for a review). These enzymes are commercially available from several suppliers and have the advantage of being able to isolate the aglycone after deglycosylation. Care needs to be taken when choosing an enzyme because of cleavage site specificity. For example, Endo F does not cleave oligosaccharides containing an
-1,3-fucose linked to the oligosaccharide core, such as are found in many plants. N-Glycosidase A removes glycans containing this group. Several endo-ß-N-acetylglucosaminidases have been described which cleave between the two GlcNAc residues of the core oligosaccharide. However, their substrate specificities limit their usefulness in the study of hybrid-type or high-mannose-type (the type found in Candida and Saccharomyces) glycans. PNGase F cleaves the bond between the asparagine residue and the first GlcNAc and so is an amidase rather than an endoglycosidase per se. Nevertheless, because of this, it can act on a wide variety of glycan structures. O-linked glycans from higher eukaryotes can be released with O-glycanase. Unfortunately, this enzyme is not useful for the analysis of Candida or S. cerevisiae glycoproteins because the fungal O-linked glycans are composed solely of mannose and O-glycanase cleaves between the serine/threonine and the attached Galß1
3GalNAc.
The glycans themselves (glucan and mannan/mannoprotein) can be extracted from cell walls by autoclaving the cell mass in citrate buffer (217) or even distilled water. Mannan, in the form of mannoprotein, can be separated from glucan by subsequent differential precipitation with detergent (Cetavlon) and alkaline borate (15, 210). Extracted and purified glycans can then be separated into fractions by chemical and enzymatic means.
One of the most common approaches to glycan fractionation is the use of acid hydrolysis. Mild-acid hydrolysis, with 0.1 N HCl in a boiling-water bath for 1 h, breaks the phosphodiester bond and releases the oligosaccharides linked through this mechanism (giving rise to the term "acid-labile," mentioned above). Stronger conditions can be used to hydrolyze carbohydrate chains down to their monosaccharide components. However, susceptibility and rates of cleavage of glycosidic bonds depend on position and anomeric configuration (216). Hence, the the conditions of hydrolysis for a given glycan must often be empirically determined. Another caveat to acid hydrolysis is the process of reversion, the acid-catalyzed polymerization of monosaccharides. This is particularly important when preparing polysaccharide acid hydrolysates for ladders or standards. Under the wrong conditions, the polysaccharide could hydrolyze but then revert, making unexpected oligosaccharides from monosaccharides or combinations of oligosaccharides and monosaccharides (157).
Another common chemical approach to glycan fractionation is acetolysis. This reaction has proven useful because under controlled conditions, the
-1,6-linkages are preferentially broken. As a result, the
-1,6-links between mannose residues in the outer-chain backbone can be cleaved, releasing the side branches intact for further structural studies (99, 281). The process of acetolysis begins with acetylation of all free alcohol groups to prevent any modification or derivatization. Sulfuric acid is then used to cleave the glycosidic bonds under conditions where the
-1,6-links are preferentially cleaved. The alcohol groups are then reverted to their previous form using alkaline sodium methoxide.
Glycans can also be fractionated by using a variety of commercially available exoglycosidases (e.g., glucosidases, mannosidases, and sialidases) with various substrate specificities and from a variety of natural sources (generally microbial or plant). However, several of the more commonly used enzymes are relatively inactive toward yeast mannan (138, 180). This difficulty prompted Ballou and colleagues to search for novel enzymes that were active on S. cerevisiae mannan. Several enzymes were isolated from soil bacteria, but they have not yet become commercially available (137-139, 199). One, however, an
-1,6-endo-D-mannanase from Bacillus circulans, has recently been cloned by Maruyama and Nakjima (190).
Exoglycosidases play an important role in strategies for determining oligosaccharide sequence (240, 241). In general, glycans are removed from their aglycones and then separated into fractions containing unique glycans. These fractions are then subjected to a battery or array of exoglycosidases for analysis. For example, if a particular exoglycosidase is capable of removing the end unit, then the treated glycan will be smaller than the parent. Other modifications or refinements to the basic scheme include in-gel deglycosylation of proteins separated by PAGE (172, 200), separation of glycans by capillary electrophoresis (322) or normal-phase HPLC (172), and separation and identification of glycans by mass spectrometry (104, 172, 247, 259, 322). In some cases, these analytical strategies require instruments that are expensive and not always available (for example, quadrupole ion trap mass spectrometry and nuclear magnetic resonance spectroscopy). However, Callewaert et al. (38) have recently described a method for profiling and sequencing glycans by using standard equipment already in use for sequencing DNA. Yanagida et al. (317) have presented a strategy for the automation of glycan removal and labeling. Further, Rudd et al. (240) note that robust and relatively rapid glycan analysis can be accomplished with an HPLC system (equipped with a fluorescence detector and normal-phase columns) and a collection of enzymes and standards. To this might be added equipment for running and imaging FACE gels.
Analysis of fungal glycans, particularly Candida glycans, has been problematic with regard to enzymes and standards. The ineffectiveness of some commercially available enzymes against yeast glycans has already been discussed. In addition, for Candida species specifically, no ß-1,2-mannosidase (exo- or endo-) has been identified. An exo-ß-1,2-mannosidase would be particularly useful for sequence and structure characterization of Candida glycans. Also missing have been oligosaccharide standards for fungal glycan side branches. Characterization of the side chains requires knowledge of both the monosaccharide sequence and the linkages in that sequence. Homopolymers of different sugars and different linkages all show different electrophoretic mobilities (J. Masuoka, B. Fraser-Reid, and K. C. Hazen, unpublished data). Thus, well-defined standards are required for glycan identification from electrophoretic separations. Because the ß anomer is less energetically favored, synthesis of ß-linked oligosaccharides has been particularly problematic. However, several groups have recently reported effective syntheses of ß-oligomannosides (55, 160, 205, 206). These synthetic oligosaccharides are conjugated through their reducing end, complicating fluorophore labeling, or have protecting groups still attached to nonbonding alcohol groups. Fraser-Reid and coworkers have recently described a method for synthesis of unconjugated ß-1,2-oligomannosides (192). Syntheses, in general, are increasing in their ability to produce oligosaccharides of increasing length and complexity of other fungal glycan groups. Ning and Kong (204) have published the synthesis of a nonasaccharide from the wall mannan of C. albicans, which represents a section of the outer-chain backbone with the beginnings of two side branches. More recently, Xing and Ning (316) have described the synthesis of the mannan antigenic determinant of C. kefyr.
|
|
|---|
It seems clear that the pathogenesis of Candida is a multifactorial process with the participation of many cellular groups; and it is the interaction of these groups with host counterparts at certain times and under certain conditions that leads to disease (31, 37, 59). Cell wall glycans are exposed to the surrounding medium and are probably the first, spatially and temporally, to interact with components on the host cell surface. As discussed in this review, these glycans can, separately or in combination with protein or lipid components, affect various aspects of the host response to the invading fungal cells. Understanding glycan structure and function could thus provide an important handle for the treatment of fungal diseases. More specifically, because of the involvement of glycans in host-pathogen interactions, it may be possible to use these interactions to provide fungus-specific protection if the interaction mechanisms were known.
Chronologically, host-fungus studies first began with the general: looking at what happens in an animal model during an infection with intact cells. From there, they progressed to looking in vitro at the action of fungal cells on specific immune system components, to breaking up the fungal cell to see which specific fractions elicited the observed effect, and now to breaking up the individual molecules into chemical fractions (e.g., glycan and protein) to see which specific molecular component of the fungal cell interacts with which receptor/molecular component of the host cell. With the advent of the methods and strategies of glycobiology (227) presented in this text, we are now at the threshold of being able to break apart the glycans and study how the component moieties influence pathogenesis, immune response, therapeutics, and diagnostics. Further, by breaking glycans down into individual, understandable pieces, we can revise our model of the fungal cell wall and use it to create a detailed map of the host-fungus interaction. From there, the road to better diagnostics and treatments should be much shorter.
Some of the work discussed in this review was supported by USPHS grant AI 043997 from NIAID, National Institutes of Health.
|
|
|---|
and IL-1ß. Int. J. Immunopharmacol. 14:1363-1373.[CrossRef][Medline]
-mannan and ß-glucan cell wall components. Infect. Immun. 62:3138-3145.
3)-ß-D-man-(1
4)-ß-D-man(1]3 subunit of the antigenic polysaccharides from Leptospira biflexa and the octameric (1
2)-linked ß-D-mannan of the Candida albicans phospholipomannan. X-ray crystal structure of the protected tetramer. J. Am. Chem. Soc. 123:5826-5828.[CrossRef][Medline]
by alveolar macrophages in response to Candida albicans mannan. J. Leukoc. Biol. 55:161-168.[Abstract]
in response to bacterial lipopolysaccharide. Immunol. Lett. 37:19-25.[CrossRef][Medline]
-mannosidase which hydrolyzes yeast mannan. J. Biol. Chem. 243:2442-2446.
-mannosidase from an Arthrobacter species. J. Biol. Chem. 244:1043-1051.
-mannosidase on yeast mannan. J. Biol. Chem. 244:1052-1059.
B activation. J. Biol. Chem. 278:25001-25008.
-mannosidase. J. Biol. Chem. 242:5474-5480.
-D-mannanase in Bacillus circulans: cloning, sequence analysis, and expression. Biosci. Biotechnol. Biochem. 64:2018-2020.[CrossRef][Medline]
1-6-D-mannanase from a soil bacterium. J. Biol. Chem. 251:174-181.
-1,6 linkages corresponding to antigenic factor 9 in the mannan of Candida guilliermondii. J. Biol. Chem. 271:9259-9266.
-1,2-linked oligomannosyl residues. Infect. Immun. 60:4100-4110.
receptor II for presentation to T lymphocytes. Infect. Immun. 70:5972-5981.
3)-ß-D-glucan, CSBG from Candida spp. Int. J. Immunopharmacol. 22:383-394.[CrossRef][Medline]
rová, T. Mayadas, and G. D. Ross. 1999. The ß-glucan-binding lectin site of mouse CR3 (CD11b/CD18) and its function in generating a primed state of the receptor that mediates cytotoxic activation in response to iC3b-opsonized target cells. J. Immunol. 162:2281-2290.
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»