Role of Ess1 in Growth, Morphogenetic Switching, and RNA Polymerase II Transcription in Candida albicans

Candida albicans is a fungal pathogen that causes potentially fatal infections among immune-compromised individuals. The emergence of drug resistant C. albicans strains makes it important to identify new antifungal drug targets. Among potential targets are enzymes known as peptidyl-prolyl cis/trans isomerases (PPIases) that catalyze isomerization of peptide bonds preceding proline. We are investigating a PPIase called Ess1, which is conserved in all major human pathogenic fungi. Previously, we reported that C. albicans Ess1 is essential for growth and morphogenetic switching. In the present study, we re-evaluated these findings using more rigorous genetic analyses, including the use of additional CaESS1 mutant alleles, distinct marker genes, and the engineering of suitably-matched isogenic control strains. The results confirm that CaEss1 is essential for growth in C. albicans, but show that reduction of CaESS1 gene dosage by half (δ/+) does not interfere with morphogenetic switching. However, further reduction of CaEss1 levels using a conditional allele does reduce morphogenetic switching. We also examine the role of the linker α-helix that distinguishes C. albicans Ess1 from the human Pin1 enzyme, and present results of a genome-wide transcriptome analysis. The latter analysis indicates that CaEss1 has a conserved role in regulation of RNA polymerase II function, and is required for efficient termination of small nucleolar RNAs and repression of cryptic transcription in C. albicans.

One strategy to overcome drug resistance while producing synergistic effects is the use of combination therapies that target distinct intracelluar pathways [22][23][24][25][26][27]. Toward this end, a number of different pathways are being investigated including those containing enzymes known as peptidyl prolyl cis-trans isomerases [28,29]. PPIases catalyze the isomerization of the peptide bond preceding prolines within protein substrates [30,31]. Three major families of PPIase have been described; cyclophilins, FK506binding proteins (FKBPs) and parvulins (reviewed in [32]. All three families are conserved from yeast to humans. Inhibitors such as cyclosporin, which targets the cyclophilins, and FK506 or rapamycin, which target FKBPs, all show potent antifungal activity [33][34][35][36]. However, these drugs are also immunosuppressive via pathways that inhibit T-cell activation, an activity that could make individuals more vulnerable to fungal infections [37,38]. An alternative PPIase target might be Ess1, a founding member of the parvulin class of PPIase, which is structurally distinct from the cyclophilins and FKBPs [39,40] and whose human ortholog Pin1 is not known to be associated with the T-cell activation pathway. The first fungal Ess1 was discovered in Saccharomyces cerevisiae and shown to be essential [41]. Ess1 and Pin1 play critical roles in gene transcription by RNA polymerase II (pol II) [42][43][44][45][46]. Ess1 isomerizes peptide bonds within the carboxy-terminal domain (CTD) of Rpb1, the largest subunit of RNA pol II, and thereby controls binding and release of transcriptional co-factors [43,[47][48][49]. The X-ray structures of the C. albicans Ess1 protein and its human ortholog, Pin1 have been solved [40,50,51], and while the enzymes show overall similarity, there are key differences including a large solvent-exposed alpha-helix within a structured linker region that is present in the fungal enzyme but absent in the human enzyme. This helix has been hypothesized to play a role in fungal-specific functions, potentially by engaging in proteinprotein interactions [50].
Homologs of Ess1 are found in all major pathogenic fungi that have been examined, including C. albicans [52], Candida glabrata (REFSEQ XP_445146), Cryptococcus neoformans [53], and Aspergillus nidulans (PinA) [54]. In C. albicans and A. nidulans Ess1 (PinA) is essential for growth [52,54]. In Cryptococcus neoformans, Ess1 is not essential for growth, but is required for expression of virulence factors melanin and urease and for virulence in a mouse model [53].
In previous work, we isolated the CaESS1 gene from C. albicans and showed it to be essential for growth in this organism using a novel temperature-sensitive mutant strategy [52]. Surprisingly, we found a heterozygous mutant strain (Caess1d/CaESS1) to be defective for filamentation in various inducing media leading us to conclude that CaESS1 gene dosage is important for morphogenetic switching [52]. However, several considerations led us to re-evaluate these findings. First, our CaESS1 dosage experiments used the URA-blaster method [55], which has largely been superseded to avoid variations in virulence phenotypes due to expression of URA3 from ectopic loci [56,57]. Second, the importance of using reconstituted strains rather than parental strains as controls is now well-established [57,58]. Finally, sequencing of the C. albicans genome [59] revealed a gene, APE2 located very close to CaESS1, whose promoter might have been disrupted in our previous strain constructions.
We therefore re-examined our previous findings using newer information and more rigorous methods. The results showed that C. albicans Ess1 is essential for growth, but that in contrast to our previous report, CaESS1 heterozygous mutants did not show a defect in morphogenetic switching, and instead the defect was traced to a host cell mutation(s). Potential effects of URA3 marker gene placement were also ruled out. We also report results of a structure-function analysis of the CaEss1 linker a-helix, and describe a conditional-lethal readthrough allele of CaESS1 that will be useful for further studies. Finally, results of a transcriptome analysis using high-throughput RNA-sequencing suggests that C. albicans Ess1, like its counterparts in budding yeast and humans plays a role in regulating RNA polymerase II function.

Results and Discussion
Reduction of CaESS1 gene dosage does not affect filamentation phenotypes In previous work, we constructed a C. albicans Caess1d/CaESS1 heterozygous mutant strain in the CAI4 (ura 2 ) parent strain using the URA-Blaster method [55] ( Figure 1A). This strain, CaGD1, was defective for filamentation in Lee's, serum-containing and Spider media when compared to the ''wild-type'' control SC5314 [52]. The CAI4 (ura 2 ) parent strain is a poorly-filamenting strain, presumably due to its uracil auxotrophy [60]. Instead, the parent strain of CAI4, clinical isolate SC5314, had been used as a control, because SC5314 is a uracil prototroph (ura + ) and could therefore be compared to CaGD1 (ura + ), even though SC5314 has two copies of URA3 and CaGD1 has only one copy.
Here, we generated a better control, isogenic to CAI4, but with one copy of the URA3 gene placed at the CaESS1 locus without disrupting it ( Figure 1E). Surprisingly, this reconstituted control strain, R6 (ura + ) (CaESS1/CaESS1), also did not filament on serum-containing medium or Spider medium even though both CaESS1 alleles were left intact ( Figure 2E). Filamentation on Lee's medium was also defective (data not shown). It is possible that the integrated URA3 construct in R6 somehow lowered CaESS1 expression levels in cis, but this was ruled out using quantitative reverse transcription real time PCR (qRT-PCR) and Western analysis, which shows the expected CaESS1 RNA and protein levels in the heterozygotes (CaGD1) and wild-type control (R6) strains ( Figure 3). Thus, placement of the URA3 gene at the CaESS1 locus or some other defect, but not CaESS1 dosage, was likely responsible for the filamentation defect.
One possibility was that the URA3 gene placed at the CaESS1 locus interrupted the promoter region of the downstream gene, APE2. The APE2 gene encodes an aminopeptidase that is thought to be secreted through the cell wall [61]. The design of our URA3based CaESS1 knockout and control constructs leaves only 15 bp of upstream sequence on one allele of APE2, leaving open the possibility that reduced expression of APE2 was responsible for the filamentation defect in both strains ( Figure 1A, 1E). As described in a later section, neither URA3 nor APE2 expression was involved.
To independently confirm that CaESS1 gene dosage did not affect filamentation we took an approach that did not rely on URA3 markers or disrupt the APE2 promoter. We used a parent strain (SN87) in which HIS1 and LEU2 are used as selectable markers [62]. Extensive studies have shown that ectopic expression of these markers does not significantly affect the virulence functions [62]. Accordingly, we replaced one allele of CaESS1 (with HIS1) generating a heterozygous strain (the other allele was marked with LEU2) ( Figure 1I). We also constructed a control strain in the same background, but leaving both CaESS1 alleles intact ( Figure 1J). To avoid any expression defects in CaESS1 or the downstream APE2 gene, the constructs were designed to have a short direct-repeat sequence (163bp) flanking the markers such that the ESS1 gene would have a total of 200 bp downstream of the coding sequence for transcription termination, and the APE2 gene would have 173bp of promoter sequence left intact ( Figure 1I, J; see also Materials and Methods). As before, qRT-PCR and Western analysis shows the expected reduction by about half of the expression of CaESS1 mRNA and protein in heterozygous mutants (CaDS-B5.5) vs. the isogenic wild-type control (CaDS-B5 ( Figure 3).
The Caess1d/CaESS1 heterozygous mutant strain and its CaESS1/CaESS1 isogenic control showed no significant differences in filamentation phenotypes on inducing media ( Figure 2I, J). In addition, no difference between these strains was observed in either germ-tube formation assays or in drug susceptibility growth assays using a number of commercially available antifungal drugs (data not shown). In summary, we find that the reduction of CaESS1 gene dosage by half does not affect major virulencerelated phenotypes of C. albicans in-vitro. We conclude that one copy of CaESS1 is sufficient for growth and morphogenetic switching consistent with the finding that Ess1 in S. cerevisiae is present in excess under standard growth conditions [47].
Neither URA3 nor APE2 levels are responsible for the filamentation defect in our CAI4 isolate If CaESS1 gene dosage was not responsible for the filamentation defects in our CAI4-derived strains, then what was responsible? To determine if ectopic URA3 at the ESS1 locus caused the filamentation defects in CaGD1 (Caess1d/CaESS1) [52] and R6 (CaESS1/CaESS1) strains (Figure 2A,E), we created two additional sets of strains. In one set, the URA3 gene in both the CaGD1 and R6 strains were removed and replaced into the native URA3 locus ( Figure 1B,C & F,G). In the other set, the URA3 gene was removed and placed at the RPS1 locus ( Figure 1B,D & F,H). The RPS1 locus is commonly used to express URA3 and other genes to avoid positional affects [56,57,62]. We then used qRT- Figure 1. Schematic representation of the strains used in this study. Strains A-H are made in the CAI4(*) background using URA3 as a selectable marker. Strains I-O are made in SN87 background with HIS1 and LEU2 selectable markers. The CaESS1 coding region and promoter region (pCaEss1) are shown in green; the APE2 gene is shown in purple. APE2 gene has two exons and the wavy line represents the intronic region. CaESS1 and APE2 are 210 bp apart in the genome (grey). The URA3 gene (blue) with the two flanking hisG direct repeats from S. typhimurium (yellow) is part PCR to compare the levels of expression of URA3 that resided at the CaESS1, native (URA3) or RPS1 loci. This was done for cells grown in selective (uracil-deficient) media at 30uC and in 10% serum medium (10% FBS in YPD) at 37uC ( Figure 4A,B). URA3 expression was essentially the same regardless of where the gene resided, thus ruling out position effects on URA3 expression as causing filamentation defects in our CAI4-derived strains. Moreover, placement of URA3 at different locations had no effect on expression of CaESS1 mRNA or protein ( Figure 3).
Note, however, that all of the above strains carry one copy of the APE2 gene with a truncated or hisG-disrupted promoter ( Figure 1A-F). To test whether reduced APE2 expression caused the filamentation defect, APE2 expression was measured in multiple strains (corresponding to Figure 1A-J) using qRT-PCR. Expression was measured in cells grown in selective media (30uC) and in serum-inducing medium (10% FBS in YPD, 37uC), and no significant differences in APE2 expression were detected ( Figure 4C,D). These results suggest that the remaining copy of APE2 with an intact promoter (strains in Figure 1A-F) compensated for the loss of upstream sequences in the second copy, or that the transcription is regulated from an internal promoter. In experiments done in the SN87 background using a strain with both APE2 promoter regions truncated (59ape2/59ape2) ( Figure 1K) neither the expression of APE2 (data not shown) nor filamentation ( Figure 2K) is affected. Results thus far indicate that APE2 expression is not correlated with the filamentation defect in CAI4-derived strains.
Finally, to characterize our parental CAI4 isolate, which we now concluded was suspect, we placed URA3 at the native locus or at the RPS1 locus and tested its ability to filament. Analogous CAI4-derived strains (e.g. CAI12, Figure 2R) made in other laboratories using similar approaches [56,63,64] were able to filament, but the strains derived from our CAI4 isolate were not ( Figure 2P,Q). Thus, not all CAI4 isolates behave similarly under inducing conditions, and we suspect that the isolate we used previously had acquired a background mutation(s) that resulted in a loss in the ability to undergo morphogenetic switching. These results explain the difference between our current results and those reported in Devasahayam et al., [52] with respect to CaESS1 gene dosage and filamentation phenotypes. They may also explain the reduction in organ load in mice injected with SC5314 vs. CAI4derivatives with mutations in CaESS1 [50].
of the construct used in the URA-Blaster method to target genes and later recycle URA3. RPS1 is shown in brown, LEU2 in red, HIS1 in pink and (p) indicates promoter sequence. The sequence upstream of APE2 that is repeated is shown in a gray-white gradient. The CAI4(*) indicates that this strain appears to have acquired a mutation that affects filamentation as per this study. Figure  CaESS1 is essential for growth in C. albicans Previously, a novel temperature-sensitive (ts) approach was used to demonstrate that CaESS1 is essential for C. albicans viability [52]. In this approach one allele of CaESS1 was deleted and the second allele was replaced with a form of CaESS1 engineered to be conditional. Specifically, a substitution in a critical histidine residue in the C. albicans Ess1 active site (H171R) was generated based on a ts-mutation (ess1 H164R ) well characterized in S. cerevisiae [42]. Prior to its use in C. albicans, the CaESS1 H171R allele was tested in complementation assays in S. cerevisiae and confirmed to be conditional; cells grew at 25uC and 30uC, but not at 37uC. When this allele was integrated into C. albicans CAI4, the resulting strain (Caess1d/Caess1 H171R strain) was ts-lethal: it grew at 30uC but not at 40uC, demonstrating that Ess1 is essential in C. albicans.
Here we sought to confirm, using this ts-strategy, that Ess1 is essential in another strain background, using a marker system other than URA3. Despite repeated attempts in the SN87 background, we were unable to generate a Caess1D/Caess1 H171R strain, consistent with CaEss1 being essential and suggesting that in this strain background the Caess1 H171R allele cannot support growth. However, this is a negative result, so we took another approach.
We noted that the analogous S. cerevisiae ess1 H164R protein has,0.01% the catalytic activity of wild-type protein [47] and decided to generate an additional allele, again based on prior work in S. cerevisiae [42]. The allele we generated alters the termination codon (TAA to TGC), resulting in translational readthrough. In S. cerevisiae, analogous mutations resulted in longer, fusion proteins that rendered the corresponding strains ts (X. Wu and S. Hanes, unpublished). The C. albicans strain we generated here, CaDS-C (ess1d/Caess1 TGC ; Figure 1L), which contains mutations in the stop codon (TAA to TGC) in CaEss1, encodes a protein that is larger than normal and is thermolabile at 42uC ( Figure 5A). The level of mutant protein is only slightly reduced after two hours at 37uC (data not shown).  We compared the growth at different temperatures of the termination-mutant strain (CaDS-C) to an isogenic, reconstituted control strain in which the TGC readthrough codon was changed back to a TAA termination codon (CaDS-FC; Caess1d/CaES-S1 TAA ). Both strains grew at lower temperatures (30uC and 37uC), but only the wild type (reconstituted) strain grew at 42uC ( Figure 5B). The results show that the Caess1 ts(TGC) allele is unable to support sustained growth at the restrictive temperature demonstrating that CaEss1 is essential for growth in the SN87 strain background. Thus, using a distinct conditional allele (Caess1 TGC ) in a different strain (SN87) with different markers (HIS1, LEU2), we confirm that CaEss1 is essential in C. albicans.

A large reduction in Ess1 levels reduces filamentation
In the course of characterizing the ts-lethal growth defect of the readthrough mutant strain (CaDS-C, Caess1d/Caess1 TGC ), we noticed that at 37uC there is a reduction in filamentation relative to the control strain (CaDS-FC; Caess1d/CaESS1 TAA ) under inducing conditions. This reduction is readily apparent for cells grown for 4 days on solid Spider medium ( Figure 5C, upper panels). Even after 7 days, colonies were unwrinkled compared to controls (data not shown). To confirm this result, we examined cells grown in liquid Spider medium for 2 hrs. Again, we find that the readthrough mutant (Caess1d/Caess1 TGC ) is reduced in germtube formation relative to reconstituted control strain (Caess1v/ CaESS1 TAA ) ( Figure 5 C.). This is an intriguing result because we do not observe a growth-rate defect in the readthrough mutant when grown in standard (non-inducing) medium, even at 37uC. It seems likely therefore that the levels of Ess1 protein in this mutant strain are sufficient for standard growth but not for specialized functions such as the induction of filamentation pathways. This is consistent with findings in S. cerevisiae showing that Ess1 is present in vast excess in cells growing in standard media (YPD, CSM), but that under various stress conditions, a significant reduction in Ess1 levels rendered cells unable to grow [47]. Structure-function analysis of the C. albicans Ess1 The C. albicans Ess1 protein contains a highly-structured linker region that locks the WW and PPIase domains together [65]. Unlike the short, flexible linker in the human Pin1 protein, this linker also contains a prominent, solvent-exposed a-helix [50]. To help understand these key structural differences, we generated two mutations in the CaEss1 protein. First, we replaced residues 37-67 in C. albicans Ess1 protein with the flexible linker found in human Pin1 (residues 40-54), effectively generating a linker-swapped mutant. Second, we substituted three residues along the surface of the a-helix that would drastically alter the charge pattern it displays (E44K, A51D, and K54E; Figure 6A). As an initial test of protein function we tested their ability to complement a S. cerevisiae ts mutant (ess1 H164R ) at non-permissive temperature ( Figure 6B) and to complement an ess1Dmutant using a plasmid-shuffle strategy ( Table 1). The C. albicans linker-swapped mutant was unable to complement in either assay, whereas the helix mutant fully complemented in both assays.
To confirm that the mutant C. albicans Ess1 proteins were expressed and stable in S. cerevisiae, we performed Western blot analysis (using antibodies to C. albicans Ess1). The linker-swapped mutant was present at very low levels ( Figure 6C), likely explaining its failure to complement in the functional assays. Due to its presumed instability, the linker-swapped mutant was not used for further studies. In contrast, the helix mutant was expressed at much higher levels, indicating the protein is stable consistent with its functionality ( Figure 6C). That the helix mutant complements S. cerevisiae ess1 mutants suggests it performs the essential catalytic function required for growth in this organism. This result is not entirely surprising given that the human and other vertebrate orthologs lacking this a-helix also rescue growth in S. cerevisiae [66]; Wilcox & Hanes, unpublished]. We interpret this result to mean either the a-helix is required solely for specialized functions, e.g. specific protein-protein interactions in C. albicans that might be lacking (or not revealed) in S. cerevisiae, or that it is not important for growth under standard conditions.
To further test the importance of this a-helix, we introduced the helix mutant (hm) alleles into C. albicans cells using the SN87 parent strain. Two strains CaDSE-5 (hm/hm) and CaDSE-5.10 (Caess1d/ hm) were constructed along with their respective isogenic controls. Both strains, CaDSE-5 and CaDSE-5.10 ( Figure 1N, O) were tested for filamentation, germ tube formation and anti-fungal drug susceptibility profiles. No significant differences were seen in any of these assays ( Figure 2N, O, and data not shown), suggesting the linker helix of CaEss1 is not important for morphogenetic switching or drug sensitivity in cultured cell assays. Thus, the prediction that the linker helix is important for fungal-specific functions has not yet been demonstrated. It is also possible that the triple mutation is not severe enough to inactivate its putative function(s), or that studies in animals models are needed to detect functional consequences of these mutations.

A conserved role for CaEss1 in RNA polymerase II transcription
Orthologs of CaEss1 in budding yeast (Ess1) and humans (Pin1) regulate RNA pol II transcription [42,45,46]. In yeast, Ess1 coordinates the recruitment of co-factors required for proper RNA synthesis and processing [43,49]. In this role, ScEss1 is required for efficient termination of at least two classes of RNA pol II products, small non-coding RNAs, e.g. small nucleolar RNAs (snoRNAs), and protein-coding mRNAs. ScEss1 is also important for repressing the transcription of cryptic unstable transcripts (CUTs) [49]. In S. cerevisiae ess1 mutants, CUTs are detected throughout the genome, many of which are the same as those detected in RNA-decay mutants [43]. To determine whether CaEss1 plays an analogous role in C. albicans, we carried out a whole-genome transcript analysis in wild-type and CaEss1 mutant strains using high-throughput RNA sequencing (RNA-seq). The strains and conditions used are summarized in Table 2. A large  Table 2), indicating that our RNA libraries were of high quality.
For each mutant and control pair analyzed, there was a significant change in the level of expression for between 2-4% of all transcribed genes ( Table 3). This is only slightly less than that observed in budding yeast (wild-type vs. ts-mutant) using standard microarrays where 3-10% of genes were misregulated depending on the temperature [43]. The results indicate that in C. albicans, as in S. cerevisiae the expression of some but not all genes is adversely affected by mutations in Ess1. Raw sequence reads will be deposited in the European Nucleotide Archive (http://www.ebi. ac.uk/ena/). Analyzed datasets will also be submitted to the NIH Gene Expression Omnibus (GEO) (http://www.ncbi.nlm.nih. gov/geo/). In this report, we will focus on the comparison between the transcriptomes of the conditional (ts) readthrough mutant (ess1D/ess1 TGC ) and its isogenic control (ess1D/ESS1 TAA ), relative to the wild-type strain (ESS1/ESS1).
To determine whether CaEss1 functions in snoRNA termination, we obtained RNA-seq data from ts-mutant and control strains and examined the abundance of sequence reads at non polycistronc, independently-transcribed snoRNA genes of the box H/ACA-class [67][68][69][70]. In total, we examined expression of 26 H/ ACA-class genes using Integrated Genome Viewer (IGV), and found that about one in five show evidence of readthrough as revealed by RNA-seq data. These genes included SNR3a, SNR8a, SNR32a, SNR43a, and SNR189) [67]. Primary transcripts of the other major class of snoRNAs (box C/D) are often encoded within other genes in C. albicans and were not examined here [68][69][70].
Each of the five H/ACA-class genes listed above contained an abundance of sequence reads (.30, but more often in the hundreds) immediately downstream of the putative termination site in the mutant background, suggesting possible transcription readthrough. The IGV profiles for three of these genes are shown (Figure 7, A-C). RNA sequence reads in the forward direction (red) and reverse direction (blue) are indicated as well as the overall number of reads (grey). For the control (ess1D/ESS1 TAA ) ( Figure 7A-C) and wild-type (ESS1/ESS1) (data not shown) strains, the large number of reads is likely to reveal the actual position of the snoRNA transcription units (dotted grey bars) more accurately than sequence annotations (solid bars) [68][69][70]. The IGV also reveals examples of increased transcript abundance in the mutant cells for some snoRNAs, e.g. SNR332a, SNR43a ( Figure 7B,C), and a protein-coding gene (ORF19.1968) and a non-coding transcript (CUT) ( Figure 7B).
To confirm the presence of snoRNA readthrough transcripts, we carried out qRT-PCR ( Figure 7D). cDNA synthesis was done using strand-specific primers, so only transcripts in the forward (''sense) direction relative to the snoRNA transcription units would serve as templates. Primer sets for amplification consisted of one primer within the snoRNA gene, and the other downstream of the putative termination site ( Figure 7A-C), therefore, only extended (readthrough) transcripts would be detected. The results show a 2-15 fold increase in the amount of readthrough transcription in the CaEss1 ts-mutant cells relative to isogenic control cells ( Figure 7D). In summary, both the RNA-seq and qRT-PCR data indicate that CaEss1 is important for efficient termination of at least some snoRNA genes in C. albicans.
To determine if CaEss1 represses cryptic transcription we used IGV to compare the transcription profiles of the ts-mutant and its isogenic control, focusing on the intergenic regions along all of the chromosomes. In the mutant cells, intergenic transcription was rampant, with large amounts detected on all eight chromosomes. Examples are shown from chromomes 2, 4 and 8 ( Figure 8A-C). Intergenic transcripts that lie between divergent genes are probably CUTs, since they cannot be due to readthrough transcription from neighboring genes (e.g. Figure 8A). Using both short and long qRT-PCR to characterize other potential CUTs revealed that only the short products (marked with an asterisk in Figure 8B,C`) were amplified ( Figure 8D, and data not shown), indicating the RNA-seq reads are not due to readthrough transcription. These data suggest that in CaEss1 mutants, cryptic promoters are activated and/or cryptic RNAs are stabilized. CUTs from each of the eight chromosomes were easily detected by  [40] and CaEss1 [50]. Substitutions made in the structured linker a-helix of CaEss1 to construct the helix mutant (hm) strains are indicated. (B) Complementation of CaEss1 linker mutants in S. cerevisiae. Plasmids encoding the indicated mutant proteins (independent clone isolates of linkerswapped plasmid, pDS426(sw) and helix substitution plasmid, pDS426(pm)) were separately transformed into a ts-mutant strain of S. cerevisiae (Scess1 H164R ) [42]. Plasmids were constructed using a pRS426 backbone [88]. Serial dilution assays (1:5) shows the growth of these independent transformants (a, b). Growth at 37uC, the restrictive temperature for the S. cerevisiae ess1 mutant, indicates complementing activity. The CaEss1 helix substitution mutant complements but the linker swap mutant does not. pGDCaEss1 [52] (pRS426 backbone) was used as a positive control and pRS426 [88] was used as an empty vector control. (C) Western analysis of whole cell extracts of S. cerevisiae expressing the indicated CaEss1 proteins. A total of 15 mg of protein was used per lane and the blot was probed using anti-CaEss1 polyclonal antibody at a 1:500 dilution. For the mutant strains, two independent clone isolates of pDS413(sw) (lanes 3 and 4) and pDS413(pm) (lanes 5 and 6) were transformed into S. cerevisae and analyzed. Plasmids were constructed using a pRS413 backbone [88]. The S. cerevisiae strain used (CBW22; [44] does not express endogenous Ess1 protein but is viable due to a suppressor mutation (ess1Dsrb10D). The linker-swapped protein appears to be absent, or present at a very low level compared to the vector control, while the helix mutant protein is easily detected. pGD-CaESS1 (pRS413 backbone) encoding the wild-type protein (Devasahayam and Hanes, unpublished) was used as a positive control, and pRS413 was used as the empty vector control. doi:10.1371/journal.pone.0059094.g006 Ess1 Prolyl Isomerase Is Essential in C. albicans PLOS ONE | www.plosone.org qRT-PCR (e.g. Figure 8D), indicating that CUT expression is widespread in CaEss1 mutant cells.
Many putative CUTs were also found within open reading frames, often in the antisense direction. One example is a potential antisense CUT within the open reading frame of FGR46 ( Figure 8E), a gene implicated in filamentation by a transposon insertion screen [71]. The reverse orientation of this CUT relative to the FGR46 ORF was confirmed by strand-specific qRT-PCR ( Figure 8F). Thus, it is possible that under conditions where Ess1 levels are strongly reduced, CUT induction might enhance or suppress expression of genes important for virulence. Additional analyses will be needed to fully explore the information in the RNA-sequencing datasets and determine the effect(s) of CaEss1 mutation on expression of virulence-related genes. Thus far, our findings are consistent with a conserved role for CaEss1 in transcription by RNA polymerase II.

Conclusions
In this study, we showed that the ESS1 gene is essential for growth in C. albicans. This result independently confirms earlier findings [52], but used more rigorous methods that included better controls and considered genomic sequence information that was not previously available. In contrast, the prior finding that reducing gene dosage by half (ess1D/ESS1) prevents morphogenetic switching, which is required for virulence in vivo [60], was not confirmed in the present study. The most likely explanation is that a background mutation(s) in the isolate of CAI4 used in the prior work caused the defect in filamentation. Thus, like other studies [72,73], our results make clear the importance of reconstituted isogenic controls, appropriate marker gene systems, and the dangers of isolate-to-isolate variability between supposedly identical strains. The results of transcriptome analysis of CaEss1 mutants support the idea that regulation of RNA polymerase II by Ess1 is conserved between the evolutionarily distant species S. cerevisiae and C. albicans. Further studies will be needed to determine whether transcriptional defects are responsible for growth and morphogenetic switching phenotypes in CaEss1 mutants, and whether the mechanism by which Ess1 functions in transcription in C. albicans is similar to that in S. cerevisiae [43,49]. With respect to C. albicans Ess1 as an antifungal drug target, data from this study suggest that elimination or strong inhibition of Ess1 enzyme activity will prevent growth. Given that Ess1 is essential not only in C. albicans, but also in A. nidulans [54], and is required for virulence in C. neoformans [53], it is possible that inhibitors of Ess1 could potentially be developed into broad-spectrum antifungal agents.

C. albicans strains, media, and transformation
All parent strains were grown in YPD media while the engineered derivative strains were grown in their respective selective media [74], unless otherwise specified. To select for URA3 loop-out events, cells were plated on YPD supplemented with 1 mg/ml 5-FOA [75]. In addition to the descriptions below, please refer to the tables for details on primers ( Table 4), plasmids ( Table 5) and strains ( Table 6) used in this study. Transforma- tions were performed according to the lithium acetate protocol [76].

C. abicans strain construction
Insertion of the URA3 gene at different genomic positions. For re-evaluating the C. albicans heterozygous Caess1 mutant (CaGD1) [52] along with its control strain, R6 (CaESS1/ CaESS1), the URA3 gene was removed from the CaESS1 locus using 5-FOA mediated loop-out events to create uracil auxotrophs CaGD2 (Caess1d/CaESS1) and R6* (CaESS1/CaESS1) respectively. One copy of URA3 was integrated at the native URA3 locus by digesting pLUBP plasmid [77] with BglII and PstI restriction enzymes and transforming into the above strains (CaGD2 and R6*). Similarly, one copy of URA3 was integrated at the RPS1 locus in these strains by digesting CIp10 plasmid [78] with StuI restriction enzyme and transforming. URA3 was also integrated into the parent strain CAI4* using the same pLUBP and CIp10 plasmids. (*) indicates a strain that may have acquired one or more mutations affecting filamentation as per this study.
Isogenic wild-type control strain, CaDS-B5 (ESS1/ ESS1). Plasmid pDS426(pm) is a derivative of pRS426 that carries the CaESS1 promoter region (2255 to 0), CaESS1 ORF (with helix point mutations) and CaESS1 termination region (+535 to +735) between EcoR1 and BamH1 sites. The LEU2 marker (from plasmid pSN40) or the HIS1 marker (from plasmid pSN52) [62] was cloned into plasmid pDS426(pm) between the BamH1 and Not1 sites. These plasmids were named pDS(b)Leu and pDS(b)His, respectively. Because in the C. albicans genome, APE2 is just 210 bp downstream of ESS1, a 163 bp sequence (+572 to +735) was repeated on either side of the selectable markers to allow enough sequence for CaESS1 termination and for the 59 APE2 promoter region. This was done by amplifying the downstream sequence of CaESS1 (+572 to +1265) with NotI ends (primers OW1075 and OW1081), digesting and inserting into pDS(b)Leu and pDS(b)His, thus positioning the NotI-NotI downstream sequence after the markers. The resulting plasmids are pDS(c)Leu and pDS(c)His. The pDS(c)Leu and pDS(c)His plasmids were digested with SacI and serially transformed into SN87 [62]. In the first step, one CaESS1 allele was marked with the LEU2 gene to generate strain CaDS-B0.5 (ESS1: LEU2/ESS1). In the next step, the second allele of CaESS1 was marked with HIS1 gene to generate a Leu+ His+ control strain CaDS-B5 (ESS1: LEU2/ESS1: HIS1). Allele-specific PCR (primer OW221 with OW769 or OW770) and DNA sequencing was used to identify clones in which the integrating DNA recombined at positions that generated the wild-type CaESS1 ORF, (i.e. that excluded the helix mutations). The gene modifications were verified using junction-PCR and Southern blot hybridization (data not shown). Temperature-sensitive strain, CaDS-C (ess1d/ess1 (TGC ) and isogenic control. The plasmid pDS(x)His was digested with SacI and transformed into SN87 to generate CaDS-87-2.3 (ess1d/ESS1), which is auxotrophic for LEU2. The Nde1 site of the vector portion in plasmid pDS(c)Leu was removed (as done for pDS(c)His), to form pDS(d)Leu. Next, the start codon (ATG) and the stop codon (TAA) of CaESS1 in plasmid pDS(d)Leu were changed to an NdeI site and a SphI site (primers OW1099 and OW1100, respectively) in two steps, to first form plasmid pDS(e)Leu and then plasmid pDS(f)Leu, respectively. The wildtype CaESS1 sequence was amplified (primers OW1354 and OW1355) with flanking Nde1 and Sph1 sites, digested and inserted into the same sites of pDS(f)Leu to give pDS(g)Leu. The Sph1 site changes the stop codon from TAA to TGC. The pDS(g)Leu plasmid was digested with SacI and transformed into CaDS87-2.3 (ess1d/ESS1) to make a prototrophic strain, CaDS-C (ess1d/ess1 TGC ). This strain was verified using junction-PCR and Southern blot hybridization (data not shown). The control strain was generated by a novel strategy. About a 900 bp PCR fragment amplified (using primers OW216 and OW1231) from a wild-type background containing the normal termination codon (TAA) was introduced into CaDS-C (ess1d/Caess1 TGC ). Transformants capable of growth at 42uC were selected, analyzed by PCR amplification and DNA sequencing of the CaESS1 alleles and The log 2 fold changes are based on results from the Cuffdiff program in the Galaxy server. Cuffdiff was used to select significant gene expression changes depending on whether the p value was greater than the allowed false discovery rate after Benjamini-Hochberg correction for multiple-testing. The highest and lowest log 2 fold changes of the genes considered to be significantly differentially expressed in each group are documented it in the the mutation of TGC to TAA was confirmed. The resulting prototrophic strain CaDS-FC (ess1d/ESS1 TAA ) produced Ess1 protein of the normal size (19 kDa) and was non-ts. Linker-helix mutations in CaEss1. PCR overlap extension (with primer sets OW748-OW1224, OW749-OW1233, and OW1224-OW1233) was used to generate a fragment that encodes CaEss1 protein bearing the linker region of human Pin1 (codon optimized for C. albicans [79]. This linker-swap fragment was digested with EcoRI and BamHI sites and cloned into the same sites of pRS426 vector to form pDS426(sw). A similar strategy was used to generate a CaEss1 mutant bearing three amino acid substitutions (E44K, A51D, and K54E) in the linker helix (using primers OW750 and OW751). The helix mutant fragment was cloned into pRS426 vector to form pDS426(pm).
APE2 promoter mutant, CaDS-51B-2 (59ape2/59ape2). A Not1-Not1 PCR fragment (primers D31 and OW1081) was cloned into plasmids pDS(b)Leu and pDS(b)His, to form plasmids  . The longer products indicated by long green arrows (B, C) did not amplify, indicating that the CUT signal is not likely due to a readthrough product from nearby open reading frames (ORF). cDNA synthesis was primed using a mixture of random hexamers and oligo(dT). Cell growth conditions were as listed in Table 2 (shift to 42uC, then serum induction at 42uC). (E) Prominent CUT identified within open reading frame of FGR46, which has been implicated in filamentation. (F) CUT is in reverse orientation relative to the FGR46 ORF as shown by strand-specific cDNA synthesis followed by qRT-PCR. Results are expressed as fold-change of ts-mutant (CaDS-C) over isogenic control (CaDS-FC), normalized to ACT1. Cells were serum-induced at 37uC. cDNA synthesis was primed using strand-specific primers that would reverse transcribe forward (UP044) or reverse (UP043) transcripts. For details, see Materials and Methods. doi:10.1371/journal.pone.0059094.g008

Growth assays
Overnight cultures of either S. cerevisiae or C. albicans were diluted to an OD 600 of 0.1 and grown with shaking at 30uC until midlog phase. Cultures were brought to an OD 600 of 0.5 either by dilution or concentration by centrifugation. For S. cerevisiae, a 1:5 dilution series of the strains were spotted onto plates containing selective media and incubated for two nights. For C. albicans strains, a 1:3 dilution series was used, with cells spotted on YPD plates and incubated overnight.

Filamentation assays
Overnight cultures of C. albicans were diluted to an OD 600 of 0.25 (,3610 6 cells/ml). From dilution, 2 ml of culture was spotted onto Spider media [80] and serum-containing media (4% FBS and 2% agar). The plates were incubated at 37uC for 4 days prior to being photographed. To test germ tube formation, a suspension of 10 5 -10 6 cells/ml was made in liquid Spider medium [80], and incubated at 37uC for 2 hrs prior to microscopic observation.

Western blot analysis
Overnight cultures were diluted to an OD 600 of 0.1, grown to an OD 600 of 0.5 and harvested by centrifugation. Protein extracts were prepared using yeast protein lysis buffer (50 mM HEPES, 140 mM NaCl, 1 mM EDTA, 1% Triton X and, 0.1% sodium deoxycholate) with proteinase inhibitor cocktail (ethanolic protease inhibitors with 0.25 M PMSF and 0.7 mg/ml pepstatin/aqueous protease inhibitors with 0.1 mg/ml leupeptin, 1 mg/ml soybean trypsin inhibitor, 0.1 mg/ml aprotinin in 10 mM Tris, pH7.5, 20 mM benzamidine, 10 mM sodium vanadate and, 500 mM sodium fluoride). A 15% polyacrylamide gel was run at 100 V for about 2.5 hrs and transferred to a PVDF Transfer Membrane (Millipore) overnight at 32 V at 4uC. A 1:500 dilution of rabbit a-CaEss1 polyclonal antibodies (Applied Biosystems) was used as primary antibody and a 1:25,000 dilution of horseradish peroxidase-linked donkey anti-rabbit IgG antibody (GE Healthcare) was used as the secondary antibody. Purified CaEss1 made in E. coli [50] was used to generate CaEss1 polyclonal antibodies at Applied Biosystems. Detection was done using the ECL-Plus Western Blotting Detection System (GE Healthcare) and exposure to X-ray film (Kodak).
High-throughput RNA sequencing RNA and protein preparation. For the ts-readthrough mutant and control strain CaDS-C (Caess1d/Caess1 TGC ) and CaDS-FC (Caess1d/CaESS1 TAA ) the following protocol was used. Cultures (10 ml) were grown overnight in selective media at 30uC, diluted 10-fold in YPD, and grown overnight again to an OD 600 of 9-11 [81]. Cultures were diluted 30 times in pre-warmed (42uC), pre-shaken fresh YPD (1 ml in 29 ml YPD) [81], and cells were grown at 42uC with shaking. After 1 hr, 2 ml of culture was collected for protein preparation as above and the remaining culture diluted 2-fold in pre-warmed (42uC), pre-shaken YPD+20% FBS (20 ml culture in 20 ml YPD+20% FBS) for serum induction. After 2 hr incubation at 42uC, cultures were collected by centrifugation for protein preparation (2 ml) and RNA extraction (10 ml). At extraction the cells were at an OD 600 of 0.4. RNA was extracted by the hot phenol method [82]. For all other strains the following protocol used. Overnight cultures in YPD were grown to an OD 600 of 9-11, and diluted 20-fold in prewarmed (37uC), pre-shaken fresh YPD+10% FBS (1 ml culture in 19 ml YPD+10% FBS) [81]. After a 2 hr incubation at 37uC, 10 ml of culture (OD 600 of 0.8) was collected for RNA extraction. RNA was quantified using a Nanodrop Spectrophotometer and aliquots of 10 mg of RNA subjected to DNase digestion (Epicentre) for 45 min at 37uC. Digested RNA was purified using an RNeasy Mini Kit (Qiagen). 400 ng of RNA was fractionated by agarose (1%) gel electrophoresis to confirm integrity (data not shown). For the ts-readthrough strain and its isogenic control, protein extracts before and after the temperature shift and serum induction were analyzed by Western blot analysis to confirm reduced amount of Ess1 protein at the restrictive temperature (42uC).  Library Preparation Kit (Epicenter). The ScriptSeq V2 Kit is an improved version of the ScriptSeq Kit and uses the same basic principal for library construction. Here the HiSeq2000 sequencer was run at 5862 cycles (per lane) resulting paired-end read lengths of about 75 bp. RNA sequence analysis. The raw FastQ data files were aligned to the C. albicans genome (Assembly 21) [83] obtained from Candida Genome Database (http://www.candidagenome.org/) using Bowtie/TopHat short sequence read alignment software programs [84,85]. Each strain resulted in a large number of reads mapped to the genome ( Table 2). The output files (in SAM format) were converted to BAM files using SAM Tools [86]. The BAM files were visually analyzed on Integrative Genomics Viewer (http://www.broadinstitute.org/igv/) and/or GenomeView (http://genomeview.org/). The files were quantitatively analyzed using the Galaxy server (https://main.g2.bx.psu.edu/), an open access platform for high-throughput data analysis. For example, the SAM or BAM files were used as Cuffdiff (a part of the opensource Cufflinks software package) [85] input files to identify differential expression of transcripts genome-wide among test strains and controls.

Quantitative reverse transcription real-time PCR
Overnight cultures were diluted to an OD 600 of 0.5 and grown to an OD 600 of 0.8-1.2 and harvested. RNA was extracted using the hot phenol method [82] and subjected to DNase digestion (Epicenter) for 45 min at 37uC. From 1 mg of RNA, cDNA was synthesized using the First Strand cDNA Synthesis Kit for Real-Time PCR(USB) using mixture of oligo dT and random hexamers. In some sets of experiments cDNA was synthesized using individual cDNA synthesis reagents purchased from USB/ Affimetrix. Quantitative real-time PCR was performed in the ABI2000 RT-PCR machine using the HotStart-IT SYBR Green qPCR Master Mix (USB). In some sets of experiments real-time PCR was carried out using the Fermentus SYBR Green Master Mix (Life Sciences). All results were normalized agains the same internal control gene, ACT1. qRT-PCR calculations were done as per Yu et al., [87] where the quantitative real-time PCR cycle numbers relative to ACT1 were summed for each biological replicate and normalized to the summed average of all samples within a given experiment. The re-normalized values were then used for average and standard deviation calculations. This method corrects for trial-to-trial variability. To calculate the fold difference between samples, the ACT1-normalized (DCt) values for the FC (WT/-) and C (ts/2) data sets were normalized against the WT/ WT (B5) data (to obtain DDCt), and the following calculation was performed: fold difference = 2 2DDCt . The values obtained were averaged, and a standard deviation calculated.