Dissection of Ire1 Functions Reveals Stress Response Mechanisms Uniquely Evolved in Candida glabrata

Proper protein folding in the endoplasmic reticulum (ER) is vital in all eukaryotes. When misfolded proteins accumulate in the ER lumen, the transmembrane kinase/endoribonuclease Ire1 initiates splicing of HAC1 mRNA to generate the bZIP transcription factor Hac1, which subsequently activates its target genes to increase the protein-folding capacity of the ER. This cellular machinery, called the unfolded protein response (UPR), is believed to be an evolutionarily conserved mechanism in eukaryotes. In this study, we comprehensively characterized mutant phenotypes of IRE1 and other related genes in the human fungal pathogen Candida glabrata. Unexpectedly, Ire1 was required for the ER stress response independently of Hac1 in this fungus. C. glabrata Ire1 did not cleave mRNAs encoding Hac1 and other bZIP transcription factors identified in the C. glabrata genome. Microarray analysis revealed that the transcriptional response to ER stress is not mediated by Ire1, but instead is dependent largely on calcineurin signaling and partially on the Slt2 MAPK pathway. The loss of Ire1 alone did not confer increased antifungal susceptibility in C. glabrata contrary to UPR-defective mutants in other fungi. Taken together, our results suggest that the canonical Ire1-Hac1 UPR is not conserved in C. glabrata. It is known in metazoans that active Ire1 nonspecifically cleaves and degrades a subset of ER-localized mRNAs to reduce the ER load. Intriguingly, this cellular response could occur in an Ire1 nuclease-dependent fashion in C. glabrata. We also uncovered the attenuated virulence of the C. glabrata Δire1 mutant in a mouse model of disseminated candidiasis. This study has unveiled the unique evolution of ER stress response mechanisms in C. glabrata.


Introduction
In eukaryotic cells, the majority of secretory and transmembrane proteins are folded and modified in the lumen of the endoplasmic reticulum (ER). Contingent on proper folding, they are either transported to the Golgi apparatus or degraded [1]. Impairment of these vital cellular machines can be caused by various factors, such as chemical compounds and mutations in genes involved in ER quality control, resulting in the accumulation of unfolded or misfolded proteins in the ER, collectively called ER stress [2,3]. Unsolved ER stress has detrimental consequences for eukaryotic cells. For example, in humans, ER stress is implicated in the pathology of various diseases including metabolic disease, inflammation, neurodegenerative disorders, and cancer [4,5]. It has also been revealed in several pathogenic fungi that ER quality control is important for antifungal resistance and virulence [6,7,8,9,10,11].
In response to ER stress, eukaryotic cells activate a signaling pathway, termed the unfolded protein response (UPR), to induce a comprehensive gene expression program that adjusts the proteinfolding capacity of the ER (reviewed in [2,12]). In Saccharomyces cerevisiae, the ER-resident transmembrane stress transducer Ire1, which contains an ER luminal sensing domain and cytosolic protein kinase/endoribonuclease domains, is responsible for detecting unfolded proteins and triggering the UPR [13,14]. Ire1 is activated by direct interaction with the unfolded proteins, resulting in the oligomerization of its luminal domain. Signaling is transduced to the cytosolic domain through the sequential steps of autophosphorylation and oligomerization of the endoribonuclease domain, leading to the activation of the Ire1 ribonucrease [15,16]. Active Ire1 cleaves the HAC1 mRNA to excise the intron, allowing translation of the basic-leucine zipper (bZIP) transcription factor Hac1 that subsequently induces transcription of the UPR target genes [17,18]. ER-stressed cells attempt to reduce the load of abnormally folded proteins in the ER by facilitating protein folding (e.g. upregulation of genes encoding ER-resident chaperones and protein-modifying enzymes) and by translocating misfolded proteins from the ER to the cytosol where they are degraded by the proteasome. The latter mechanism is called ERassociated protein degradation (ERAD) (reviewed in [3]). An alternative mechanism of degradative response is autophagy, which degrades organelles including damaged ER. In addition to ER-resident chaperones and protein-modifying enzymes, many of the components that mediate ERAD and autophagy have also been identified as UPR targets [19,20,21].
In pathogenic fungi, the molecular basis of ER quality control has been poorly understood, but several recent studies in Aspergillus fumigatus and Cryptococcus neoformans have found that Ire1 and Hac1 homologs are key components of the UPR and are indeed required for the ER stress response [7,8,10]. Interestingly, these studies have also discovered that the UPR is implicated in fungal pathogenicity and antifungal resistance. In A. fumigatus, the loss of HacA/Hac1 results in increased susceptibility to caspofungin, amphotericin B, and azole antifungals, and decreased virulence in mouse models of invasive aspergillosis [10]. Recently, the necessity of IreA/Ire1 for A. fumigatus virulence has also been reported [8]. C. neoformans mutant strains lacking Ire1 or its downstream transcription factor Hxl1 display increased azole susceptibility, failure to grow at 37uC, and avirulence in a mouse model of systemic cryptococcosis [7]. It is also known in Candida albicans that Hac1 plays a role in hyphal development [11], and a Dire1 mutant is hypersusceptible to caspofungin [6]. These observations shed light on the UPR as an attractive target for the development of novel antifungal therapies.
Candida glabrata has emerged as an important fungal pathogen due in part to its intrinsic or rapidly acquired resistance to azole antifungals such as fluconazole [22,23]. In addition, recent surveillance data have revealed an increase of C. glabrata clinical isolates that display resistance to not only azoles, but also echinocandin-class antifungals [24,25]. Considering the limitations of the currently available antifungals in clinical settings, there is an urgent need to develop an effective antifungal strategy for a broad range of fungal pathogens, including C. glabrata.
Since how C. glabrata cells deal with ER stress has not been explored, we first functionally characterized the C. glabrata Ire1 and Hac1 orthologs. It has been believed that the UPR mediated by the Ire1-Hac1 linear pathway is evolutionarily conserved in most eukaryotic species, but surprisingly, we found that C. glabrata Ire1 plays a role in the ER stress response in a Hac1-independent manner, despite the presence of an apparent HAC1 ortholog. The present study revealed that C. glabrata has lost the canonical Ire1-Hac1 UPR, but has developed alternative mechanisms for ER quality control. In addition, our comprehensive analyses of Dire1 mutant phenotypes revealed significant diversities of Ire1-mediated stress response mechanisms between C. glabrata and other fungi.
Here, we describe the unique evolution of ER quality control systems in C. glabrata.

Results
Comparison of growth ability in the presence of ER stress between S. cerevisiae, C. neoformans, and Candida spp The ability of fungal cells to cope with ER stress was assessed by monitoring cell growth in the presence and absence of two wellknown ER stress inducers that interfere with protein folding in the ER by different mechanisms: tunicamycin (TM), an N-linked glycosylation inhibitor, and dithiothreitol (DTT), an inhibitor of disulfide bond formation. Compared to S. cerevisiae and C. neoformans wild-type strains, C. glabrata was relatively tolerant to both TM and DTT independently of its strain backgrounds and culture media, although S. cerevisiae displayed strain dependent susceptibilities ( Figure S1). Other representative Candida species also exhibited higher tolerance to these agents than S. cerevisiae and C. neoformans with intriguing exceptions: Candida krusei was highly susceptible to TM, but not to DTT, while Candida tropicalis was hypersusceptible to DTT, but not to TM. These results imply that diverse aspects of ER stress response mechanisms may exist even in closely related yeast species. In this report, the following studies focused on C. glabrata, which is phylogenetically close to S. cerevisiae, but has gained increased tolerance to ER stress and pathogenic potential to humans.
Identification and structural characterization of the IRE1 and HAC1 orthologs in C. glabrata The C. glabrata IRE1 and HAC1 orthologs were identified by BLASTp searches using the NCBI (http://www.ncbi.nlm.nih. gov/BLAST/) and Genolevures (http://cbi.labri.fr/Genolevures/ elt/CAGL) databases. The respective amino acid sequences of S. cerevisiae IRE1 (YHR079c) and HAC1 (YFL031w) were used as queries. The deduced amino acid sequence of C. glabrata IRE1 (NCBI accession No.: XP_446111, Genolevures ID: CAGL0F03245g) displayed 53.6% similarity and 35.8% identity with that of S. cerevisiae IRE1. As presented schematically in Figure 1A, C. glabrata Ire1 consists of typical Ire1-domain structures [16,26], including an N-terminal hydrophobic signal sequence, an ER luminal domain, a transmembrane domain, and a serine-threonine protein kinase followed by a nuclease domain. The signal sequence and the ER luminal domain are important for ER localization and unfolded protein sensing, respectively [14,27,28,29]. The most conserved segment is the cytoplasmic C-terminal region containing the protein kinase linked to the nuclease domain ( Figure 1B). The protein kinase domain is required for autophosphorylation and concomitant activation of the nuclease domain, which provides the endoribonuclease activity [30,31,32]. C. glabrata IRE1 is syntenic with S. cerevisiae IRE1 situated on chromosome VIII, but the 39 region of C. glabrata IRE1 is orthologous to S. cerevisiae chromosome IV ( Figure S2). This suggests that a genomic arrangement between S. cerevisiae chromosomes IV and VIII has occurred to evolve the IRE1surrounding region on chromosome F in C. glabrata.
A BLASTp search using the NCBI and Genolevures databases also identified a single HAC1 ortholog (NCBI accession No.: XP_448761, Genolevures ID: CAGL0K12540g) in the C. glabrata genome, exhibiting 36.9% similarity and 24.2% identity with the deduced amino acid sequence of the unspliced form of S. cerevisiae HAC1. It has been known that functional orthologs of transcription factors often display limited sequence similarity outside their DNA-binding domains [11]. Indeed, C. glabrata HAC1 is not highly conserved at the overall sequence level, but contains a bZIP

Author Summary
The majority of secretory and transmembrane proteins are structurally matured in the endoplasmic reticulum (ER). The accumulation of misfolded proteins in the ER (ER stress) activates the ER-resident stress transducer Ire1, which has two distinct outputs: the unfolded protein response (UPR) and regulated Ire1-dependent decay (RIDD). The UPR is a transcriptional response to increase the protein folding capacity of the ER. RIDD induces degradation of ER-localized mRNAs to reduce the ER load. To date, the UPR has been believed to be an evolutionarily conserved pathway in almost all eukaryotic species, while RIDD has been found only in metazoans. Recent studies in several pathogenic fungi revealed that the UPR is implicated in antifungal resistance and virulence, and thus it has attracted attention as a therapeutic target. Here, we demonstrate that the important fungal pathogen Candida glabrata has lost the canonical UPR, but instead possesses the RIDD pathway and is relatively tolerant to ER stress. The transcriptional response to ER stress was dependent mainly on calcium signaling mediated by the protein phosphatase calcineurin in C. glabrata. Our results provide novel insights into ER quality control mechanisms and are useful for understanding evolutionary biology and the development of antifungal agents targeting the UPR. The asterisk indicates the conserved two catalytic residues in the nucleotide-binding pocket of Ire1 kinase (D797 and K799 in S. cerevisiae and D723 and K725 in C. glabrata). The predicted nuclease domain is underlined. Ten residues (boxed with dotted line) including the highly conserved threenuclease active sites (arrowheads) are deleted in C. glabrata IRE1-ND.  (Figure 2A). In addition, there is a clear syntenic relationship between S. cerevisiae and C. glabrata HAC1 loci ( Figure S2). Based on these sequence similarities and synteny conservation, we considered the genes CAGL0F03245g and CAGL0K12540g to be C. glabrata IRE1 and HAC1, respectively.
It has been known in S. cerevisiae [33,34,35] that the uninduced form of HAC1 (HAC1 u ) contains the 252 nucleotides intron, which is excised by Ire1 to generate the induced form of HAC1 (HAC1 i ) ( Figure 2B). The well-conserved Ire1 recognition motifs, CNG9CNGN and CNG9AAGC at the 59 and 39 boundaries of the intron, respectively [36], locate in the loop regions of the HAC1 mRNA stem-loop structures [34,37]. Based on the conservation of common features around the predicted intron ( Figure 2C), C. glabrata HAC1 mRNA is also likely to form stem-loop structures [36]. If excision of the predicted intron in C. glabrata HAC1 mRNA occurs, it changes the last 15 amino acids of the putative ORF and extends it by only 9 amino acids ( Figures 2B and 2C). However, C. glabrata HAC1 has mutations in the consensus sequences of the Ire1 recognition motifs at both splice sites: a C to A mutation at the 4th position of the 59 splice site and an A to U mutation at the 5th position of the 39 splice site ( Figure 2C).
Ire1, but not Hac1, is required for ER stress response in C. glabrata To examine involvement of Ire1 and Hac1 in the ER stress response in C. glabrata, we constructed a mutant of each in which the entire open reading frame (ORF) of IRE1 or HAC1 was deleted, and examined growth on plates containing TM and DTT ( Figure 3). Both S. cerevisiae Dire1 and Dhac1 mutants were used as controls and they exhibited severe growth defects in the presence of TM or DTT. In C. glabrata, the Dire1 mutant displayed (B) A schematic representation of the putative splice sites and ORF lengths in C. glabrata and S. cerevisiae HAC1. The HAC1 ORFs and bZIP domains were shown in grey and blue boxes, respectively. In S. cerevisiae, the intron (purple box) in the uninduced form of HAC1 (HAC1 u ) is excised in the induced form of HAC1 (HAC1 i ). Red arrowheads and asterisks indicate putative splice sites and stop codons, respectively. Blue arrows indicate primers that were used for RT-PCR assays to examine HAC1 splicing. (C) Alignment of the RNA sequence surrounding the predicted intron in C. glabrata and S. cerevisiae HAC1. In S. cerevisiae, HAC1 mRNA is known to form stem-loop structures that enclose the intron between the two loops of seven residues (shown in red) held in place by short stems (yellow and green boxes). The predicted 59 and 39 splice sites and intron lengths in the S. cerevisiae and C. glabrata HAC1 mRNAs are indicated. doi:10.1371/journal.ppat.1003160.g002 decreased tolerance to both TM and DTT compared with the wild-type strain, whereas surprisingly, the Dhac1 mutant exhibited wild-type growth in the presence of these ER stress-inducing agents ( Figure 3). Southern blot analysis demonstrated that HAC1 was truly absent and there was no ectopic integration of the deletion construct in the Dhac1 mutant ( Figure S3). The results indicate that Ire1, but not Hac1, is required for the ER stress response in C. glabrata. This is quite a contrast to S. cerevisiae, which is dependent equally on Ire1 and Hac1 to survive ER stress [38], implying that C. glabrata may possess a unique mechanism, which is mediated by Ire1 independently of Hac1, to cope with ER stress.
Ire1 does not cleave HAC1 mRNA in C. glabrata Ire1-mediated splicing of HAC1/XBP1 mRNA was first uncovered in S. cerevisiae [32], and similar mechanisms in HAC1/XBP1 homologs were later described in some fungi, including Trichoderma reesei, Aspergillus nidulans [39], C. albicans [11], Pichia pastoris [40], Yarrowia lipolytica [41], A. fumigatus [8,10], and C. neoformans [7], as well as in metazoans including Caenorhabditis elegans, mice [37,42], humans [43], and Drosophila melanogaster [44]. We examined HAC1 mRNA splicing in ER-stressed S. cerevisiae and C. glabrata cells by RT-PCR using the HAC1 specific primer pairs denoted in Figure 2B. As expected, S. cerevisiae HAC1 mRNA was spliced in response to treatment with 1.5 mg/ml TM in wild-type control but not in the Dire1 strain ( Figure 4A, left panel). As C. glabrata wild-type cells were relatively tolerant to both TM and DTT ( Figure S1), they were treated with higher concentrations of TM and DTT for 1 and 3 h. However, contrary to S. cerevisiae, splicing of HAC1 mRNA was not induced in C. glabrata under any of the conditions tested in this study ( Figure 4A, right panel). The absence of a HAC1 splicing event was also verified by sequencing the RT-PCR products of the entire HAC1 mRNA (data not shown), suggesting that Hac1 is not a downstream target of Ire1 in C. glabrata.
To further confirm this aspect, we also performed Northern blot analysis using the probes directed against the 59 region of the HAC1 ORF in S. cerevisiae and C. glabrata. In S. cerevisiae, HAC1 mRNA was efficiently spliced upon ER stress induced by 1.5 mg/ ml TM or 5 mM DTT ( Figure 4B, left panel). In addition to the unspliced and spliced HAC1 mRNAs, detection of two smaller bands (indicated with asterisks in Figure 4B) were consistent with previous reports [17,45,46], although it is unclear whether these bands correspond to splicing intermediates or dead-end products [17,46]. The C. glabrata wild-type and Dire1 cells were incubated under the normal conditions and in the presence of 10 mg/ml TM or 10 mM DTT for 3 h. Under all conditions tested, the C. glabrata wild-type strain somehow exhibited 2 bands but these bands were also detected similarly in the Dire1 mutant ( Figure 4B, right panel), indicating that it was not due to Ire1-mediated splicing. These results and the phenotypic differences between the Dire1 and Dhac1 mutants strongly suggest that C. glabrata Ire1 plays a role in the ER stress response in a Hac1-independent manner.

Exploration of other bZIP transcription factors as potential downstream targets of Ire1 in C. glabrata
To address the possibility that a HAC1-like gene encoding a bZIP transcription factor may exist as a downstream target of Ire1 in C. glabrata, we performed a low-stringent BLASTp search (e-value = 10 22 ) using only the bZIP domain sequences of S. cerevisiae Hac1, C. neoformans Hxl1, and Homo sapiens Xbp1 as described previously [7]. In addition to HAC1, the following eight putative genes were identified in the C. glabrata genome (Table S1): CAGL0H04631g (YAP1), CAGL0J06182g (SKO1), CAGL0K02585g (YAP3), CAGL0M10087g (YAP3), CAGL0L02475g (GCN4), CAGL0F03069g (CAD1/YAP2), CAGL0F01265g (YAP7), and CAGL0M08800g (YAP6). We sequenced the RT-PCR products of these mRNAs, which were obtained from C. glabrata wild-type cells treated with TM at concentrations of 1.5 and 5 mg/ml for 3 h, but none of them displayed evidence of a splicing event (data not shown). The results suggest that another Hac1 homolog is unlikely to exist in C. glabrata. Complementation assays of the IRE1 and HAC1 orthologs between S. cerevisiae and C. glabrata To investigate whether C. glabrata Ire1 could complement S. cerevisiae Ire1 function, C. glabrata IRE1 was expressed under the S. cerevisiae ADH1 promoter in the S. cerevisiae Dire1 mutant. After treatment with 1.5 mg/ml TM for 3 h, a spliced form of S. cerevisiae HAC1 was generated in the wild-type strain, but not in the Dire1 strain regardless of the presence of C. glabrata IRE1 ( Figure 4A). The absence of a HAC1 splicing event in the S. cerevisiae Dire1 mutant containing C. glabrata IRE1 was also verified by sequencing analyses of the RT-PCR products (data not shown). As expected, C. glabrata Ire1 did not rescue the growth of the S. cerevisiae Dire1 mutant in the presence of TM ( Figure 5A). These results indicate that C. glabrata Ire1 is unable to initiate the UPR due to its inability to cleave the HAC1 mRNA in S. cerevisiae, accounting for the growth defects of the S. cerevisiae Dire1 mutant containing C. glabrata IRE1 in the presence of TM.
Although HAC1 was dispensable for C. glabrata cells to cope with ER stress (Figure 3), the fact that the bZIP domain is conserved in this gene prompted us to investigate whether C. glabrata Hac1 behaves as a transcription factor similarly to S. cerevisiae Hac1. To address this question, C. glabrata HAC1 was constitutively expressed under the ADH1 promoter in the S. cerevisiae Dhac1 mutants, and the transcription of representative UPR target genes, including KAR2 (ER-resident chaperone), PDI1 (protein disulfide isomerase), DER1 (ER-associated degradation), and FPR2 (ER protein trafficking), was examined. Compared to the wild-type strain, the expression levels of these genes were decreased in the Dhac1 mutant, but reversed by the heterologous expression of C. glabrata HAC1 in the mutant ( Figure 5B). Although a direct interaction between C. glabrata Hac1 and UPR targets in S. cerevisiae was not examined, the overexpression of C. glabrata HAC1 conferred the increased expression levels of these genes (e.g. 3.5-fold induction of KAR2 expression relative to the wild-type control). Since sustained activation of the UPR is toxic to cells [47], the constitutive overexpression of C. glabrata HAC1 and the induced form of S. cerevisiae HAC1 (ScHAC1 i ) similarly impaired cell growth even under normal growth conditions ( Figures 5C and 5D). Furthermore, growth of the S. cerevisiae Dhac1 mutant in the presence of TM was rescued by C. glabrata Hac1 to the wild-type level ( Figure 5C). The sequences of C. glabrata HAC1 obtained by RT-PCR from TMtreated and -untreated cells were identical (data not shown). These results indicate that C. glabrata Hac1 was matured without splicing in S. cerevisiae. In agreement with this, C. glabrata Hac1 also rescued growth of the S. cerevisiae Dire1 mutant in the presence of TM ( Figure 5C). These results suggest that C. glabrata Hac1 retains a function as a transcription factor to activate the UPR targets; however, unlike S. cerevisiae HAC1, the C. glabrata HAC1 mRNA does not need to be spliced by Ire1 before translation.
We also examined the effects of heterologous expression of S. cerevisiae HAC1 in C. glabrata. Since C. glabrata Ire1 was unable to splice S. cerevisiae HAC1 ( Figure 4A), ScHAC1 i was expressed under the control of the PGK1 promoter in C. glabrata. Although C. glabrata Hac1 complemented S. cerevisiae Hac1 functions (Figures 5B and 5C), ScHAC1 i did not rescue growth of the C. glabrata Dire1 mutant in the presence of TM ( Figure 5E). ScHAC1 i partially rescued growth of the S. cerevisiae Dhac1 and Dire1 mutants in the presence of TM, confirming its functionality ( Figure 5D). Since C. glabrata Hac1 induced transcription of the UPR targets in S. cerevisiae ( Figure 5B), the Hac1 orthologs in C. glabrata and S. cerevisiae may share a common binding motif in the promoter regions of the target genes. These results support the idea that the impaired ER stress response of the C. glabrata Dire1 mutant is not associated with Hac1 functions.
The Ire1, calcineurin, and Slt2 signaling pathways are coordinately required for ER stress response in C. glabrata We next investigated the possibility that the cellular response to ER stress might involve potential crosstalk between Ire1 and other signaling pathways in C. glabrata. In addition to the UPR, the serine-threonine-specific protein phosphatase calcineurin and the Slt2 mitogen-activated protein kinase (MAPK) pathway are also required for the ER stress response in S. cerevisiae [48,49,50,51]. We examined whether this is the case in C. glabrata using four C. glabrata mutant strains lacking a key component of these signaling pathways, including Ire1, the regulatory B subunit of calcineurin Cnb1, the calcineurin-regulated transcription factor Crz1, and the last member of the PKC1-MAPK cascade Slt2. While all mutants displayed wild-type growth on a drug-free control plate, the Dire1, Dcnb1, and Dslt2 mutants exhibited decreased tolerance to both TM and DTT ( Figure 6A). Growth of the Dcrz1 mutant was at nearly wild-type levels in the presence of TM, but drastically impaired on the plate containing DTT. These results were in agreement with a previous proposal that C. glabrata calcineurin is involved in various stress responses via Crz1-dependent andindependent mechanisms depending on the type of stress [52]. All mutant phenotypes were recovered to wild-type levels by reintroducing the corresponding wild-type genes into the mutants ( Figure 6A). Taken together, the results indicate that Ire1, calcineurin, and Slt2 are required for the ER stress response in C. glabrata, consistent with previous findings in S. cerevisiae.
Next, we deleted IRE1 in the Dcnb1, Dcrz1, and Dslt2 backgrounds and monitored the viabilities of the single and double deletants after exposure to 1.5 mg/ml TM in liquid medium for up to 5 h. The double deletants displayed more rapid and greater reduction of viability than each single deletant in the presence of TM ( Figure 6B). The results suggest that Ire1, calcineurin, and Slt2 function in parallel to cope with ER stress in C. glabrata.

Gene expression profiling of C. glabrata mutants in the presence of ER stress
To examine the transcriptional response to ER stress in C. glabrata, the wild-type, Dire1, Dcnb1, Dcrz1, and Dslt2 mutants were exposed to 1.5 mg/ml TM for 0.5 and 3 h, and genome-wide analyses were conducted using DNA microarrays. The complete dataset can be found at the NCBI Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) with accession number GSE29855. It has been reported that at least 381 genes, corresponding to more than 5% of the ORFs in the genome, were induced in response to TM and DTT under the control of the UPR in S. cerevisiae [20]. In our assay, a total of 325 genes were differentially regulated (greater than 2.0-fold change) and 75 genes of them were upregulated in the C. glabrata wild-type strain after treatment with TM for 3 h (Figure 7 and Table S2). In contrast to previous findings in S. cerevisiae [20], the induced dataset in C. glabrata was not enriched for genes involved in folding capacity of the ER (Table S2). Therefore, to rule out the possibility that unfolded protein stress might not be induced sufficiently by the conditions used in our study, C. glabrata wild-type cells were treated with higher concentrations of TM and DTT, and then expression levels of five C. glabrata genes (KAR2, PDI1, DER1, FPR2 and HAC1), which are well-known UPR targets in S. cerevisiae [20,53], were examined by qRT-PCR. The induction levels of KAR2 in the presence of high concentrations of TM and DTT were similar to those observed in the presence of 1.5 mg/ml TM ( Figure S4A). In addition, expression levels of the other 4 genes were not increased even in cells treated with high concentrations of TM and DTT ( Figure S4B). Collectively, these results suggest that ER quality may be controlled differently in C. glabrata than S. cerevisiae, which relies primarily on transcriptional induction of genes that increase the protein folding capacity of the ER. Furthermore, in contrast to S. cerevisiae, the vast majority of genes induced in the C. glabrata wild-type strain were also similarly upregulated in the Dire1 mutant, but at less than two-fold change, or were downregulated in the Dcnb1 mutant ( Figure 7 and Table  S2). The loss of Crz1 and Slt2 had partial effects on the upregulation of those genes. Many of the genes induced by TM overlapped with calcineurin-regulated genes reported in a recent study of calcineurin signaling in C. glabrata [54], indicating the activation of calcineurin signaling in response to TM. A subset of genes that displayed calcineurin-dependent upregulation in the presence of 1.5 mg/ml TM in our microarray experiments were further evaluated by qRT-PCR after treatment with a higher concentration of TM (10 mg/ml) for 3 h (Figure 8). All the genes examined in this assay were upregulated upon TM treatment in a calcineurin-and Crz1-dependent manner but independently of Ire1 and Hac1. In addition to calcineurin, Slt2 was also required for the induction of CAGL0I07249g, consistent with the microarray data. Although these assays cannot differentiate a direct from an indirect effect, the results suggest that calcineurin, but not Ire1, plays important roles in transcriptional response to ER stress in C. glabrata.
Transcriptional induction of the ER-resident chaperone KAR2 is a well-known marker for UPR activation in yeast [55,56]. In response to TM treatment, expression levels of C. glabrata KAR2 were increased in the wild-type, Dslt2, and Dire1 strains, but not in the Dcnb1 and Dcrz1 mutants ( Figure 8). These results are in contrast to previous observations in S. cerevisiae, where Ire1-Hac1 signaling is required for upregulation of KAR2 in response to ER stress [13,17]. A Hac1-binding site, termed the unfolded protein response element (UPRE), was originally defined as a 22-bp sequence element of the S. cerevisiae KAR2 promoter and subsequently refined to a seven-nucleotide consensus, 59-CAGNGTG-39 [57,58]. Mutations in any of the six conserved nucleotides, or deletion of the central nucleotide, cause it to lose its function as an autonomous upstream activating sequence [57]. Consistent with the KAR2 expression data, the consensus UPRE sequence was absent in the 1-kb upstream region of C. glabrata KAR2. On the other hand, it is known in S. cerevisiae and C. glabrata that the promoters of most calcineurin-dependent genes contain one to six copies of the Crz1-binding sequence, 59-GNGGC(G/ T)-39 [59,60]. Five copies of the Crz1-binding sequence, including one copy of the full consensus sequence, 59-GNGGCTCA-39 [60], were found within the 1-kb upstream region of C. glabrata KAR2. These results support our KAR2 expression data, collectively suggesting that transcriptional induction of KAR2 in response to ER stress is mediated by the calcineurin-Crz1 pathway, but not by Ire1 signaling, in C. glabrata.
In our microarray analyses, there was a subset of ,33 genes whose mRNA abundance was decreased in the wild-type strain, but not in the Dire1 mutant at a relatively early phase (0.5 h) of ER stress (Figures S5 and Table S4). This cluster predominantly contained genes encoding membrane proteins involved in transferase activity (Table S5). Most of these gene products are known to be ER-resident or pass through the ER before translocating to their final destinations. Further analyses were made to interpret this phenomenon, which will be discussed below.
Functional dissection of the protein kinase and nuclease domains in C. glabrata Ire1 To further investigate how Ire1 is involved in the ER stress response in C. glabrata, we examined the contribution of the protein kinase and nuclease functions of C. glabrata Ire1 to cell growth in the presence of TM and DTT. In S. cerevisiae, mutations of two catalytic residues (D797 and K799) in the nucleotide-binding pocket of Ire1 kinase to asparagines abolish phosphorylation, but preserve RNase activity [61]. Therefore, these mutations (D797N, K799N) allow uncoupling of kinase and nuclease activities of Ire1. Since these two residues are highly conserved in the Ire1 orthologs of fungi, including C. glabrata ( Figure 1B, asterisks), we created C. glabrata kinase-dead IRE1 (IRE1-KD) harboring the corresponding mutations (D723N, K725N) in the Ire1 kinase domain. The growth defects of the Dire1 null mutant in the presence of TM and DTT were complemented by wild-type IRE1, but not by IRE1-KD (Figure 9), indicating that the protein kinase function of Ire1 is required for the ER stress response in C. glabrata.
Unlike in other eukaryotes, C. glabrata Ire1 did not cleave HAC1 mRNA in response to ER stress; we therefore hypothesized that the nuclease function of Ire1 would be dispensable for the ER stress response in C. glabrata. To test this hypothesis, we engineered C. glabrata nuclease-dead IRE1 (IRE1-ND) containing a 10 residue (D973-Y982) internal deletion within the nuclease domain ( Figure 1B). This deleted region contains three highly conserved residues (indicated by arrowheads in Figure 1B) that are the nuclease active sites of S. cerevisiae Ire1 [16]. A recent study has shown that deletion of the corresponding region (D1076-R1085) in A. fumigatus IreA abrogates its nuclease activity without an apparent effect on any nuclease-independent functions [8]. Contrary to our expectations, the Dire1 mutant carrying IRE1-ND displayed growth defects in the presence of TM and DTT, similar to the Dire1 null mutant in C. glabrata (Figure 9). These results suggest that both protein kinase and nuclease functions of Ire1 are required for the ER stress response in C. glabrata.
The nuclease activity of Ire1 is required for degradation of ER-associated mRNAs It has been found in higher eukaryotes that Ire1 induces degradation of mRNAs encoding proteins that are translocated into the ER lumen [62,63,64]. This cellular response is called regulated Ire1-dependent decay (RIDD) and represents an Xbp1independent posttranscriptional mechanism to selectively relieve the burden of incoming proteins on the ER [63]. In our microarray analysis, a subset of genes displayed Ire1-dependent ''downregulation'' in response to TM ( Figure S5 and Table S4). There was a strong enrichment for genes encoding GPI-anchored cell wall and membrane proteins in this cluster (Table S5). Thus, we further analyzed the expression levels of 3 representative genes, GAS2 encoding a GPI-anchored cell wall protein that functions as b-1,3-glucanosyltransferase, GAS4 encoding a member of the GAS family similar to GAS2, and ECM33 encoding a GPI-anchored membrane protein involved in apical bud growth, by qRT-PCR using C. glabrata cells treated with DTT and TM. In response to the treatment with 10 mM DTT for 2 h, RNA abundance of these genes was significantly decreased in the wild-type strain compared to the results with the Dire1 mutant ( Figure 10A). Similar results were obtained for cells treated with 1.5 mg/ml TM in the microarray analysis (Table S4) and by a confirmatory qRT-PCR assay (data not shown). This Dire1 phenotype was reversed by intact IRE1, but not by IRE1-ND ( Figure 10A), indicating that the degradation of these mRNAs was dependent on the nuclease activity of Ire1. C. glabrata IRE1-KD harboring the simultaneous mutations D723N and K725N had moderate effects on the mRNAs decay, implying that Ire1's nuclease activity may be partially impaired by these mutations in C. glabrata.
Blocking transcription with 1,10-phenanthroline (PHEN) inhibited the transcriptional upregulation of CAGL0K11946g and CAGL0L06776g by DTT but had no or little effect on the mRNA dacay of GAS2, GAS4 and ECM33 ( Figure 10A), indicating that the decrease in the mRNA abundance was not due to transcriptional repression. This was further supported by the observation that bgalactosidase activities of the C. glabrata GAS2 promoter-lacZ fusion gene were increased, not decreased, in both the wild-type and Dire1 strains in response to DTT exposure ( Figure 10B). Although it is unclear why the activity of the GAS2 promoter is increased in response to DTT, this paradoxical behavior suggest that transcriptional regulation and Ire1-dependent decay may not coordinately function in ER-stressed C. glabrata cells.

Phenotypic differences between the Dire1 mutant in C. glabrata and UPR defective mutants in other fungal species
To further investigate the roles of Ire1 in stress response in C. glabrata, we examined the growth of the Dire1 mutants under various stress conditions. Pathogenic fungi commonly encounter a variety of adverse environmental conditions, such as low oxygen availability and nutrient depletion, in the infected host. It has been reported in A. fumigatus that IreA is required for growth in hypoxia and under irondepleted conditions, which is induced by adding the Fe 2+ chelator bathophenantroline disulphonate (BPS) to the culture medium [8]. In C. glabrata, the loss of Ire1 had no apparent effect on cell growth under low oxygen tension ( Figure S6A) and on plates containing BPS or the bacterial siderophore desferrioxamine (DFO) ( Figure S6B). C. glabrata cells are unable to utilize DFO, which therefore induces irondepletion for this fungus [65]. Growth of both C. glabrata wild-type and Dire1 strains in the presence of BPS or DFO was rescued by supplementation of the media with 1 mM FeCl 3 .   It is also reported in several pathogenic fungi, including C. albicans, C. neoformans, and A. fumigatus, that mutant strains lacking either Ire1 or Hac1 display growth defects in the presence of cell wall-damaging agents such as caspofungin, Congo red, and calcofluor white [6,7,8,10,11,66]. However, in contrast to those fungi, C. glabrata strains lacking IRE1, HAC1, or both did not exhibit decreased tolerance to these agents at various concentrations tested (data not shown), suggesting that neither Ire1 nor Hac1 is primarily involved in cell wall stress response in C. glabrata.
Importantly, disruption of Ire1/IreA confers a drastic increase in azole susceptibility in C. neoformans and A. fumigatus [7,8]. In contrast to these findings, deletion of IRE1 alone did not affect azole susceptibility in C. glabrata ( Figure 11A). However, we found that the Dcnb1 Dire1 double mutant, but not the Dcrz1 Dire1 double mutant (data not shown), was more susceptible to azole antifungals than either single mutant ( Figure 11A). The results are in agreement with our previous report that calcineurin is required for azole tolerance through a Crz1-independent mechanism in C. glabrata [52]. Furthermore, deletion of either CNB1 or IRE1 alone did not affect cell growth in the presence of caffeine and NaCl, whereas the simultaneous loss of these genes resulted in severe growth defects ( Figure 11B). None of these effects was observed when IRE1 was deleted in the Dslt2 background (data not shown). These results suggest that calcineurin and Ire1 serve redundant roles in cell growth under certain stress conditions in C. glabrata. Collectively, these phenotypic analyses revealed that loss of Ire1 alone does not induce diverse phenotypes in C. glabrata unlike in other fungi.

Ire1 is required for virulence in C. glabrata
We have previously demonstrated in C. glabrata that calcineurin and Slt2 are required for virulence in murine models of disseminated candidiasis [52,67], but the involvement of Ire1 in C. glabrata virulence has not been examined. The C. neoformans Dire1 and A. fumigatus DireA mutants are avirulent in murine models of systemic cryptococcosis and invasive aspergillosis, respectively, which can be explained primarily by their growth defects at 37uC [7,8]. In C. glabrata, the loss of Ire1 did not affect cell growth at 37uC (generation time of the wild-type and Dire1 strains in Yeastpeptone-dextrose (YPD) broth was 75 and 78 min, respectively). Therefore, we compared the virulence of the C. glabrata wild-type, Dire1, and IRE1-complemented strains in a mouse model of disseminated candidiasis. Immunocompetent mice infected with the Dire1 mutant displayed significantly reduced fungal burden in both kidney and spleen relative to those infected with the wild-type and Ire1-complemented strains ( Figure 12A). To evaluate mortality of mice with disseminated candidiasis, mice were immunosuppressed by cyclophosphamide prior to C. glabrata infection. Consistent with the results of organ fungal burden, mice infected with the wild-type and IRE1-complemented strains exhibited higher mortality rates than mice infected with the Dire1 mutant ( Figure 12B). These results suggest that Ire1 is important for virulence in C. glabrata.

Discussion
According to the current evidence, the UPR regulated by Ire1-Hac1 signaling is highly conserved and is a key pathway to cope with ER stress in eukaryotes [2,12]. However, the present study demonstrates the lack of this pathway and the development of alternative mechanisms for the ER stress response in C. glabrata. The gene CAGL0K12540g, named C. glabrata HAC1, seems to be the sole HAC1 ortholog in C. glabrata, since it is the only gene showing sequence similarity and conserved synteny to S. cerevisiae HAC1. In addition, C. glabrata Hac1 was able to complement S. cerevisiae Hac1 functions. However, surprisingly, C. glabrata Hac1 did not necessitate Ire1-mediated splicing of mRNA before translation, and thus could recover the growth of the S. cerevisiae Dire1 mutant in the presence of ER stress. A recent study has comprehensively characterized the RNA structures of the Hac1/ Xbp1 orthologs in various eukaryotic species [36]. In C. glabrata HAC1, although the overall RNA structure is maintained, the predicted intron (379 nucleotides) is much longer than those in other species and has mutations in the Ire1 recognition motifs at both splice sites, suggesting that the Ire1-dependent unconventional splicing mechanism may not be present in C. glabrata [36]. This has been experimentally confirmed in our studies: the heterologous expression of C. glabrata HAC1 complemented S. cerevisiae Hac1 functions without splicing, and C. glabrata HAC1 mRNA was not spliced by Ire1 even in ER-stressed C. glabrata cells. Furthermore, Hooks and Griffiths-Jones [36] revealed that the non-canonical Hac1/Xbp1 intron structure is not conserved in 28 out of 156 eukaryotic species searched, including several Candida species such as Candida parapsilosis, Candida lusitaniae, and Candida guilliermondii, despite the fact that the HAC1 ORFs are intact in these Candida species. These findings suggest that these species may have lost the unconventional splicing mechanism, but instead acquired an alternative mechanism to regulate the UPR. Future studies are warranted to determine whether Hac1 is involved in the ER stress response in these species.
Although some Hac1-independent functions of Ire1 have been reported in some eukaryotic species [2,7,8,63], mutants lacking Hac1 or its homolog always phenocopy Dire1 strains with respect to growth defects in the presence of ER stress-inducing agents (e.g. TM and DTT) in all eukaryotic species tested so far, and the UPR mediated by Ire1-Hac1 signaling has been believed to be a phylogenetically conserved pathway in eukaryotes [2,68,69]. However, our study demonstrated that C. glabrata Ire1 plays a role in the ER stress response in a Hac1-independent manner. Exploration of potential Ire1 targets in this fungus by a genomewide analysis will be of interest in future investigations (e.g. RNAseq analysis of ER-stressed and unstressed cells of wild-type and Dire1 strains to detect Ire1-dependent spliced RNAs). The results of our gene expression assays suggest that Ire1 does not mediate transcriptional response to ER stress in C. glabrata, raising the question of how Ire1 is involved in the ER stress response in this fungus.
In metazoans, ER stress triggers two distinct outputs of Ire1's nuclease activity, XBP1 splicing and RIDD [69,70]. The latter is an Xbp1-independent pathway that selectively degrades a small subset of ER-associated mRNAs and remodels the repertoire of proteins translated in ER-stressed cells [63,64]. This cellular response is predicted to reduce the ER load by limiting protein influx and unfolded protein load into the ER lumen. The RIDD pathway was first uncovered in D. melanogaster [64], and later, its conservation was confirmed in mammalian cells [62,63], but has yet to be fully determined in fungi. Interestingly, in S. cerevisiae, RIDD does not occur [63,70], but the classic Ire1-Hac1 UPR mediates downregulation of some mRNAs encoding membrane proteins [64,71]. Here, we have demonstrated that mRNA abundance of a subset of genes encoding GPI-anchored cell wall and membrane proteins was diminished during the response to ER stress in the wild-type strain dependent on the nuclease activity of Ire1. There was no overlap between these repressed C. glabrata genes and the previously identified S. cerevisiae gene that were transcriptionally downregulated by the UPR in response to TM [71]. It is thought that RIDD targets are nicked by the Ire1 endoribonuclease at sites that do not display an identifiable consensus sequence, in contrast to Ire1-mediated splicing of HAC1/XBP1 mRNA at the highly conserved splice junctions [69]. As excessive activation of RIDD seems to be detrimental to cell integrity [62], reducing the load of proteins entering the ER must be balanced with the need to sustain synthesis of essential proteins. In contrast to Ire1 in metazoans, C. glabrata Ire1 may not need to switch between specific and nonspecific modes of cleavage. We are currently investigating how the RIDD activity is regulated in this fungus.
Although the only known substrate of the Ire1 kinase is Ire1 itself in S. cerevisiae [30,31], we also addressed the possibility that Ire1 may integrate with other signaling pathways in C. glabrata. In S. cerevisiae, calcineurin and Slt2 are also involved in the ER stress response through different mechanisms [48,49,50,51]. Slt2 plays a role in the ER stress surveillance (ERSU) pathway that ensures transmission of only functional ER to daughter cells during cell division [72]. Upon ER stress, the ERSU pathway delays ER inheritance and cytokinesis to prevent death of both mother and daughter cells. Calcineurin mediates the calcium cell survival pathway by regulating intracellular Ca 2+ homeostasis. TM increases Ca 2+ uptake by stimulating the Cch1-Mid1 high affinity Ca 2+ channel, while calcineurin dephosphorylates the Cch1 subunit of the channel to inhibit Ca 2+ influx and prevents nonapoptotic cell death in S. cerevisiae [48,51]. In C. glabrata, loss of either calcineurin or Slt2 resulted in decreased tolerance to ER stress, and the additional deletion of IRE1 in the Dcnb1 and the Dslt2 backgrounds had synthetic effects on viability loss in the presence of TM. The results suggest that Ire1 functions in parallel with the calcineurin and Slt2 MAPK pathways to cope with ER stress in C. glabrata, consistent with previous findings in S. cerevisiae [48,49,50]. However, the possibility that Ire1 might interact with other signaling pathways has not been completely ruled out. Whether the Ire1 kinase domain has an effector other than Ire1 itself remains one of the major unsolved questions in this field [69].
The term UPR is derived from a cellular response to chemicals that interfere with proper protein folding (e.g. TM and DTT), but it is now known that the signaling pathway is involved in various stress responses from yeast to humans [73,74]. Indeed, downstream targets of the UPR include not only ER chaperones, but also genes involved in diverse functions, including protein trafficking and quality control, lipid and sterol metabolism, heme biosynthesis, and cell wall biogenesis in S. cerevisiae [20]. To extend our understanding of Ire1 functions in C. glabrata, phenotypes of the Dire1 mutant were investigated in various stress conditions in view of clinical significance. The UPR is critical for azole tolerance in C. neoformans and A. fumigatus, and thus loss of either Ire1/IreA or Hxl1/HacA leads to increased susceptibility to azole antifungals in these species [7,8,10]. In contrast, loss of Ire1, Hac1, or both did not affect azole tolerance in C. glabrata ( Figure 11A and data not shown), further supporting the idea that the classic UPR mechanism regulated by Ire1-Hac1 signaling is not conserved in this fungus. However, the additional deletion of IRE1 in the Dcnb1 background increased azole susceptibility in C. glabrata. One possible explanation is that azole antifungals induce severe ER stress in calcineurin-defective C. glabrata cells, which therefore require Ire1 to survive this stress condition.
Previous studies in several yeasts and filamentous fungi have demonstrated that the UPR plays a role in maintaining cell wall integrity, and that mutants lacking Ire1 or Hac1 exhibit decreased tolerance to a variety of cell wall perturbing agents [6,7,8,10,11,66]. TM also induces cell wall stress, thereby explaining why certain genes implicated in cell wall biogenesis are upregulated during the UPR in TM-treated S. cerevisiae and C. albicans cells [11,20,75]. In contrast to other fungi investigated to date, the loss of Ire1 alone did not confer a cell wall-defective phenotype, and upregulation of some cell wall-related genes in response to TM exposure was mainly dependent on calcineurin, but not on Ire1, in C. glabrata (Table S2). These results provide additional evidence that there are diversities in Ire1-dependent stress response mechanisms between C. glabrata and other fungal species. In addition, Dire1/ireA mutants of C. neoformans and/or A. fumigatus exhibit growth defects under various environmental conditions, such as growth temperature at 37uC, hypoxia, and iron limitation, which are commonly encountered by pathogenic fungi in the infected host [7,8]. However, none of these phenotypes were observed in the C. glabrata Dire1 mutant. In agreement with our proposal that the canonical Ire1-Hac1 UPR has been lost in C. glabrata, phenotypes of the C. glabrata Dire1 mutant were confined relative to those of UPR-defective mutants in other fungi where the UPR is involved in various stress responses.
Diversity in the transcriptional induction system in response to ER stress has developed during evolution [68]. For instance, the Ire1-Hac1 pathway is required for upregulation of KAR2 in yeast while the pathway is dispensable for induction of the Kar2 homolog GRP78/BiP in mammalian cells [17,76]. In C. glabrata, the majority of genes, including KAR2, induced in response to ER stress were dependent on the calcineurin-Crz1 pathway, but not on Ire1 signaling. In addition, several lines of evidence described above demonstrate that C. glabrata has lost the classic Ire1-Hac1 UPR, but instead possesses an alternative mechanism, RIDD, like metazoans. To our knowledge, these unusual characteristics have been demonstrated for the first time in any eukaryote, and perhaps they were developed in C. glabrata after its divergence from S. cerevisiae due to its ecological niche. Our results provide novel insights into ER stress response mechanisms as well as fundamental information for evolutionary biology to further understand how eukaryotic cells have developed ER quality control systems.

Strains and culture conditions
The C. glabrata and S. cerevisiae strains used in this study are listed in Table S6. Cells were routinely propagated at 30uC in YPD, synthetic complete (SC) medium lacking appropriate amino acids, or synthetic defined (SD) medium [77], unless otherwise indicated. The nitrogen starvation medium, SD-N, consists of 0.17% yeast nitrogen base without amino acids and ammonium sulfate (BD Biosciences, Franklin Lakes, NJ) plus 2% glucose as described previously [78]. The Anaeropack system (Mitsubishi Gas Chemical Company Inc, Tokyo, Japan) was used for cultures under conditions of low oxygen tension (hypoxic). To induce irondepleted conditions, the Fe 2+ chelator BPS (MP Biomedicals, Solon, OH) or the bacterial siderophore DFO (EMD Chemicals, San Diego, CA) were added to SC media at final concentrations of 20 and 50 mM, respectively.

Plasmid and strain construction
The primers and plasmids used in this study are listed in Tables S7 and S8, respectively. Transformation of C. glabrata and S. cerevisiae was performed using a lithium acetate protocol as described previously [79]. C. glabrata deletion mutants were constructed using a one-step PCR-based technique as described previously [67]. Briefly, a deletion construct was amplified from pBSK-TRP or pBSK-HIS using primers tagged with the 100 bp sequences homologous to the flanking regions of the target ORF. C. glabrata parent strains were transformed with the deletion construct, and the resulting transformants were selected by tryptophan or histidine prototrophy. PCR and Southern blotting were performed to verify that the desired homologous recombination occurred at the target locus.
To generate complementation plasmids, C. glabrata and S. cerevisiae genes were amplified from the genomic DNA of CBS138 [80] and BY4742 [81], respectively. Procedures for each plasmid construction are summarized in Table S8. The spliced form of S. cerevisiae HAC1 mRNA was obtained from BY4742 cells treated with 1.5 mg/ml TM for 3 h. The extracted RNA was then reversetranscribed using a QuantiTect Reverse Transcription kit (Qiagen, Valencia, CA), and the resulting cDNA was used as a template to amplify the mature S. cerevisiae HAC1, ScHAC1 i (''i'' for induced).
The plasmid pCgACT-PIRE was used as the template to generate pCgACT-PIRE-KD and pCgACT-PIRE-ND. pCgACT-PIRE-KD was generated by mutating two residues, D723 and K725, in the kinase domain of C. glabrata Ire1 to asparagines using the KOD-Plus-Mutagenesis Kit (Toyobo, Osaka, Japan), and mutagenic primers CgIRE1-mut-F2171 and CgIRE1-mut-R2170. Similarly, pCgACT-PIRE-ND was created by introducing a 10 residue (D973-Y982) internal deletion within the nuclease domain of C. glabrata Ire1 using mutagenic primers, CgIRE1-F2947 and CgIRE1-R2916. All of the plasmids constructed using PCR products were verified by sequencing before use.

Drug susceptibility assay
Spot dilution tests and time-kill assays were performed as described previously [52]. Voriconazole was kindly provided by Figure 12. C. glabrata Ire1 is required for virulence in a murine model of disseminated candidiasis. (A) Groups of 8 immunocompetent mice were intravenously inoculated with 8610 7 cells for each C. glabrata strain. Bilateral kidneys and spleen were excised 7 days after injection. Appropriate dilutions of organ homogenates were plated, and the numbers of CFU were counted after 2 days of incubation at 30uC. Numbers of recovered CFU from each organ are indicated for individual mice in the scatter plots. The geometric mean is shown as a bar. Statistical analyses were performed using the Kruskal-Wallis test with Dunn's multiple comparison post-test. Asterisks indicate statistically significant differences (*: P,0.05; **: P,0.01; ***: P,0.001). NS indicates no significance (P.0.05). Representative data of two independent experiments are shown. C. glabrata strains: Wild-type, TG11; Dire1, TG122; and Dire1+IRE1, TG123. (B) Groups of 7 mice were immunosuppressed by intraperitoneal administration of cyclophosphamide (200 mg/kg/day) 72, 48, and 24 h before challenge with C. glabrata cells. The mice were infected intravenously with C. glabrata cells (2610 7  Pfizer (New York, NY). Other drugs were purchased from Sigma (St Louis, MO). TM, voriconazole, and caffeine were dissolved in dimethyl sulfoxide (DMSO), and others were dissolved in distilled water. DMSO alone did not interfere with cell growth at the final concentrations used in this study. MICs of azole antifungals were determined by a broth microdilution test using a commercially prepared colorimetric microdilution panel (ASTY; Kyokuto Pharmaceutical Industrial Co., Ltd.) [82]. All sensitivity tests were performed on at least two separate occasions to ensure reproducibility.
Analysis of HAC1 mRNA splicing by RT-PCR and Northern blotting HAC1 mRNA splicing was analyzed using a modification of a reported protocol [66]. Logarithmic-phase S. cerevisiae cells grown in SC medium lacking leucine were treated with 1.5 mg/ml TM or 5 mM DTT for 3 h. Logarithmic-phase C. glabrata cells grown in SC medium were treated with either TM (5 and 10 mg/ml) or DTT (5 and 10 mM) for 1 and 3 h. These cells were harvested, and total RNA was extracted using a FastRNA Red Kit (Qbiogene, Carlsbad, CA) according to the manufacturer's instructions. cDNA was synthesized from 1 mg of total RNA using a QuantiTect Reverse Transcription kit (Qiagen) in a final volume of 20 ml, and 5 ml of resulting cDNA were then used as the template for individual PCR. The amplified PCR products were analyzed by both electrophoresis on a 1% agarose gel and DNA sequencing. For Northern blot analysis, twenty micrograms total RNA was separated on a 2% agarose gel and transferred to a Hybond-N + membrane (GE Healthcare, Buckinghamshire, UK), which was incubated in PerfectHyb Hybridization Solution (Toyobo) with probes directed against the 59 region of the HAC1 ORF. All probes were generated by PCR with primers listed in Table S7 and were labeled with [a-32 P]dCTP using a Random Primer DNA Labeling Kit Ver.2.0 (Takara Bio Inc., Shiga, Japan). Images were obtained using Image Reader FLA-5000 (FUJIFILM Corporation, Tokyo, Japan).

qRT-PCR
Total RNA extraction and cDNA synthesis were performed as described above, and 3 ml of resulting cDNA were used as the template for individual PCR with a QuantiTect SYBR Green PCR kit (Qiagen). qRT-PCR was carried out in triplicate in a 96well plate format, using a 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA). The mRNA abundance of the target genes was normalized to 18S rRNA (CAGL0L13398r). The qRT-PCR assays were repeated at least twice on independent occasions.

Microarray analysis
C. glabrata cells were grown in SC medium to exponential phase (OD600 = 0.8) and exposed to 1.5 mg/ml TM at 30uC. Total RNAs were extracted using the FastRNA Red Kit (Qbiogene). The quality of RNA was checked with a RNA 6000 Nano Kit and Agilent 2100 Bioanalyzer. Double-stranded cDNA was synthesized using the Invitrogen SuperScript Double-Stranded cDNA Synthesis Kit and oligo (dT) primers. The resulting cDNA samples were labeled with Cy3 using the NimbleGen One-Color DNA Labeling Kit and subsequently hybridized to a custom-made 4672 K C. glabrata array (Roche NimbleGen, Tokyo, Japan) wherein each chip measures the expression levels of 5,217 genes from C. glabrata CBS138 with six 60-mer-probe pairs per gene, with two-fold technical redundancy. The arrays were washed using the NimbleGen Wash Buffer Kit and scanned with a NimbleGen MS 200 Microarray Scanner. Data were quantified using the NimbleScan v2.6 software and normalized as described previously [83,84]. Hierarchical clustering was performed using Gene Cluster 3.0 and visualized using Java TreeView ver. 1.1.5r2. Gene Ontology (GO) enrichment was searched using the Candida Genome Database GO term finder (http://www.candidagenome. org/cgi-bin/GO/goTermFinder) with default parameters [85].

b-galactosidase assay
A 699-bp DNA fragment containing the 59 untranslated region (UTR) and the first 17 codons of the C. glabrata GAS2 gene was amplified from the genomic DNA of CBS138 and fused to the lacZ reporter in pEM14 [86] to construct pEM14-GAS2 as described in Table S8. Logarithmic-phase cells of the C. glabrata wild-type and Dire1 strains containing pEM14-GAS2 were grown in SC-ura broth, adjusted to 1610 7 cells/ml, and then incubated in the presence and absence of 3 mM DTT for 2 h. b-galactosidase assay was performed as described previously [59]. The cell cultures were harvested and washed twice with ice-cold phosphate-buffered saline. Cells (100 ml) were re-suspended in 300 ml Reporter Lysis Buffer (Promega, Madison, WI) containing 5 ml Protease Inhibitor Cocktail (Sigma). Cell extracts were prepared using acid-washed glass beads (Sigma) and cleared by centrifugation at 14,0006 g for 30 min at 4uC. Protein concentrations were determined by the Bio-Rad Protein Assay (Bio-Rad, Richmond, CA) using bovine serum albumin as a standard. b-galactosidase activities were measured using the b-Galactosidase Enzyme Assay System (Promega) according to the manufacturer's instructions and calculated in Miller units (nmoles/min/mg of protein) at 37uC [87]. All assays were performed in triplicate on separate days.

Virulence assay
Animal experiments were performed as described previously [52,67]. Briefly, to prepare cells for injection, logarithmic-phase C. glabrata cells were harvested, washed, resuspended in sterile saline, and adjusted to 4610 8 cells/ml after counting the number of cells using a hemocytometer. The actual CFU in the inocula were confirmed by plating serial dilutions of cell suspension on YPD plates. Groups of 8 female, 8-week-old, BALB/c mice (Charles River Laboratories Japan Inc., Japan) were injected with 0.2 ml of the C. glabrata cell suspension via the lateral tail vein. The mice were euthanized 7 days after injection and the spleen and bilateral kidneys were then excised. No mice died before euthanasia in this experiment. Appropriate dilutions of organ homogenates were plated on YPD plates. Colonies were counted after 2 days of incubation at 30uC and the CFUs per organ were calculated. Statistical analyses were performed using the Kruskal-Wallis test with Dunn's multiple comparison post-test (GraphPad Prism 5, La Jolla, CA).
To examine mortality of mice with disseminated candidiasis due to C. glabrata, groups of 7 female BALB/c mice (20-23 g body weight, mean 20.6-20.9 g for each group) were immunosuppressed by intraperitoneal administration of cyclophosphamide at a concentration of 200 mg/kg/day on days 23, 22, and 21 of infection, and housed under sterile conditions. The mice were inoculated intravenously with 0.2 ml of C. glabrata cell suspensions (1610 8 and 1610 7 cells/ml) on day 0. Checks were made for dead or moribund mice in a blinded fashion twice daily over a period of 16 days. A group of 4 immunosuppressed mice was injected with sterile saline instead of C. glabrata cell suspension, and no mice died due to immunosuppression only. Survival was plotted on a Kaplan-Meier curve for each C. glabrata strain, and the log rank (Mantel-Cox) test was used for pairwise comparison of percent survival (GraphPad Prism 5). A P value of ,0.05 was considered statistically significant. These animal experiments were conducted on two separate occasions to ensure reproducibility.
Accession numbers for genes and proteins mentioned in the text (NCBI Entrez Gene ID or GenBank accession number) Figure S1 Comparison of cell growth in the presence of ER stress between various fungal species. Logarithmic-phase cells of each fungal strain were harvested after appropriate incubation in YPD broth. Cell concentration was adjusted with optical density at 600 nm, and then 5 ml of serial 10-fold dilutions were spotted onto YPD or synthetic complete (SC) plates containing either tunicamycin (TM) or dithiothreitol (DTT) at the indicated concentrations. Plates were incubated at 30uC for 48 h. (TIF) Figure S2 Synteny relationships of the IRE1 and HAC1 loci between S. cerevisiae and C. glabrata. (A) Schematic representation of the C. glabrata IRE1 (CAGL0F03245g) locus compared to syntenic regions of the S. cerevisiae genome. Colored arrows indicate orthologs in the two species. White arrows indicate S. cerevisiae genes of which orthologs are not located around IRE1 in the C. glabrata genome. (B) Synteny analysis of the C. glabrata HAC1 (CAGL0K12540g) and S. cerevisiae HAC1 loci. (TIF) Figure S3 Disruption of HAC1 in C. glabrata. The entire open reading frame (ORF) of HAC1 was replaced with the deletion construct containing a TRP1 marker. Southern blot analysis of ClaI-digested genomic DNA using a HAC1 probe, which was designed within the HAC1 ORF, identified the predicted 4.3 kb band in the parent strain 2001HT (Dhis3 Dtrp1) but no band in the Dhac1 mutant TG141 (Dhac1::TRP1 Dhis3). A second Southern blot analysis using a TRP1 probe confirmed that the desired homologous recombination had occurred at the HAC1 locus without ectopic integration of the deletion construct in the mutant. (TIF) Figure S4 qRT-PCR analysis to validate expression profiles of representative UPR target genes in the presence of tunicamycin (TM) and dithiothreitol (DTT). C. glabrata wild-type (CBS138) cells were treated with TM or DTT at the indicated concentrations for 1 and 3 h. Expression levels of KAR2 (A) and other known UPR targets (B) were examined by qRT-PCR as described in Materials and Methods. The means and standard deviations for three independent experiments are shown. (TIF) Figure S5 Genome-wide gene expression profiles in response to tunicamycin (TM) exposure for 0.5 h. Hierarchical clustering of genes whose expression levels were changed more than 1.5-fold after treatment with 1.