Nicotinic Acid Limitation Regulates Silencing of Candida Adhesins During UTI

See allHide authors and affiliations

Science  06 May 2005:
Vol. 308, Issue 5723, pp. 866-870
DOI: 10.1126/science.1108640


The adherence of Candida glabrata to host cells is mediated, at least in part, by the EPA genes, a family of adhesins encoded at subtelomeric loci, where they are subject to transcriptional silencing. We show that normally silent EPA genes are expressed during murine urinary tract infection (UTI) and that the inducing signal is the limitation of nicotinic acid (NA), a precursor of nicotinamide adenine dinucleotide (NAD+). C. glabrata is an NA auxotroph, and NA-induced EPA expression is likely the result of a reduction in NAD+ availability for the NAD+-dependent histone deacetylase Sir2p. The adaptation of C. glabrata to the host, therefore, involves a loss of metabolic capacity and exploitation of the resulting auxotrophy to signal a particular host environment.

In the United States, Candida albicans and C. glabrata are the primary and secondary causes of both bloodstream and mucosal candidiasis (1, 2). Candida accounts for about 25% of all urinary tract infections (UTIs) related to indwelling catheters, with C. glabrata accounting for approximately 15% of all Candida isolates (3). Adherence to host cells is likely important in the virulence of Candida species (4). In C. glabrata, adherence to epithelial cells in vitro is mediated by a lectin that is encoded by the EPA1 gene (5). EPA1 is part of a gene family in C. glabrata, most members of which are encoded in subtelomeric loci, where they are subject to SIR-dependent transcriptional silencing (6). In Saccharomyces cerevisiae, silencing at telomeres occurs by recruitment of the Sir complex (Sir2p, Sir3p, and Sir4p) to the telomeric repeats, followed by spread into adjacent subtelomeric regions (7). In C. glabrata sir3Δ mutants, two normally silent subtelomeric EPA genes (EPA6 and EPA7) are transcribed and contribute strongly to the overall adherence of the mutant strain (8). To understand the role of chromatin silencing in the regulation of C. glabrata adherence and virulence, we asked specifically whether EPA6 is normally transcribed during the course of an infection.

To assess EPA6 expression during the course of animal infections, we used a recombinational in vivo expression technology approach (9, 10). We constructed a reporter strain, BG1087, in which we replaced the chromosomal EPA6 open reading frame (ORF) with the S. cerevisiae FLP1 ORF encoding the Flp1p site-specific recombinase. In this same strain, we modified the TRP1 locus by placing the PGK1 promoter upstream of the TRP1 coding region, followed by a promoterless hph gene from Klebsiella pneumoniae [encoding hygromycin B resistance (HygR)]. We engineered FRT sites flanking the TRP1 gene (Fig. 1A). Flp1-mediated recombination results in loss of the TRP1 ORF, rendering the cells Trp- HygR, in contrast to the Trp+ Hyg-sensitive (HygS) parent strain. Consistent with the low level of EPA6 expression, we found that only 1% of cells were HygR in cultures of the EPA6::FLP1 grown for 40 generations in vitro; as expected, if the EPA6::FLP1 cassette was carried on a plasmid (where it is not subject to silencing), the entire population was HygR by the time cultures could be analyzed (11).

Fig. 1.

FLP recombinase reporter. (A) Schematics of the FLP1 replacement at the EPA6 locus and the PGK1/FRT/TRP1/FRT/hph reporter, showing pre- and post-recombination configurations. (B) HygR phenotype (expressed as a percentage of total cells recovered) of C. glabrata strain BG1087 recovered from organs after bloodstream infection or UTI. Each symbol (circles) corresponds to the C. glabrata cells recovered from one animal. Bars represent the mean.

We used this system to study the expression of EPA6 in two infection models. First, mice were infected intravenously with strain BG1087, and yeast were recovered from target organs 4 and 7 days after infection. There was no increase in the percentage of HygR colonies recovered from target organs as compared with the percentage present in the initial inoculum (Fig. 1B). We also adapted an established murine model of UTIs (12) to study C. glabrata UTI. After delivery of an inoculum transurethrally, the bladder was colonized and infection ultimately ascended to the kidney (11). In this model, nearly all mice infected with strain BG1087 showed a significant increase in the percentage of HygR colonies recovered from both the bladder and kidney (Fig. 1B). These data suggest that EPA6 is transcribed during UTI but not during disseminated bloodstream infection. To test the hypothesis that urine might induce EPA6 expression, we used a strain in which green fluorescent protein (GFP) replaces the EPA6 ORF at the normal chromosomal locus; as expected, EPA6::GFP was silenced in a wild-type background but was strongly transcribed in a sir3Δ background (fig. S1). Expression of EPA6::GFP or of the EPA6 gene itself was induced when the strain was grown in human urine samples (Fig. 2A and fig. S2). In order to determine what components of urine might be important for this induction, we examined EPA6 induction in a defined synthetic urine medium (13). This is essentially a nutritionally poor medium [5% synthetic complete (SC)] supplemented with urine-specific salts and urea (14). EPA6::GFP expression was induced in synthetic urine, and this induction was independent of urine-specific components but occurred simply in 5% SC (fig. S3A) as a result of limitation of the vitamin niacin [or nicotinic acid (NA)] (Fig. 2B). Limitation of other components of SC did not induce EPA6::GFP transcription (fig. S3, B and C). These results were confirmed by direct measurement of EPA6 transcript levels (fig. S3D).

Fig. 2.

Transcriptional induction of EPA6 during growth in urine or NA-limited media. (A) S1 nuclease protection assay of EPA1, EPA6, and EPA7 transcripts in strain BG2 (wild-type) grown in human urine with or without supplemental NA. (B) Fluorescence of strain BG1045 (EPA6::GFP) cultured in SC media limited (5% of normal) in individual vitamins. (C) HygR phenotype (expressed as a percentage of total cells recovered) of strain BG1087 recovered from a mouse bladder 4 days after infection. Each symbol (circles) corresponds to yeast recovered from a single animal. Results are shown for infections of mice fed normal diets or supplemented daily with NA (90 mg/kg). Bars represent the mean.

Because NA limitation induced EPA6 expression, we examined the NA requirements for C. glabrata growth and found that C. glabrata is an NA auxotroph (Fig. 3A). The closely related S. cerevisiae is an NA prototroph because it can synthesize nicotinic acid mononucleotide (NaMN) from tryptophan via the kynurenine pathway [the enzymes of which are encoded by the BNA genes (15)] (fig. S4). Inspection of the recently published C. glabrata genomic sequence (16) showed that C. glabrata is lacking all of the functional BNA genes (fig. S5). C. glabrata is also auxotrophic for thiamine and pyridoxine (fig. S6A), but equivalent growth limitation of C. glabrata by limiting NA, thiamine, or pyridoxine showed that only NA limitation resulted in induction of EPA6 (fig. S6B).

Fig. 3.

Growth and silencing in C. glabrata tna1Δ, sir2Δ, hst1Δ, and hst2Δ strains. (A) Strain BG1045 and strain BG1176 grown on SC NA plates supplemented with a range of NA concentrations (from 0.325 to 32.5 μM). WT, wild type. (B) Phase-contrast and fluorescence microscopy showing EPA6::GFP fluorescence of strains BG1045 and BG1176 grown in SC (3.25 μM NA). The table below shows fluorescence-activated cell sorter (FACS) quantitation of cells grown on SC (3.25 μM NA) plates. The percent of all cells that are fluorescent is shown, as well as the average fluorescence intensity of those cells that are fluorescent. (C) EPA6 transcript levels, measured by S1 nuclease protection, in strains BG2 (WT), BG1048 (sir2Δ), BG1073 (hst1Δ), and BG1112 (hst2Δ). Also indicated is the fold increase of the EPA6 transcript (normalized to ACT1 levels) relative to levels in BG2.

After demonstrating that NA limitation in yeast medium or in synthetic urine cultures resulted in EPA6 induction, we confirmed that EPA6 induction in urine itself was also the result of NA limitation, because addition of excess NA or the related compound nicotinamide (NAM) suppressed the induction of EPA6 (Fig. 2A and fig. S2). To test whether NA limitation contributed to EPA6 induction during experimental UTI, we tested whether alteration of NA levels in the urine of mice could change the induction of EPA6 during infection. Rodents fed an excess of NA excrete the excess NA in the form of NA and NAM [as well as various metabolites such as nicotinuric acid and methyl nicotinamide (17)]. In our laboratory, mice fed a diet supplemented daily with 90 mg of NA per kg of feed (mg/kg) (versus approximately 15 mg/kg in daily consumption of normal mouse feed) had combined urine levels of NA and NAM about five times higher than in mice kept on a control diet (14). When we infected these mice with the EPA6::FLP1 strain, we found that, on average, 5% of the cells recovered from the bladders of mice fed the high-NA diet were HygR, compared with 12% of cells recovered from the bladders of mice fed a normal diet (Fig. 2C). This difference was marginally significant (P = 0.055) and consistent with a role for NA limitation during UTI, contributing to induction of EPA6.

Because C. glabrata is an NA auxotroph, it relies solely on exogenous sources of NA for growth. In S. cerevisiae, NA is transported from the environment by the high-affinity transporter Tna1p (18). We deleted the C. glabrata TNA1 ortholog in the EPA6::GFP reporter strain to measure the effect on EPA6 induction. The tna1Δ EPA6::GFP strain cannot grow at concentrations of NA sufficient for growth of the wild-type strain (Fig. 3A), which is consistent with a role in high-affinity NA transport. For cells growing in SC media (3.25 μM NA), EPA6::GFP was essentially not expressed in wild-type cells, but it was expressed in the majority of tna1Δ cells (Fig. 3B), consistent with a decrease in cellular NA levels signaling a loss of EPA6 silencing.

NA is a precursor of nicotinamide adenine dinucleotide (NAD+), and our results suggested that limiting environmental NA levels leads to reduced intracellular NAD+ levels, thus affecting Sir2p NAD+-dependent histone deacetylase activity and ultimately, EPA6 silencing. To test this model, we examined the role of SIR2 and two related NAD+-dependent histone deacetylases, encoded by the HST1 and HST2 genes, which have been implicated in S. cerevisiae in repression of transcription (7, 19, 20). We deleted the C. glabrata orthologs of SIR2, HST1, and HST2 and assessed the derepression of EPA6 in each background. EPA6 was derepressed most significantly in the sir2Δ background (Fig. 3C); moreover, there was no further increase in EPA6 expression when the sir2Δ strain was grown under NA-limited conditions, consistent with NA limitation signaling EPA6 derepression via its effect on Sir2p. If NA limitation induces EPA6 via Sir2p, then multiple genes regulated by subtelomeric silencing should be derepressed under those growth conditions. We found that three telomeric EPA genes—EPA1, EPA6, and EPA7—were induced by growth in urine and that this induction could be suppressed by the addition of excess NA (Fig. 2A). We also saw a modest increase in the levels of EPA4 and EPA5 (11). This pattern of EPA gene derepression was essentially identical to that seen in strains in which SIR3 was deleted (8) and suggests that NA-limited growth results in a general loss of silencing, mimicking the absence of SIR3.

Because at least some of the EPA genes are adhesins (8), we assessed whether growth in urine had any impact on in vitro adherence to uroepithelial cells. We assessed adherence to T24 cells (human bladder epithelial cells) and A498 cells (human proximal tubular renal epithelial cells) and found that growth of wild-type and epa1Δ cells in urine or NA-limited medium increased adherence to both cell types. By contrast, an epa(1,6,7)Δ strain did not adhere significantly to either cell type (Fig. 4A), suggesting that the three EPA genes that were induced by growth in urine also mediate most of the uroepithelial adherence induced by NA limitation. These data were mirrored by experiments with a sir3Δ strain (table S1). Lastly, we found that the epa(1,6,7)Δ strain showed a decreased (P = 0.018) ability to colonize the bladder in the murine UTI model when compared to the wild-type parent strain, consistent with a functional role for EPA genes in colonization of the urinary tract (Fig. 4B).

Fig. 4.

Adherence of C. glabrata cells grown in urine. (A) Adherence to uroepithelial T24 and A498 cells of C. glabrata strains BG2 (WT), BG178 (epa1Δ), and BG1026 [epa(1,6,7)Δ] grown in SC; SC with limited NA; urine; or urine supplemented with NA, expressed as the percent of adherent cells present in each culture. (B) Colonization of murine bladder by strains BG462 (WT) and BG1026 [epa(1,6,7)Δ]. Circles represent the colony-forming units (CFU) recovered from individual animals and bars represent the geometric mean.

It has been hypothesized that changes in cell metabolism that alter cellular NAD+ levels might modulate Sir2p activity (21, 22). Here we provide data suggesting that cellular NAD+ levels, which in C. glabrata are tied to environmental NA levels, modulate Sir2p-mediated silencing of subtelomeric EPA genes in response to the host environment. The cellular responses of C. glabrata and S. cerevisiae to NA limitation differ markedly. NA limitation in S. cerevisiae does not result in loss of silencing (22), because the de novo NAD+ biosynthesis pathway (encoded by the BNA genes) acts to maintain a constant cellular NAD+ level independent of environmental NA (23). Inspection of genome sequences ( showed that the BNA pathway is broadly distributed among fungi (being present, for example, in C. albicans, Neurospora crassa, Aspergillus nidulans, and Cryptococcus neoformans). C. glabrata has deleted all the BNA genes [with the exception of BNA3 (24)], which is striking because the S. cerevisiae and C. glabrata genomes are highly related; of the nearly 6000 S. cerevisiae ORFs, only approximately 200 have no obvious ortholog in C. glabrata (25). The BNA genes are present in Eremothecium ( and Kluyveromyces ( in blocks that are syntenic relative to the S. cerevisiae genome, consistent with the pathway having been lost in the C. glabrata lineage rather than recently acquired in the S. cerevisiae lineage.

That C. glabrata is an NA auxotroph suggests that there must be microbiologically available NA in C. glabrata's ecological niche. C. glabrata is not normally isolated from the environment but is found only in association with the mammalian host (26). Relevant to the experiments in this paper, urine is known to contain both NA and NAM (27). Urinary tract isolates of Escherichia coli are also often NA auxotrophs (28, 29). Our data also imply that the NA levels sensed by C. glabrata in the urinary tract are sufficient to support growth but insufficient for full Sir2p function.

What is the impact of NA limitation on the virulence of C. glabrata? NA limitation induces three adhesins—EPA1, EPA6, and EPA7—which we show have roles in adherence to uroepithelial cells in vitro and in bladder colonization in the mouse UTI model. For C. glabrata, and indeed C. albicans, colonization of the urinary tract is seen in the hospital setting primarily in the context of indwelling catheters on which Candida can form biofilms. EPA6 has recently been implicated in biofilm formation in vitro (30). Perhaps NA limitation in the urinary tract, or specifically on catheter surfaces, might increase EPA6 expression and stimulate the formation of catheter-associated biofilms.

We suggest that C. glabrata's evolutionary adaptation to the host included two sets of events: the loss of the BNA genes (and resulting NA auxotrophy) and the encoding of the EPA adhesin family at subtelomeric loci. We provide evidence that these aspects of the C. glabrata genome, combined with the NAD+ dependence of Sir2p, are key features governing the regulation of the EPA genes in response to particular NA-limited host environments.

Supporting Online Material

Materials and Methods

Figs. S1 to S6

Tables S1 to S4


References and Notes

View Abstract

Navigate This Article