Structural basis for promiscuity and specificity during

Structural basis for promiscuity and specificity during Candida glabrata invasion of host epithelia Manuel Maestre-Reynaa, Rike Diderrichb, Maik Stefan Veeldersa, Georg Eulenburga, Vitali Kalugina, Stefan Brücknerb, Petra Kellerc, Steffen Ruppc, Hans-Ulrich Möschb,1, and Lars-Oliver Essena,d,1 a Biomedical Research Center/Department of Chemistry, Philipps-Universität Marburg, D-35032 Marburg, Germany; bDepartment of Genetics, Philipps-Universität Marburg, D-35043 Marburg, Germany; cFraunhofer-Institut für Grenzflächen und Bioverfahrenstechnik (IGB), D-70569 Stuttgart, Germany; and dSchool of Biological Sciences, Nanyang Technological University, Singapore 637551, Singapore

The human pathogenic yeast Candida glabrata harbors more than 20 surface-exposed, epithelial adhesins (Epas) for host cell adhesion. The Epa family recognizes host glycans and discriminates between target tissues by their adhesin (A) domains, but a detailed structural basis for ligand-binding specificity of Epa proteins has been lacking so far. In this study, we provide high-resolution crystal structures of the Epa1A domain in complex with different carbohydrate ligands that reveal how host cell mucin-type O-glycans are recognized and allow a structure-guided classification of the Epa family into specific subtypes. Further detailed structural and functional characterization of subtype-switched Epa1 variants shows that specificity is governed by two inner loops, CBL1 and CBL2, involved in calcium binding as well as by three outer loops, L1, L2, and L3. In summary, our study provides the structural basis for promiscuity and specificity of Epa adhesins, which might further contribute to developing anti-adhesive antimycotics and combating Candida colonization. molecular recognition

| T-antigen | lectin | fungal pathogen

T

he strict mammalian commensal yeast Candida glabrata can act as a health care associated opportunistic pathogen. In immune-compromised patients, colonization results in lethal systemic infections and sepsis (2, 3). Besides that, C. glabrata is the causative agent of several less severe, mucosa-related, oral or urogenital diseases (4, 5). Due to its resistance against traditional antimycotics (6), this pathogen is considered to be highly emergent (4), and new therapeutic tools are being actively sought. One putative target for drug development is the epithelial adhesin (Epa) machinery, which is represented in C. glabrata by up to 23 different genes. Apart from EPA1, which confers adhesion of fungal cells to mammalian surfaces in vitro (7) and in vivo (8), the function of other EPA genes is only partly understood. For example, EPA6 and EPA7 have been identified as virulence factors, and EPA6 is further linked to biofilm formation and colonization of the vaginal epithelium (6). Interestingly, most of the EPA genes are clustered in subtelomeric regions, indicating rapid genetic adaptation to varying environmental conditions during host colonization (1). Also, expression of EPA1, EPA6, and EPA7 is regulated by a distinct silencing mechanism via the Sir machinery and mutation of SIR genes causes hypervirulence and enhanced kidney colonization (9). Members of the Epa family are highly glycosylated mannoproteins with a modular architecture common to fungal GPIanchored cell wall proteins (GPI-CWP) (10). While the N-terminal region (adhesion, or A domain) resembles C-type lectins mediating calcium-dependent adhesion to carbohydrate structures, the central serine- and threonine-rich segment (B domain) comprises a variable number of highly O-glycosylated repeats. Finally, a C-terminal anchor (C domain) cross-links the adhesin covalently to the cell wall (7, 11). Epa A domains are distantly related to flocculins from Saccharomyces cerevisiae, which also harbor A domains belonging to the PA14/Flo5-like family (12, 13). For the A domain of the flocculin Flo5, mannoside recognition involves a novel motif not found in other C-type lectins www.pnas.org/cgi/doi/10.1073/pnas.1207653109

(12), the DcisD motif, whose role in galactose recognition by Epa1 has been recently confirmed (14). Other structural features determining sugar specificities by Flo5-like domains are so far unclear and do not allow correlating the different Epa proteins to their distinct and mostly still unknown specificities. Recent, semiquantitative functional characterizations of Epa1, Epa6, and Epa7 showed that the specificity of these three adhesins is related (15). All three adhesins require a terminal galactose containing carbohydrate ligand for host recognition. Epa6 binds to oligosaccharides comprising galactose linked via 1–3 or 1–4 glycosidic bonds to glucose, galactose, or their N-acetylated derivates. Furthermore, Epa6 does not discriminate between α- and β-glycosidic linkages. In contrast, Epa1 has a narrower specificity, mainly excluding α-linked carbohydrates. Epa7 is the most specific of the three adhesins and almost exclusively binds to lactose (Galβ1–4Glc), Galβ1–3Gal, or their respective N-acetylated derivatives (15). Finally, adhesion specificity of EpaA domains is claimed to be governed by hypervariable stretches comprising amino acids 114–139 and 227–231 (15). However, a structural basis correlating these findings to Epa specificities has not been established. Here, we present high-resolution crystal structures of the Epa1A domain complexed to cognate disaccharide ligands including Galβ1–3Glc or the T-antigen (Galβ1–3GalNAc). We further provide a structure-guided classification of the Epa family into specific subtypes and present structures of subtypeswitched Epa1 variants with modified binding sites that correspond to Epa2, Epa3, and Epa6. Together with high-throughput carbohydrate-binding screens and fluorescence titration analyses these structures uncover the critical differences within the Epa protein family in regard to specificity and promiscuity. In vivo, subtype-specific binding of Epa proteins to epithelial surfaces in part depends on the specificity-determining loop region CBL2. Overall, Candida species can generate a large repertoire of Epa adhesins for colonizing various host niches by exerting subtle structural changes to a common motif. Results and Discussion Epa1A Preferentially Recognizes Galβ1–3 Glycans. To determine the carbohydrate specificity profile of Epa1, the recombinant, fluorescence-labeled Epa1A domain was analyzed by glycan array screening [Consortium for Functional Glycomics (CFG)]. Binding of Epa1A to such arrays at 200 μg/mL showed a marked

Author contributions: H.-U.M. and L.-O.E. designed research; M.M.-R., R.D., M.S.V., G.E., V.K., S.B., and P.K. performed research; M.M.-R., R.D., M.S.V., S.R., H.-U.M., and L.-O.E. analyzed data; and M.M.-R., R.D., H.-U.M., and L.-O.E. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. Data deposition: The atomic coordinates and structure factors have been deposited in the Protein Data Bank, www.pdb.org (PDB ID codes 4AF9, 4AFA, 4AFB, 4AFC, and 4ASL). 1

To whom correspondence may be addressed. E-mail: [email protected] or [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1207653109/-/DCSupplemental.

PNAS Early Edition | 1 of 6

BIOCHEMISTRY

Edited by Gerald R. Fink, Whitehead Institute for Biomedical Research/Massachusetts Institute of Technology, Cambridge, MA, and approved September 4, 2012 (received for review May 9, 2012)

preference for glycans containing galactose at nonreducing ends (Fig. 1 A and B). Both α- and β-linked galactoses are bound by Epa1A, but the binding capability for Galβ1-terminated glycans is significantly higher than for those with Galα-terminal residues. Previous binding studies with Epa1A presented at the surface of S. cerevisiae indicated that this adhesin does not discriminate between mucin-relevant Galβ1–3 and Galβ1–4 disaccharides (15). Here, also no obvious preference was found when comparing Galβ1–3 and Galβ1–4-containing glycans, but all of the best binders within the Galβ1–4 group contain terminal Galβ1–3 branches (Fig. 1B). To further discriminate between stringent primary and nonstringent secondary specificities, binding at lower protein concentrations was conducted. At 50 μg/mL Epa1A, a moderate decrease in the group of Galβ1–4-containing glycans was observed that sharply drops at 1 μg/mL. Likewise, oligosaccharides with terminal α-linked galactose moieties show diminished binding at restricting Epa1A concentrations, further pointing to a preference for β-linkage. Best binders are Galβ1–3 group members, especially those with a Galβ1–3GalNAc motif including the T-antigen (glycans 137 and 140, Fig. 1A). In contrast, β1–4-linked lactosides, which are known to inhibit Candida adherence to host cells in the millimolar range (7), show only weak (lactose, glycans 169 and 170) or very moderate (N-acetyllactosamine, glycans 167 and 168) binding to Epa1A. In summary, our data provide a refined specificity profile for Epa1 and offer compelling evidence that this adhesin prefers Galβ1–3 disaccharides, but lacks high affinity for Galβ1–4 disaccharides. This conclusion is further supported by our fluorescence titration analysis (Fig. S1), which demonstrates that Epa1A has at least a 16-fold higher binding affinity for the T-antigen (KD = 2.1 ± 0.3 µM, Galβ1–3GalNAc) than for milk-derived lactose (34.6 ± 6.1 µM), which is likely to contain significant amounts of a Galβ1– 3Glc contaminant as delineated by our structural analysis below.

Overall Epa1A Structure. To determine the structure of Epa1A, crystals were obtained at low pH, but only in the presence of high concentrations of lactose. The Epa1A crystal structure was solved by molecular replacement, using a truncated homology model based on the S. cerevisiae Flo5A structure (12). Despite a pairwise sequence identity of less than 25%, the resulting structure (Tables S1 and S2) reveals a high similarity with Flo5A, as indicated by an rmsd of 1.4 Å over 121 common Cα atoms (Fig. 2A). The barrel-shaped Epa1A domain has an overall dimension of 57 × 42 × 36 Å3, consists of 15 β-strands, and is complexed via a calcium ion to its respective disaccharide ligand (Fig. 2A). Eleven of the strands form an antiparallel β-sandwich motif with PA14/Flo5-like topology (16). An L-shaped stretch, composed of the remaining four β-strands and the N and C termini, wraps around the β-sandwich and protects its bottom end from solvent access (12). This end of Epa1A is adjacent to the repetitive stalk of the Epa1 B domain. The two disulfide bridges C50–C179 and C180–C263, respectively, which covalently tether the N and C termini to the β-sandwich domain (Fig. 2A), are preserved in Flo5, suggesting that this rigid linkage between A and B domains is a general feature of GPI-CWP adhesins. Which structural motifs define the ligand-binding site in Epa1A? In contrast to S. cerevisiae flocculins, Epa1A lacks the Flo5-like subdomain, which in budding yeast confers specific interactions with terminal α-1,2-mannobioside ligands (12). Surprisingly, the difference electron density map within the binding pocket of the Epa1A/lactose cocrystals clearly ruled out lactose, i.e., Galβ1–4Glc, as bound ligand. At 1.5 Å resolution only a Galβ1–3Glc disaccharide, a contaminant from commercially available lactose, was consistent with the electron density observed (Fig. 2B). This serendipitous finding corroborates the strong preference of Epa1A for Galβ1–3 over excess of Galβ1–4linked disaccharides as indicated by our glycan profiling and

Fig. 1. Glycan binding by the Epa1A domain. (A) Binding profiles of Epa1A at different protein concentrations (1, 50, and 200 μg/mL), using the CFG array V4.1 harboring 451 different glycan structures. Relative fluorescence units as monitored reflect relative affinities toward the corresponding glycan. Glycans bound by Epa1A are indicated by their CFG array numbers. Overall, Epa1A recognizes exclusively terminal galactosides. Green, Galα group; red, Galβ1–3 group; blue, Galβ1–4 group. (B) Groups that are recognized by Epa1A are shown; their locations within the CFG array are given in parentheses. Structural formulas are described according to the CFG nomenclature. At 200 μg/mL, Epa1A recognizes both Galβ1–3 (red) and Galβ1–4 (blue) glycans. However, the latter can harbor terminal β1–3 ramifications (red dotted lines). At low concentrations, Epa1A has a much narrower specificity profile.

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1207653109

Maestre-Reyna et al.

fluorescence titration analyses. The binding site of the Galβ1–3 disaccharide is formed of five loops: two inner, short ones involved in calcium binding and ligand recognition (CBL1 and CBL2) and three outer, longer loops (L1, L2, and L3), which shield CBL1 and CBL2 from solvent access (Fig. 2D). CBL1 links β−strand 8 with 9 and is hallmarked by an unusual cispeptide between D165 and D166, the DcisD motif. This motif is crucial for the C-type lectin function of Flo5A (12) and is conserved throughout the entire Epa family (Fig. S2). CBL2 links β-strands 12 and 13 and directly contributes to specific glycan binding by four of its side chains, R226, E227, Y228, and D229 (Fig. 2 D and E), which we further refer to as CBL2 positions I– IV (Fig. 3). Finally, the Ca2+-binding site comprises the side chain of N225, the carbonyl groups of peptide bonds within CBL2, and the carboxylic side chains of D165 and D166, the DcisD motif of CBL1. Moreover, the Ca2+ ion directly interacts with the nonreducing galactose moiety over the 3- and 4-hydroxyl groups (Fig. 2B) and is essential for glycan binding (Fig. S1). Disaccharide Recognition by the Inner and Outer Subsites of Epa1A.

What is the structural basis for disaccharide discrimination by Epa1A? The Epa1A·Galβ1–3Glc complex indicates that at least two subsites are crucial for glycan recognition: (i) the inner subsite, which is formed of CBL1, the Ca2+ ion, R226 from CBL2, and W198 from loop L3 and confers specific binding to the terminal galactose, and (ii) the outer subsite, which consists of the side chains of CBL2 together with loops L1, L2, and L3 and interacts with the peripheral hexose moiety (Fig. 2 A and C). Moreover, loop L1 is not part of the core PA14/Flo5A-like domain, but connects β-strands 2 and 3. Also, a disulfide bridge formed by C78 and C119 cross-links L1 and L2 to arrest their conformation for shielding the binding site from bulk solvent. The Epa1A·Galβ1–3Glc complex further reveals that the galactose moiety in the inner subsite adopts a flipped orientation, which differs from the orientation of mannose in Flo5A (12). This flipped binding mode is crucial for coaligning the pyranose moiety with residue W198 from loop L3 as well as for hydrogen bonding between the R226 side chain and the 2- and 3-hydroxyls of galactose (Fig. 2D). This suggests that the indole side chain of W198 together with R226 is crucial for selecting the terminal galactose (Figs. S3 and S4), a conclusion that is supported by our finding that an Epa1W198A variant shows only marginal binding activity (Fig. S5). We also performed molecular dynamics studies for ligand binding by Epa1A (Movies S1, S2, and S3), which indicate that W198 may exert an organizing effect on loop L3, because they show that the stretch W198–T202 of loop L3 adopts the observed degree of order only upon complexation of galactose. The outer subsite apparently selects the type of glycosidic bondage and nature of the hexose moiety linked to the terminal galactose. The Epa1A·Galβ1–3Glc structure shows that the site for recognition of Galβ1–3-linked hexoses not only comprises E227 and Y228 of CBL2 (Fig. 2 B and D), but also loops L1 and Maestre-Reyna et al.

L2 lining the outer site. The glutamic acid side chain projected from position II forms hydrogen bonds with the 2-hydroxyls of both hexoses of Galβ1–3Glc as well as with loop L2 via the peptide group of G118–C119. Likewise, the L2 loop wraps with A115–G118 around Y228 and thereby stabilizes its packing with the pyranose ring of the glucose within the Galβ1–3Glc ligand. To obtain further insights into disaccharide binding specificity, the structure of the cognate Epa1A·T-antigen complex was solved at 1.24 Å resolution. This structure reveals the same mode of glycan recognition at its inner subsite as found for the Epa1A·Galβ1–3Glc complex, but major differences at the outer subsite (Fig. 2 C and E). Whereas in the Epa1A·Galβ1–3Glc complex the glucose moiety is coplanar to galactose, the N-acetylgalactosamine moiety of the T-antigen adopts an orthogonal conformation when complexed to Epa1A. This diverging conformation is apparently caused by the hydrogen bond between the axial 4-hydroxyl group and E227 of CBL2. As shown above, the Epa1A binding site is highly specific by disfavoring Galβ1–4-linked glycans. Molecular modeling of a lactose complex and subsequent molecular dynamics studies show that a β1–4 linkage fails to position the second hexose in a defined conformation within the outer subsite, e.g., by packing with Y228 (Fig. S4 and Movies S3 and S4). Nevertheless, the outer subsite harbors sufficient flexibility to allow different binding modi, as shown for the T-antigen. This suggests that EpaA domains permit a certain degree of promiscuity for disaccharide binding in their outer, but not in their inner ligand-binding subsite. Differential Promiscuity of Epa Subtypes Is Dominated by CBL2. We next generated a sequence alignment of 19 EpaA domains that is based on the structures of Epa1A and Flo5A (Fig. S2) and provides a structure-guided phylogenetic tree of the Epa family (Fig. 3A). Four prominent branches appear in this tree, which cluster 11 EpaA domains into four subtypes, which were named Epa1 (Epa1A, Epa6A, Epa7A), Epa2 (Epa2A, Epa4A, Epa5A, Epa19A), Epa3 (Epa3A, Epa22A), and Epa9 (Epa9A, Epa10A), respectively. Remarkably, the CBL2 positions I–IV are mostly sufficient to discriminate between the different subtypes (Fig. 3B). For Epa1 subtype members, this motif consists of REYD or RDND, and the Epa2 and Epa9 subtype adhesins contain related RDNN or RDYH motifs. Only Epa3 subtype proteins show a highly diverging IGKD motif. In addition, the Epa9 subtype differs from others by a prominent elongation within the L1 loop. To explore how CBL2 affects the specificity of different Epa subtype proteins, the CBL2 motif of Epa1A was exchanged for the motif of Epa2A and Epa3A (Epa1→2A and Epa1→3A variants). Because the Epa1 subtype includes two different CBL2 signatures, an Epa1→6A variant was also generated. However, the Epa9 subtype was not further investigated due to its diverging L1 loop. In the next step, fluorescence titration analysis of the subtype-switched Epa1A variants was performed. This analysis revealed that the binding affinities of the Epa1→6A variant for lactose and the T-antigen resemble those of Epa1A PNAS Early Edition | 3 of 6

BIOCHEMISTRY

Fig. 2. Glycan-binding site of the Epa1 adhesin. (A) Structure of the Epa1A domain (gray) with Galβ1– 3Glc (Gal, yellow; Glc, blue). The outer subsite for glycan binding is formed by loops L1 and L2 (red), and the inner subsite comprises the CBL loops (cyan) and a Ca2+ ion (orange). The Flo5A structure is shown as a green transparent overlay, for comparison purposes. The Galβ1–3Glc (B) and T-antigen (C) ligands are shown within the Epa1A-binding site with their SIGMAA-weighted omit electron density maps, respectively (contouring level: 1σ, 0.37electrons/ Å3 and 0.8σ, 0.31electrons/Å3). (D and E) Perpendicular views on the Epa1A-binding site either with the Galβ1–3Glc ligand (D) or with the T-antigen (E). Loops L1 and L2 are covalently linked via a cysteine bridge. W198 from L3 and Ca2+ interact directly with the galactose moiety, whereas the side chains from the CBL2 tip interact with glucose or GalNAc hydroxyls.

outer subsite is promiscuous, being occupied by glucose, N-acetylglucosamine, galactose, or N-acetylgalactosamine (Figs. S6 and S7), a finding that explains previous observations (15). Next, the adhesion behavior of different Epa subtypes was analyzed in vivo, using a heterologous S. cerevisiae system presenting EpaA domains on the cell surface (Fig. S8) and the human colorectal, epithelial cell line Caco-2. The T-antigen is a typical mammalian surface glycan, which is present at high densities on Caco-2 cell surfaces, but not on fungal cells. As expected, S. cerevisiae strains that lack an EpaA domain failed to adhere to epithelial cells (Fig. 5A). In contrast, strains presenting Epa1A or subtype-switched Epa1→6A strongly adhere to Caco-2 cell layers. Obviously, T-antigen–presenting epithelial cells are well recognized by the Epa1 subtype independently of the used sequence of its CBL2 motif (Fig. 5A). Furthermore, the 15-fold lower affinity of Epa1→3A to recognize the T-antigen in vitro is mirrored in vivo, because strains presenting either Epa1→3A or native Epa3 only inefficiently bind to Caco-2 cells (Fig. 5A). Epa1→2A that clearly mediates binding to Caco-2 cells presents a different scenario. Given an only fourfold reduced in vitro binding affinity to lactose compared to Epa1A and Epa1→6A (Fig. S1), one may expect that the affinity or at least koff of Epa1→2A against the Tantigen is similar to that of the Epa1A subtypes. However, the native Epa2A domain fails to confer significant epithelial cell binding, suggesting that in this case switching the specificity and binding characteristics between Epa subtypes seems to require more than exchanging the CBL2 motif. Additional structural features of the outer subsite like the L1 and L2 loops may hence contribute to binding and specificity. Nevertheless, our results indicate that the CBL2 loop plays a prominent role in conferring specificity and promiscuity to Epa family adhesins and thereby significantly affects efficient host cell recognition.

Fig. 3. Phylogeny of C. glabrata EpaA domains. (A) Phylogenetic tree of EpaA domains. The tree is based on a structure-guided multiple-sequence alignment (Fig. S2), using the crystal structures of C. glabrata Epa1A and S. cerevisiae Flo5A (PDB ID: 2XJP). The Epa family shows four major, distinct subtypes: Epa1 (gray), Epa2 (green), Epa3 (blue), and Epa9 (orange). The scale bar indicates phylogenetic distances in number of amino acid substitutions per position. (B) Alignment of the CBL2 regions (cyan box). The Epa1, Epa2, Epa3, and Epa9 subtypes are color-shaded as in A. The Roman numerals indicate the four variable residues within CBL2.

(KD = 33.5 ± 6.6 µM and 1.7 ± 0.4 µM). In contrast, the Epa1→2A variant has a fourfold lower affinity for lactose (KD = 128 ± 16.3 µM). Unfortunately, this variant rapidly aggregates in the presence of the T-antigen, thus preventing quantification of binding despite its obvious interaction with this glycan. For the Epa1→3A variant, we detected only extremely low affinity toward lactose (KD = 27.3 ± 5.7 mM) and a 15-fold less efficient binding of the T-antigen (KD = 30.0 ± 6.1 µM) when compared to Epa1. To further explore the role of the CBL2 motif, the subtype-switched Epa1A variants were subjected to CFG glycan arrays, revealing further differences in their ligand-binding specificity (Fig. S6). In contrast to Epa1, the Epa1→6A variant was found to recognize glycans within the Galβ1–3 group without clear discrimination between α- and β-galactosides (Fig. 4). In addition, Epa1→6A weakly prefers glycans harboring β1–3linked galactose moieties within the Galβ1–4 group, with glycans 164 and 165 found as best binders, as well as Galβ1–4 terminated glycans like N-acetyllactosamine (glycan 168). As found in the fluorescence titration analysis, lactose is bound with much weaker affinity by this variant, similar to Epa1 (Fig. 4). For Epa1→2A and Epa1→3A, we found strongly diminished binding compared to Epa1A and Epa1→6A, which were highly biased toward glycans with terminal, α-linked galactose moieties. In summary, all variants analyzed by glycan array profiling are highly specific for galactose in the inner subsite, whereas the 4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1207653109

Subtle Structural Changes Cause Promiscuity. To obtain further detailed insights into Epa protein functionality, the structures of the subtype-switched Epa1A variants were determined by using the orthorhombic crystal form obtained from cocrystallization with lactose. This analysis revealed structural alterations of the Epa1A variants, which were restricted to their glycan-binding sites when compared to Epa1 (Fig. 4). Similar to Epa1, the Epa1→2A and Epa1→6A variants were found to harbor the Galβ1–3-linked disaccharide with a well-defined positioning of the galactose moiety in the inner subsite. In contrast to Epa1, however, the terminal glucose residue was present in two alternative conformations within the outer subsite of Epa1→6A (Fig. 4). This suggests that the changes at positions II and III in CBL2 (E227D, Y228N) cause a wider and more flexible outer subsite when compared to that in Epa1A (Fig. 4). As a consequence, the reducing end of the disaccharide can almost freely rotate due to increased entropic contribution to binding. Thus, sterically demanding glycans appear to be more easily accommodated in Epa6-like binding sites, a conclusion that is supported by our glycan profiling, which shows increased promiscuity for Epa1→6A, because it efficiently binds to Galβ1–3 and Galβ1–4 as well as to Galα joined glycans (Fig. 5B and Fig. S6C). Interestingly, Epa6 not only confers epithelial cell adhesion, but also is involved in the formation of biofilms, a prerequisite for C. glabrata to colonize inert surfaces such as clinical catheters (17, 18). The reduced Epa6 specificity and tendency to biofilm formation may thus promote host persistence due to improved resistance to antifungal agents. Accordingly, Epa6 has been assigned a role in urinary tract infections (19). In the Epa1→2A·Galβ1–3Glc complex, the glucose adopts an orthogonal conformation similar to that of the Epa1→6A·Galβ1– 3Glc complex (Fig. 4). Having the same exchanges at CBL2 positions II and III, the Epa1→2A variant recognizes Galαjoined glycans as well. In addition, the change in position IV (D229N) reduces the negative charge next to the galactose ligand. As a consequence, Epa1→2A exhibits a marked preference for terminal galactosides that are sulfated at the 6-hydroxyl position (Fig. 5B). These glycan structures are prevalent in sulfomucins, which are in turn predominant components of Maestre-Reyna et al.

intestinal mucosa (20). The structure of the Epa1→2A·Galβ1– 3Glc complex offers sufficient room in the inner subsite to position the 6-sulfate of these terminal sulfo-galactosides close to N229. In contrast, the Epa1 subtype is hindered to recognize sulfo-galactosides due to possible electrostatic repulsion by D229 at position IV. Finally, the Epa1→3A variant that is strongly impaired for in vitro and in vivo ligand binding harbors a glycerol molecule in its binding pocket that stems from the cryo buffer (Fig. 4). The glycerol is coordinated to Ca2+ via two vicinal hydroxyl groups similar to the galactose moiety in Epa1. The Epa3 subtype is characterized by a highly divergent CBL2 region with positions I– III being replaced by IGK. Apparently, a change at position I from arginine to isoleucine (R226I) is sufficient to weaken the interaction with galactose in the inner subsite and to cause impaired discrimination of glycerol from hexoses (Fig. S6B). Nevertheless, Epa1→3A still has residual affinity toward the T-antigen, i.e., 15fold lower than that of other Epa1A variants (Fig. 4). Finally, Epa1→3A binds more efficiently to 6-sulfated galactosides than Epa1. This indicates that the lysine residue at position III of the

Epa3 subtype CBL2 is suitably positioned to form a salt bridge with the 6-sulfate of the sulfo-galactoside bound to the inner subsite. This may overcome the electrostatic interference by aspartate at position IV that is observed for the Epa1 subtype. Conclusions Our detailed structural and functional characterization of Epa1A and subtype-switched variants demonstrates that the specificity of C. glabrata Epa adhesins for different glycan structures is exerted by a distinct region of their A domain. This region comprises the calcium-coordinating loops CBL1 and CBL2 as well as parts of loop L3 and bears modular characteristics to generate Epa subtype variability. The specificity profiles of Epa1A and its variants strengthen and refine the link between these C. glabrata epithelial adhesins and the cores of mucin-type O-glycans (15). Importantly, we found that the T-antigen, which represents the core 1 of mucin-type O-glycans (21) and is part of most mucin-derived glycans (22), is one of the best ligands for all EpaA variants investigated in this study. Other glycans that we found to be recognized by Epa1A and variants are also related to mucins and

Fig. 5. In vivo binding of EpaA domains and the distribution of mucin-derived glycan ligands. (A) Relative adhesion to epithelial cells as conferred by different EpaA domains was determined by using a heterologous S. cerevisiae expression system. S. cerevisiae strains presenting different EpaA domains on Flo11 BC stalks were incubated with a monolayer of Caco-2 cells and adhesion was determined after 2 h. Colors correspond to the different EpaA subtypes as shown in Fig. 3A. (B) Pie charts show glycan types of the CFG array for which binding signals exceed either 20% or 50% of the signal of the best binder. For each glycan type (color as above), the core structures are shown below in CFG standard notation. Nonassigned glycans are grouped as “other sugars” (yellow).


PNAS Early Edition | 5 of 6

BIOCHEMISTRY

Fig. 4. Specificity profiles and binding pockets of subtype-switched Epa1A variants. (Upper) CFG glycan array profiles for Epa1A variants at 200 μg/mL protein concentration. Strong binders are indicated by their CFG array number. Unlike Epa1A, Epa1→6A lacks the strict Galβ1–3 specificity and instead binds galactose-comprising glycans within the Galα, Galβ1–4, and Galβ1–3 groups (green, blue, and red regions, respectively), indicating increased promiscuity. Epa1→2A and Epa1→3A have narrowed specificity profiles with Epa1→3A preferring terminal α-joined galactose-comprising glycans. The lower fluorescence signals of Epa1→2A and Epa1→3A indicate weaker binding to glycans than that of Epa1A and Epa1→6A. (Lower) Top view of subtype-switched Epa1Abinding sites with SIGMAA-weighted omit electron densities for ligands complexed to Epa1→6 (anthracite gray, contouring level: 0.8σ, 0.29electrons/Å3), Epa1→3A (blue, 0.8σ, 0.28electrons/Å3) and Epa1→2A (green, 0.6σ, 0.19electrons/Å3). The orientation of the disaccharide within the binding pocket is shown (Upper Right Inset). CBL loops of the inner subsite are marked in cyan and L loops of the outer subsite in orange.

include other mucin-type O-glycan cores or N-glycans, e.g., sulfogalactose–containing glycans. Our study further reveals that the structural basis for the specificity and promiscuity in Epa–glycan interactions is significantly affected by the properties of CBL2. This finding is consistent with previous domain-swapping experiments between Epa6 and Epa1/7, which identified a larger, CBL2-covering region for controlling specificity (15). Specificity of the inner subsite for galactose is dictated by recognition of the cis-diol formed by the 3- and 4-hydroxyls. Furthermore, position I of CBL2 is occupied by an arginine in all Epa subtypes except Epa3. This crucial residue forms two hydrogen bonds with the 2-hydroxyl of the bound galactose and packing to loop L1 via W79 and Y83 stabilizes its conformation. In the outer subsite, position III of CBL2 presents either large aromatic residues (Y, Epa1, Epa9, Epa10, Epa12, Epa15, and Epa20; F, Epa23) or polar amino acids like asparagine (Epa2, Epa4, Epa5, Epa6, and Epa 19). Whereas the former promote packing with planar glucose/galactose derivatives, resulting in specific interactions with either terminal Galβ1–3Glc(NAc) or Galβ1–3Gal(NAc), the latter can increase promiscuity (Epa6) or enable additional binding to, e.g., sulfated galactoses (Lys: Epa3, Epa22). Position IV appears to mainly control the degree of allowed modification in the terminal galactose moiety. Together, CBL2 positions I–IV may be proposed to form a simple structural code for Epa specificity. However, the differences found for binding to host cell monolayers by Epa1→2A and Epa2A as well as the high-resolution structure of the Epa1A·T-antigen complex with its orthogonal binding mode for the GalNAc moiety challenge this notion and imply that the periphery formed by the loops L1 and L2 adds to the specificity of Epa adhesins. In summary, our study suggests that variable presentation of Epa family members with diversified outer subsites at the cell surface of C. glabrata is one of the key elements for efficient and tissue-specific host invasion and could explain the clinical behavior of this human pathogenic fungus (23). Our structurebased insights that Candida–host interaction crucially depends on the conserved nature of an inner galactose-binding site of Epa adhesins may further contribute to the development of tailored antimycotics to combat this emerging pathogen. 1. Kaur R, Domergue R, Zupancic ML, Cormack BP (2005) A yeast by any other name: Candida glabrata and its interaction with the host. Curr Opin Microbiol 8:378–384. 2. Gulia J, Aryal S, Saadlla H, Shorr AF (2010) Healthcare-associated candidemia–a distinct entity? J Hosp Med 5:298–301. 3. Komshian SV, Uwaydah AK, Sobel JD, Crane LR (1989) Fungemia caused by Candida species and Torulopsis glabrata in the hospitalized patient: Frequency, characteristics, and evaluation of factors influencing outcome. Rev Infect Dis 11:379–390. 4. Buitrón García-Figueroa R, Araiza-Santibáñez J, Basurto-Kuba E, Bonifaz-Trujillo A (2009) Candida glabrata: An emergent opportunist in vulvovaginitis. Cir Cir 77:423–427. 5. Pignato S, et al. (2009) Persistent oral and urinary Candida spp. carriage in Italian HIVseropositive asymptomatic subjects. J Prev Med Hyg 50:232–235. 6. Mundy RD, Cormack B (2009) Expression of Candida glabrata adhesins after exposure to chemical preservatives. J Infect Dis 199:1891–1898. 7. Cormack BP, Ghori N, Falkow S (1999) An adhesin of the yeast pathogen Candida glabrata mediating adherence to human epithelial cells. Science 285:578–582. 8. Filler SG (2006) Candida-host cell receptor-ligand interactions. Curr Opin Microbiol 9: 333–339. 9. Castaño I, et al. (2005) Telomere length control and transcriptional regulation of subtelomeric adhesins in Candida glabrata. Mol Microbiol 55:1246–1258. 10. Linder T, Gustafsson CM (2008) Molecular phylogenetics of ascomycotal adhesins– a novel family of putative cell-surface adhesive proteins in fission yeasts. Fungal Genet Biol 45:485–497. 11. Verstrepen KJ, Reynolds TB, Fink GR (2004) Origins of variation in the fungal cell surface. Nat Rev Microbiol 2:533–540. 12. Veelders M, et al. (2010) Structural basis of flocculin-mediated social behavior in yeast. Proc Natl Acad Sci USA 107:22511–22516. 13. Rigden DJ, Mello LV, Galperin MY (2004) The PA14 domain, a conserved all-beta domain in bacterial toxins, enzymes, adhesins and signaling molecules. Trends Biochem Sci 29:335–339. 14. Ielasi FS, Decanniere K, Willaert RG (2012) The epithelial adhesin 1 (Epa1p) from the human-pathogenic yeast Candida glabrata: Structural and functional study of the carbohydrate-binding domain. Acta Crystallogr D Biol Crystallogr 68:210–217.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1207653109

Methods Protein Production and Purification. Recombinant Epa1A and variants were generated in a thioredoxin and glutathione reductase-deficient Escherichia coli strain. Epa1A was purified by NiNTA-affinity and size-exclusion chromatography. Phylogenetic Analysis. Phylogenetic analysis was performed by the neighborjoining method, using Clustal X2.0 (24), COBALT (25), or a local copy of tcoffee (26) implemented with 3DCoffee (27) for structure-based alignments. Preliminary targets were selected with the help of BLAST. Crystallization and Structure Determination. Epa1A/lactose cocrystals belonging to space group C2221 were obtained from ammonium sulfate-containing conditions. Epa1A·T-antigen cocrystals were obtained by soaking Ca2+-stripped Epa1A/lactose cocrystals. The structure of Epa1A was then solved by molecular replacement and refined using REFMAC5 or PHENIX. High-Throughput Glycan-Binding Assays. The CFG glycan array consists of different groups of oligosaccharides that are presented by mammalian cells. Recombinant, fluorescently labeled Epa1A domains were applied to the CFG array V4.1 chips at concentrations ranging from 1 μg/mL to 200 μg/mL. Chip surfaces were repeatedly washed; remaining fluorescence was measured and quantified. Fluorescence Titrations. Fluorescence titrations of Epa1A and variants were performed against lactose and the T-antigen. Binding was followed at an emission wavelength of 346 nm by excitation of intrinsic tryptophan fluorescence at 295 nm. Fluorescence quench was recorded during titration and fitted using a one-site plus unspecific-binding model. In Vivo Adhesion Assays. In vivo adhesion of Epa proteins to epithelial cells was determined using a nonadhesive S. cerevisiae strain presenting the different EpaA domains at the cell surface (Tables S3–S5). Adhesion assays were performed as previously described (28). ACKNOWLEDGMENTS. The authors thank Tobias Klar and Alexander Popov for support at the European Synchrotron Radiation Facility, Grenoble (beamlines ID14-4 and ID23); Petra Gnau and Lisa Ludewig for technical support; Uwe Linne for mass-spectrometric analyses; and David Smith and the Consortium for Functional Glycomics for performing in vitro screening of glycan specificity (Consortium for Functional Glycomics Request 2080). This research is supported by Grants ES152/7, MO 825/1-4, and GRK 1216 from the Deutsche Forschungsgemeinschaft and by the Landes-Offensive zur Entwicklung Wissenschaftlich-ökonomischer Exzellenz (LOEWE) Center for Synthetic Microbiology (H.-U.M. and L.-O.E.).

15. Zupancic ML, et al. (2008) Glycan microarray analysis of Candida glabrata adhesin ligand specificity. Mol Microbiol 68:547–559. 16. Petosa C, Collier RJ, Klimpel KR, Leppla SH, Liddington RC (1997) Crystal structure of the anthrax toxin protective antigen. Nature 385:833–838. 17. Ramage G, Mowat E, Jones B, Williams C, Lopez-Ribot J (2009) Our current understanding of fungal biofilms. Crit Rev Microbiol 35:340–355. 18. Iraqui I, et al. (2005) The Yak1p kinase controls expression of adhesins and biofilm formation in Candida glabrata in a Sir4p-dependent pathway. Mol Microbiol 55: 1259–1271. 19. Domergue R, et al. (2005) Nicotinic acid limitation regulates silencing of Candida adhesins during UTI. Science 308:866–870. 20. Nieuw Amerongen AV, Bolscher JG, Bloemena E, Veerman EC (1998) Sulfomucins in the human body. Biol Chem 379:1–18. 21. McEntyre J, et al. (2009) Essentials of Glycobiology. Cold Spring Harbor Lab Press, Cold Spring Harbor, NY), 2nd Ed. 22. Linden SK, Sutton P, Karlsson NG, Korolik V, McGuckin MA (2008) Mucins in the mucosal barrier to infection. Mucosal Immunol 1:183–197. 23. Arendrup MC (2010) Epidemiology of invasive candidiasis. Curr Opin Crit Care 16: 445–452. 24. Larkin MA, et al. (2007) Clustal W and Clustal X version 2.0. Bioinformatics 23: 2947–2948. 25. Papadopoulos JS, Agarwala R (2007) COBALT: Constraint-based alignment tool for multiple protein sequences. Bioinformatics 23:1073–1079. 26. Notredame C, Higgins DG, Heringa J (2000) T-Coffee: A novel method for fast and accurate multiple sequence alignment. J Mol Biol 302:205–217. 27. O’Sullivan O, Suhre K, Abergel C, Higgins DG, Notredame C (2004) 3DCoffee: Combining protein sequences and structures within multiple sequence alignments. J Mol Biol 340:385–395. 28. Dieterich C, et al. (2002) In vitro reconstructed human epithelia reveal contributions of Candida albicans EFG1 and CPH1 to adhesion and invasion. Microbiology 148: 497–506.


Supporting Information Maestre-Reyna et al. 10.1073/pnas.1207653109 SI Methods Yeast Strains. Yeast strains used in this study are described in

Table S3. For isolation of the EPAA domains the Candida glabrata stain ATCC2001 was used. In vivo adhesion assays were performed in the nonadhesive Saccharomyces cerevisiae strain BY4741 carrying appropriate plasmids (Table S5). Standard methods for yeast culture medium and transformation were used as described in ref. 1. Plasmid Construction. All plasmids used in this study are listed in

Table S4. Numbering of amino acid residues refers to sequences described in the UniProt database (www.uniprot.org). BHUM1829 was obtained by PCR amplification of the EPA1A domain from genomic DNA of C. glabrata strain ATCC2001, using primers fwdEpa1-aa31 and rev-Epa1-aa271, and subsequent insertion of the resulting XhoI/NdeI fragment into the vector pET-28a(+). BHUM1804, BHUM1805, and BHUM1806, carrying variants of the EPA1A domain with exchanged amino acid in CBL2, were generated via site-directed mutagenesis, using primers pET28_Epa1→6A Fwd and Rev, pET28_Epa1→2A Fwd and Rev, pET28_Epa1→3A Fwd_1 and Rev_1, and pET28_Epa1→3A Fwd_2 and Rev_2 and BHUM1829 as a template. To generate BHUM1835, BHUM1836, BHUM1889, and BHUM1892, the corresponding EPA1A domains were amplified using primers Epa_1.2_SacII and Epa_1.2_SacI together with BHUM1804, BHUM1805, BHUM1806, or BHUM1829 as a template and subsequent cloning of the resulting SacII/SacI fragments into BHUM1760. To obtain BHUM1983, BHUM1984, BHUM2016, and BHUM2017, the corresponding EPA1A domains including the FLO11 secretion signal and the 3HA-tag were amplified using primers SalI-SS-3HA-EpaXA and Epa_1.2-SacI together with BHUM1835, BHUM1836, BHUM1889, or BHUM1892 as a template and subsequent insertion of the resulting SalI/SacI fragments into BHUM1964. For the construction of BHUM1871 and BHUM1877, EPA2A and EPA6A wild-type domains were amplified using either primers Epa_2_SacII and Epa_2_SacI or Epa_6,7_SacII and Epa_6,7_SacI as well as genomic DNA from C. glabrata strain ATCC2001 as a template. The resulting SacII/ SacI fragments were then inserted into BHUM1760. For the construction of BHUM1990, the EPA3A wild-type domain was amplified using primers Epa_3.2_SacII and Epa_3_SacI together with genomic DNA from C. glabrata strain ATCC2001 as a template. The resulting SacII/SacI fragment was then inserted into BHUM1964. To construct plasmids BHUM1985 and BHUM2018, DNA fragments were PCR amplified from plasmids BHUM1871 and BHUM1877, using primers SalI-SS-3HAEpaXA together with either Epa_2-SacI or Epa_6,7_SacI, respectively. Resulting fragments carrying 3HA-tagged versions of the EPAA domains were subsequently cloned into BHUM1964, using restriction enzymes SalI and SacI. BHUM1760 carrying (i) the FLO11 promoter, (ii) the FLO11 secretion signal spanning amino acid residues 1–30, (iii) the FLO11BC domain encompassing amino acids 214–1,360, and (iv) the FLO11 terminator was generated by whole-vector PCR, using primers 1601-A2-SacI-SacII and Flo11-5 together with BHUM1601 as a template and subsequent ligation. BHUM1601 was obtained by PCR amplification of the FLO11 genomic region from S. cerevisiae strain WY423, using primers HUM193 and HUM194 and subsequent cloning into the XbaI/XhoI-digested backbone of BHUM778 by homologous recombination in S. cerevisiae strain RH2662. BHUM1964 carrying (i) the PGK1 promoter, (ii) the FLO11 secretion signal covering amino acids 1–25, (iii) the FLO11BC doMaestre-Reyna et al. www.pnas.org/cgi/content/short/1207653109

main spanning amino acid residues 214–1,360, and (iv) the FLO11 terminator was constructed by replacing the FLO11 promoter in BHUM1327 with the HindIII/SacI fragment of BHUM1962 carrying the PGK1 promoter. BHUM1962 was generated by a combination of three fragments: (i) the SacI/HindIII backbone fragment of BHUM1505, (ii) the HindIII/SalI fragment carrying the PGK1 promoter obtained from BHUM1043 using primers PGK Pr fw and PGK Pr rev, and (iii) the SalI/SacI FLO11 secretion signal from BHUM1879. Recombinant Overproduction and Crystallization of Epa1A Domains.

Both the wild-type Epa1A and all subtype-switched variants were overproduced using the low-temperature protocol developed by Veelders et al. (2). The only modification to the protocol was the use of Escherichia coli strain shuffle T7 express (NEB) instead of E. coli Origami 2, slightly improving yields. After lysis and clarification of the supernatant, the protein was purified by Ni-NTA affinity chromatography (Qiagen) and subsequent size exclusion chromatography, using Superdex 200 prep grade material (GE Healthcare), initially in AM buffer (20 mM Tris·HCl, pH 8.0, 200 mM NaCl). Epa1A interacted strongly with the Superdex 200 material under these conditions, resulting in very poor yields. This issue could be solved by adding either 50 mM lactose (AML buffer) or 10 mM EDTA (AME buffer) to the AM buffer. Initial crystal screening was performed in a 600-nL sitting-drop setup, using commercially available screens (Qiagen) with a Microsys SQ4000 dispensing system (Genomic Solutions), and yielded several positive conditions at 18 °C. Optimizations of original hits took place in a similar, 96 conditions, 18 °C, 600-nL sitting-drop setup. Finally, optimized hits were reproduced in a 2-µL hanging-drop setup. Drops were composed of 50% (vol/vol) protein solution in AML buffer (6 or 12 mg/mL) and 50% (vol/vol) reservoir solution. To obtain Epa1A·T-antigen crystals, Epa1A crystals were grown as described above. The crystals were then soaked in mother liquor supplemented with 1 mM EDTA. Crystals underwent this process three times sequentially, for 2 h, 1 h, and 30 min, respectively. Next, crystals were picked again and given into a drop containing mother liquor supplemented with 2 mM Tantigen and 2 mM CaCl2. Crystals were soaked with T-antigen for 2–24 h. All crystals were frozen in mother liquor supplemented with 20% glycerol. Data Collection and Structure Solution. Datasets for structure solution were recorded at the European Synchrotron Radiation Facility (ESRF) beamlines ID14-2 for Epa1A, ID23-2 for Epa1→3A, and ID14-4 for Epa1→6A. The Epa1A·T-antigen dataset was collected at Bessy II beamline 14.1. Finally, the dataset for Epa1→2A was recorded on a mar345dtb area detector system (Marresearch), using an FR591 rotating anode (Bruker/Nonius, copper target) as X-ray source (Table S1). The structure of Epa1A was solved via molecular replacement, using a carefully trimmed homology model based on Flo5A generated with the program CHAINSAW (3). The homology model was based on a multiple alignment as published by Veelders et al. (2) and is similar to the one presented in Fig. S1. Subtype-switched Epa1A variants crystallized isomorphously and were solved using the Epa1A structure by molecular substitution and subsequently exchanging the mutated amino acids with Coot (4). The very same process was applied to the Epa1A·T-antigen dataset. Phase solution was performed with PHASER (5); data processing with XDS, 1 of 12

PHENIX, and CCP4 (3, 6, 7); and final refinement with REFMAC5 (8), phenix.refine, and Coot (4) (Table S2). Secondary structure assignment was performed with STRIDE (9) as shown in Fig. 2. Figures of protein structures were generated with the molecular graphics program PyMol v1.4 (10). Phylogenetic Analysis. Phylogenetic analysis was performed by the neighbor-joining method, using Clustal X2.0 (11), COBALT, or a local copy of t-coffee implemented with 3DCoffee (12, 13) for structure-based alignments. Preliminary targets were selected with the help of BLAST (14). High-Throughput Glycan-Binding Assays. The CFG glycan array used consists of different groups of oligosaccharides that are presented by mammalian cells. Recombinant Epa1A domains were fluorescently labeled using an AlexaFluor 488 SPD kit (Invitrogen) and applied to CFG array V4.1 chips at concentrations ranging from 1 μg/mL to 200 μg/mL. Chip surfaces were repeatedly washed and remaining fluorescence was measured and quantified. Fluorescence Titrations of Epa1A and Variants. Fluorescence titrations were performed in AM buffer, pH 8, which had been supplemented with 1 mM EDTA and 5 mM CaCl2. Five hundred microliters of protein solution (0.3 mg/mL protein in supplemented AM buffer) was titrated against analyte solution containing either lactose or T-antigen. Tryptophans were excited at a wavelength of 295 nm, and the quench was followed at the emission maximum, which was always observed between 340 and 345 nm. Measurements were done as triplicates, averaged, and fitted with Qtiplot (15), using Eq. S1,

dilution of 1:10,000 in PBS/1% BSA for 20 min at RT. After three further washing steps, a Zeiss Axiovert 200 M microscope was used to (i) visualize S. cerevisiae cells with differential interference contrast and (ii) detect EpaA domains at the cell surface, using a rhodamine filter set (AHF Analysentechnik). Cells were photographed with a Hamamatsu Orca ER digital camera and pictures were processed and analyzed using the Volocity software (Improvision). Fluorescence signals were then quantified using the ImageJ software (16). Epithelial Cell Cultures. For EpaA-directed adhesion assays, the human epithelial cell line Caco-2 (American Type Culture Collection HTB-37) was used, which is a continuous line of heterogeneous human epithelial colorectal adenocarcinoma cells. To gain a confluent monolayer, Caco-2 cells were first grown in 75cm2 tissue culture flasks (Greiner) and split 1:3 every second or third day, depending on the confluence, which did not exceed 80%. Once the cell culture was initiated, a periodic medium change was performed using Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 1 mM sodium pyruvate, and 1% gentamicin (Invitrogen). After approximately 20 subcultures, cells were seeded into 24-well polystyrene plates (Greiner) and incubated at 37 °C under 5% CO2 for 1–2 d until a confluent monolayer was formed.

Detection of EpaA Domains by Fluorescence Microscopy. In a first step, the presence of EpaA domains at the S. cerevisiae cell surface was quantified by immunofluorescence microscopy. For this purpose, cultures of plasmid-carrying strains were grown in low fluorescence yeast medium to an optical density at 595 nm of 1, before cells were washed three times in PBS/1% BSA. Then, cells were incubated with a monoclonal mouse anti-HA antibody (H3663; Sigma Aldrich) at a dilution of 1:1,000 in PBS/1% BSA for 30 min at room temperature (RT). After three wash steps, cells were incubated in darkness with a Cy3-conjugated secondary goat anti-mouse antibody (C2181; Sigma Aldrich) at a

Adhesion Assay. Adhesion assays of S. cerevisiae on human epithelial Caco-2 cell lines were performed as previously described (17). Briefly, 24-well polystyrene plates with a confluent monolayer of Caco-2 cells were used after removal of the culture medium and addition of 250 μL fresh prewarmed DMEM without gentamicin. S. cerevisiae strains carrying appropriate plasmids were grown in YPD medium to exponential phase at 30 °C and diluted in DMEM/10% FBS/1 mM sodium pyruvate to a concentration of approximately 6,000 cells per milliliter of medium. Fifty microliters of these yeast cell suspensions were then added to each well with a confluent layer of Caco-2 cells. Plates were incubated at 37 °C under 5% CO2 for 0, 30, 60, 120, or 180 min, respectively. The complete supernatant containing the nonadherent S. cerevisiae cells was removed and plated on YPD agar to determine the colony forming units (cfu). To determine the adherent yeast cells, wells were washed twice with 300 μL phosphate-buffered saline (PBS) before the epithelial cells together with the attached S. cerevisiae cells were scratched off the polystyrene surface. The resulting suspension was also plated on YPD plates to determine the cfu of adherent cells. After incubation for 2 d at 30 °C, cfu values for nonadherent and adherent cells were determined using an aCOLyte colony counter (7510 DWS; Synbiosis). The average values for nonadherent and adherent cells were determined on the basis of 10 independent experiments. Outliers were eliminated with the help of the standard deviation, the standard error, and a t test. Relative adhesion values (A) were calculated by using the formula A = cfu (adherent cells)/cfu (adherent cells) + cfu (nonadherent cells).

1. Guthrie C, Fink GR (1991) Guide to Yeast Genetics and Molecular Biology (Academic, San Diego). 2. Veelders M, et al. (2010) Structural basis of flocculin-mediated social behavior in yeast. Proc Natl Acad Sci USA 107:22511–22516. 3. CCP4 (1994) The CCP4 suite: Programs for protein crystallography. Acta Crystallogr D Biol Crystallogr 50:760–763. 4. Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of Coot. Acta Crystallogr D Biol Crystallogr 66:486–501. 5. McCoy A, et al. (2007) Phaser Crystallographic Software. J Appl Cryst 40:658–674. 6. Kabsch W (2010) XDS. Acta Crystallogr D Biol Crystallogr 66:125–132. 7. Adams PD, et al. (2010) PHENIX: A comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr D Biol Crystallogr 66: 213–221.

8. Murshudov GN, Vagin AA, Dodson EJ (1997) Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr D Biol Crystallogr 53:240–255. 9. Heinig M, Frishman D (2004) STRIDE: A web server for secondary structure assignment from known atomic coordinates of proteins. Nucleic Acids Res 32:W500–W502. 10. Delano WL (2002) The PyMOL Molecular Graphics System (DeLano Scientific, San Carlos, CA). 11. Larkin MA, et al. (2007) Clustal W and Clustal X version 2.0. Bioinformatics 23: 2947–2948. 12. Notredame C, Higgins DG, Heringa J (2000) T-Coffee: A novel method for fast and accurate multiple sequence alignment. J Mol Biol 302:205–217. 13. O’Sullivan O, Suhre K, Abergel C, Higgins DG, Notredame C (2004) 3DCoffee: Combining protein sequences and structures within multiple sequence alignments. J Mol Biol 340:385–395.

qðcÞ ¼

qmax · c ; KD þ c

[S1]

where q is quench, qmax is maximum quench, c is ligand concentration, and KD is dissociation constant. Adhesion of S. cerevisiae to Human Epithelial Cells. For adhesion tests of S. cerevisiae to human epithelial cells, strain BY4741 was used, carrying plasmids with the appropriate PPGK1-3HA-EPAAFLO11BC constructs (Table S4). Specifically, different EPAAencoding domains were tagged with a HA epitope and fused to the BC domain of the flocculin gene FLO11. They were expressed in S. cerevisiae from the PGK1 promoter.

Maestre-Reyna et al. www.pnas.org/cgi/content/short/1207653109

2 of 12

14. Altschul SF, et al. (1997) Gapped BLAST and PSI-BLAST: A new generation of protein database search programs. Nucleic Acids Res 25:3389–3402. 15. Vasilef I (2009) QtiPlot: Data Analysis and Scientific Visualization (ProIndep Serv, Romania). 16. Abramoff MD, Magalhaes PJ, Ram SJ (2004) Image processing with ImageJ. Biophotonics Int 11:36–42.

17. Dieterich C, et al. (2002) In vitro reconstructed human epithelia reveal contributions of Candida albicans EFG1 and CPH1 to adhesion and invasion. Microbiology 148: 497–506.

Fig. S1. Examples and table summary for fluorescence titrations of Epa domains against lactose or T-antigen. Several examples are shown for fluorescence titrations, and a summary of dissociation constants (KD) and maximum quench (Qmax) is shown.


3 of 12

Fig. S2. Structure-based sequence alignment of different PA14/Flo5-like A domains. Shown is sequence alignment of PA14/Flo5-like A domains from C. glabrata Epa proteins, S. cerevisiae Flo5, and putative adhesins from Pichia pastoris/Komagataella pastoris. A local copy of t-coffee implemented with 3DCoffee (12, 13) and the 3D structures of Epa1A and Flo5A were used to generate the structure-based sequence alignment. The Epa1A secondary structure is shown over the alignment. The outer binding pocket is composed of the loops L1, L2, and L3, which are highlighted in light red. The inner binding pocket, which is composed of the CBL1 and CBL2 loops, is highlighted in cyan. The DcisD motif and N225 of the CBL2 loop, which are complexed to a Ca2+ ion for carbohydrate binding, are marked in red. Cysteines conferring disulfide formation are marked in yellow. W198 in L3, which confers galactoside specificity of the inner subpocket of the Epa adhesins, is shown in green. W196 of Flo5 is also in the L3 loop (in green), but is not involved in ligand-binding specificity.


4 of 12

Fig. S3. Epa1A outer binding site. Orange: L loops solvent accessible surface. Cyan: CBL loops solvent accessible surface (I, III, and IV mark visible CBL2 positions). Galactose and glucose are represented in yellow and blue, respectively. Red mesh: surface that is occluded by W198 side chain. Around 50% of the galactose moiety is rendered inaccessible by the presence of the tryptophane, whereas the 6-hydroxyl is still solvent exposed, indicating that modification of galactose, e.g., by sulfate or other glyco-moieties, is possible at this position.

Fig. S4. Molecular dynamics analysis of carbohydrate binding by the Epa1A domain. The Epa1A domain was subjected to molecular dynamics, using AMBER11 with the ff99sb and GLYCAM06 force fields. The Epa1A/Ca2+ complex was positioned in a periodic, water-filled and neutralized box either alone (A, Movie S1), in complex with galactose (B, Movie S2), complexed to Galβ1–3Glc (C, Movie S3) or bound to lactose (Galβ1–4Glc; D, Movie S4). The box size was chosen as 80 × 62 × 66 Å and the molecular dynamics were performed at 300 K, step size 2 fs, using an isothermal-isobaric ensemble (NPT). After minimization and equilibration for 2 ns, trajectories were collected for a further 18 ns and analyzed with VMD. The figures and movies depict regions of the glycan-binding site crucial for recognition.


5 of 12

Fig. S5. Epa1A W198A mutant fails to bind T-antigen. (Upper) Isothermal titration calorimetric (ITC) data (black) and calculated baseline (red). (Lower) Processed data. No fitting values are shown, because the minor heats observed are most consistent with dilution enthalpy. Accordingly, an erroneously assumed 1:1 binding model results in calculated errors ∼15 times larger than estimated values for KD, ΔH, or ΔS.


6 of 12

Fig. S6. Binding analysis of Epa1A and cluster-converted variants to CFG array presented mono- and disaccharides. (A) Concentration-dependent binding of Epa1A to monosaccharides. Binding intensities have been normalized to Galα-Sp8. Binding was always best for either Galα-Sp8 or Galβ-Sp8. For comparison, the next best ligand intensity is given next. The average intensity of all monosaccharides presented on the array, including both galactose isoforms, is presented on the fourth position. (B) Binding of cluster-converted Epa1A variants to monosaccharides. (C) Concentration-dependent binding of Epa1A to disaccharides, as well as of subtype-switched Epa1A variants. T-antigen (Galβ1–3GalNAc) and α-N-acetyl-lactosamine (Galα1–4GlcNAc) are highlighted as green dots and red dots, respectively. (Left) Binding intensities are normalized to the binding to the T-antigen. Overall, Epa1A binds preferentially to the T-antigen, whereas several other binders are better for subtype-switched variants (indicated by a black frame). (Right) Binding intensities are normalized to binding to the best binder of each variant to delineate variant-specific promiscuity. Epa1A and Epa1→3A present few, well-defined high binders with many low binders. Epa1→6A and Epa1→2A, on the other hand, show a fairly spread out distribution, indicating a significant degree of promiscuity, and the capacity to bind many different disaccharides without major differences in affinity.


7 of 12

Fig. S7. Best binders for EpaA subtype-switched variants. EpaA subtype-switched variants best binders highlighted in Fig. 4, and not present already in Fig. 1, are presented here in CFG standard notation.

Fig. S8. Presence of different EpaA domains on S. cerevisiae. S. cerevisiae strain BY4741 carrying BHUM1983 (Epa1A), BHUM1984 (Epa1A→2A), BHUM1985 (Epa2A), BHUM2017 (Epa1A→3A), BHUM2016 (Epa1A→6A), or BHUM2018 (Epa6A) was grown to logarithmic phase and EpaA domains were detected by immunofluorescence microscopy, using anti-HA primary and Cy3-conjugated secondary antibodies. (Scale bar, 10 µm.)


8 of 12

Table S1. Data collection statistics for Epa1A and subtype-switched variants

X-ray source Wavelength, Å Space group Cell parameters, Å a b c Resolution, Å Completeness, % Total reflections Unique reflections Multiplicity Rmerge (%) Mean I/σ

Epa1A

Epa1→6A

Epa1→3A

Epa1→2A

Epa1A·T-antigen

ID14-2, ESRF 0.9330 C2221

ID14-4, ESRF 0.9395 C2221

ID23-2, ESRF 0.8762 C2221

Rotating anode (FR591, CuKα) 1.5418 C2221

ID14.1, Bessy II 0.9180 C2221

74.64 104.33 69.79 1.50–19.9 (1.50–1.58) 98.8 (98.7) 174,640 43,249 4 7.0 (65.5) 14.6 (2.6)

74.43 103.89 69.28 1.55–19.6 (1.55–1.63) 99.6 (99.9) 186,996 39,090 4.8 4.0 (47.4) 26 (3.5)

75.74 103.53 70.48 1.90–46.2 (1.90–2.00) 99.5 (99.4) 115,032 22,061 5.2 14.5 (49.3) 7.5 (2.5)

74.56 104.04 69.15 2.00–19.8 (2.00–2.11) 97.7 (99.7) 55,079 18,024 3.1 12.3 (51.9) 7.4 (2.2)

74.60 103.90 69.40 1.24–10.0 (1.24–1.31) 98.7 (97.5) 309,225 75,251 4.1 3.4 (55.3) 22.7 (2.7)

Data in parentheses correspond to the highest resolution shell.

Table S2. Refinement statistics for Epa1A and subtype-switched variants

Resolution, Å Rwork, % Rfree, % Reflections, all Test set No. atoms Water molecules B-factor, Å2 rms deviations Bond length, Å Bond angle, °

Epa1A

Epa1→6A

Epa1→3A

Epa1→2A

Epa1A·T-antigen

19.9–1.50 15.7 18.6 41,751 1,498 2,122 274 18.1

19.6–1.55 14.4 17.1 37,731 1,358 2,089 208 20.6

37.9–1.90 16.9 21.7 22,056 782 3,735 160 26.0

19.6–2.05 20.4 23.9 16,152 571 2,017 140 17.3

10.0–1.24 11.9 15.5 72,672 2,579 2,484 392 15.0

0.015 1.601

0.015 1.571

0.014 1.275

0.01 1.296

0.013 1.524

Table S3. Yeast strains Strain BY4741 ATCC2001 RH2662 WY423

Relevant genotype

Source

S288c MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 C. glabrata wild-type strain CBS138 Σ1278b MATa ura3-52 flo11Δ::kanR Σ1278b MATa ura3-52 his3::hisG leu2::hisG 3HA-FLO11

Euroscarf www.atcc.org (1) (2)

1. Braus GH, Grundmann O, Bruckner S, Mosch HU (2003) Amino acid starvation and Gcn4p regulate adhesive growth and FLO11 gene expression in Saccharomyces cerevisiae. Mol Biol Cell 14:4272–4284. 2. Guo B, Styles CA, Feng QH, Fink GR (2000) A Saccharomyces gene family involved in invasive growth, cell-cell adhesion, and mating. Proc Natl Acad Sci USA 97:12158–12163.


9 of 12

Table S4. Plasmids Plasmid pET-28(a)+ BHUM0778 BHUM1043 BHUM1327 BHUM1505 BHUM1601 BHUM1760 BHUM1804 BHUM1805 BHUM1806 BHUM1829 BHUM1835 BHUM1836 BHUM1871 BHUM1877 BHUM1879 BHUM1889 BHUM1892 BHUM1962 BHUM1964 BHUM1983 BHUM1984 BHUM1985 BHUM1990 BHUM2016 BHUM2017 BHUM2018

Relevant genotype

Source

PT7 6xHis lacI KanR PFLO11-FLO11-TFLO11 in YCplac33 PPGK1-FLO11-TPGK1 in YEplac181 PFLO11-FLO11(aa1–25)-FLO11(aa214–1,360)-TFLO11 in YCplac33 PFLO11-FLO11(aa1–24)-FLO5(aa272–1,075)-TFLO11 in YCplac33 PFLO11-FLO11(aa1–30)-3HA-FLO11(aa31–1,360)-TFLO11 in YCplac33 PFLO11-FLO11(aa1–30)-3HA-FLO11(aa214–1,360)-TFLO11 in YCplac33 EPA1(aa31–271;E220D;Y221N;D222N) in pET-28a(+) EPA1(aa31-271;R219I,E220G;Y221K) in pET-28a(+) EPA1(aa31–271;E220D;Y221N) in pET-28a(+) EPA1(aa31–271) in pET-28a(+) PFLO11-FLO11(aa1–30)-3HA-EPA1(aa31–271)-FLO11(aa214–1,360)-TFLO11 in YCplac33 PFLO111-FLO11(aa1–30)-3HA-EPA1(aa31–271;R219I,E220G;Y221K)-FLO11(aa214–1,360)-TFLO11 in YCplac33 PFLO11-FLO11(aa1–30)-3HA-EPA2(aa32–262)-FLO11(aa214–1,360)-TFLO11 in YCplac33 PFLO11-FLO11(aa1–30)-3HA-EPA6(aa26–271)-FLO11(aa214–1,360)-TFLO11 in YCplac33 SalI-20bp-FLO11(aa1–25)-SacII-GCA-SacI in “pANY-Amp” PFLO11-FLO11(aa1–30)-3HA-EPA1(aa31–271;E220D;Y221N;D222N)-FLO11(aa214–1,360)-TFLO11 in YCplac33 PFLO11-FLO11(aa1–30)-3HA-EPA1(aa31–271;E220D;Y221N)-FLO11(aa214–1,360)-TFLO11 in YCplac33 PPGK1-FLO11(aa1–25)-FLO5(aa272–1,075)-TFLO5 in YCplac33 PPGK1-FLO11(aa1–25)-FLO11(aa214–1,360)-TFLO11 in YCplac33 PPGK1-FLO11(aa1–30)-3HA-EPA1(aa31–271)-FLO11(aa214–1,360)-TFLO11 in YCplac33 PPGK1-FLO11(aa1–30)-3HA-EPA1(aa31–271;E220D;Y221N;D222N)-FLO11(aa214–1,360)-TFLO11 in YCplac33 PPGK1-FLO11(aa1–30)-3HA-EPA2(aa32–262)-FLO11(aa214–1,360)-TFLO11 in YCplac33 PPGK1-FLO11(aa1–25)-EPA3(aa28–266)-FLO11(aa214–1,360)-TFLO11 in YCplac33 PPGK1-FLO11(aa1–30)-3HA-EPA1(aa31–271;R219I,E220G;Y221K)-FLO11(aa214–1,360)-TFLO11 in YCplac33 PPGK1-FLO11(aa1–30)-3HA-EPA1(aa31–271;E220D;Y221N)-FLO11(aa214–1,360)-TFLO11 in YCplac33 PPGK1-FLO11(aa1–30)-3HA-EPA6(aa26–271)-FLO11(aa214–1,360)-TFLO11 in YCplac33

Merck Germany (1) (2) (3) (3) This study This study This study This study This study This study This study This study This study This study Mr Gene This study This study This study This study This study This study This study This study This study This study This study

1. Braus GH, Grundmann O, Bruckner S, Mosch HU (2003) Amino acid starvation and Gcn4p regulate adhesive growth and FLO11 gene expression in Saccharomyces cerevisiae. Mol Biol Cell 14:4272–4284. 2. Volschenk H, et al. (1997) Engineering pathways for malate degradation in Saccharomyces cerevisiae. Nat Biotechnol 15:253–257. 3. Veelders M, et al. (2010) Structural basis of flocculin-mediated social behavior in yeast. Proc Natl Acad Sci USA 107:22511–22516.

Table S5. Primers Primers Mutagenesis pET28_Epa1→6A Fwd pET28_Epa1→6A Rev pET28_Epa1→2A Fwd pET28_Epa1→2A Rev pET28_Epa1→3A Fwd_1 pET28_Epa1→3A Fwd_1 pET28_Epa1→3A Fwd_2 pET28_Epa1→3A Rev_2 Cloning fwd-Epa1-aa31 rev-Epa1-aa271 HUM193 HUM194 1601-A2-SacI-SacII Flo11-5 PGK Pr fw PGK Pr rev Epa_1.2_SacII Epa_1.2_SacI Epa_2_SacII Epa_2_SacI Epa_3.2_SacII Epa_3_SacI Epa_6,7_SacII Epa_6,7_SacI SalI-SS-3HA-EpaXA

Sequence 5′→3′ ccctattaggttattttataataacagagataatgatggcgcgctcagttttac gtaaaactgagcgcgccatcattatctctgttattataaaataacctaataggg ccctattaggttattttataataacagagataataatggtgcgctcagttttac gtaaaactgagcgcaccattattatctctgttattataaaataacctaataggg ccctattaggttattttataataacataggatatgatggcgcactcagttttac gtaaaactgagtgcgccatcatatcctatgttattataaaataacctaataggg ccctattaggttattttataataacataggaaaggatggcgcgctcagttttac gtaaaactgagcgcgccatcctttcctatgttattataaaataacctaataggg ccatatgacatcttccaatgatatcag actcgagttaagaagaatcgtagctg ccggaattcgtggcgcggtgccaatactaccggtacttg acgcgtcgacccccaattcaagaatacaattacttagcgtgg aaagagctcccgcgggaaggagggggatccactag aaagagctcatagattgtgacaacaattgtgctccagtacc aaaaagcttatcttgttttgcaagtaccactg aaagtcgacgttttatatattgttgtaaaaagtagataattacttcc aaaccgcggacatcttccaatgatatcag aaagagctcagaagaatcgtagctg aaaccgcggcctaaatccaaggatc aaagagctcgcagtggtagttatag aaaccgcggaagcgagaattaagttccc aaagagctccctgcatgtagtatcg aaaccgcggaaggatgactattcttcc aaagagctccgaagtatcataactaac aaagtcgacatgcaaagaccatttccattcgc


10 of 12

Movie S1. Molecular dynamics simulation of the Epa1A/Ca2+ complex without carbohydrate ligand. The Epa1A domain was subjected to molecular dynamics, using AMBER11 with the ff99sb and GLYCAM06 force fields. The Epa1A/Ca2+ complex was positioned in a periodic, water-filled and neutralized box alone. The box size was chosen as 80 × 62 × 66 Å and the molecular dynamics were performed at 300 K, step size 2 fs, using an isothermal-isobaric ensemble (NPT). After minimization and equilibration for 2 ns, trajectories were collected for a further 18 ns and analyzed with VMD. The figures and movies depict regions of the glycan-binding site crucial for recognition (Fig. S4).

Movie S1

Movie S2. Molecular dynamics of Epa1A/Ca2+ complexed with galactose. Epa1A/Ca2+, in complex with galactose, was treated as in Movie S1. Movie S2


11 of 12

Movie S3. Molecular dynamics of Epa1A/Ca2+ complexed with Galβ1–3Glc. Epa1A/Ca2+, in complex with Galβ1–3Glc, was treated as in Movie S1. Movie S3

Movie S4.

Molecular dynamics of Epa1A/Ca2+ complexed with lactose. Epa1A/Ca2+, in complex with lactose (Galβ1–3Glc), was treated as in Movie S1.

Movie S4


12 of 12