Hyperadhesive isolate PEU1221 was obtained from a central venous catheter (CVC) by routine clinical diagnostic procedure in University Medical Center Göttingen (Germany) and identified using MALDI-TOF (Bruker). Strain ATCC 2001 [karyotype I (Bader et al., 2012)] was used as reference strain. Unless mentioned otherwise, cultures grown overnight in liquid YPD (1% yeast extract, 2% peptone and 2% glucose) at 37°C were used as starting material for the assays described.

PEU1221 and ATCC 2001 were inoculated from overnight cultures in fresh YPD and grown at 37°C to logarithmic phase. From these cultures, 20 OD₆₀₀ units in YPD were seeded in polystyrene Petri dishes (20 for PEU1221; 40 for ATCC 2001) in a volume of 20 mL per plate. After incubating for 24 h at 37°C in a moist environment, the settled cell layers were gently washed with milli-Q water to remove unbound cells, and adhered cells were collected by scraping. Cell wall isolations were performed using a Fastprep 24 instrument (MP Biomedicals) as described (De Groot et al., 2004; De Groot et al., 2008). Complete cell breakage was verified using a microscope. Cell walls were washed extensively with 1 M NaCl, and subsequently two times with SDS extraction buffer (50 mM Tris-HCl, 2% SDS, 100 mM Na-EDTA, 150 mM NaCl, and 0.8% β-mercaptoethanol, pH 7.8) at 100°C for 10 min. Finally, the cell walls were washed extensively with water, lyophilized, and stored at -20°C until use.

Preparation of cell walls and digestion with Trypsin Gold (Promega) for mass spectrometric analysis was performed as described (Yin et al., 2005). The peptide material was freeze-dried and taken up in 50% acetonitrile, 2% formic acid. After measuring the OD₂₁₄ and calibrating against a peptide mix control sample (Thermo Scientific), samples were diluted in 0.1% trifluoroacetic acid solution to reach a peptide concentration of about 0.25-0.5 pmol µl^-1. The amount of sample used per run was 0.5-1.0 pmol peptide material. Duplicate biological samples were analyzed using an amaZon Speed Iontrap with a CaptiveSpray ion source (Bruker) coupled to an EASY-nLC II (Thermo Scientific) chromatographic system, employing identical settings as described previously (Gómez-Molero et al., 2015). Precursor ions were automatically selected for low-energy collision-induced dissociation (CID) to obtain fragmentation spectra for peptide identification.

After processing the raw MS/MS data with Data Analysis software (Bruker), database searching was performed against C. glabrata protein sequences in NCBI using Mascot software (Version 2.5.1). Simultaneous searching was performed against a “common contaminants database” (compiled by Max Planck Institute of Biochemistry, Martinsried) to minimize false identifications. Mascot search parameters were: trypsin with allowance of one missed cleavage, fixed modification of cysteine with carbamidomethyl, variable modification of oxidized methionine, peptide charge states +2, +3, and +4, a peptide cut-off score of 20, peptide and MS/MS mass error tolerances of 0.3 Da, decoy database activated, and 1% false discovery rate as output threshold. Protein identifications based on a single peptide MS/MS match were only taken into consideration if the protein had been identified previously in cell walls of C. glabrata, and in such case, the validity of the identified peptide was verified by manual inspection of MS/MS spectra in the raw data using the Data Analysis software. Unmatched peptides were subjected to a second Mascot search with different settings: semitrypsin as the protease, including N/Q deamidation as variable modification, and a cut-off peptide score of 40. Identified semitryptic peptides solely served to extend the sequence coverage of proteins identified in the first trypsin search. Mass spectrometric details of identified individual peptides identified in both PEU1221 samples are given in Supplementary Table S1 . The detailed proteomics data of the ATCC 2001 samples have already been presented previously (Gómez-Molero et al., 2015), therefore, ATCC 2001 data is here only used for comparing identified proteins in both strains ( Table 1 ) and for peptide counting ( Table 2 ).

AWP14 deletion mutants were generated in hyperadhesive strain PEU1221 and reference strain ATCC 2001 using SAT1-flipper system plasmid pSFS1a (Reuß et al., 2004), designed in pBluescript II backbone for use in C. albicans but also functional in C. glabrata. This plasmid contains the SAT1 flipper gene, conferring resistance to nourseothricin (NT), for transformant selection, and an inducible FLP1 gene under control of the CaSAP2 promoter for excision of the integrated cassette by recombination of palindromic flippase recognition sequences on both sides of the construct. Regions of about 0.6 kb upstream and within the AWP14 coding sequence were amplified using proofreading KAPA polymerase and cloned into the KpnI and XhoI (upstream fragment) and NotI and SacI (downstream fragment) sites of the pSFS1a vector. Correctness of the deletion cassette was verified by Sanger sequencing. Genomic incorporation of this construct leads to deletion of 595 bp of the A-domain of AWP14. C. glabrata transformation was performed by electroporation of competent cells prepared using lithium acetate (De Backer et al., 1999), and transformants were selected on YPD agar containing 200 µg/mL NT. Deletion mutants among transformants were identified by PCR using combinations of primers that cover the integration junctions and combining external and internal AWP14 primers. Deletion cassettes were removed from the genome by inducing the flippase gene by growth in YCB-BSA-YE (1.17% yeast carbon base, 0.4% bovine serum albumin, 0.2% yeast extract, pH 4.0) and selecting slow-growing colonies on plates containing a low concentration of 10 µg/mL NT. Loss of the deletion cassettes was verified by PCR ( Supplementary Figure S1 ). In each background, at least two deletion mutants from independent transformation experiments were obtained and in all phenotypic assays data of mutants therefore represent the average of at least two mutants. Oligonucleotides used in this work are listed in Supplementary Table S2 .

Adhesion to polystyrene was measured using two different assays depending on the time allowed to adhere, 4 or 24 h, the latter allowing more time for biofilm formation. For the 24 h experiment, overnight cultures in YPD at 37°C were adjusted to a cell density OD₆₀₀ = 0.5 in fresh YPD, and 200 µL of the cell suspension was pipetted into a polystyrene 96-well plate and incubated at 37°C for 24 h in a humid environment. Unattached cells were removed by gentle washing with milli-Q water, and the remaining adhered cells were stained with 0.1% crystal violet (CV) solution for 30 min followed by washing with milli-Q water. Finally, CV was solubilized in 33% glacial acetic acid and quantified by measuring the OD₅₉₅ using a microplate reader (Molecular Devices). Data for each strain are the average of six technical and at least two biological replicates.

For the 4 h adhesion experiment, overnight cultures were diluted to OD₆₀₀ = 0.05 in phosphate-buffered saline (PBS), and 0.5 mL (~1 × 10⁵ cells) was pipetted in 12-wells plates and incubated for 4 h at 37°C. Unbound cells were removed by two washes with PBS, after which adhered cells were treated for 10 min at room temperature (RT) with a 2.5% porcine pancreas trypsin (Sigma) in PBS solution. Finally, PBS was added to a final volume of 0.5 mL, cells were resuspended and measured using a MACSQuant (Miltenyi Biotec) flow cytometer. Data are the average of at least two biological replicates measured in triplicate.

Binding capacity of C. glabrata cells to fungal cell wall components and extracellular matrix (ECM) collagen was also evaluated by flow cytometry following a similar procedure as described above for 4 h adhesion to polystyrene, using the same amount of cells. However, before incubation with C. glabrata, the microtiter plates were coated with chitin (Sigma; 250 μg/mL in 1% acetic acid), laminarin (β-1,3-glucan, Thermo Fisher; 500 μg/mL in milli-Q water), pustulan (β-1,6-glucan, Calbiochem; 500 μg/mL in 50 mM potassium acetate buffer) or bovine collagen (Sigma; 250 μg/mL in 0.2 M bicarbonate buffer, pH 9.6) by adding 500 μL of the respective solutions, and allowing for passive adsorption (1 h at 30°C followed by overnight incubation at 4°C), in the case of chitin and collagen, or evaporation (overnight at 37°C), in the case of laminarin and pustulan.

For determination of growth rates, cells from overnight cultures were inoculated 1/100 in fresh YPD and incubated in 200 µL volumes in 96-well plates at 37°C with agitation in a SpectraMax 340PC microplate reader. The increase in OD₆₀₀ was followed in time. For determination of zymolyase sensitivity, cells were grown in YPD until log phase, collected, and resuspended at an OD₆₀₀ of 2.0 in 10 mM Tris-HCl, 0.25% of β-mercaptoethanol, pH 7.5. After 1 h of incubation at RT, 180 µL of cell suspensions and 20 µL of up to 100 U/mL zymolyase 20T were mixed in a 96-well plate and placed in the microplate reader at 37°C. Decrease in OD₆₀₀ was measured each minute after a short mixing pulse. Curves represent averages of five technical and two biological replicates.

Overnight cultures in YPD at 37°C were transferred to glass tubes. Every 10 min 20 µL of sample was carefully taken 2 cm below the surface of the cell suspension for OD₆₀₀ measurements. Cell aggregation of overnight cultures in YPD was observed with a Leica DM1000 microscope mounted with a MC170 HD digital camera and by flow cytometry measuring size (FSC-A) and granularity (SCC-A) of 20,000 cell particles.

Overnight cultures in YPD at 37°C were washed two times with PBS and resuspended at an OD₆₀₀ of 0.7. Cell suspensions were mixed with hexadecane (Sigma) in glass tubes at a 15:1 volume ratio. Upon 1 min of gentle vortexing, the phases were allowed to settle for 10 min after which the OD₆₀₀ of the aqueous phase was measured. Each strain was assayed twice with two technical replicates each.

Susceptibility to antifungal agents was tested in 96-well plates following EUCAST guidelines (EUCAST, 2020). Twofold serial dilutions of compounds were prepared in YPD and mixed 1:1 with cells from overnight cultures diluted to an OD₆₀₀ of 0.01. Plates were incubated for 24 h at 37°C. Minimal inhibitory concentrations (MIC) were determined after reading the OD₆₀₀ in a microplate reader. Antifungals were Caspofungin (Merck Sharp & Dohme), Micafungin (Astellas), and Nikkomycin Z (Sigma). Sensitivity to cell wall perturbants Calcofluor white (CFW) and Congo red (CR) was tested in spot assays. Tenfold serial dilutions were prepared from overnight cultures, and 4 µL were spotted on YPD plates containing 100 µg/mL CFW or 100 µg/mL CR. Growth was monitored after 24 and 48 h of growth at 37°C.

Statistical significance of phenotypic data was analyzed by Student’s t-tests or one-way ANOVA followed by post hoc Delayed Matching to Sample (DMS) tests. P values <0.05 were considered statistically significant.

N-terminal ends of identified CWPs produced by signal peptide cleavage were analyzed using SignalP (http://www.cbs.dtu.dk/services/SignalP/). The three-dimensional structure of the ligand-binding or A-domain of Awp14 (residues 22-400) was modeled using ColabFold (Mirdita et al., 2021) and visualized using iCn3D (Wang et al., 2020). Structural similarity to proteins in the PDB database was analyzed using the PDBeFold, with a default cut-off for lowest acceptable similarity match of 70% (Krissinel and Henrick, 2004), and DALI (Holm, 2020) servers.

C. glabrata strain PEU1221 was isolated from an infected central venous catheter during routine diagnostic procedure. When measuring adhesion to polystyrene after 4 h of incubation in PBS using flow cytometry, only slightly higher adhesion was observed for PEU1221 compared to reference strain ATCC 2001, however, PEU1221 showed an about tenfold higher amount of adhered biomass onto polystyrene after 24 h of incubation in YPD measured by Crystal violet staining, demonstrating the high biofilm-forming (HBF) capacity of this clinical isolate ( Figure 1 ).

Previous studies have revealed correlations between the capacity to form biofilms and the incorporation of cell wall adhesins in Candida (Gómez-Molero et al., 2015; Moreno-Martínez et al., 2021). This prompted us to investigate the incorporation of cell wall adhesins in strain PEU1221 in comparison to ATCC 2001, a low-biofilm-forming (LBF) strain, using a proteomic approach. Cell walls of PEU1221 and ATCC 2001 were purified from 24 h biofilms on polystyrene, subjected to trypsin digestion, and analyzed by tandem mass spectrometry, presented in Tables 1 , 2 . Consistent with previous analyses of other strains (Gómez-Molero et al., 2015), from comparing the cell wall proteomes of the two strains it is clear that the identified proteins can be divided in two groups: (i) a stable core wall proteome present in both strains, and (ii) a strain-dependent variable set of putative adhesins that is amplified in the HBF strain. The identified non-adhesin core wall proteome consisted of 19 proteins ( Table 1 ), including carbohydrate-active enzymes from Crh, Gas, and Bgl2/Scw4 families involved in cell wall polysaccharide remodeling, non-enzymatic proteins of Cwp1, Pir, and Srp1/Tip1 families with suggested roles in crosslinking of β-glucan chains, Plb proteins with presumed lysophospholipase activity, and proteins with unknown functions including widely present Ecm33 family proteins and Ssr1. Except Plb1, which was identified only in PEU1221, all core wall proteins were identified in both strains ( Table 1 ). Counts of identified MS/MS spectra showed that the Cwp1 family proteins Cwp1.1 and Cwp1.2 account for about half of all the identified peptides and thus are by far the most abundant proteins in cell walls of both strains ( Table 2 ). For more information on these core wall proteins, we refer to our previous studies (De Groot et al., 2008; Gómez-Molero et al., 2015).

In contrast to the core wall proteome, the proteomic data showed a clearly elevated incorporation of adhesins in strain PEU1221. Thirteen different adhesins were unambiguously identified in PEU1221 compared to only five in ATCC 2001 biofilms ( Table 1 ). Consistent with this, the number of different peptides identified from adhesins in PEU1221 (78 peptides, see Supplementary Table S1 ) was also higher than in ATCC 2001 [23 peptides, for details, see Gómez-Molero et al. (2015)]. Moreover, adhesin peptides represented 24% of the total amount of identified peptides in PEU1221, higher than the 10% in ATCC 2001 ( Table 2 ). Altogether, our proteomic data indicate that incorporation of adhesin protein molecules was about two- to three-fold higher in cell walls of HBF strain PEU1221 than in LBF strain ATCC 2001 under the tested biofilm-forming conditions.

Breaking down the adhesins in different families, Epa cluster I adhesins identified in PEU1221 were Epa3, 6, 7 and 22, whereas in ATCC 2001 only Epa3 and 6 were unambiguously identified. Epa22 was unambiguously identified for the first time in C. glabrata while Epa7 has been identified before in two hyperadhesive isolates (Gómez-Molero et al., 2015). Both the total number of Epa peptides and their proportion among all adhesin peptides (42% of 228 in PEU1221 versus 16% of 92 in ATCC 2001, Table 2 ) was higher in PEU1221 than in ATCC 2001. This is in accordance with the documented upregulation and importance of EPA genes, in particular EPA6 and EPA7, during biofilm formation (Iraqui et al., 2005).

PEU1221 cell walls also showed increased incorporation of cluster V adhesins: whereas Awp2 and 4 were identified in both strains, Awp8, 9, 10, 11, and Awp2e were identified only in PEU1221. Awp8 - 11 have also been identified previously in other HBF clinical isolates (Gómez-Molero et al., 2015), but Awp2e, was identified for the first time. Adhesin cluster VII protein Awp12 was identified in both strains albeit with four different peptides in PEU1221 and only with a single peptide in ATCC 2001. Finally, another putative adhesin that was only detected in PEU1221 is CAGL0A04851g. This cluster III protein, which we named Awp14, was identified for the first time in cell walls of C. glabrata. Cluster III includes 13 putative adhesins ( Figure 2 ), mostly uncharacterized, of which only Aed1 and Awp13 have been previously identified in cell walls by mass spectrometry. The total (23) and different (10) number of Awp14 peptides identified suggests it is one of the most abundant adhesins in PEU1221 biofilm cells. We therefore focused our further studies on functional characterization of this protein.

AWP14 is located on chromosome A, with its 3` end close to the right-hand telomere, and spans more than 26 kb according to the ATCC 2001 genome update published by Xu and colleagues (Xu et al., 2020). However, this published sequence appears to contain a sequencing error (in a repeat region), leading to premature translation termination. By comparing to other repeat sequences and to the more recently published BG2 genome (Xu et al., 2021), the error is corrected by inserting a T at position 15,180, resulting in an extremely large protein of 8,759 aa. Like all putative C. glabrata adhesins, this protein sequence starts with a signal peptide for secretion and ends with a GPI anchor addition signal, both removed during its translocation to the cell surface as result of processing steps. The remaining mature protein consists of an A-domain of about 375 aa, immediately followed by an extremely long tail comprising megasatellites (repeat sequences) that contain VSHITT (114) or SFFIT (4) cores, which are present in all cluster III adhesins ( Figure 2 ) as well as in adhesins from other clusters (De Groot et al., 2013).

We generated AWP14 deletion mutants in both PEU1221 and reference strain ATCC 2001. To avoid problems with correct integration of the deletion cassette and PCR verification of the mutants, we deleted the largest part of the A-domain but staying away from repeat sequences in the C-terminal domain. Obtained deletion mutants were phenotypically characterized, thereby mostly focusing on cell surface-related features. First, we determined if deletion of AWP14 led to altered adhesion or biofilm formation on polystyrene. However, neither in ATCC 2001 nor PEU1221 background significant changes were observed upon deleting AWP14 ( Figure 1 ), indicating that Awp14 is not responsible for the high biofilm formation of the latter. Then, we tested if the protein might perhaps play a role in cell-cell interactions with neighboring cells through binding to cell wall molecules by measuring cells extracted after binding to laminarin (β-1,3-glucan), chitin, and pustulan (β-1,6-glucan). In these flow cytometry assays, the awp14Δ mutants did not show alterations in binding to glucan, however, a small but significant decrease (95% versus 89%) in binding to chitin was observed for the awp14Δ mutants in PEU1221 background ( Figure 3 ). In a similar assay, we also tested binding to collagen as model for possible host interactions but found no effect of deleting AWP14. Further, except for pustulan, these in vitro assays did not show differences in affinity between ATCC 2001 and PEU1221, indicating that the different biofilm-formation capacities of the two strains do not affect their binding to these cell surface molecules. For pustulan, however, a lower binding for PEU1221 (60%) than for ATCC 2001 (75%) was observed, and this was not affected by deleting AWP14.

Previous work has shown that HBF phenotypes may coincide with increased cell surface hydrophobicity, increased cell-cell aggregation, and faster sedimentation in liquid (Gómez-Molero et al., 2015; Moreno-Martínez et al., 2021). This is also the case for PEU1221. It has higher surface hydrophobicity, sediments faster, and showed more cell clumping than the reference strain as observed by microscopy and flow cytometry ( Figure 4 ). Probably because of the cell clumping, OD₆₀₀ measurements of liquid cultures indicate lower growth rates during exponential phase and reached a lower maximum cell density. Deleting AWP14 did not have an evident effect on these phenotypes albeit that in PEU1221 background the awp14Δ mutant appeared to sediment slightly slower than the parental strain and flow cytometry indicated increased complexity but not size of the measured cell particles. Further cell wall-related phenotypes studied were minimal inhibitory concentrations of the antifungal compounds caspofungin, micafungin, and nikkomycin using microbroth dilutions, growth inhibition by Calcofluor white and Congo Red in spot assays, and resistance to cell lysis by zymolyase (a β-1,3-glucanase-containing enzyme preparation). These assays did not detect any phenotype caused by deleting AWP14 (not shown) but noteworthy is the high resistance to zymolyase of the HBF strain as compared to ATCC 2001 ( Figure 4E ). Similar higher resistance to zymolyase as compared to ATCC 2001 has also been observed previously for other HBF strains (Gómez-Molero et al., 2015), however, what architectural cell wall changes underlie this phenotype and if this is related to adhesin incorporation, is currently unknown.

Finally, the 3D structure of the Awp14 A-domain was predicted using ColabFold ( Figure 5 ). The obtained structural model revealed a right-handed β-helix containing three parallel β-sheets per coil followed by a C-terminal β-sandwich structure. Similarity searches against the PDB database revealed several proteins with similar three-faced β-helices, including a variety of polysaccharide lyases from different bacteria [e. g. the pectate lyase Bsp165PelA from Bacillus sp. N165 (Zheng et al., 2012) and alginate lyase FPU-7 from Paenibacillus sp. str. (Itoh et al., 2019)], adhesin HMW1 (Yeo et al., 2007) and heme-hemopoxin binding HxuA (Baelen et al., 2013) from Haemophilus influenza, and the cell-wall attached adhesin PfbA (plasmin- and fibronectin-binding protein A) of Streptococcus pneumoniae (Suits and Boraston, 2013). Root mean square deviation (RMSD) values between Awp14-A and the search results ranged from 2.79 to 5.92 Å, indicating high structural similarity.