Adhesion and Morphogenesis in Candida albicans

From its first appearance as a superficial infection in the mouths of babes, the yeast Candida albicans holds primacy among all fungi for virulence and versatility. Its subsequent predilection to penetrate mucosal surfaces or to invade the bloodstream depends on its ability to“read” the host's microenvironment and to execute three strategies of pathogenesis: adhesion, colonization, and invasion.

Throughout this pathogenic cascade, morphology determines pathology (Fig. 1). Adhesion involves the attachment of blastospores to an epithelial surface (left panel); we see them here, budding as they divide and colonize. As host defenses fail and invasion supervenes, we see the malignant transformation of yeast to germ tube or pseudohypha with the extended cell body that drives through the epithelial barrier (right panel). This cunning capacity to change shape enables a supremely well equipped microorganism to read and respond to its environment and to press its advantage against the enfeebled host.

Figure 1

(Left panel) C. albicans blastospores. Reprinted from Ray TL, Payne CD 1988 Scanning electron microscopy of epidermal adherence and cavitation in murine candidiasis: a role for Candida acid proteinase. Infect Immun 56:1942-1949, with the permission of the publishers. (Right panel). C. albicans germ tubes. Reprinted from Wilborn WH, Montes LF 1980 Scanning electron microscopy of oral lesions in chronic mucocutaneous candidiasis. JAMA 244:2294-2297, with the permission of the publishers.

Over the past few years, studies from our laboratory and others have implicated integrin-like proteins in the processes of candidal adhesion and invasion. Last year I reviewed for you some of these studies in the biology of adhesion⁽¹⁾. This year, I will touch upon these previous concepts and then describe one such gene in C. albicans and its unexpected functions.

Before we talk about integrin-like proteins in C. albicans, we should re-acquaint ourselves with their more recent descendants, the vertebrate integrins, a family of 14 α-subunits and 8 β-subunits, which are expressed as heterodimeric transmembrane proteins on a wide variety of eukaryotic cells-epithelium, leukocytes, fibroblasts, and tumor cells⁽²⁾. In present taxonomy, integrins are grouped in subsets on the basis of their β-subunits. Two of the the β2 integrins-also known as the leukocyte adhesion glycoproteins-were described by Dr. Donald Anderson, a previous winner of this award⁽³⁾ and are now known to share antigenic, structural, and functional homologies with candidal surface proteins (Fig. 2, left panel):α_M, which is also known as Mac-1, CD11b, or complement receptor type 3 and α_X, which is known as p150,95, CD11c, or complement receptor type 4. α_M and α_X are polypeptides of 165 and 150 kD, respectively; in leukocytes, they are expressed in association with a common β-subunit of 95 kD.

Figure 2

(Left panel) Model of functional domains of the α-subunits of the β2 integrins in mammalian cells. Modified from Albelda SM, Buck CA 1990 Integrins and other cell adhesion molecules. FASEB J 4:2868-2880. (Right panel) Predicted functional domains of the integrin-like protein in C. albicans. Both figures are reprinted from Hostetter MK 1994 Interactions of C. albicans with eukaryotic cells. ASM News 60:370-374, with the permission of the publishers.

There are several functional domains on vertebrate integrins that deserve special mention (Fig. 2, left panel)

• In the extracellular region, α_M and α_X exhibit an inserted domain, or I domain, that is essential for binding of integrin ligands in the extracellular matrix⁽⁴⁾. Within the I domain, α_M and α_X contain a conformationally dependent cation binding site, or MIDAS motif⁽⁵⁾.

• Just C-terminal to the I domain are three linear divalent cation binding sites that are required for ligand binding and conform to the EF-hand motif⁽⁴⁾.

• Both the α- and β-subunits have highly conserved transmembrane domains, and both have distinct cytoplasmic tails with conserved lysine and tyrosine residues. The cytoplasmic tails link the cell's exterior to the cytoskeleton via interactions with α-actinin, talin, or paxillin.

It is presently thought that the extracellular domains of both the α- and β-subunits cooperate in the recognition and binding of integrin ligands in the extracellular matrix; many but not all of these ligands are distinguished by the tripeptide sequence arginine, glycine, aspartic acid, or RGD. Two ligands for α_M and α_X are fibrinogen and the C3 degradation fragment iC3b.

Incubation of C. albicans blastospores, germ tubes, or pseudohyphae with anti-α_M MAb such as OKM1 results in circumferential immunofluorescence, whereas incubation with isotype controls yields no fluorescence^(6–9). These same antibodies do not bind to yeast species such as Saccharomyces cerevisiae, seldom a pathogen. MAb to integrin β-subunits do not bind, either to C. albicans or to S. cerevisiae. These studies demonstrate that C. albicans bears surface proteins that share antigenic determinants with α_M and α_X.

Structural similarities occur as well. You will recall that α_M and α_X had molecular masses of 165 and 150 kD, respectively. When the integrin analog in C. albicans is isolated by affinity chromotography with anti-integrin MAb, a single band of 185 kD elutes under nonreducing conditions.

Functional analogies are even more illuminating. Both the leukocyte integrins and the integrin analog in C. albicans bind the C3 fragment iC3b noncovalently with identical affinity constants⁽¹⁰⁾. This competition for iC3b binding sites allowsC. albicans to evade the neutrophil and suggests that both candidal and vertebrate integrins contain an I domain. Moreover, just as integrins on vertebrate cells mediate adhesion to other eukaryotic surfaces, so too do integrin-like proteins in C. albicans directly facilitate the attachment of this yeast to human epithelium and endothelium.

Using cervical epithelial cells, Dr. Catherine Bendel developed anin vitro model of epithelial adhesion and correlated species-specific attachment with surface expression of integrinlike proteins⁽¹¹⁾. The percent yeast cells fluorescing with integrin MAb and adhering to cervical epithelial cells were greatest for C. albicans, next highest for Candida tropicalis, reduced for less virulent Candida species, and virtually absent in S. cerevisiae. Thus, surface expression of integrin-like proteins correlates directly with epithelial adhesion.

Dr. Bendel next focused on C. albicans and C. tropicalis to dissect the species-specific regulation of epithelial adhesion. First she wanted to know what C. albicans recognized on the epithelial cell, what was the ligand for attachment? Simple protein blockade studies showed very clearly that albumin, C3d, and fibronectin did not inhibit C. albicans adhesion but that iC3b, a C3 degradation fragment that contains the RGD tripeptide, inhibited adhesion by nearly 80%. The same was true for RGD peptides. Fibronectin peptides, even very large ones(23-mer), failed to inhibit C. albicans adhesion, as did RGD peptides from iC3b that had fewer than 9 amino acids. But iC3b peptides of 9, 10, or 15 amino acids inhibited adhesion incrementally⁽¹¹⁾.

But why did C. tropicalis adhere almost as well as C. albicans while expressing so little of the α_M/α_X analog? Was there another integrin-like protein in C. tropicalis? The answer of course is yes, and Dr. Bendel found that fibronectin and its RGD peptides were inhibitory for C. tropicalis-the first evidence for a fibronectin receptor in this species.

The results of Dr. Bendel's experiments are summarized in Figure 3. On the left is C. albicans with its integrin analog. The surface of the epithelial cell is studded with RGD-containing ligands. C. albicans recognizes one of these ligands-the RGD site and specific flanking residues in iC3b-and adhesion occurs. C. tropicalis, in contrast, recognizes the RGD site and flanking residues in fibronectin, and there is virtually no interchangeability. The RGD sites in these epithelial substrates promote adhesion, and half a dozen flanking residues confer precise specificity. Two distinctly different mechanisms of adhesion and two distinctly different integrin-like proteins as well: a 185 kD α_M/α_X analog in C. albicans and, as Dr. Greg DeMuri has shown in his abstract, a 115-kD β1 analog in C. tropicalis⁽¹²⁾. Ladies and gentlemen, the tap-dancing yeasts: Fred Astaire and this year, Ginger Rogers.

Figure 3

Distinct mechanisms of epithelial adhesion in C. albicans (left panel) and C. tropicalis (right panel). Reprinted from Hostetter MK 1994 Interactions of Candida albicans with eukaryotic cells. ASM News 60:370-374, with the permission of the publishers.

So, these many studies by our laboratory and others gave ample evidence of integrin-like proteins in the most pathogenic Candida species. But how best to get our hands on one of these genes? On the basis of the foregoing studies, we decided to focus on C. albicans first. We predicted that any integrin analog in C albicans must have an I domain, because both yeast and leukocyte integrins bound iC3b; there must be a transmembrane region, because both yeast and vertebrate integrins are surface-expressed; and there might well be divalent cation binding sites (Fig. 2,right panel), even though iC3b binding in C. albicans is not calcium-dependent.

We therefore obtained from Dr. Dennis Hickstein of Seattle a cDNA clone encoding the transmembrane region of human α_M and used aSma I/EcoRI fragment as a probe to screen a library ofC. albicans genomic DNA. We were betting that the most important structural motifs had been preserved across 2 billion years of evolution with the best of biologic economy. The five clones that we isolated from the initial screen were then further refined by hybridization with a degenerate oglionucleotide encoding the sequence KVGFFK, which marks the division between transmembrane domain and cytoplasmic tail in virtually all α-integrin subunits. We reasoned that many C. albicans proteins might have a transmembrane domain, but that only integrin-like proteins should have this cytoplasmic sequence.

The net result was one clone that encodes an open reading frame of 4,992 bp, sufficient to encode a protein of approximately 1664 amino acids with a molecular mass of 188 kD. The protein encoded by αINT1(“ain't one” to unbelievers) exhibits several motifs common toα_M and α_X, and to some other integrinα-subunits as well (Fig. 2, right panel). There is a putative I domain with approximately 18% identity to the amino acid sequence in α_M. There are two divalent cation binding sites, and both conform to the EF-hand loop motif. There is a hydrophobic transmembrane domain of 25 amino acids and a cytoplasmic tail with a single tyrosine residue. And best of all, the Candida protein has its own RGD site(amino acids 1149-1151).

One might predict that less pathogenic yeast species would be devoid ofαINT1 at the genomic level and indeed, this is confirmed by Southern blot. Under conditions of high stringency, hybridizing fragments are evident only in C. albicans and not in C. tropicalis orS. cerevisiae. We have now examined more that two dozen isolates ofC. albicans, C. tropicalis, and S. cerevisiae;αINT1 is present in all C. albicans strains and in neither of the other species.

Now what does this gene do? Our prediction, of course, was that theαINT1 gene product was an epithelial adhesin for C. albicans, but we wanted to test this function using nonadherent S. cerevisiae. This part of the project was executed by Dr. Cheryl Gale, funded in our laboratory by the Pediatric Scientist Development Program. Dr. Gale's strategy called for insertion of the open reading frame ofαINT1 behind the upstream activating sequence of theGAL1/GAL10 promoter of S. cerevisiae. Her plasmid, pCG01, could then be used to transform Saccharomyces, and expression of theCandida gene would be induced with 2% galactose. The control wasS. cerevisiae bearing vector alone, without the candidal insert⁽¹³⁾.

Growth curves for S. cerevisiae transformants bearing the candidal gene or vector alone were identical. The corresponding Northern showed message of 5.5 kb at 6 and 24 h after galactose induction in S. cerevisiae transformants bearing the Candida gene, but no message in the transformants bearing vector alone. No message was seen in transformants grown in glucose, under which condition the Candida gene is repressed. These of course were the expected results that showed how nicely this system could be manipulated.

However, when Dr. Gale took a look under the microscope at her transformants, we had a remarkable surprise. The parent strain ofSaccharomyces transformed with plasmid alone contains no candidal insert and exhibits the typical spheroid shape with an occasional budding yeast. This is perfectly normal. But in the Saccharomyces transformants expressing the candidal gene we noted germ tube-like projections that we have christened “noses” (Fig. 4). Under low power magnification, one can see noses on approximately 90% of allS. cerevisiae transformants expressing αINT1; these bear a startling resemblance to germ tubes in C. albicans.

Figure 4

Germ tube formation in S. cerevisiae transformed with the plasmid containing the open reading frame ofαINT1, the gene that encodes the integrin-like protein inC. albicans.

We performed a number of controls to ensure that it is expression ofαINT1 that engenders this morphologic change. If one curesSaccharomyces of the plasmid-no noses. If one growsSaccharomyces in glucose-no noses. Or in a noninducing concentration of galactose-no noses. If one expresses a gene from Chlamydomonas instead of the Candida gene behind the same promoter-no noses. Thus, from these studies we can say conclusively that synthesis of theCandida gene product αInt1p in S. cerevisiae induces germ tubes⁽¹³⁾. Our preliminary studies have confirmed that αInt1p is surface expressed in S. cerevisiae and can be recognized by polyclonal antibodies to candidal peptides and by MAb to vertebrate integrins⁽¹³⁾.

Now, how might αInt1p participate in the three steps of candidal pathogenesis: adhesion, colonization, and invasion? First of all, the presence of an I domain that recognizes RGD sites in proteins of the epithelial extracellular matrix would enable the C. albicans integrin to mediate adhesion. Second, the internal RGD site in the candidal gene product could be a ligand for the I domain, thereby allowing yeast cells to attach to each other and to colonize the host. Last, of course, the germ tubes that we see upon expression of αInt1p in S. cerevisiae may well occur in C. albicans as well. This is eminently testable by creation of a homozygous null mutant, and Dr. Gale's experiments are underway. In each case, the protein encoded by αINT1 is the antenna, the extracellular hook, that initiates the organism's response to environmental conditions.

So, this gene may well help us to dissect the mechanisms of candidal pathogenesis, but let us not ignore the effect in Saccharomyces. This strange Durante effect, these schnozzolas-we have now taken a harmless yeast and made it sticky and nosey; what use for them? InSaccharomyces, of course, these noses can answer for us myriad questions regarding signal transduction. Many examples of morphologic change and gene induction in Saccharomyces are directed by proteins of the MAP kinase cascade, so well defined by Fink, Herskowitz, and others: the mating and “invasive” pathways in haploids and the pseudohyphal response to nitrogen deprivation in diploids^(14–16). Each of these morphologic variants depends upon the sequential activation of a series of sterile genes: fromSTE2 or 3 at the cell surface, through STE20, STE11, STE7, and STE5, and finally to the phosphorylation ofSTE12, a yeast transcription factor. David Finkel of our laboratory disrupted STE12, transformed his null mutant with theCandida gene, and still got noses⁽¹³⁾. Thus, the induction of noses independently of STE12 may help us to elucidate a novel signaling pathway in S. cerevisiae.

But there are potential medical applications as well. First, because we have a yeast that should stick to epithelial surfaces but has not the machinery for invasion, we may be able to colonize patients at risk forCandida infection with a noninvasive Saccharomyces species transformed with a single Candida gene. Secondly, we may be able to use this system to deliver other gene products to epithelial surfaces (Fig. 5). Make Saccharomyces stick to epithelium by expressing the integrin gene, and then add a second plasmid for expression of a phosphate-binding protein; this strategy could decrease the dietary phosphate load in patients with chronic renal failure. Or use a second plasmid to provide a source of vaccine antigen for gastrointestinal pathogens like cholera. In the genitourinary tract, expression of spermicides in sticky yeasts could provide a cheap and infrequent method of contraception. And synthesis of protein-based antiretroviral agents could help to reduce transmission of HIV in the birth canal.

Figure 5

Use of adherent S. cerevisiae transformed with αINT1 as a gene-delivery system in the gastrointestinal or genitourinary tract.

I am very grateful to the committee for having chosen me for this honor. I have been fortunate to have the most encouraging of mentors, the most generous of collaborators, and the most inspiring fellows and technicians(Table 1). Were it not for the width of this platform, they should all be up here with me. I have also, of course, as an English major, had the most prescient of guides, for although one might wish to present these findings as novel, I must admit that they were in fact predicted more than 230 years ago by Laurence Sterne, in his marvelous satireTristram Shandy, published in 1764:

Table 1 Acknowledgment

“It so happens and ever must, that the excellency of the nose is in a direct arithmetical proportion to the excellency of the wearer's fancy.”

And what is our fancy? Our fancy is the belief that science will enlighten, that those whom we train will be better than ourselves, that they will see more broadly, shine more brightly, succeed surpassingly. And so I say, follow that unexpected result, that elusive footfall, that shadowy coattail as it disappears around the next corner and the next, follow your intuition, follow your nose.