Conservation and Divergence in the Candida Species Biofilm Matrix Mannan-Glucan Complex Structure, Function, and Genetic Control

ABSTRACT Candida biofilms resist the effects of available antifungal therapies. Prior studies with Candida albicans biofilms show that an extracellular matrix mannan-glucan complex (MGCx) contributes to antifungal sequestration, leading to drug resistance. Here we implement biochemical, pharmacological, and genetic approaches to explore a similar mechanism of resistance for the three most common clinically encountered non-albicans Candida species (NAC). Our findings reveal that each Candida species biofilm synthesizes a mannan-glucan complex and that the antifungal-protective function of this complex is conserved. Structural similarities extended primarily to the polysaccharide backbone (α-1,6-mannan and β-1,6-glucan). Surprisingly, biochemical analysis uncovered stark differences in the branching side chains of the MGCx among the species. Consistent with the structural analysis, similarities in the genetic control of MGCx production for each Candida species also appeared limited to the synthesis of the polysaccharide backbone. Each species appears to employ a unique subset of modification enzymes for MGCx synthesis, likely accounting for the observed side chain diversity. Our results argue for the conservation of matrix function among Candida spp. While biogenesis is preserved at the level of the mannan-glucan complex backbone, divergence emerges for construction of branching side chains. Thus, the MGCx backbone represents an ideal drug target for effective pan-Candida species biofilm therapy.

Here, we present evidence of a conserved matrix mannan and glucan complex backbone (MGCx) which contributes to profound drug resistance exhibited by the four most prevalent Candida species. We show that select matrix synthesis C. albicans orthologs play similar roles across Candida species to synthesize a common polysaccharide backbone. However, structural and molecular divergence in matrix assembly is suggested by subtle differences in matrix branching and absence of the involvement of matrix modification enzyme orthologs. Our findings argue that broad-spectrum Candida biofilm drug discovery should target the level of MGCx backbone synthesis as opposed to side chain synthesis.

RESULTS
Comparison of Candida species biofilm growth and architecture. We determined basic parameters of biofilm development for three non-albicans Candida species: C. parapsilosis, C. tropicalis, and the more distantly related C. glabrata. We also included wild-type C. albicans as a standard for comparison. Biofilm formation begins with adherence of cells to the substrate, and we found marked differences among the species in this property (Fig. 1A). Despite similar inocula, the yield of adherent cells was approximately fivefold lower for C. parapsilosis and C. glabrata than that measured for either C. albicans or C. tropicalis. Biofilm maturation over the next 24 h erased these differences in cell number estimates (Fig. 1B), thus indicating that all four species reach similar sessile equilibria under in vitro growth conditions. Scanning electron microscopy (SEM) of mature biofilms revealed that each of the species, with the exception of C. glabrata, demonstrated a preponderance of filamentous or elongated cell growth (Fig. 1C). In addition, abundant extracellular matrix material encased the cells within each biofilm (marked by white arrows). We extended our comparison to examine in vivo biofilm formed in the lumen of a rat vascular catheter. Basic architecture in this model recapitulated the biofilm characteristics observed in vitro: filamentous cells were present in C. albicans, C. parapsilosis, and C. tropicalis biofilms, and extracellular matrix material was abundant. In fact, extracellular matrix appeared more pronounced in vivo than in vitro, likely due to the contribution of host components (48). These results indicate that mature biofilms of the three non-albicans Candida species resemble those of C. albicans in terms of cellular content, matrix accumulation, and for C. parapsilosis and C. tropicalis, presence of filamentous cells.
To determine the chemical nature of biofilm extracellular matrix material from each species, we used a large-scale, roller bottle apparatus for biofilm production and matrix isolation as described previously (49). Under these conditions, net biofilm biomass was comparable among the species ( Fig. 2A), as it was in the small-scale experiments described above. The amounts of biofilm matrix mass (dry weight) in relation to the total biomass were relatively similar across species (Fig. 2B), but the relative abundance of each macromolecular matrix component varied among the species (Fig. 2C to G). Specifically, the protein component was greatest for C. albicans, while the carbohydrate component was greatest for C. parapsilosis and C. glabrata. The extracellular DNA (eDNA) component was comparable among the species (Fig. 2F). These results are consistent with those of a prior investigation (47) and indicate that macromolecular components vary slightly in their relative contributions to overall matrix composition.
Detection of matrix MGCx in non-albicans Candida species. The mannan-glucan complex (MGCx) is a signature feature of the C. albicans biofilm (40). The MGCx is comprised of an ␣-1,6-mannan backbone and ␤-1,6-glucan (40). Based on our previous report, the MGCx constitute approximately 20% of the total carbohydrate pool in the C. albicans matrix. The MGCx isolated from the non-albicans Candida species (NAC) biofilm matrices constituted 13.5%, 34.8%, and 17.0%, in C. tropicalis, C. parapsilosis, and C. glabrata, respectively (Table 1). Further gas chromatography revealed the presence of both mannan and glucan in the matrix of each non-albicans Candida biofilm (Fig. 3A). The C. albicans MGCx has a mannan/glucan ratio of 89:11 (40). We found mannan/ glucan ratios of 64:25 for C. tropicalis, 83:12 for C. parapsilosis, and 93:7 for C. glabrata ( Fig. 3A and Table 1). These findings are consistent with the production of an MGCx by each non-albicans Candida species, though the MGCx of each species may have distinct structural features.
Nuclear magnetic resonance (NMR) analyses performed on total purified NAC MGCx revealed compelling structural similarities as well as striking differences among the neutral matrix polysaccharides of the species (Fig. 3B and C and Table 2; see Table S1 in the supplemental material). Control analysis of C. albicans samples revealed several ␣and ␤-Man spin systems that resembled the high-and low-molecular-weight F2 and F17 glucomannans we reported previously (40). Of particular relevance, we found multiple signals specific for ␣-1¡2-Man␣-1¡2 residues enclosed within branched regions, such as those found in the MGCx side chains. Several of these spin systems ( Fig. 3A and B, peaks B, D, H, I, K, and L) were common to C. albicans, C. parapsilosis, and C. tropicalis. Fewer spin systems were common to C. albicans and C. glabrata (Fig. 3B, FIG 2 NAC form biofilm matrix of variable quantity and quality. (A) Total biofilm mass was assessed by measurements (dry weight) of in vitro biofilms grown in polystyrene roller bottles (three replicates of 20 bottles per species). The four species studied were C. albicans (CA), C. tropicalis (CT), C. parapsilosis (CP), and C. glabrata (CG). The single asterisks indicate statistically significant lower values (P Ͻ 0.001) for C. tropicalis and C. glabrata based upon ANOVA using the Holm-Sikak method for pairwise comparison. (B) Biofilm matrix biomass was quantified by the dry weight following matrix separation from biofilm cells. Biofilms were grown in polystyrene roller bottles (three replicates of five bottles per species). Two asterisks indicate statistically lower values for C. tropicalis (P ϭ 0.003) and C. glabrata (P ϭ 0.005) between strains based upon ANOVA using the Holm-Sikak method for pairwise comparison. (C) Biofilm matrix total carbohydrate concentration was assessed using the phenol-sulfuric acid assay. The results were normalized by matrix biomass. The single asterisks indicate statistically lower concentrations for C. parapsilosis (P ϭ 0.009) and C. glabrata (P ϭ 0.002) based upon ANOVA using the Holm-Sikak method for pairwise comparison. (D) Relative percent monosugar composition (Rha, rhamnose; Rib, ribose; Man, mannose; Glu, glucose) in the biofilm matrix of C. albicans, C. tropicalis, C parapsilosis, and C. glabrata. The single asterisks indicate statistically significant differences (P Ͻ 0.001) between strains based upon ANOVA. (E) Biofilm matrix total protein concentration was assessed using the BCA protein assay kit. The results were normalized by matrix biomass. The single asterisks indicate statistically lower concentrations for C. tropicalis, C. parapsilosis, and C. glabrata (P Ͻ 0.001) than for C. albicans based upon ANOVA using the Holm-Sikak method for pairwise comparison. (F) Biofilm matrix total eDNA. The results were normalized by matrix biomass. The single asterisks indicate statistically lower concentrations for C. tropicalis, C. parapsilosis, and C. glabrata (P Ͻ 0.001) than for C. albicans based upon ANOVA using the Holm-Sikak method for pairwise comparison. (G) Biofilm matrix total lipid concentration was assessed by gas chromatography. The results were normalized by matrix biomass. Dominguez et al. ® peaks C, I, K, and L), as may be expected from their greater phylogenetic distance. We also found peaks characteristic of ␤1,6-glucan in two-dimensional (2D) NMR analysis (Fig. 3B, peaks P and O). In addition to differences in the presence or absence of specific peaks, differences among the polysaccharide compositions were also evident from the quantitative differences among individual anomeric peaks ( Fig. 3B and Table 2).
We used a biochemical approach to test for a physical interaction among the matrix polysaccharide components. Specifically, we assessed the glycosyl composition of matrix polysaccharides following a series of neutral sugar purification and fractionation steps. For each species fraction, we identified both mannan and glucan in the highmolecular-weight (HMW) and low-molecular-weight (LMW) fractions. Comigration of both mannan and glucan components strongly suggests the presence of covalent bonds between those two polymers; however, their definite nature is not completely understood. This observation is consistent with coelution of each monosugar as a distinct mannan-glucan complex polysaccharide in the NAC biofilm matrix (Table 1). However, in keeping with analysis described above, the ratios of mannan and glucan varied somewhat across the species, suggesting structural differences in this matrix component among the NAC. In addition, there were several distinct mannan-glucan complexes of different sizes for C. glabrata. These cofractionation results support the model that each non-albicans Candida species produces a biofilm matrix MGCx.
We also used pharmacological and enzymatic approaches to test predictions of the MGCx model. Both polysaccharides are required for MGCx assembly in this model; hence, disruption of either mannan or glucan alone would be predicted to decrease matrix accumulation of the other (40). To inhibit matrix mannan production, we utilized tunicamycin, an inhibitor of N-glycosylation, and ␣-mannosidase, an enzyme that hydrolyzes and degrades mannan (39). Either tunicamycin or ␣-mannosidase treatment of biofilm reduced the matrix mannan content on average by 28% and 25%, respectively ( Fig. 3C and D). Each treatment caused a concomitant decrease in the concentration of glucan that was nearly identical in magnitude (25% and 27%, respectively) ( Fig. 3C and D). The finding that matrix glucan accumulation is dependent upon matrix mannan supports the model that polysaccharide interaction leading to an MGCx is conserved across these Candida species (39). The carbohydrate type indicates whether carbohydrate is associated either with the uncharged neutral fraction or the charged bound fraction (glycoproteins). c Values are percentages of the listed fractions in the total matrix carbohydrate pool. Content (%) represents the percent content within the neutral pool for each species. d MWF is the molecular weight fraction for each isolated polymer within the neutral carbohydrate pool. HMWF and LMWF are the high-molecular-weight-reaction and low-molecular-weight fraction, respectively.  Table  2. (B) Comparison of C. albicans NMR spectra spin systems to each of the NAC species (C. tropicalis, C. parapsilosis, and C. glabrata, respectively) as reflected by heteronuclear single quantum coherence (HSQC) and nuclear Overhauser effect spectroscopy (NOESY) data. (C and D) Carbohydrates in the matrix of wild-type biofilms treated with tunicamycin (C) and ␣-mannosidase (D) were quantified by gas chromatography. Data are presented as percentages of the reference strain (Ref), with means Ϯ standard errors (SEs) (error bars) shown. All values were significantly lower than the reference value according to ANOVA as indicated by the single asterisks.

Mannan-glucan matrix function in NAC species.
Prior studies have linked a majority of the C. albicans biofilm resistance phenotype to sequestration of antifungal drugs by the extracellular matrix (28,34,36,37,39,(50)(51)(52). The MGCx appears strongly linked to this drug sequestration phenomenon. Therefore, we examined the drug susceptibility of biofilms both in vitro and in vivo using a rat vascular catheter model for each of the Candida species (53, 54) ( Fig. 4A and B). For C. albicans, C. tropicalis, and C. glabrata, the highest soluble concentration of the antifungal fluconazole (1,000 g/ ml, which is more than 1,000 times the planktonic MIC) did not appreciably impact biofilm cell burden in either model, consistent with prior biofilm antifungal testing for these species (55)(56)(57)(58). For the strain of C. parapsilosis studied, treatment was more effective compared to the other species. However, the fluconazole concentration needed for effective treatment remained 100 times greater than that associated with planktonic efficacy. These results confirm that biofilms of these species are dramatically less sensitive to fluconazole than planktonic cells. To evaluate drug sequestration as a mechanism underlying the biofilm-associated drug resistance, we tracked radiolabeled fluconazole within the biofilm of each Candida species. Nearly all of the antifungal accumulated in the biofilm matrix for each species (Fig. 4C). Interestingly, despite the fact that each species was exposed to a similar concentration of fluconazole, the total sequestered drug concentrations were lower in the NAC biofilms. Because matrix levels are relatively similar among the species (Fig. 2B), we speculate that differences in matrix-drug binding affinity among the species account for differences in drug accumulation.
We further explored the capacity of NAC matrices to interact with fluconazole using one-dimensional 1 H NMR. In our experiments, interactions of the antifungal drug with the matrix were evident by changes in the intensity of 1 H peaks upon increasing concentrations of matrix material. Similarly to our previous report (40), the signals of both aromatic and azole protons decreased as a function of increasing concentration of matrices, suggesting that both the aromatic ring and the heterocyclic triazole rings were involved in interactions with matrix components. Interestingly, each NAC matrix had a distinct drug binding dynamics profile (Fig. 4D). The degree of interaction in this assay was greatest for C. albicans, consistent with differences in matrix affinity among the species.  a CA, C. albicans; CG, C. glabrata; CT, C. tropicalis; CP, C. parapsilosis. Blank cells indicate that the residue was not found. b F2 and F17 refer to high-and low-molecular-weight MGCx previously reported in the biofilm matrix of C. albicans (40). Blank cells indicate that the MGCx was not found. c (l), residues within a linear region, i.e., residues whose neighbors are not branching residues. d (b), residues within a branched region, i.e., residues whose neighbors are branching residues.
If the MGCx were responsible for functional drug sequestration, then MGCx disruption should increase biofilm drug susceptibility. We tested this prediction through pharmacological and enzymatic treatments. Mannan accumulation was reduced by either tunicamycin or ␣-mannosidase treatment as shown above. In addition, both mannan and ␤-1,6-glucan accumulation were reduced by brefeldin A treatment (59). These treatments alone did not influence biofilm cell viability. However, each treatment FIG 4 NAC biofilm drug resistance phenotype and mechanism. (A) Biofilm antifungal susceptibility (fluconazole [125 or 1,000 g/ml] for 48 h) of wild-type C. albicans, C. tropicalis, C. parapsilosis, and C. glabrata was assessed using an XTT assay in a 96-well polystyrene plate assay. The asterisk indicates a statistically significant difference (P Ͻ 0.001) between strains based upon ANOVA using the Holm-Sikak method for pairwise comparison. (B) Biofilm antifungal susceptibility (fluconazole [250 g/ml] after 24-h exposure) of wild-type C. albicans, C. tropicalis, C. parapsilosis, and C. glabrata was assessed using viable counts from the rat vascular catheter biofilm model. (C) Fluconazole sequestration and binding to the Candida species biofilm extracellular matrix. Sequestration of 3 H-labeled fluconazole was assessed using in vitro intact biofilms as well as the extracellular matrix and intracellular components. (D) Fluconazole binding to the NAC biofilm extracellular matrices. Fluconazole interactions with the tested matrices studied by one-dimensional 1 H NMR at 600 MHz were determined as decreases in the intensity of chemical shift peaks characteristic of protons present either in the heterocyclic azole rings or the aromatic ring of the drug. Spectra were recorded at the constant fluconazole concentration of 0.653 mM, and matrix concentrations ranged from 0 up to 8 mg/ml. (E) Biofilms were treated with pharmacological inhibitors of mannan or glucan or a mannan hydrolysis enzyme both with and without 1,000 g/ml fluconazole for all species except C. parapsilosis for which we used 250 g/ml. Efficacy was assessed in a 96-well plate format for quantification with the XTT assay. The asterisks indicate statistically significant differences (P Ͻ 0.001) between the combination and either treatment alone based upon ANOVA using the Holm-Sikak method for pairwise comparison. FLUC, fluconazole; TM, tunicamycin; BFA, brefeldin A; ␣-MS, ␣-mannosidase.
augmented the activity of fluconazole, resulting in profound killing of biofilm cells (Fig. 4D). These effects were biofilm specific, as these agents did not enhance the activity of fluconazole against planktonic cells (Fig. S1). In sum, these results support the hypothesis that MGCx contributes to sequestration of antifungals to promote biofilm resistance across Candida species.
Genetic determinants of NAC matrix MGCx structure and function. In order to identify genetic determinants of matrix production across Candida species, we used a candidate gene approach. Hypothesizing genetic conservation among species, we identified orthologs of 12 C. albicans genes shown to be involved in production or modification of matrix mannan or ␤-1,6-glucan (Table 3) (34,39). We successfully constructed homozygous deletion mutations for the genes in each species with the exceptions of ALG11 and VRG4 in C. tropicalis, BIG1, KRE5, and VRG4 in C. parapsilosis, and KRE5 and VRG4 in C. glabrata. We speculate that these genes may be essential in the respective species. Most of the mutants formed biofilms similar to their reference strains. However, three mutants formed biofilms with reduced cell numbers (Fig. S2), including C. parapsilosis mnn11Δ/Δ, pmr1Δ/Δ, and phr1Δ/Δ mutants and C. glabrata mnn11Δ and big1Δ mutants.
We examined each of the mutant biofilms by electron microscopy to visualize biofilm architecture and matrix deposition. Biofilm architecture seemed largely unaffected by the mutations, but there was a striking reduction of visible extracellular matrix in seven of the mutant biofilm strains (Fig. 5). This suggests that each of these genes is required for biofilm matrix production in the respective NAC species.
We next examined the influence of the genetic disruptions on the biofilm resistance phenotype in vitro ( Fig. 5 and Fig. S3 and S4). Surprisingly, despite the importance of  Conservation and Divergence of Candida Biofilm Matrix ® each of these genes for biofilm resistance in C. albicans, only a small subset impacted susceptibility in the NAC biofilms. Specifically, we found a total of 7 of 29 deletion mutants exhibiting enhanced biofilm drug susceptibility (Fig. 6A). For C. tropicalis, fluconazole-susceptible strains included mannan synthesis mutants, i.e., mnn9Δ/Δ and van1Δ/Δ mutants, and the ␤-1,6-glucan synthesis mutant, the kre5⌬/⌬ mutant. Among the C. parapsilosis mutants screened, we identified three mannan synthesis mutants, mnn9Δ/Δ, van1⌬/⌬, and pmr1⌬/⌬ mutants. Interestingly, two of the C. parapsilosis orthologs were also relevant for C. tropicalis (mnn9⌬/⌬ and van1⌬/⌬). Only a single mutant from C. glabrata exhibited a biofilm susceptibility phenotype: a putative enzyme for ␤-1,6-glucan synthesis (big1⌬). The susceptibility to fluconazole appeared specific to the biofilm state, as planktonic MICs for all mutants were unchanged (Fig. 6A). The biofilm susceptibility phenotype was reversed for all strains in which we reintroduced a wild-type copy of the deleted gene (Fig. S5). For three C. parapsilosis mutants (mnn9⌬/⌬, pmr1⌬/⌬, and van1Δ/Δ mutants), complementation was not successful; consequently, we examined multiple independent deletion mutant strains to verify the enhanced antifungal susceptibility phenotype (Fig. S6). Antifungal biofilm susceptibility was similarly assessed for two additional drugs, amphotericin B and micafungin (Fig. S4). Enhanced susceptibility was observed for the majority of mutants, consistent with a pan-antifungal mechanism. These results show that inferred defects in mannan or ␤-1,6-glucan production can cause increased biofilm drug susceptibility in NAC species.
We explored the clinical relevance of these findings using the rat catheter in vivo biofilm model. Enhanced fluconazole efficacy was observed for representative mannan or ␤-1,6-glucan synthesis mutants from each species (Fig. 6B). These results support the hypothesis that the increased biofilm drug susceptibility that results from impaired mannan or ␤-1,6-glucan synthesis is relevant in a model infection environment.
To determine directly whether matrix mannan and glucan are required for drug sequestration in the NAC biofilm mutants, we conducted [ 3 H]fluconazole binding assays. Compared to the respective wild-type strain, each of the mannan and glucan mutant strains had decreased matrix sequestration of radiolabeled drug (Fig. 6C). These  findings strengthen the conclusion from our pharmacological and enzymatic matrix disruption treatments that mannan and glucan contribute to matrix drug sequestration. However, variability in the degree of sequestration relative to the change in susceptibility suggests the potential of additional mechanisms as previously described (28).
To identify matrix changes associated with increased biofilm susceptibility, we harvested matrix from biofilms of each mutant and assayed for total carbohydrate and specifically for mannan and glucan ( Fig. 6D and 5E). We found that all seven mutants had significantly lower levels of the corresponding polysaccharide than the reference strain with remaining amounts less than 50% of the level in the wild type on average (Fig. 6E). Furthermore, genetic disruption of either a mannan or glucan synthase was associated with a lower content of both mannan and glucan. This observation is consistent with the model that the MGCx accounts for the drug resistance phenotype of the NAC biofilms. Unfortunately, matrix quantities did not allow for complementary NMR analysis. Because the synthesis pathways for the carbohydrates are distinct, inhibition of one pathway would not be expected to lead to a biochemical reduction in the nonimpacted pathway components. Our findings suggest that there is an extracellular physical interaction among the matrix components that is conserved among Candida species.

DISCUSSION
We theorized that the MGCx, recently found in C. albicans, is conserved and promotes the drug resistance phenotype across Candida species. In order to address this question, in this study, we first utilized large-scale biofilm growth and matrix composition analysis. We find that each of the major matrix polysaccharide constituents is required for assembly and function of matrix across Candida species. At the structural level, several complementary assays corroborate the presence of an MGCx in diverse Candida species. Specifically, we identified an ␣-1,6-mannan backbone with ␣-1,2 branches and ␤-1,6-glucan in the matrix. Analysis of the glycosyl composition of matrix after fractionation also revealed coelution of mannan and glucan components. Additionally, both pharmacological and genetic studies revealed codependence of accumulation of matrix mannan and glucan. Phenotypic studies documented the central role of this polysaccharide complex in matrix sequestration and biofilm-specific drug resistance under both in vitro and in vivo conditions. However, the results of refined NMR analysis of the complex and phenotypic and genotypic assays suggested meaningful differences in the MGCx among the Candida species. Specifically, the relative ratios of the mannan and glucan components varied among the four species. Additionally, while NMR analysis revealed similar mannan and glucan backbone features, there were differences in branching pattern and length. The C. parapsilosis matrix mostly resembled low-molecular-weight glucomannans of C. albicans. The abundance of ␣-1¡6-linked Man residues along with a substantially increased content of terminal Man␣-1¡2 residues indicated the presence of a similar ␣-1¡6-linked Man backbone structure but with shorter side chains consisting of a blend of ␣-1¡2and ␣-1¡3-linked Man residues. The lack of ␤-Man for C. parapsilosis matrix was unique, as ␤-Man was observed for the other species. The matrix of following treatment. The values are means Ϯ standard deviations (error bars) from three replicates. The asterisks indicate that the CFU values were significantly different from the CFU for the reference strain (P Ͻ 0.001) based upon ANOVA using the Holm-Sikak method for pairwise comparison. (C) Intact biofilms grown from the wild-type and mutant strains were exposed to [ 3 H]fluconazole, washed, and harvested. Scintillation counting was performed in triplicate to determine the fluconazole content in the intact biofilms and the isolated matrix. Standard deviations are shown. (D) Mature in vitro biofilms from the wild-type strain and null mutants were assayed for matrix carbohydrate concentration using the phenol sulfuric acid method. The value for each mutant is presented as a percentage of the value for the wild-type strain. The graph shows data from three biological replicates and three assay replicates. The asterisks indicate that glucan measurements were significantly different (P Ͻ 0.0001) from the wild-type measurement based on ANOVA. (E) Mature in vitro biofilms from the wild-type strain and null mutants were assayed for matrix glucan and mannan concentrations by gas chromatography. The value for each mutant is presented as a percentage of the value for the wild-type strain. The graph shows data from three biological replicates and three assay replicates. The asterisks indicate that glucan measurements were significantly different (P Ͻ 0.0001) from the value for the wild-type strain based upon ANOVA. C. tropicalis exhibited characteristics similar to those of C. albicans and C. parapsilosis matrix. Like the latter, this pool contained amounts of ␣-1¡2-Man␣-1¡2and 1¡6linked Man as well as of ␤-Man residues that were comparable to those measured in C. albicans. On the other hand, C. tropicalis matrix neutral carbohydrates did not contain any 3-linked Man residues, whereas there was a larger terminal Man␣-1¡2 residue content, along with more 1¡6-linked Man. This pattern suggests more branching but relatively shorter side chains. On the basis of measured distribution of 2,6-Man␣-1¡6 and ␣-1¡2-Man␣-1¡2 residues, it appeared that the overall branching pattern is bit less diverse than that observed in C. albicans matrix. The analysis of C. glabrata matrix neutral carbohydrates revealed the most distinct highly branched structure, which also contained well-defined linear regions without any branching Man residues present. This observation suggested an uneven and infrequent distribution of side chains in this carbohydrate pool as reflected by the presence of only one type of ␣-1¡2-Man␣-1¡2 signal, whereas the amount of the terminal Man␣-1¡2 residue indicated a side chain length distribution type resembling the distribution determined in C. albicans. Unlike the latter, C. glabrata matrix neutral carbohydrates did not contain 3-linked Man residues, while the content of ␤-Man was significantly reduced but still detected. Overall, the major Man residues were similar among all four tested Candida species. However, the relative proportion of the residues varied significantly, likely due to differences in the degree of branching, the length of side chains, and the presence of unique residues such as ␤-Man or 3-linked Man. We speculate that these structural differences in the MGCx and potentially other noncarbohydrate constituents determine the unique sequestration properties of the NAC.
Our genetic studies reveal both similarities and differences among the species with regard to the genetic pathways responsible for production of matrix. It was initially surprising that the only orthologous mutations in the NAC that impacted the antifungal resistance phenotype were those linked to component synthesis alone. Conversely, the mutations with putative glucan and mannan modification function based upon the C. albicans MGCx structure had no apparent impact on NAC matrix production. A plausible explanation for this observation is divergence in the genetic pathways responsible for the differences in the impact of C. albicans orthologous mutations in the NAC species. Researchers have often assumed that if a yeast species is related to another yeast species (especially within the same genus), the underlying molecular and cellular mechanisms must also be closely related. However, even within a Candida clade, the genetic relatedness between any two NAC species is often larger than the genetic distance between humans and reptiles (60). There is ample precedent for genetic rewiring among the Candida species, including for biofilm formation pathways (13,15,16,(61)(62)(63). While sequence identity is not a great predictor of function, the amino acid similarity across species for these C. albicans matrix resistance genes was only modest.
Another potential explanation for the lack of genetic conservation is the distinct differences in MGCx branching structure. We favor this model due to the biochemical differences that support a model of altered matrix branch chain structure across species. The latter model suggests that enzymes needed for production of the matrix backbone may be useful for pan-Candida biofilm development of useful therapies, while targeting modification enzymes would be predicted to be less effective across species. In future work, careful characterization of the remaining genetic components of the NAC biofilm synthesis pathway will help to elucidate additional therapeutic targets. in YPD, and biofilms were grown in RPMI 1640 buffered with morpholinepropanesulfonic acid (RPMI-MOPS). Strains and strain construction. Strains used for this study are listed in Table 3, and the genotypes of strains constructed in the present studies are shown in Table S2 in the supplemental material. The parent strains CAY3764, CPL2H1, and HTL were used to create homozygous deletion strains using fusion PCR disruption cassettes as previously described for C. tropicalis, C. parapsilosis, and C. glabrata, respectively (6,(65)(66)(67). Correct integration was confirmed by PCR. At least two independent mutants were created for each gene of interest. The primers used for strain construction and confirmation are listed in Table S3.

MATERIALS AND METHODS
Construction of C. tropicalis mutant strains was conducted using the background strain of CAY3764, which is auxotrophic for LEU2 and HIS1 (67). The primers used for strain construction and confirmation are listed in Table S3. For C. parapsilosis strain construction, we used the background strain of CPL2H1, which is auxotrophic for LEU2 and HIS1, as described by Holland et al. (6). The first allele was deleted by replacing one allele with HIS1 from C. dubliniensis, and the second with LEU2 from C. maltosa. C. glabrata transformations were conducted using auxotrophic background strain HTL (HIS3, LEU2, TRP1) (63). Candidate genes were deleted by replacing the allele with the gene conferring resistance to the antibiotic nourseothricin. All mutant strains were confirmed by PCR using primers inside nourseothricin and a primer outside the integration sites at both the 5= and 3= ends of the gene.
Complementation of C. tropicalis and C. parapsilosis mutant strains with a single wild-type gene copy used selection for nourseothricin resistance, whereas complementation of C. glabrata mutant strains used selection for hygromycin B resistance. A single copy of the gene was reintroduced to its endogenous location within the genome. Briefly, each open reading frame (ORF) (plus 1 kb upstream and downstream) was amplified by PCR, and using the PCR fusion method, a cassette was created with either hygromycin B or nourseothricin at the tail end of the PCR product. This fusion PCR product was then transformed into their respective species using the same method as previously described for mutant construction. Colony PCR was used to verify all genotypes; the primers are listed in Table S3.
In vitro biofilm models. Biofilms were grown in one of four models: 96-well or 6-well polystyrene plate, polystyrene roller bottle, or glass coverslip. Ninety-six-well flat-bottom polystyrene plates were used to assess biofilm adherence, maturation, and treatment effect as previously described (68,69). The Candida species inocula (10 6 cells/ml) were prepared by growth in YPD with uridine overnight at 30°C, followed by dilution in RPMI-MOPS based on hemocytometer counts. The six-well plate assay was used to assess matrix composition. For this assay, 1 ml of culture was inoculated in each well. After a 60-min adherence period at 30°C, the nonadherent inoculum was removed and 1 ml of fresh medium (RPMI-MOPS) was applied to each well. The plates were incubated at 37°C for 48 h on an orbital shaker set at 50 rpm. The medium was removed, and fresh medium was added midway through the incubation period. The coverslip assay was used for in vitro biofilm imaging. Briefly, in vitro biofilms were grown on sterile coverslips (Thermanox) in sterile 12-well plates that had previously been coated with 10 l of human NaEDTA plasma each and allowed to dry at 30°C. Forty microliters of yeast in RPMI was counted and diluted as in the biofilm models described above and added to each coverslip for 60 min at 30°C. The initial inoculum was then removed, 1 ml of RPMI-MOPS plus 5% NaEDTA human plasma was added to each well, and the plates were incubated for 20 h at 37°C and 50 rpm on an orbital shaker for an additional 24 h. A rolling bottle system was used to generate matrix for analyses (49). Briefly, aliquots of C. albicans grown in RPMI-MOPS were used to inoculate a polystyrene roller. Bottles were placed on a roller apparatus (Wheaton Science Products, Millville, NJ), rolling at the rate of 20 rpm at 37°C. After 24 h, the biofilm culture medium was replaced with fresh medium, and the bottles were incubated for another 24 h. At least three biological replicates were performed for each assay.
In vitro biofilm and planktonic antifungal susceptibility testing. A tetrazolium salt XTT [2,3-bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide inner salt] reduction assay was used to measure in vitro biofilm drug susceptibility (70,71). Biofilms were formed in the wells of 96-well microtiter plates as described above. After a 6-h biofilm formation period, the biofilms were washed with phosphate-buffered saline (PBS) twice to remove nonadherent cells. Fresh RPMI-MOPS and drug dilutions were added, followed by additional periods of incubation (48 h). The antifungal studies included fluconazole at 4 to 1,000 mg/ml, amphotericin B at 125 to 0.5 g/ml, and micafungin at 125 to 0.5 g/ml. For experiments with tunicamycin (1.0 g/ml) and brefeldin A (0.6 g/ml), biofilms were treated alone or in combination with fluconazole after an initial 6-h growth phase. Biofilms treated with ␣-mannosidase (0.78 U/ml; jack bean; Sigma) were grown for 24 h before a 24-h dose either alone or in combination with fluconazole. Drugs were reapplied after 24 h, and the plates were incubated for an additional 24 h. Following treatment with 90 l XTT (0.75 mg/ml) and 10 l phenazine methosulfate (3.20 mg/ml) for 30 min, absorbance at 492 nm was measured using an automated plate reader. The percent reduction in biofilm growth was calculated using the reduction in absorbance compared to that of controls with no antifungal treatment. Assays were performed in triplicate, and significant differences were measured by analysis of variance (ANOVA) with pairwise comparisons using the Holm-Sidak method. The CLSI M27 A3 broth microdilution susceptibility method was used to examine the activities of fluconazole against planktonic Candida. sp. (72). Endpoints were assessed after 24 h by visible turbidity in triplicate assays.
Biofilm SEM. In vitro biofilms from sterile coverslips were grown as described above. Following a 24-h incubation period, the medium was replaced with 1 ml of fixative (4% formaldehyde, 1% glutaraldehyde in PBS), and the coverslips were incubated at 4°C for 24 h. The coverslips were then washed with PBS and treated with 1% osmium tetroxide for 30 min at ambient temperature. After a series of alcohol washes (30 to 100%), final desiccation was performed by critical point drying. Coverslips were mounted, palladium-gold coated, and imaged in a scanning electron microscope (SEM) (LEO 1530) at 3 kV. The images were assembled using Adobe Photoshop 7.0.1.
Matrix isolation from roller bottle and six-well biofilms. A rolling bottle system was used to generate matrix for composition analyses (49). After incubation for 48 h, the medium was removed, and the Candida biofilms were dislodged with a spatula and gently sonicated to avoid cell wall disruption (sonication with a 6-mm microtip at 20 kHz with an amplitude of 30% for 8 min). The aggregate biofilm was then centrifuged to separate fungal cells and matrix. The supernatant-containing matrix was then collected and lyophilized. The sample was resuspended in water and dialyzed (molecular size cutoff of 3.5 kDa) for 5 days and again lyophilized, yielding the "crude" biofilm matrix. Overall, a total of 400 bottles of the matrix corresponding to the biofilm area of 59.5 m 2 were collected for analysis.
Matrix was similarly collected from six-well plates as described previously (34). Following incubation, biofilms were harvested by removing and discarding the medium, washing each well with 1 ml of double-distilled water (ddH 2 O), and then removing the biofilms using a spatula. Biofilms were collected in 1 ml of ddH 2 O per well and then sonicated for 20 min. The soluble matrix was then separated from the cells by centrifuging the samples at 2,880 ϫ g for 20 min at 4°C.
Biofilm matrix analysis. (i) Dry weight. Following biofilm matrix collection described above, matrix was lyophilized, and weighed to obtain total biomass for each biofilm.
(ii) Carbohydrate analysis. The carbohydrate concentration of crude matrix was determined colorimetrically (492 nm) using the phenol-sulfuric acid method (73). Structural analysis was performed after a series of purification and fractionation steps, including size exclusion chromatography.
(iii) Protein analysis. The protein concentration of the crude matrix sample was assessed colorimetrically at 562 nm using the BCA protein assay kit (Pierce Biotechnology, Rockford, IL) (74).
(v) Monosaccharide analysis. Sugars were detected and quantified by gas-liquid chromatography with a flame ionization detector (GLC-FID) on a Shimadzu GC-2010 system after conversion to alditol acetate derivatives as previously described (76). A 50% cyanopropylmethyl-50% phenylmethyl polysiloxane column was used (Restek) under the GLC conditions previously described (77). Data for each monosaccharide were calculated and presented as a percentage of the total detected sugars.
(vi) Lipid analysis. Lipids were extracted from the desalted lyophilized matrix powder with a mixture of CHCl 3 and methanol (MeOH) (2:1, by volume) as described elsewhere (78). Methylation of fatty acids was performed using 0.5 ml of 14% boron trifluoride (BF3) in MeOH, and methyl esters were recovered with hexane. Fatty acid methyl esters were analyzed by gas chromatography using a Hewlett-Packard 5890 instrument (Hewlett Packard, Palo Alto, CA).
(vii) eDNA analysis. Extracellular DNA (eDNA) was isolated from non-albicans Candida species (NAC) matrices using the MasterPure yeast DNA purification kit (Epicentre Biotechnologies, Madison, WI). Nucleic acid concentrations were measured spectrophotometrically with a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, San Jose, CA). The average measured ratio of absorbance at 260/ 280 nm was about~1.8, which indicated that the DNA was pure and free of contaminants.
NMR of neutral carbohydrates. After isolation, matrix samples were resuspended in 1 ml of 20 mM bis-Tris HCl (pH 6.5) loading buffer and fractionated on a HiPrep 26/10 desalting column prepacked with Sephadex G-25 Fine (GE Healthcare Life Sciences, Uppsala, Sweden). Column-dialyzed fractions were then separated on an anion exchanger HiPrep 16/10 DEAE FF column (GE Healthcare Life Sciences) equilibrated with 20 mM bis-Tris HCl (pH 6.5). Elution was performed in a 20 mM bis-Tris HCl (pH 6.4)-0.5 M NaCl buffer system at a flow rate of 1 ml/min in a linear gradient of salt from 0 to 100% in 20 column volumes. Neutral free carbohydrates were detected in flowthrough fractions, which were then pooled, lyophilized, resuspended in 2 ml of 150 mM NH 4 HCO 3 , and applied to gel filtration on a Superdex 200 10/300 GL column (GE Healthcare Life Sciences). Matrix components were eluted at a flow rate of 0.5 ml/min, and 1-ml fractions were collected. All chromatographic separation steps were performed at room temperature on the high-performance liquid chromatography Äkta-Purifier 10 system (GE Healthcare Life Sciences). All buffers used were filtered through 0.2-m nylon membrane filters (Nalgene, Rochester, NY) and degassed prior to use. Isolated polysugar fractions were lyophilized, resuspended in a small volume of water, and incubated at 55°C overnight in order to decompose and remove any remaining ammonium bicarbonate. These steps were repeated until all of the salt was removed and the isolated sugars appeared as an anamorphous cotton-like material after the final lyophilization. The molecular sizes of biofilm matrix neutral carbohydrates were estimated using size exclusion column calibration with a set of Leuconostoc species dextran standards, which included 100-kDa, 70-kDa, 40-kDa, 25-kDa, and 6-kDa polymers.
To elucidate the structure of the isolated carbohydrates, we utilized a combination of onedimensional (1D) 1 H nuclear magnetic resonance (NMR) and 2D heteronuclear single quantum coherence (HSQC) NMR experiments as well as known chemical shift assignments characteristic of individual mannosyl and glucosyl motifs in mannan and glucan structures based on previously published studies (79)(80)(81)(82)(83). All data were collected at 70°C on a Bruker Biospin Avance III 500-MHz NMR spectrometer (Bruker BioSpin GmbH, Rheinstetten, Germany) equipped with a 5-mm triple resonance, cryogenic probe, CPTXI 500 H-C/N-D. One-dimensional spectra were collected with 32 acquisitions using a standard one-pulse experiment. The spectral width was 10 ppm centered at 4.7 ppm. The relaxation delay time was 2 s with an acquisition time of 3.3 s (32,768 data points). Thirty-two acquisitions were collected. Multiplicity-edited, phase-sensitive, echo-antiecho 1 H HSQC spectra were obtained using four acquisitions per indirect time point with 1 H decoupling during acquisition (84). Matched swept adiabatic 13 C inversion pulses were used. The raw data matrix size was 2,048 ϫ 128 blocks. Spectra were collected with Conservation and Divergence of Candida Biofilm Matrix ® a relaxation time delay of 2 s and an acquisition time of 0.2 s with sweep widths of 10 ppm ( 1 H) and 65 ppm ( 13 C), respectively. The centers of the spectra were 4.7 ppm for 1 H and 82 ppm for 13 C.
Matrix carbohydrate fractionation and analysis for an MGCx. Additional structural analysis was performed after a series of purification and fractionation steps to further ascertain the presence of a mannan-glucan complex. These steps include size exclusion chromatography followed by separation on an anion exchanger HiPrep 16/10 DEAE FF column (GE Healthcare Life Sciences). Neutral free carbohydrates were collected in flowthrough fractions, which were pooled and applied to gel filtration on a HighPrep 16/60 Sephacryl S-300 HR column (GE Healthcare Life Sciences), yielding 22 individual polysaccharide peaks, F1 to F22. The molecular weight of biofilm matrix neutral carbohydrates was determined using size exclusion column calibration with a set of Leuconostoc species dextran standards (polymers with molecular weights [in thousands] of 100, 70, 40, 25, and 6). Both high-molecular-weight (HMW) and low-molecular-weight (LMW) fractions were examined by GC for monosugar analysis. Matrix monosugar composition and quantification was performed on alditol acetate derivatives by GLC-FID (Shimadzu GC-2010 system; Shimadzu, Kyoto, Japan).
In vivo Candida venous catheter biofilm model. A jugular vein rat central venous catheter infection model was used for in vivo biofilm studies (53). Candida strains were grown to logarithmic phase in YPD at 30°C. After a 24-h conditioning period after catheter placement, infection was achieved by intraluminal instillation of 500 ml of C. albicans (10 6 cells/ml). After an adherence period of 6 h, the 500-ml volume was withdrawn, and the catheter was flushed with heparinized saline. Following a 24-h incubation period, the catheters were removed, and the animals were prepared for SEM imaging as described above. The images were assembled using Adobe Photoshop 7.0.1.as described above. For drug treatment experiments, fluconazole (250 g/ml) was instilled in the catheter after 24 h of biofilm growth. After a 24-h drug treatment period, the posttreatment viable burden of Candida biofilm on the catheter surface was measured by viable plate counts on Sabouraud's dextrose agar (SDA) following removal of the biofilm by sonication and vortexing. We utilized three replicates for each condition.

Sequestration of [ 3 H]fluconazole in biofilms.
Radiolabeled fluconazole was used in an assay to assess drug retention in biofilms formed in six-well plates (85). Biofilms were grown for 48 h in six-well polystyrene plates as described above, washed, and then incubated with 8.48 ϫ 10 5 cpm of [ 3 H]fluconazole (Moravek Biochemicals; 50 M, 0.001 mCi/ml in ethanol) in RPMI-MOPS for 30 min at 37°C with orbital shaking at 50 rpm. Unlabeled fluconazole (20 M) in RPMI-MOPS was added for an additional 15-min incubation period. After washing, biofilms and matrix were collected and isolated as described above. For a subset of biofilm cells, cells were disrupted by bead beating to yield cell wall and intracellular portions. Samples were added to a Tri-Carb 2100 TR liquid scintillation analyzer after adding ScintiSafe 30% liquid scintillation counting reagent (LSC) mixture to each sample fraction. The values for three technical replicates were averaged, the standard errors (SEs) were calculated, with values compared to the reference strain using pairwise comparisons with ANOVA by the Holm-Sidak method.
Fluconazole-matrix interaction determination. Interactions between the biofilm matrix and fluconazole were also probed using one-dimensional 1 H NMR. Data were collected at 37°C on a Bruker Biospin Avance III 600-MHz NMR spectrometer (Bruker BioSpin GmbH) equipped with a 1.7-mm triple resonance, cryogenic probe, CPTXI 500 H-C/N-D. One-dimensional spectra were collected with 512 acquisitions using a one-pulse sequence experiment with water suppression and excitation sculpting with gradients (zgesgp). The spectral width was 16 ppm centered at 4.7 ppm. The relaxation delay time was 2 s. The approach was based on monitoring chemical shifts of fluconazole-specific protons in the presence and absence of the biofilm matrix under different pH conditions. In this study, all tested reactions were prepared in PBS (pH 7.2) and fluconazole was used at a constant concentration of 0.653 mM. In this drug-matrix system, interactions were represented by decreasing in signal intensities of the chemical shift peaks of protons present in fluconazole.

ACKNOWLEDGMENTS
This study made use of the National Magnetic Resonance Facility at the University of Wisconsin-Madison, which is supported by NIH grant P41GM103399 (NIGMS) (old number, P41RR002301). Equipment was purchased with funds from the University of Wisconsin-Madison, NIH (P41GM103399, S10RR02781, S10RR08438, S10RR023438,