The Epithelial adhesin 1 tandem repeat region mediates protein display through multiple mechanisms.

The pathogenic yeast Candida glabrata is reliant on a suite of cell surface adhesins that play a variety of roles necessary for transmission, establishment and proliferation during infection. One particular adhesin, Epithelial Adhesin 1 [Epa1p], is responsible for binding to host tissue, a process which is essential for fungal propagation. Epa1p structure consists of three domains: an N-terminal intercellular binding domain responsible for epithelial cell binding, a C-terminal GPI anchor for cell wall linkage and a serine/threonine-rich linker domain connecting these terminal domains. The linker domain contains a 40-amino acid tandem repeat region, which we have found to be variable in repeat copy number between isolates from clinical sources. We hypothesized that natural variation in Epa1p repeat copy may modulate protein function. To test this, we recombinantly expressed Epa1p with various repeat copy numbers in S. cerevisiae to determine how differences in repeat copy number affect Epa1p expression, surface display and binding to human epithelial cells. Our data suggest that repeat copy number variation has pleiotropic effects, influencing gene expression, protein surface display and shedding from the cell surface of the Epa1p adhesin. This study serves to demonstrate repeat copy number variation can modulate protein function through a number of mechanisms in order to contribute to pathogenicity of C. glabrata. © FEMS 2020.

INTRODUCTION

Systemic infections by yeast of the Candida genus, or invasive candidiasis, have been an emerging public health concern since the 1980s and are a particular concern to hospitalized patients suffering from immune deficiencies such as HIV/AIDS (Fidel, Vazquez and Sobel 1999; Lin, Schranz and Teutsch 2001; McNeil et al. 2001; Pfaller and Diekema 2007). Candida glabrata follows Candida albicans as the second most common cause of invasive candidiasis in the United States and tends to cause particularly severe infections as a result of ever-increasing rates of antifungal resistance (Pfaller and Diekema 2007; Alexander et al. 2013; Spettel et al. 2019). In addition, C. glabrata infections are made worse by the highly plastic haploid genome that allows for the pathogen to adapt to a wide range of environments, such as those experienced in different tissues and on various abiotic surfaces (Fidel, Vazquez and Sobel 1999; Ahmad et al. 2014; Lopez-Fuentes et al. 2018).

There are over 60 cell-wall associated proteins, members of a large radiation of the adhesin superfamily, many of which are essential for the establishment and propagation of C. glabrata infection (Timmermans et al. 2018). These adhesins can bind a variety of substrates and serve many functions including transmission of the pathogen, establishment of infection and formation of antibiotic-resistant biofilms. First identified by Cormack and colleagues, deletion of the gene which codes for one prominent adhesin, Epithelial Adhesin 1 [EPA1, Epa1p], results in a 95% reduction in adherence to cultured epithelial cells. Heterologous expression of EPA1 in Saccharomyces cerevisiae confers the ability of cells to bind to epithelium (Cormack, Ghori and Falkow 1999). Epa1p-mediated binding to host epithelial cells is reliant on the interaction between the N-terminal Epa1p lectin-like binding domain and a terminal galactose β1–3 glucose residue of a host cell surface glycan (Maestre-Reyna et al. 2012).

To facilitate binding, Epa1p structure enables the presentation of the lectin-like domain (A domain) on the cell surface. A hydrophobic N-terminal signal peptide is located at amino acid residues 6–24 and is responsible for the translocation of the protein into the secretory pathway and eventual targeting to the cell surface (Schekman 1982; Cormack, Ghori and Falkow 1999). Once at the surface, Epa1p is anchored by a C-terminal glycosylphosphatidylinositol (GPI) anchor, which is responsible for covalent linkage of the protein to cell wall glycans (Cormack, Ghori and Falkow 1999; Frieman, McCaffery and Cormack 2002; Klis et al. 2010). Epa1p contains a serine/threonine-rich linker (B domain) of roughly 600-amino acids between the binding domain and GPI anchor (Cormack, Ghori and Falkow 1999). This domain has been shown to be highly modified by both O- and N-linked glycosylation (Frieman, McCaffery and Cormack 2002). This is a process which is predicted to further crosslink GPI-linker cell wall proteins (GPI-CWP) to the glycan network of the cell wall to affect protein structure and function (Frieman, McCaffery and Cormack 2002; Klis et al. 2010). Based on models of other GPI-CWPs, glycosylation of Epa1p has been predicted to stabilize the protein into an extended rod structure to enhance A-domain clearance of the cell wall (Jentoft 1990; Frieman, McCaffery and Cormack 2002). Alternatively, stability of Epa1p in the cell wall may be regulated in a glycosylated-mediated mechanism similar to Flocculation 11 [Flo11p], a related GPI-CWP in S. cerevisiae. When Flo11p is under-glycosylated, its stability in the cell wall decreased compared to controls as a result of increased susceptibility to extracellular proteases such as Kexin 2 (Karunanithi et al. 2010; Meem and Cullen 2012).

Within the Ser-/Thr-rich B domain of Epa1p, there is a tandem repeat region of a 40-amino acid sequence (T/M)VRSTLPSSAGSNETSINVPFSSSTESNTSTSSTSTSNSK, which repeats three times in the reference genome of C. glabrata (Cormack, Ghori and Falkow 1999). DNA minisatellite sequences, such as that which codes for this repetitive domain, are known to be genetically unstable and prone to expansions and contractions (Richard and Dujon 2006; Gemayel et al. 2010; Thierry, Dujon and Richard 2010). Variation in repeat regions, both in repeat copy number and mutations within repeats, has been shown to have an important impact on phenotype and serve as loci for population-wide genetic variability (Babokhov et al. 2018a,b). While six of the nine genes in the EPA family contain repetitive regions, the link between repeat variation and function of these proteins has not been widely characterized (Thierry, Dujon and Richard 2010; Ahmad et al. 2014). However, increased repeat copy number in the FLO genes, which code for similar GPI-CWPs in S. cerevisiae, improves adherence, so it has been postulated that a similar effect will be seen in the Epa adhesins (Verstrepen et al. 2005; Fidalgo et al. 2006; Thierry, Dujon and Richard 2010). While studies of Epa1p have not yet specifically focused on the repeat region, the overall size of the linker domain is positively correlated to Epa1p-mediated binding to epithelial cells (Frieman, McCaffery and Cormack 2002). As discussed above, there may be a variety of mechanisms at play resulting in this enhanced adhesion conferred by adhesin proteins with extended linker domains including increased clearance of the cell wall and higher protease protection of longer repeat variants.

In this study, we aimed to identify whether repeat copy number variation exists in the EPA1 gene across C. glabrata isolates collected from a variety of clinical samples and determine whether this variation influenced Epa1p function. Here, we show that several length variants exist, and that in a clinical population, repeat variants of between three and five repeats are most common. We determine that repeat copy number variation has complex effects on protein function. In our recombinant system, the effect of repeat copy number on the surface display of Epa1p occurs through modulation of both the expression of EPA1 and stability of Epa1p at the cell surface. In sum, these data show a link between repeat copy number variation and C. glabrata pathogenicity, which potentially offers a new angle for therapeutic intervention during C. glabrata infections.

METHODS

Yeast strains and clinical isolates

Clinical isolates of C. glabrata were provided by Dr. Yoav Golan from the Tufts Medical School (Boston, MA). Strains were named numerically (CG1-CG24; Table S1). The C. glabrata reference strain BG14 (ura3::Tn903 G418R; (Cormack and Falkow 1999)) was a gift from Professor Brendan Cormack (Johns Hopkins University). For all experiments involving S. cerevisiae, strain BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) was used. BY4741 and BG14 were grown at 30˚C in YPD media supplemented with 2% agar when required. For experiments requiring S. cerevisiae carrying plasmid DNA, strains were grown with Synthetic Complete dropout media lacking one or more amino acids for selection, supplemented with 2% agar when required. For short term usage, cultures were kept on plates at 4˚C. For long term storage, cultures were supplemented with 15% glycerol, and cells were stored at −80 ˚C.

Hela cell culture

Human cervix adenocarcinoma (HeLa) cells were cultured in Dulbecco's Modified Eagle Medium supplemented with 10% heat inactivated fetal bovine serum. Cells were cultured at 37˚C with 5% CO2 and were passaged every 2–4 days before reaching 100% confluence.

Analysis of EPA1 repeat variation

Genomic DNA was isolated from C. glabrata through a rapid yeast genomic prep (Hoffman and Winston 1987). We then amplified the repetitive region of EPA1 through PCR with primers EPA1repF and EPA1repR (see Table S2 for primer sequences) which anneal to EPA1 36 and 58 bp upstream and downstream of the repeat region, respectively. The DNA isolated from the repeat region was then analyzed by gel electrophoresis to determine repeat copy number and confirmed by sequencing. For DNA sequencing analysis, we isolated the EPA1 gene from C. glabrata genomic DNA with the OCR014 and OCR015 primers, and the sequencing was carried out by Eton Biosciences Inc. (Charlestown, MA) with the EPA1repR primer. Sequences were then aligned utilizing the translation alignment tool in Geneious Prime (Biomatters Ltd).

Plasmid construction

For expression of EPA1 in S. cerevisiae, the pSPHA expression vector was used. (See Fig. S1 for plasmid maps and cloning strategy.) The pSPHA expression vector allows for constitutive expression of a protein product with an N-terminal signal peptide from Epa1p and 3x hemagglutinin (HA) tag. The vector was maintained in S. cerevisiae by growth in Synthetic Complete minus Leucine media and an ampicillin resistance marker allows for selection and cloning in Escherichia coli. To build the pSPHA expression vector, the pRS415 expression vector (pADH; Addgene 87 374) was modified through the addition of the 254 bp SPHA cassette (GeneArt; Table S2) coding for the signal peptide, 3x HA tag and a 3’ multiple cloning site into the SmaI (EC 3.1.21.4) restriction site via Gibson Assembly. The final construct was confirmed by DNA sequencing.

To build the EPA1 expressing construct pSPHA-EPA1, the fragments of the EPA1 gene containing the A, B and GPI domains were isolated using the OCR035 and OCR036 primer set. A total of four different EPA1 fragments were created; EPA1 was amplified from BG14, CG3, CG15 and CG2 with three, four, five and ten repeats respectively [EPA1 (3 rep), EPA1 (4 rep), EPA1 (5 rep), EPA1 (10 rep); Fig. 1C]. These fragments were then inserted into SmaI-linearized pSPHA through Gibson Assembly to create the final expression vector, which was confirmed via sequencing. pSPHA-EPA1(0 rep) was created by deleting the repeat region from EPA1 by QuickChange Mutagenesis (Agilent), amplifying EPA1 with OCR035 and OCR036 primers and cloning via Gibson Assembly as described above.

Figure 1.

EPA1 codes for an adhesin with variable repeat copy number. (A)EPA1 repeat region amplified from gDNA of 24 C. glabrata clinical isolates (1–24) and BG14 reference strain (R) through PCR and visualized on an agarose gel. (B)EPA1 repeat copy number frequency from the 24 clinical isolates. (C) pSPHA-EPA1 expression vector cloned into S. cerevisiae allows for constitutive expression of HA-tagged Eap1p (A domain—purple, B domain—light blue, GPI anchor—dark blue) with a 3x HA-tag (green) directed to the surface by the Epa1p signal sequence (pink) under the control of the ADH promoter (grey arrow) and CYC1 terminator (grey box). Five repeat copy number variants of EPA1 were cloned; the black arrows represent each individual tandem repeat. pADH is the empty vector backbone.

Flow cytometry analysis

Measurement of surface display by HA-tagged Epa1p was achieved through extracellular staining of cells and flow cytometry analysis. pSPHA-EPA1 carrying strains were grown to log-phase and stained with chicken anti-HA antibody (Gallus Immunotech AHA) in PBS + 0.1% BSA (PBSA). Cells were then stained with AlexaFluor 488 conjugated goat anti-chicken secondary antibody (Abcam 150 169) in PBSA. Stained cells were resuspended in PBSA and analyzed on an Attune NxT Flow Cytometer (Thermo Fisher, Waltham MA).

The flow cytometry gating strategy is outlined in Fig. 2A. Briefly, debris and doublets were gated out, and 10 000 events within the singlet population were randomly selected through the FlowJo DownSample plugin (FlowJo, LLC). The BL-1 laser (excitation 488 nm/emission 530 nm bp 30 nm) was used to measure AlexaFluor 488 signal, and the population of HA + cells was determined by setting the gate on the pADH negative control. From this population we determined the percentage of cells stained for HA (%HA+) by setting an HA + gate on the negative population. The mean fluorescent intensity of the HA + population (MFI) was determined by taking the mean of the normally distributed events within the positive gate.

Figure 2.

Epa1p repeat copy number affects surface display. HA-Epa1p with 0, 3, 4 and 5 repeats was expressed in S. cerevisiae under constitutive expression under the ADH promoter, and surface display was measured via ɑHA flow cytometry. (A) Flow cytometry gating strategy. Single cells were selected from debris and doublets, and the negative control (pADH) was utilized to set the gate for the HA + population detected by positive AlexaFluor 488 fluorescence. (B) Surface display of HA-Epa1p with various repeat copy numbers. Y-axis counts are normalized by division by the count of the mode fluorescent intensity of each sample. Plot is representative of multiple experiments. (C) Percentage of cells stained positively for HA-Epa1p. (D) Mean fluorescence intensity (arbitrary units, AU) of HA + stained cells. Mean + SEM is shown. n = 3. Representative experiment from multiple independent experiments. ** = P < 0.01, **** = P < 0.0001 as determined by paired Bonferroni corrected T-tests between all experimental groups.

Reverse transcriptase quantitative PCR

To measure levels of EPA1 expression at the mRNA level, S. cerevisiae transgenic strains were grown to log phase, and BG14 was measured as cells were exiting stationary phase. Cells were spheroplasted in 0.02 U/mL zymolyase in 1 M sorbitol + 100 mM EDTA; total RNA was extracted with the illustra RNAspin Mini Kit (GE Healthcare, Chicago IL). Complementary DNA was then synthesized using the SuperScript III First Stand Synthesis System (Invitrogen) with random hexamer primers. Quantitative PCR (qPCR) was then performed with SYBR Green 2x Master Mix (Applied Biosytems, Foster City CA) or Brilliant III Ultra-Fast SYBR Green QPCR Master Mix (Agilent, Santa Clara CA) utilizing the EPA1qPCR F and EPA1qPCR R primer set to amplify the EPA1 amplicon, and the cycle threshold (CT) was determined within the linear range of amplification. CT values for each reaction were then normalized to internal ACT1 controls, amplified with ACT1qPCR F and ACT1qPCR R primers.

Detection of shed protein via western blot

To analyze the shedding of HA-Epa1p from the yeast cell wall, we collected and analyzed growth media through western blotting. To prepare samples, pSPHA-EPA1 carrying strains were grown to log phase and equivalent volumes of culture based on cell OD600 were collected. Cultures were then spun down; the media supernatant was collected (cells were discarded) and immediately frozen at −80 ˚C. Frozen media samples were lyophilized to a film and reconstituted at 40x concentration in 50 mM Tris, 5 mM EDTA, pH8 buffer. Laemmli SDS-PAGE sample buffer was added, and the sample was heated for 5 minutes at 95˚C.

The supernatant from the media collected from 9 × 106 cells was run on a 10% SDS-PAGE gel and transferred to a PVDF membrane (GE Lifesciences). The membrane was probed with monoclonal mouse anti-HA antibody (Invitrogen 326 700) freshly diluted (1:333) in PBST + 2.5% non-fat milk followed by incubation with HRP-conjugated goat anti-mouse secondary antibody (Invitrogen 626 520) in PBST + 2.5% non-fat milk (1:1000). Signal on the blot was detected with Amersham ECL Prime western blotting Detection Reagent (GE Lifesciences, Chigo IL) and imaged with autoradiography film (Andwin Scientific, Tryon NC).

Epithelial cell adherence assay

The adherence of various yeast strains to epithelial cells was measured through a binding assay based on those outlined by (Cormack, Ghori and Falkow 1999; Halliwell et al. 2012; Diderrich ; Fig. 5A). First, HeLa cells were grown to confluence in a 24-well dish and fixed with 2% formaldehyde in phosphate buffered saline (PBS). After fixation, Neuraminidase was utilized to cleave sialic acid from the cell surface glycans becayse cell lines deficient in sialic acid allow for better binding of Eap1p to its ligand (Cormack, Ghori and Falkow 1999). HeLa cells were treated overnight with 0.2 U/mL of ɑ2–3,6,8,9 Neuraminidase in cell culture media (New England Biolabs).

Figure 5.

Quantifying HA-Epa1p mediated epithelial cell adherence. (A) Binding of S. cerevisiae expressing HA-Epa1p repeat copy number variants (0, 3, 5 repeats) to epithelial cells was measured via an epithelial cell adherence assay. Unbound yeast were then removed via washing in PBS, and HeLa cells were chemically lysed. Bound yeast were quantified by counting colony forming units (CFUs) and comparing them to the CFUs on input control plates. (B) Binding normalized to HA-Epa1p (3 rep). Values are mean + SEM deviation. n = 5–6. Data pooled from multiple experiments.

To adhere yeast to HeLa cells, S. cerevisiae carrying pSPHA-EPA1 with various repeat copy numbers or carrying the pADH empty vector were grown to log phase and 1 × 106 yeast cells were added to treated HeLa cells in fresh HeLa culture media. The co-culture was spun for 2 min at 200 rcf to induce contact between yeast and HeLa cells and incubated for 30 minutes on the benchtop. Unbound yeast were removed by five washes with PBS. Lysis buffer (10 mM EDTA + 0.1% Triton X-100 in PBS) was used to lyse mammalian cells, and yeast were diluted and plated on fresh YPD plates. Colony-Forming Units (CFUs) were then counted and compared to CFUs on input control plates, plated directly from yeast cultures at the same dilution. Data were normalized between multiple experiments by division of each sample's % binding by the mean % bound in the HA-Epa1p (3 rep) expressing strain of S. cerevisiae.

Statistical analysis

A Single Factor Analysis of Variance (ANOVA) was utilized to determine significant differences between experimental samples. Once ANOVA identified that there was variation between groups, one-tailed paired t-tests with Bonferroni Correction were run between all populations to determine which were the significantly different groups. The statistically significant α value was set at P < 0.05.

RESULTS

EPA1 codes for a repetitive and variable adhesin

To explore natural variation in EPA1 repeat length, we amplified the repeat region from the genomic DNA of 24 clinical isolates and the BG14 reference strain by PCR. By agarose gel electrophoresis we observed four distinct band sizes, which corresponded to three (two isolates and BG14), four (seven isolates), five (14 isolates) repeats (Fig. 1). Repeat length was confirmed via Sanger sequencing from a representative collection of clinical isolates and BG14 (Fig. S2). One isolate (CG2) had a larger EPA1 repeat region that could not be fully sequenced due to limits of Sanger sequencing read length; we calculated that this strain had ten tandem repeats in the EPA1 repeat region by DNA migration (Fig. 1, data not shown).

HA-Epa1p surface display is modulated by repeat copy number

To study the effect of the EPA1 repeat region on the characteristics of the protein, we designed a transgenic system to study Epa1p in the related yeast, S. cerevisiae. This system was implemented as a means of normalizing expression of EPA1 in a non-adherent background strain. HA-Epa1p variants with different repeat copy numbers were expressed in S. cerevisiae under the constitutive ADH promoter (Fig. 1C). Surface display of HA-Epa1p with various Epa1p repeat copy numbers in S. cerevisiae was measured via flow cytometry (Fig. 2). The percentage of cells stained positive for HA-Epa1p increases as Epa1p repeat copy number increases (Fig. 2C). A one-way ANOVA was conducted between HA-Epa1p expressing strains, and there was found to be significant variation between groups (P < 0.001), with the repeat region knockout (0 rep) having significantly fewer cells stained positive for HA (HA+) than all other repeat variants (P < 0.01). When per cell density of HA-Epa1p on the cell surface was analyzed via MFI, the opposite trend was seen (Fig. 2D). HA-Epa1p with the repeat region deleted had a significantly higher MFI than the repeat copy numbers (three, four and five) seen in the natural population (P < 0.0001).

Heterologous EPA1 expression under constitutive expression decreases with increase repeat copy number

To study the underlying mechanisms which may control the variable Epa1p surface display we observed by flow cytometry, EPA1 mRNA levels were measured by qPCR (Fig. 3). In general, strains expressing EPA1 under the ADH promoter expressed the transcript at a level 3-fold higher than under the native promoter in BG14 when grown in rich media (Fig. 3A). Overall, increasing repeat copy number inversely correlated with EPA1 mRNA transcript levels, with the five-repeat copy expressing at about 50% of the level of EPA1 with the repeat region deleted (Fig. 3B). While ANOVA found significant variation between the different copy number variants (P < 0.01), the increase in expression of EPA1 (0 rep) compared to the other variants failed statistical significance. Rather, the only significant changes were found when comparing other variants to EPA1 (10 rep) (P < 0.05), which showed no expression in our system (Fig. 3B).

Figure 3.

EPA1 mRNA expression levels decrease with increased repeat copy number. (A) mRNA expression levels of EPA1 in S. cerevisiae carrying the empty pADH vector or pSPHA-EPA1 (3 rep) or BG14, a lab strain of C. glabrata measured by rt-qPCR. Expression is expressed as fold change over Actin (ACTI). (B) RT-qPCR-quantified expression levels of HA-EPA1 with various repeat copy numbers (0, 3, 4, 5, 10) of the EPA1 repeat region expressed through the pSPHA-EPA1 plasmid system. n.d. = no detection. Expression is expressed as fold-change over expression of Actin (3 rep). Mean + SEM is shown. n = 2. Representative from multiple independent experiments.

Repeat region protects HA-Epa1p from cell surface shedding

We hypothesized that the differences surface display observed via flow cytometry could be explained by more rapid turnover of Epa1p with fewer repeats at the cell surface. There is evidence for a protective role of protein glycosylation in related proteins (Karunanithi et al. 2010; Meem and Cullen 2012), and we wondered whether a key function of the repeat region might be to mediate this protection. To test this, we probed the growth media of HA-Epa1p expressing cells for the presence of HA-tagged fragments. As expected, when the media was analyzed via western blot, there were two major HA + protein species present, both migrating at apparent sizes smaller than the predicted cell wall associated protein size of ∼115 kDa (Fig. 4A). These products ran at an apparent molecular weight of 70 kDa and 40 kDa, respectively and likely represent proteolytically cleaved products that are shed from the cell surface (Fig. 4B). Qualitatively, as repeat copy number increases, there is a decrease in both fragments present in media.

Figure 4.

Shedding of HA-Epa1p fragments increases with reduced repeat copy number. (A) HA-Epa1p shedding assay. Briefly, cells were grown to log phase and their growth media was collected by centrifugation. Media was then concentrated by lyophilization and resuspended at 40x in Tris-EDTA. (B) ɑHA western blot of concentrated media from cells expressing HA-Epa1p with various repeat copy numbers (0, 3, 4, 5 rep) or the empty pSPHA vector. Media were collected from an equivalent number of cells and concentrated via lyophilization before western blot analysis. Blot is representative of three replicates.

Epa1p repeat variation does not affect binding to epithelial cells

HA-Epa1p was expressed in S. cerevisiae under the ADH promoter and bound to neuraminidase-treated HeLa cells to test the clinical relevance of differences in repeat copy number. The pADH negative control conferred no binding to epithelial cells while all three HA-Epa1p expressing strains tested showed binding at a similar level (Fig. 5B).

DISCUSSION

Epa1p repeat copy number variation exists within the clinical population

Repetitive regions of DNA serve as important loci for population-wide strain to strain variability in yeast which leads to a variety of different phenotypes (Babokhov et al. 2018a), so we hypothesized that there would be variability in the copy number of the 120 bp EPA1 minisatellite. Based on observations from other adhesins with similar repetitive domains, we hypothesized that binding would be improved by expanding repeat regions (Verstrepen et al. 2005; Fidalgo et al. 2006). Indeed, when we analyzed the region's size in various clinical isolates by PCR, we found that there was a considerable amount of variation at this locus (Fig. 1). We concluded that the majority of strains had an EPA1 gene with between three and five repeats, with a single isolate (CG2) having ten repeats (Fig. 1) and confirmed that the variation was the result of expansions of the tandem repeat region via sequencing (Fig. S2).

In general, these data are in line with previous findings. Frieman, McCaffery and Cormack (2002) found that in a similar collection of 25 clinical isolates, EPA1 genes contained between three and six repeats, however this is the first know documentation of an EPA1 gene with as many as ten repeats. In line with previous research in the field of repetitive domains in GPI-CWPs, there appears to be selection for Epa1p regions with expanded repeat regions. However, it was very interesting that there was only one strain whose EPA1 gene coded for a repeat region with greater than five repeats, suggesting that unchecked expansion of this repeat actually is detrimental for function.

The tandem repeat region plays a role in regulating Epa1p display through multiple mechanisms

To test the functional differences of EPA1 genes coding for various repeat copy numbers, we first attempted to examine the differences in EPA1 expression in clinical strains with different repeat copy numbers. However, upon screening our available strains, no trends could be identified (data not shown); it became clear that there were other factors affecting the expression of the gene that were beyond the scope of the present study.

Instead of studying Epa1p function in the native system, we elected to express HA-tagged Epa1p with various repeat copy numbers in the related yeast, S. cerevisiae. Expressing HA-tagged Epa1p in S. cerevisiae under a constitutive promoter is a widely utilized system (Cormack, Ghori and Falkow 1999; Timmermans et al. 2018) and allowed us to uncouple our studies from the highly regulated gene expression of adhesin genes natively in C. glabrata. This approach had the other advantages in that S. cerevisiae has no inherent adhesive properties toward mammalian cells (Lopez-Fuentes et al. 2018) and shares a similar cell wall structure to C. glabrata (Cormack, Ghori and Falkow 1999; de Groot et al. 2008). The HA-tag allowed us to visualize the protein utilizing readily available reagents and had no measurable effect on Epa1p mediated binging to epithelial cells (data not shown).

To test how variation in the EPA1 repetitive region affects phenotype, the cell surface display of HA-Epa1p was analyzed via flow cytometry (Fig. 2). We tested HA-Epa1p variants with three, four, or five repeats, or with the repeat region deleted; the ten-repeat variant was omitted because we were unable to identify expression at the mRNA level (Fig. 3). Significantly higher mean fluorescent intensity was observed for the cells expressing HA-Epa1p with the repeat region deleted than any of the naturally occurring repeat copy number variants (Fig. 2D). This trend suggests there is decreased rate of protein surface display of Epa1p as a result of increased size of the repetitive region. However, HA-Epa1p with increased repeat copy number was displayed on a higher percentage of the total population (Fig. 2C). Taken together, these findings suggest that variation of the Epa1p repeat region may exert control on surface display through multiple mechanisms.

To identify these mechanisms, we first measured the transcription of all of repeat variants (Fig. 3). We observed that EPA1 with increased repeat copy number expressed at lower levels than the gene with fewer copies of the tandem repeat. While not statistically significant, EPA1 with the repeat region deleted appears to be more highly expressed, roughly 1.5-fold higher, than the native forms of the gene. Furthermore, no expression of EPA1 (10 rep) was detected. This trend fits with previous findings that transcription of repetitive DNA is impaired as repeat copy number increases through a variety of mechanisms (Feng, Goldgar and Corbex 2007; Biscotti et al. 2015; Madireddy and Gerhardt 2017). This trend may serve as an explanation as to why only one clinical isolate had a repeat copy number greater than five; without the expression of EPA1, pathogenicity of C. glabrata would be severely impaired.

While the difference in mRNA levels may explain the increased density of Epa1p on the surface of the cells when the repeat region is deleted, it does not explain why fewer of these cells are stained positively for HA-Epa1p. The repeat region is rich in glycosylated Ser/Thr residues which may contribute to stability within the yeast cell wall (Klis, Brul and De Groot 2010; Meem and Cullen 2012). To test this, we probed the cell culture media for differential loss of HA-Epa1p from the cell wall across repeat copy number variants (Fig. 4). The total amount of HA-tagged product in the culture media decreases as repeat copy number increases, which suggests that indeed greater copy numbers stabilize Epa1p in the cell wall (Fig. 4B). We attempted to blot for the cellular fraction of Epa1p, to test how these differences compared to the total amount of HA-Epa1p present in the cell wall, however, due to technical limitations we were unable to visualize HA-Epa1p in the cell.

Neither of the fragments observed by western blot were the expected molecular weight of fully intact HA-Epa1p (∼115 kDa), supporting that HA-Epa1p is shed from the cell surface by cleavage by extracellular proteases. While our study did not directly assess the glycosylation status of the repeat region, we conclude that this is the most likely mechanism that drives increased stability in the cell wall as this process occurs for other GPI-CWPs (Karunanithi et al. 2010; Meem and Cullen 2012). We predict that as the repeat copy number, and therefore Ser-/Thr-associated glycosylation, increases, HA-Epa1p stability in the cell wall improves, resulting in less loss of HA-Epa1p via proteolytic shedding.

Epa1p mediated cellular binding is not affected by copy number variation

Epa1p-mediated adherence to epithelial cells is an essential step in the establishing C. glabrata infection, and we looked to determine how repeat copy number would affect this process, given its complicated influence on Epa1p expression, display and retention. We tested the ability of our HA-Epa1p expressing S. cerevisiae to bind to sialic acid-depleted epithelial cells in an adherence assay based on those established by previous groups (Cormack, Ghori and Falkow 1999; Halliwell et al. 2012; Diderrich et al. 2015). We observed that transgenic expression of the adhesin confers the otherwise non-adherent yeast the ability to bind to epithelial cells in culture (Fig. 5). Surprisingly, regardless of the Epa1p repeat copy number there was no trend between repeat copy number and adherence at the population level (Fig. 5B). We hypothesize that the differences in protein display and shedding of Epa1p counteract each other, resulting in relatively stable Epa1p function regardless of repeat copy number.

In addition, the ability of HA-Epa1p (0 rep) to mediate binding to epithelium provides an interesting foil to previous findings show that large amino acid deletions from the linker domain impaired Eap1p-mediated binding (Frieman, McCaffery and Cormack 2002). While we did not test these other deletions in our system, it is possible that deletion of the repeat region is not large enough to impair the ability of the binding domain to clear from cell wall, the mechanism shown by Frieman, McCaffery and Cormack (2002).

Structural variation within the repeat region of Epa1p influences protein function by simultaneously influencing mRNA expression, protein trafficking and protein turnover. In our heterologous system, these tradeoffs are functionally balanced such that only the extremes of repeat copy number (either no repeats or 10 repeats) demonstrate impaired Epa1p-mediated cell adhesion. While still speculation, differences in repeat copy number in the native C. glabrata likely have similar tradeoffs. Furthermore, it is also possible that repeat-length differences may mediate subtle differences which modulate cell adhesion, systemic migration and cell wall integrity in the native system during infection, which we would be unable to detect in our system. Indeed, genomic analysis of the same C. glabrata infection before and after a course of antifungal treatment revealed lability in tandem repeat regions, perhaps especially so in the complement of cell wall proteins (Vale-Silva et al. 2017). Thus, we predict that this variation may be a natural mechanism for tuning individual protein properties. As such, considering these mechanisms could prove crucial in treating infections of C. glabrata and understanding how certain infections achieve drug resistance.

Click here for additional data file.