Expression of Candida glabrata adhesins following exposure to chemical preservatives

Abstract

In Candida glabrata, an opportunistic yeast pathogen, adherence to host cells is mediated in part by the Epa family of adhesins, which are encoded largely at subtelomeric loci where they are subject to transcriptional silencing. In analyzing the regulation of the subtelomeric EPA6 gene, we found that its transcription is highly induced after exposure to methylparaben, propylparaben or sorbate. These weak acid-related chemicals are widely used as antifungal preservatives in many consumer goods, including over-the-counter (OTC) vaginal products. Culture of C. glabrata in a variety of vaginal products induced expression of EPA6, leading to increased adherence to cultured human cells as well as primary human vaginal epithelial cells. We present evidence that paraben/sorbate-induction of EPA6 expression involves both preservative stress and growth under hypoxic conditions. We further show that activation of EPA6 transcription depends on the Flo8 and Mss11 transcription factors and does not require the classical weak acid transcription factors War1 or Msn2/Msn4. We conclude that exposure of C. glabrata to commonly used preservatives can alter expression of virulence-related genes.

INTRODUCTION

Parabens (methyl, ethyl or propyl esters of parahydroxy benzoic acid) and sorbic acid are weak acid related compounds that have been used for decades as preservatives to prevent the growth of spoilage yeasts in consumer goods [1, 2]. Their action is related to cytosolic acidification as well as effects on energy metabolism through inhibition of glycolytic enzymes [3].

Candida species are the cause of disseminated infections in hospitalized patients as well as superficial mucosal infections. Up to 75% of women will experience at least one bout of Candida vulvovaginitis in their lifetime, with up to 10% having recurring infections [4–6]. C. albicans is the leading cause of these infections, but C. glabrata has emerged over the last two decades as an frequent isolate and accounts for approximately 15% of both mucosal and disseminated infections.

C. glabrata encodes a family of at least 23 EPA genes, which encode cell surface proteins capable of mediating adherence to epithelial cells [7]. The EPA genes are encoded largely in sub-telomeric regions where they are subject to transcriptional silencing, dependent on the SIR2, SIR3 and SIR4 genes. Some of the silenced EPA genes can be transcriptionally induced under particular environmental conditions: for example, EPA6 is de-repressed by growth in urine and during murine urinary tract infection. This transcriptional de-repression is thought to occur in response to limiting environmental levels of vitamin precursors of NAD⁺, which leads to a reduction in cellular NAD⁺. Since Sir2 is a NAD⁺-dependent enzyme, this results in a reduction in Sir2-mediated histone deacetylation and loss of silencing [8].

We show here that EPA6 transcription is also induced by exposure to hypoxic conditions and to two different preservatives widely used in consumer products, and that this expression leads to increased adherence in vitro to vaginal epithelial cells. We have characterized the transcription factors required for preservative-induced EPA6 expression. Unexpectedly, expression is independent of “classic” weak acid-responsive transcription factors and depends rather on the Flo8 and Mss11 transcription factors.

METHODS

Strains, plasmids and primers

All strains used in this study are shown in Table S1. Plasmids for construction of strains are summarized in Table S2. Plasmids were propagated in E. coli strain DH10 (Invitrogen). Primers used for cloning are listed in Table S3. All PCR products were sequenced before use.

Media

SC media supplemented with casamino acids (6g l⁻¹) was used unless otherwise indicated; YPD media was used for growth and analysis of Candida albicans [9].

Deletion strain construction

Yeast transformation was carried out as previously described [10]. Some deletion strains were constructed with the two-step method [10, 11]. Other strains were generated using a one step [8]. In brief, a cassette was generated consisting of the Hph resistance gene from Klebsiella pneumoniae flanked by the C.g. PGK1 promoter sequence and the 3’UTR of HIS3 from S. cerevisiae. This was cloned between PCR-amplified 5’ and 3’ intergenic regions for the ORF being disrupted. In order to allow recycling of the marker, we placed FRT sites flanking the cassette Hph cassette. Marker removal is accomplished using a Ura⁺ plasmid (pRD16) containing FLP expressed from the EPA1 promoter. Flanking regions for deletion of ORFs were generated by PCR.

Plate and liquid growth assays

SC plates/media was supplemented with chemicals and adjusted to a pH of 4.5 after addition. We used concentrations of chemicals or OTC vaginal products that empirically resulted in similar levels of growth inhibition (OD₆₀₀ after 14h between OD 3–7, compared to OD₆₀₀ =30 for control cultures): parabens (1.5 mM methylparaben/165uM propylparaben), sorbic acid (5mM), acetic acid (50mM), benzoic acid (6mM), lactic acid (400mM), propionic acid (20mM). For other preservatives, concentrations (w/v) used are as follows: benzalkonium chloride 0.00075%, benzethonium chloride 0.0025%, EDTA 0.0125%, glycerin 5%, phenethyl alcohol 0.29%, propylene glycol 5%. For OTC products, concentrations ranged from 15% to 25% of the culture volume for gels, liquids, and foams. For the Today Sponge, 1/10 of one sponge was used in the 10 ml culture.

FACS analysis of GFP expression

To analyze EPA6::GFP expression, cells were grown, washed in PBS, analyzed using a FACScan7600 (Becton Dickinson).

RNA preparation and S1 nuclease protection assay

S1 nuclease protection assays and RNA preparation was as previously described [12]. Oligonucleotide probes (Table S3) were labeled with ³²P[ATP]γ. For the experiments in figure 3, the strain was cultured sequentially for two overnights in SC media or SC media supplemented with parabens or sorbic acid. This optimizes the signal amplitude and does not select for genetically altered mutants since subsequent culture in the absence of preservative returns the cells to a non-expressing state from which preservative re-exposure results in EPA6 induction with the same kinetics and amplitude as seen for unexposed cells.

Figure 3. Growth and expression patterns in C. glabrata transcription factor mutant backgrounds.

(A) Plate growth assay of war1Δ, msn2/4Δ and msn2/4Δwar1Δ mutants under weak acid stress. (B) S1 nuclease protection assay of EPA6 and PDR12 transcripts under (lane 1) control, (lane 2) paraben (1.5 mM methylparaben/165uM propylparaben) or (lane 3) sorbic acid (5mM) stress in transcription factor mutant strains. For war1Δ and msn2/4Δ war1Δ strains, concentrations of sorbic acid were 1.7 mM since war1Δ mutants are hypersensitive to this compound. After quantitation by phosphorimager, transcript levels were normalized to ACT1, and fold induction relative to levels in wild type unstressed cells is reported.

Adherence assays

Cells were labeled by growing at 30°C in media supplemented with 25µl ³⁵S[Met] protein labeling mix (New Life Sciences Products, Boston cat# NEG072). Cells were collected at OD₆₀₀ = 4–8, washed (3X) and resuspended in HBSS + 5mM CaCl₂ at a density of 10⁷ cells/ml. Cells were then assayed for adherence (MOI 100:1) to mammalian Lec2 cell line [12]. For assessing adherence of additional clinical isolates, we screened a collection of 23 isolates from various anatomical sits. Most were already adherent under control conditions and two others flocculated when exposed to preservatives, precluding accurate assessment of adherence. We analyzed the 6 strains (including our lab strain) that showed the least adherence to Lec2 cells when grown under control conditions in YPD.

For assessing adherence to primary human cells, adherence assays were conducted essentially as described in [13]. Vaginal epithelial cells were collected and pooled from healthy female volunteers by gentle scraping of the mucosal surface of the vagina using sterile swabs. Collected cells were pooled, washed and suspended in HBSS + 5mM CaCl₂ at a final concentration of 10⁵ cells/ml. Yeast cells were labeled and prepared as above. Epithelial cells and yeast (MOI 100:1) were mixed and incubated at 37°C for 2 hours on a rotary platform. Samples were then suspended in 10mL HBSS + 5mM CaCl₂ and filtered over 12µM polycarbonate filters (Millipore) using a vacuum manifold system. Filters were washed with 50mL of HBSS + 5mM CaCl₂ and placed in 0.5%SDS/0.1% Triton X100 in PBS at 50°C for 15 minutes; samples were diluted into 4mL of scintillation fluid and counted. Background adherence of yeast to filters was calculated from yeast-only controls, and subtracted from all experimental values.

RESULTS

Preservatives present in vaginal health products induce EPA6 expression

Transcription of the EPA6 adhesin gene is strongly silenced by chromatin modification mediated by the Sir proteins. We hypothesized that environmental signals might stimulate expression of EPA6 even in the context of intact Sir2-mediated silencing.

Approximately half of vaginal health products contain parabens or sorbic acid as preservatives [14]. We assessed the effects of these preservatives on C. glabrata EPA gene transcription using a reporter strain (BG1045) in which the EPA6 ORF has been replaced by GFP. When the EPA6::GFP strain is cultured in media supplemented with several OTC vaginal products, expression of EPA6 is strongly induced in a subpopulation of cells (Figure1A and 1B). All vaginal health care products that contained parabens or sorbic acid induced EPA6 expression. The single product tested that did not contain either compound (Very Private Moisturizer) failed to induce EPA6 transcription (Figure 1B).

Figure 1. EPA6 expression under conditions of OTC product preservative stress.

(A) FACS analysis of EPA6::GFP (BG1045) EPA1::GFP (BG215) or EPA7::GFP) (BG1393) expression during growth in SC supplemented with 1.5 mM methylparaben/165uM propylparaben or sorbic acid (5mM). (B) Expression of EPA6::GFP expression after growth in SC media (pH 5.2) supplemented with addition of 15–25% (v/v) OTC vaginal products, parabens, sorbic acid, benzoic acid, parahydroxybenzoic (PHB) acid or lactic acid.

To demonstrate that the preservative compounds were responsible for EPA6 induction, we tested the effects of adding either parabens or sorbic acid to normal growth media. The concentration of preservatives in OTC vaginal products are 0.1%–0.18% (wt/vol) for methylparaben (6.6–11.8mM), 0.02–0.1% for propylparaben(1.1–5.5mM), and 0.1–0.2% for potassium sorbate (3.3–13.3mM) [14]. Since methyl- and propyl- paraben are usually used in combination, we used a mixture of these two compounds. In SC media, in the presence of 1.5 mM methylparaben/165µM propylparaben (15% of the concentration present in most formulations) or 5mM potassium sorbate (culture pH 5.2) yeast growth was inhibited to 25% of control (data not shown) and EPA6 expression was strongly induced (Figure 1A, B). EPA6 expression is also induced by exposure to growthlimiting concentrations of either methylparaben or propylparaben alone (data not shown). EPA6 induction in the presence of parabens or sorbate was not affected by supplementation of SC media with a 1000× excess of Nicotinic Acid (3.25 mM) (data not shown).

We next assessed whether two other EPA adhesins, EPA1 and EPA7, are transcriptionally induced after paraben or sorbic acid exposure. EPA1 is expressed under paraben, but not sorbic acid stress, and transcription of EPA7 does not seem to be affected by either condition (Figure 1A).

Specificity of EPA6 induction by weak acid related preservatives

To assess the specificity of paraben or sorbic acid induction of C. glabrata adherence, we examined EPA6 expression under additional stress conditions. These conditions included exposure to other preservatives, including benzalkonium chloride, benzethonium chloride, EDTA, glycerin, phenethyl alcohol, propylene glycol and lactic acid [14], other weak acids, including boric acid, benzoic acid, propionic acid and acetic acid. Of these, only benzoic acid, which is structurally related to parabens, induced EPA6 expression (Figure 1B). Culture at pH ranging from pH 3.0 – pH 4.5 or in the presence of the uncoupling agent, 2,4 dinitrophenol (DNP) also did not induce EPA6 expression (data not shown). Notably, parahydroxybenzoic acid (pHB acid), the parent acid for the esterified parabens, did not inhibit growth even at concentrations of 100mM (culture pH of 4.5 or 5.2) and did not induce EPA6 expression (Figure 1B).

Paraben and sorbic acid stress results in increased adherence to vaginal tissues

EPA1, EPA6 and EPA7 are lectins that mediate adherence of C. glabrata to epithelial cells in vitro [12, 15]. In Table 1, we show that in C. glabrata, exposure to parabens or sorbic acid results in increased adherence to the Lec2 epithelial cell line. In addition, we collected and pooled primary human vaginal epithelial cells and assessed the adherence of C. glabrata after paraben or sorbic acid exposure [13]. As with Lec2 cells, we found that paraben or sorbic acid exposure increased adherence to vaginal epithelial cells (Table 1 and Figure 2). As a control, we used a C. glabrata sir3Δ strain (BG676) which is disrupted for silencing and hyperadherent because it constitutively expresses some sub-telomeric EPA genes [12, 15]; the sir3Δ strain adhered to both Lec2 cells and primary vaginal epithelial cells at levels that were comparable to that elicited in wild type C. glabrata by preservative exposure. EPA1 and EPA6 are both transcriptionally induced by paraben exposure (Figure 1A); accordingly, preservative-exposed epa1Δ epa6Δ cells were non-adherent implicating these two genes in increased adherence and providing evidence that EPA7 has no role in adherence following preservative exposure (Table 1).

Table 1.

Adherence of C. glabrata strains to Lec2 cells, or primary human vaginal cells (M.O.I. 100:1). C. glabrata strains are wild type (BG2), epa(1,6)Δ (BG1044), and sir3Δ (BG808). The yeast strains were grown in SC media (pH 5.2) alone or supplemented with parabens (1.5 mM methylparaben/165uM propylparaben) or sorbic acid (5mM). Data is expressed as adherent yeast per 100 mammalian cells and represents the average of three independent assays.

Strain	Condition	Lec2 cells	Primary vaginal cells
		Adherence (yeast/100 cells)
WT	Control	37±6	28±4
	Parabens	370±80	150±60
	Sorbic Acid	240±10	70±9
epa1,6Δ	Control	3.8±2.5	1.9±1.1
	Parabens	6.4±3.9	0.23±0.20
	Sorbic Acid	5.4±1.8	3.7±2.9
sir3Δ	Control	730±10	250±70
	Parabens	650±50	160±60
	Sorbic Acid	610±100	250±10

Figure 2. C. glabrata adherence under conditions of OTC product preservative stress.

Phase contrast microscopy of primary human vaginal cells and adherent yeast after removal of non-adherent yeasts by washing. Strains were cultured in SC, supplemented as indicated with parabens (1.5 mM methylparaben/165µM propylparaben) or sorbic acid (5mM).

We next analyzed adherence in 23 additional C. glabrata clinical isolates (gifts of J. Sobel and M. Pfaller). Adherence of these strains under control culture conditions varied, with many strains adhering strongly even without preservative exposure. We selected the 5 C. glabrata clinical isolates which showed the least adherence to Lec2 cells under control growth conditions. For four of these additional strains, preservative exposure resulted in a marked increase in adherence; in the case of strain MF5046, for reasons we do not understand, adherence was marginally increased under conditions of paraben stress but decreased under sorbic acid stress (Table 2).

Table 2.

Adherence of C. glabrata clinical strains to Lec2 cells. The yeast strains were grown in SC media (pH 5.2) alone or supplemented with parabens (1.5 mM methylparaben/165uM propylparaben) or sorbic acid (5mM). Data is expressed as adherent yeast per 100 mammalian cells and represents the average of three independent assays.

Strain	Origin	Control	Parabens	Sorbic Acid
	Adherence (yeast/100 cells)
BG2	vaginal	38 ± 4	350 ± 20	320 ± 20
BG1	vaginal	54 ± 4	510 ± 5	240 ± 18
BG901	vaginal	45 ± 9	410 ± 10	370 ± 10
MF4566	blood	33 ± 3	110 ± 20	110 ± 10
MF4930	blood	260 ± 16	830 ± 20	580 ±18
MF5046	urine	400 ± 10	660 ± 40	250 ± 25

Regulation of EPA6 transcription in response to preservative stress

What transcriptional regulators control EPA6 transcription in response to preservative stress? Studies of the sorbic acid response in S. cerevisiae have implicated the War1 (Weak Acid Resistance) transcription factor as well as the general stress response transcription factors Msn2 and Msn4 [16]. To determine if any of these factors are implicated in the induction of EPA6, we deleted MSN2, MSN4 and WAR1 alone and in combination and assessed their role in EPA6 transcription.

In S. cerevisiae, war1 mutants are hypersensitive to sorbic acid, because they fail to induce transcription of the PDR12 gene, which encodes an ABC transporter required for resistance to sorbic acid and other lipophilic weak acids [17]. Accordingly, the C. glabrata war1Δ strain is hypersensitive to sorbic acid stress and propionic acid stress; the strain is not more sensitive to acetic acid stress, nor is it more sensitive to parabens (Figure 3A). C. glabrata PDR12 was transcriptionally induced under conditions of sorbic acid stress, but not paraben stress; induction of PDR12 transcription was, as expected, absolutely dependent on WAR1 (Figure 3B). However, in the war1Δ strain, EPA6 transcription was still induced under conditions of both paraben and sorbic acid stress (Figure 3B).

In the C. glabrata msn2Δ/msn4Δ strain, EPA6 transcription was still induced under conditions of both paraben and sorbic acid stress (Figure 3B). A triple msn2Δ msn4Δ war1Δ mutant is, like the war1Δ mutant, hypersensitive to sorbic acid stress (Figure 3A) and unable to induce PDR12 (Figure 3B). However, EPA6 is still expressed in the triple mutant background under paraben and sorbic acid stress, and is also transcribed at appreciable levels in the absence of these stresses (Figure 3B). In sum, our data suggest strongly that EPA6 transcription in response to sorbic acid or paraben stress occurs by a pathway that is independent of War1, Msn2 and Msn4.

EPA6 expression depends on FLO8 and MSS11

Since orthologues of the transcription factors known to be implicated in the S. cerevisiae weak acid stress response do not seem to be involved in preservative induction of EPA6, we considered other possibilities. The EPA-encoded proteins are analogous to the S. cerevisiae FLO genes [7], whose transcription depends on the transcription factors Flo8 and Mss11 [18]. We generated C. glabrata flo8Δ and mss11Δ strains. Deletion of these genes did not alter sensitivity to parabens or sorbic acid (data not shown), and did not affect PDR12 expression, but resulted in a total loss of EPA6 expression after paraben or sorbic acid exposure (Figure 3B), strongly implicating these factors in the positive regulation of this gene.

Hypoxic conditions in culture are required for preservative induction of EPA6 transcription

In determining the dynamics of EPA6 induction under paraben and sorbic acid stress, we noticed that EPA6 gene expression is maximally induced only when the OD₆₀₀ increased above 1.0. In a time course following acute paraben or sorbic acid stress, EPA6 expression increases with density (Figure 4A). We hypothesized that low dissolved oxygen levels, found in cultures at high cell density but not low density cultures might be responsible for the density dependence of EPA6 induction in response to parabens. In support of this hypothesis, we found that in cultures maintained at 0.1% oxygen, even low-density cultures expressed EPA6 under conditions of paraben or sorbic acid stress. Conversely, maintenance of atmospheric levels of dissolved oxygen in high-density cultures reduced EPA6 expression (Figure 4B). Thus, our data indicates that EPA6 induction in response to paraben or sorbic acid stress requires a combination of preservative stress and hypoxia.

Figure 4. EPA6 expression under conditions of high density and low oxygen.

(A) EPA6 transcript levels, as measured by S1 nuclease protection, after acute stress with parabens or sorbic acid collected over a range of culture densities. EPA6 transcript levels were normalized to ACT1, and fold induction relative to zero time point for each culture is reported. (B) S1 nuclease protection assay of EPA6 transcript after paraben stress in low or high-density cultures in cultures aerated with excess air or 0.1% oxygen. EPA6 transcript levels were normalized to ACT1, and fold induction relative to un-aerated, low density cultures is reported.

Effect of Preservative stress on morphology in Candida albicans

Lastly, to explore the generality of our findings, we examined the effects of parabens or sorbic acid on C. albicans. We found that exposure of C. albicans to paraben but not sorbic acid stress results in an altered morphology from primarily yeast to primarily pseudohyphal form (Figure 5), suggesting that there may be parallels in C. albicans to the transcriptional effects of parabens in C. glabrata.

Figure 5. Effects of paraben stress on C. albicans morphology.

Phase contrast microscopy of C. albicans strain 5314 grown in YPD (pH 4.5) (A), supplemented with parabens (1.5 mM methylparaben/165uM propylparaben) (B), or sorbic acid (2.7mM) (C).

DISCUSSION

C. glabrata encodes at least 23 EPA gene family members [7]. Encoding so many similar genes may be related to the pathogen’s ability to adapt to differing host environments with selective expression of different EPA genes, encoding adhesins with different ligand specificities in different host environments. Our data here suggests that some EPA genes are de-repressed under conditions of paraben or sorbic acid stress. We have previously shown that EPA genes are expressed under limiting NAD⁺ conditions due to a general loss of telomeric silencing [8]. The specificity of EPA transcriptional induction in response to paraben or sorbic acid stress (induction of EPA1 and EPA6 but not EPA7) and the fact that paraben-induced EPA6 expression is unaffected by addition of excess nicotinic acid (a precursor of NAD⁺), suggests that paraben or sorbic acid exposure likely does not result in a general loss of telomeric silencing.

EPA6 induction under paraben or sorbic acid stress conditions does not depend on the War1 and Msn2/Msn4 stress response transcription factors. Interestingly, in a triple msn2Δ msn4Δ war1Δ mutant, EPA6 was expressed at appreciable levels in the absence of preservative stress (Figure 3B), suggesting that in the absence of Msn2, Msn4 and War1, endogenous stress may activate the EPA6 induction pathway even without the addition of exogenous sorbic acid or parabens. While, Msn2, Msn4 and War1 seem to be dispensible for preservative-induced induction of EPA6, our data implicate two other transcription factors, Flo8 and Mss11. One possibility, consistent with our data, is that in response to paraben and sorbic acid stress, direct modification of Mss11 and Flo8 permits them to activate the Sir2-modified, epigenenetically-repressed EPA6 promoter.

In our experiments, EPA6 activation occurs in response to a combination of two distinct signals (paraben or sorbic acid stress and hypoxia). With regard to induction of EPA6 under hypoxic conditions, our data are consistent with previous results showing that EPA6 is transcriptionally induced under conditions of biofilm formation and at high culture density [19]. Whether preservative stress and hypoxia affect one pathway or multiple pathways is not clear, and further work is needed to identify the mechanism by which preservatives and hypoxia affect transcription in C. glabrata,

Parabens and sorbate have been used for decades in a wide array of products and, aside from some reports of skin sensitivity, extensive studies have not revealed safety concerns with their use. We have shown that two C. glabrata adhesin genes, EPA1 and EPA6, are expressed after exposure to these compounds. We have shown that adherence to vaginal epithelial cells of at least some clinical isolates can be induced by paraben exposure in vitro. To extend these findings, the effect of paraben exposure on EPA gene expression in additional clinical isolates should be analyzed. Whether Epa-mediated adherence plays a role in C. glabrata vaginal colonization, and whether preservative exposure affects EPA gene expression in colonized patients is not known, but should be explored to determine if paraben or sorbic acid exposure could affect the etiology of C. glabrata vaginitis.