|Home | About | Journals | Submit | Contact Us | Français|
Biofilms are surface-associated microbial communities with significant environmental and medical impact. Here we focus on an adherence mechanism that permits biofilm formation by Candida albicans, the major invasive fungal pathogen of humans.
The Als surface protein family has been implicated in biofilm formation, and we show that Als1 and Als3 have critical but redundant roles. Overexpression of several other Als proteins permits biofilm formation in a biofilm-defective als1/als1 als3/als3 strain, thus arguing that Als protein function in this process is governed by their respective expression levels. The surface protein Hwp1 is also required for biofilm formation, and we find that a mixture of biofilm-defective hwp1/hwp1 and als1/als1 als3/als3 strains can form a hybrid biofilm both in vitro and in vivo, in a catheter infection model. Complementary function of Hwp1 and Als1/3 seems to reflect their interaction, because expression of Hwp1 in the heterologous host S. cerevisiae permits adherence to wild-type C. albicans but not to an als1/als1 als3/als3 strain.
The complementary roles of Hwp1 and Als1/3 in biofilm formation are analogous to the roles of sexual agglutinins in mating reactions. This analogy suggests that biofilm adhesin complementarity may promote formation of monospecies biofilms.
Many microorganisms exist in surface-associated communities called biofilms . Biofilms can form on both natural and artificial surfaces, with substantial consequences for both industry  and human health [3, 4]. Biofilm formation on implanted medical devices is associated with bloodstream infection, causing over 107 device-associated infections per year in the United States [3, 4]. The diverse adherence mechanisms that permit accumulation of surface-bound biomass are of particular interest [5, 6] as they present an opportunity for development of new antivirulence therapies .
The fungal pathogen C. albicans is a frequent cause of device-associated infection . Its evolutionary distance from well-studied bacterial pathogens suggests that C. albicans biofilm formation may employ distinct gene products and mechanisms. Insight into these mechanisms has come from several approaches. First, studies of the model yeast S. cerevisiae have shown that both cell-surface and cell-cell adherence depend upon a set of surface proteins, the FLO gene products, that share an N-terminal signal sequence, numerous internal serine/threonine-rich repeats, and a C-terminal glycophosphatidyl inositol (GPI) anchor attachment sequence [8, 9]. Adherence of the fungal pathogen Candida glabrata depends upon the EPA genes [10, 11], which specify proteins of this class as well. Second, gene expression profiling has revealed induction during biofilm formation of the C. albicans ALS gene family [12-15], a family of eight genes whose products resemble Flo and Epa surface proteins . Indeed, several Als proteins have been shown to function as adhesins using a variety of assay conditions and substrates, even when expressed in the heterologous host S. cerevisiae [17-23]. These lines of reasoning have made the Als proteins excellent candidates for C. albicans biofilm adhesins.
Direct evidence indicates that Als3 may function as a biofilm adhesin under certain circumstances. For example, Hoyer and coworkers examined biofilm formation in vitro of an als3/als3 homozygous deletion mutant . The strain formed a fragile biofilm with abnormal architecture, in which elongated hyphal filaments were parallel rather than intertwined. A second line of evidence comes from analysis of Bcr1, a transcription factor required for ALS3 expression and biofilm formation . Functional significance of this regulatory relationship was underscored by the finding that increased expression of ALS3 restored biofilm formation in the bcr1/bcr1 mutant background in both in vitro and in vivo assay systems . However, Als3 was required for biofilm formation only in vitro, not in vivo, in our studies. Hence the in vivo and in vitro situations may have different functional requirements, or there may be additional Als proteins expressed in vivo, such as the closely related Als1, that compensate for the absence of Als3.
The analysis of Bcr1 pointed to a second candidate adhesin, Hwp1, a surface protein with little similarity to the Als proteins . Hwp1 is well known as a substrate for host transglutaminases, permitting covalent attachment of C. albicans to epithelial cells [26, 27]. We found that Hwp1 also has a significant role in biofilm formation: overexpression of HWP1 in a bcr1/bcr1 mutant improved biofilm formation, and an hwp1/hwp1 mutant has a partial biofilm defect in vitro and a severe biofilm defect in vivo [25, 28]. At first glance, then, Als3 and Hwp1 appear to have similar roles, despite their sequence divergence.
Here we test the possibility that Als1, Als3, and Hwp1 function redundantly to promote biofilm formation. Our results lead us to the model that Als1/3 and Hwp1 may function as complementary adhesins, a role akin to that of mating agglutinins of sexually active fungi. C. albicans may have redeployed its mating functions to support biofilm adherence, thus favoring its long-term survival.
Fungal strains were grown at 30°C in either YPD (2% Bacto Peptone, 2% dextrose,1% yeast extract) for Ura+ strains or in YPD+uri (2% Bacto Peptone, 2% dextrose,1% yeast extract, and 80 μg/mL uridine) for Ura- strains. Ura- selection was done on 5-FOA media . C. albicans transformants were selected for on synthetic medium (2% dextrose, 6.7% YNB with ammonium sulfate, and auxotrophic supplements) or on YPD+clonNAT (2% Bacto Peptone, 2% dextrose,1% yeast extract, and 400 μg/mL clonNAT (WERNER BioAgents)) for Nat+ strains. Biofilms were grown in Spider medium .
Yeast strains are listed in Supplemental Table 1. C. albicans strains were derived from BWP17 (ura3Δ::λimm434/ura3Δ::λimm434 arg4::hisG/arg4::hisG his1::hisG/his1::hisG)  except for the following, which were derived from CAI4 : CAYC2YF1U, the als1/als1 mutant strain , CAYC2QTP1U, the als1/als1 + pALS1 complemented strain, and CAH7-1A1E2 , the hwp1/hwp1 mutant strain. S. cerevisiae strains HAS149 and HAS152 were derived from BY4741 PMA1-GFP (MATa his3Δ1 leu2Δ1 met15Δ0 ura3Δ0 PMA1-GFP; Invitrogen cat. # 95700). HAS149 and HAS152 strains were constructed as follows. The HWP1 ORF was PCR-amplified from plasmid pCJN541, a plasmid containing the entire HWP1 ORF from the start codon to 500bp of UTRin pGEMT-Easy (Promega), using primers T7-F-HWP1-atg-HAS and HWP1-R-500. (See Supplemental Table 2 for primer sequences.) These primers contain flanking homology to plasmid pYES2.1 (Invitrogen) in order to transform the HWP1 ORF into pYES2.1 by homologous recombination in BY4741 PMA1-GFP. The resulting strain was S. cerevisiae strain HAS149. As a control, the pYES2.1 vector alone was transformed into BY4741 PMA1-GFP tocreate strain HAS152.
Construction of the als1/als1 +pALS1 complemented strain, CAYC2QTP1U, was done as follows. An intact ALS1 allele was amplified from genomic DNA using the primers ALS1-F and ALS1-R using high fidelity PCR and ligated into pGEM-T (Promega) to yield plasmid pGEM-T-ALS1. Next, the recyclable URA3-dpl200 cassette was amplified from plasmid pDDB57 using the 5-DR and 3-DR primers  and ligated into pGEM-T-EASY to yield pGEM-T-UUU. The URA3-dpl200 cassette was then liberated by NotI digestion and ligated into the unique NotI site of pGEM-T-ALS1. The resulting plasmid pGEM-T-ALS1UUU was linearized with BglII and used to transform the previously described Ura- als1/als1 mutant strain of C. albicans . The resulting strain which contained a single copy of ALS1 integrated at its native locus was then rendered Ura- by plating onto 5-FOA medium. Finally, to restore the URA3-IRO1 locus, this strain was transformed with a 3.9kb URA3-IRO1 fragment released from pBSK-URA3 by NotI/PstI digestion as previously described . PCR confirmed integration of the intact ALS1 gene at its native locus and reconstruction of the URA3-IRO1 locus. Construction of the als1/als1 als3/als3 double mutant, CJN1348, was done by deleting each allele of ALS3 in a Ura- als1/als1 mutant using the URA3-dpl200 cassette , followed by a reintroduction of URA3 and the adjacent IRO1 locus using plasmid pBSK-URA3 . To complement the als1/als1 als3/als3 double mutant with a wild-type copy of ALS1 or ALS3, a full–length version of ALS1 or ALS3 was digested from pGEMT-Easy with PvuI and SphI , and then subcloned into pDS10 for ALS1 or pDS11 for ALS3 . The constructs were linearized and integrated into the ALS1 locus of the als1/als1 als3/als3 Ura- strain or into the ALS3 locus of the als1/als1 als3/als3 Ura- strain, selecting Ura+. Excision of URA3-dpl200 was then selected on 5-FOA medium. URA3 and the adjacent IRO1 locus was restored in the als1/als1 als3/als3::ALS1 and als1/als1 als3/als3::ALS3 strains to create als1/als1 als3/als3::ALS1 prototrophic strain, CJN1352, and als1/als1 als3/als3::ALS3 prototrophic strain, CJN1356, as follows. Ura- derivatives of these mutants were selected by plating on 5-FOA medium. A 3.9-kb URA3-IRO1 fragment was released from pBSK-URA3 and used to transform the Ura- strains .
The NAT1-TDH3 promoter plasmid pCJN542 (Nobile et al, submitted) was used for gene overexpression. The TDH3-ALS2 overexpression strain, CJN1455, was constructed by transforming CJN1348, the als1/als1 als3/als3 double mutant, using PCR products from template plasmid pCJN542 and primers ALS2-F-OE-Ag-NAT-Ag-p-CJN and ALS2/4-R-OE-Ag-NAT-Ag-TDH3p-CJN. These primers amplify the entire A. gossypii TEF1 promoter, the C. albicans NAT1 ORF, the A. gossypii TEF1 terminator, and the C. albicans TDH3 promoterwith 100 bp of homology to the region 500 bp upstream into the promoter of ALS2 for the forward primer and 100 bp of homology from exactly the start codon of ALS2.
The homology in these primers allows for homologous recombination of the entire cassette directly upstream of the natural locus of ALS2 so that its expression is driven by the TDH3 promoter instead of its natural promoter. By the same method, primers ALS4-F-OE-Ag-NAT-Ag-p-CJN and ALS2/4-R-OE-Ag-NAT-Ag-TDH3p-CJN were used for overexpression of ALS4 to produce strain CJN1456; ALS5-F-OE-Ag-NAT-Ag-p-CJN and ALS5-R-OE-Ag-NAT-Ag-TDH3p-CJN for overexpression of ALS5 (strain CJN1460); ALS6-F-OE-Ag-NAT-Ag-p-CJN and ALS6-R-OE-Ag-NAT-Ag-TDH3p-CJN for overexpression of ALS6 (strain CJN1464); ALS7-F-OE-Ag-NAT-Ag-p-CJN and ALS7-R-OE-Ag-NAT-Ag-TDH3p-CJN for overexpression of ALS7 (strain CJN1468); ALS9-F-OE-Ag-NAT-Ag-p-CJN and ALS9-R-OE-Ag-NAT-Ag-TDH3p-CJN for overexpression of ALS9 (strain 1472); and HWP1-F-OE-Ag-NAT-Ag-TEF1p and HWP1-R-OE-Ag-NAT-Ag-TDH3p for overexpression of HWP1 (strain HAS113). The transformation into C. albicans strains and selection on YPD+clonNAT (2% Bacto Peptone, 2% dextrose,1% yeast extract, and 400 μg/mL clonNAT (Werner BioAgents) plates was done as previously described . Integration of the constructs was verified by colony PCR using a forward primer annealing to a sequence within the promoter of each gene, respectively, in combination with the reverse primer Nat-OE-R-det2-CJN annealing to a sequence found in the NAT gene. Function of this overexpression strategy was verified for each gene by real-time RT-PCR (Supplemental Figure 1).
Overnight cultures of C. albicans strains DAY185 (reference) and CJN1348 (als1/als1 als3/als3) and S. cerevisiae strains HAS149 (PGAL1-HWP1) and HAS152 (PGAL1-vector alone) were grown in SC-Ura with 0.5% glucose. Cultures were spun down, washed with water, resuspended in YPgal (containing 4% galactose), and allowed to grow to OD600=0.2 in order to induce expression from the GAL1 promoter. Cells were pelleted, washed with water, resuspended in M199 pH 8 medium to an OD600=0.2 for C. albicans and OD600=0.1 for S. cerevisiae, mixed together in a 1:1 ratio of C. albicans cells to S. cerevisiae cells, and grown together for 2.5 hours. Cells were then pelleted gently at low speed, washed with water, and visualized by fluorescence microscopy. The number of fluorescent S. cerevisiae cells per C. albicans hyphal cell was counted for 35-50 hyphal cells and normalized by hyphal length. Statistical analysis was performed with the Wilcoxon Rank Sum test.
In vitro biofilm growth assays were carried out in Spider medium with auxotrophic supplements and visualized by confocal microscopy as described previously . Biomass measurements were determined for at least 4 independent 1 cm2 samples as described previously ; statistical significance (p values) was determined with a two tailed T-test.
A rat central venous catheter infection model was selected for in vivo biofilm studies, as described previously . Catheters from 2 animals were removed at 48 hr after C. albicans infection to determine biofilm development on the internal surface of the intravascular devices. The distal 2 cm of the catheter was cut from the entire catheter length, and biofilms were imaged by scanning electron microscopy (SEM).
Als1 is closely related to Als3 in sequence, in regulation, and in function as determined in diverse assay systems [14, 20, 22, 24]. Als1 can promote biofilm formation in an in vivo catheter model, because overexpression of ALS1 in a bcr1/bcr1 mutant restored biofilm formation ability (Figure 1A-F). However, an als1/als1 deletion mutant formed a substantial biofilm in vivo (Figure 1G,H). We previously reported similar results for overexpression, deletion, and complementation of ALS3 . One simple explanation for all of our observations with Als1 and Als3 is that these two proteins act interchangeably to promote biofilm formation in vivo. This model predicts that an overall reduction of functional ALS1 and ALS3 alleles should cause a biofilm defect in vivo. We observed that strains with only one functional ALS1 allele (als1/als1 als3/als3 +pALS1) or one functional ALS3 allele (als1/als1 als3/als3 +pALS3) produce only a sparse layer of adherent cells in the in vivo catheter model (Figure 1K-N). Furthermore, two independently constructed strains lacking all functional ALS1 and ALS3 alleles (als1/als1 als3/als3) failed to produce any detectable adherent cells in the in vivo catheter model (Figure 1O,P and data not shown). We made analogous observations with an in vitro biofilm model (Table 1). Our findings indicate that C. albicans needs at least two functional ALS1 or ALS3 alleles to produce a biofilm. These observations support the model that Als1 and Als3 function redundantly to promote biofilm formation.
Als1 and Als3 are members of an 8-member C. albicans protein family. To determine whether any other Als family members have the capacity to promote biofilm, we created derivatives of the als1/als1 als3/als3 strain in which individual ALS genes were expressed from the strong TDH3 promoter. We observed that overexpression of any Als protein restored substantial biofilm formation in vitro (Table 1). Failure of TDH3-HWP1 (discussed below; Table 1) or TDH3-PGA13 (data not shown) to restore biofilm production in this assay argued against a nonspecific effect of cell wall protein overexpression in vitro. However, while TDH3-ALS6, TDH3-ALS7, and TDH3-ALS9 each rescued biofilm formation of the als1/als1 als3/als3 mutant in vivo (Figure 2G-L), we found that TDH3-ALS5 rescued only weakly (Figure 2E,F), and TDH3-ALS2 and TDH3-ALS4 did not rescue at all (Figure 2A-D). Failure of Als2 and Als4 to support biofilm development in this situation is consistent with their existence in a distinct subclass of the Als family . These results argue that all Als proteins can function equivalently to support biofilm formation in vitro, but only a subset of Als proteins function equivalently to Als1 and Als3 to support biofilm formation in the in vivo catheter model.
The surface protein Hwp1 is required along with Als1 and Als3 to promote biofilm formation in vivo . Hwp1 has little sequence similarity to the Als proteins, thus suggesting that it may have a distinct function from Als1 and Als3. In addition, we observed that a TDH3-HWP1 hybrid gene could not promote biofilm formation in the als1/als1 als3/als3 background in vitro (Table 1) or in vivo (Figure 2M,N). The hybrid gene did stimulate some adherence of a basal cell layer, though. These observations suggest that Hwp1 has a distinct function from the Als proteins in biofilm formation.
C. albicans biofilm formation thus depends upon two types of surface proteins – Hwp1 and Als1/3. We considered the possibility that Hwp1 and Als1/3 function as complementary surface adhesins, much like the S. cerevisiae a and α cell mating agglutinins. This model predicts that, whereas mutants lacking either Hwp1 or Als1/3 are biofilm defective, a mixture of the two mutant strains should be biofilm competent. We first tested this prediction through CSLM analysis of biofilm formation in vitro. The wild-type strain produced a biofilm of greater than 100 μm in depth, and the hwp1/hwp1 and als1/als1 als3/als3 mutant strains each produced rudimentary biofilms of 5-20 μm in depth (Figure 3A-C). A 1:1 mixture of the two mutant strains produced a biofilm of greater than 100 μm in depth (Figure 3D). We also conducted mixed biofilm assays with the in vivo catheter model. Each individual mutant strain had a severe biofilm defect, with few if any adherent cells on the catheter surface and no clear accumulation of extracellular matrix material (Figure 1O,P, Figure 4A,B). In contrast, a mixture of the two mutants produced a confluent biofilm comprised of yeast cells, hyphal filaments, and extracellular matrix (Figure 4C,D). These results indicate that Als1/3 and Hwp1 have complementary roles in biofilm formation in vitro and in vivo.
To determine whether surface Hwp1 may promote cell-cell adherence directly, we compared the binding properties of S. cerevisiae cells that expressed PGAL1- HWP1 to those carrying the PGAL1 vector alone (Figure 5). We tested the binding of these S. cerevisiae yeast form cells to C. albicans hyphae, which express high levels of Als1/3 [39, 40] and are readily distinguishable from yeast form cells by microscopic examination. Expression of Hwp1 increased adherence of S. cerevisiae cells to C. albicans hyphae approximately 10-fold (p<0.0001). Adherence of the Hwp1-expressing strain was 2-fold better to wild-type C. albicans than to the als1/als1 als3/als3 mutant (p<0.0001). These results argue that surface Hwp1 promotes interaction with the C. albicans cell surface directly, and that this interaction is substantially improved by the presence of Als1/3.
Biofilm formation by C. albicans creates the nidus of device-associated infection. The surface molecules that govern biofilm formation are crucial as the mediators of contacts that promote biomass accumulation and biofilm resilience. Our studies here reveal that the Als1 and Als3 surface proteins are together required for biofilm formation in vivo. We propose that they function by binding to the surface protein Hwp1 on neighboring cells. These observations suggest an analogy between these C. albicans biofilm adhesins and S. cerevisiae mating agglutinins, particularly because Als1/3 has a similar overall structure to S. cerevisiae cell type α agglutinin. The analogy between mating and biofilm formation was foreshadowed by work from Soll and colleagues that showed that mating factor can stimulate biofilm formation in genetically responsive C. albicans strains .
Previous observations have implicated Als proteins in biofilm formation (see Introduction), but it has thus far been uncertain which if any family members function in biofilm formation in vivo. Our findings indicate that Als1 and Als3 have a major role in this process in vivo, because reductions in overall ALS1/3 gene dosage yielded a progressively more severe biofilm defect. The unusual morphology of the als1/als1 mutant biofilm in vivo lends further support to our contention that Als1 is functionally significant for biofilm formation. These observations are consistent with the properties of the bcr1/bcr1 transcription factor mutant: its reduced expression of both ALS1 and ALS3 seems to be critical for its biofilm defect, because overexpression of either ALS1 or ALS3 restores biofilm formation by the bcr1/bcr1 mutant. Therefore, the crucial contribution of Als1 and Als3 to biofilm formation in vivo is consistent with many observations.
The fact that Als1 and Als3 have a shared function is consistent with their similarity in sequence (88% amino acid identity of their N-terminal 772 residues) and of their predicted N-terminal domain structures . Moreover, Als1 and Als3 have very similar binding specificities in vitro , and the N-terminal domain of either protein is sufficient to promote N- or E-cadherin-dependent endocytosis . Thus the finding that Als1 and Als3 share a function in biofilm formation is entirely consistent with their structural and biochemical properties.
We infer that the broadly shared features of Als proteins, rather than the features specific to subfamilies, are critical for biofilm formation. This inference comes from the finding that increased expression of several different ALS genes rescues the biofilm defect of the als1/als1 als3/als3 double mutant in vivo. Binding studies of S. cerevisiae cells that express individual Als proteins argue that Als6, Als7, and Als9 have quite different binding specificity from Als1 and Als3 , yet their overexpression rescued double-mutant biofilm formation well. These studies also showed that Als5 has similar binding properties to Als1 and Als3, yet Als5 overexpression rescued double-mutant biofilm formation weakly. The shared Als features include predicted N-terminal immunoglobulin folds that are similar to those of the S. cerevisiae α cell mating agglutinin, Sag1 . Mutational analysis of the Sag1 N-terminal region has defined seven residues that are critical for binding of Sag1 to the a cell agglutinin . In general, these residues are not conserved between Sag1 and the Als proteins. However, a Clustalw alignment reveals that several of the corresponding positions are well conserved among all Als proteins, including F216, V/I232, Y292, and N294 (with reference to positions in Als3). This comparison suggests that the Als proteins may have a different specificity from Sag1, but that they may have similar affinities for a C. albicans protein that is analogous to the S. cerevisiae a cell agglutinin.
This analysis raises the possibility that a C. albicans protein, analogous to S. cerevisiae a cell agglutinin, may function in biofilm formation. Our findings indicate that Hwp1 is that analog, based on three lines of evidence. First, Hwp1 is required for biofilm formation . Second, mixed inocula of biofilm-defective cells that lack either Als1/3 or Hwp1 are capable of robust biofilm formation. This result is particularly striking in view of previous studies showing that other mixtures of biofilm-defective cells cannot restore biofilm formation . Third, the idea that Hwp1 and Als proteins function as complementary adhesins is supported by our finding that S. cerevisiae cells expressing Hwp1 display adherence to C. albicans cells, which is dependent upon Als1 and Als3. Therefore, we propose that Als1/3 and Hwp1 undergo a complementary binding reaction that supports C. albicans cell-cell adherence in biofilms.
We note that the analogy between Hwp1 and the a cell mating agglutinin is further strengthened by the work of Daniels et al. , who found that surface Hwp1 accumulates only on the a/a cell side of the C. albicans conjugation bridge. Although the functional basis for this localization pattern is unknown, we reflect that this is the site of accumulation one might expect for a C. albicans a/a mating agglutinin.
In addition to Als1/3 and Hwp1, several cell surface proteins are required for biofilm formation, including Eap1, Sun41, and the CFEM family (Pga10/Rbt51, Rbt5, and Wap1/Csa1) [44-48]. One possibility is that some of these proteins are required for Als or Hwp1 synthesis or maturation. Mutants in such functions might phenocopy the als1/als1 als3/als3 or hwp1/hwp1 mutants, particularly in a hybrid biofilm assay. Another possibility is that these proteins include yeast-form cell adhesins, a class of protein (if they are proteins) yet to be defined. A third possibility is that one of these proteins is a partner for Hwp1. This idea comes from the analogy between Hwp1 and the S. cerevisiae a cell agglutinin: that agglutinin is a heterodimer of Aga1 and Aga2, bound by disulfide bonds. The N-terminal region of Hwp1 includes roughly 10% cysteine residues, so there is the opportunity for disulfide formation with an as yet undefined partner.
Our findings here strengthen one prevailing idea about C. albicans biofilm formation: that Als family members may have overlapping functions. That is, all Als family members have the capacity to support biofilm formation in vitro, and many have that capacity in vivo as well. Moreover, we do not see a precise correlation between sequence similarity and functional interchangeability. These observations all point to the critical role of ALS gene regulation in defining the manifested, as opposed to potential, biological function of each family member. Interestingly, this theme echoes findings with adhesins of C. glabrata [10, 11] and S. cerevisiae . Recent advances in dissection of ALS regulation [14, 49] hold promise to yield relevant insight in this regard. Our findings also point to a new idea about C. albicans biofilms: that the breadth of biofilm adhesins may not simply satisfy a need for diverse substrate attachment. Rather, we suggest that adhesin complementarity is critical for biofilm formation. Adhesin complementarity may be only an evolutionary relic that reflects retooled gene products from a more sexually active ancestor of modern day C. albicans. But an interesting possibility is that adhesin complementarity supports exclusion of foreign species from C. albicans biofilm communities, because integrity of the biofilm is dependent upon intraspecies contact. Thus fungal intraspecies adhesin interaction may accomplish much the same goal as bacterial quorum-quenching activities , protecting a niche from competitors.
We thank all members of the Mitchell lab for ideas and advice on this work. We thank Aaron Hernday for help with analysis of the quantitative real time RT-PCR data, and Frank Smith and Jessica Hamaker for technical assistance. We are grateful for the availability of the Candida Genome Database, without which this work would not have been possible. This study was supported by NIH grants R01 AI067703 (A.P.M.),K08 AI01767 (D.R.A.), T32 HL07899 (J.E.N.), and R01 AI054928 (S.G.F.).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.