The inositol regulon controls viability in Candida glabrata

Abstract

Inositol is essential in eukaryotes, and must be imported or synthesized. Inositol biosynthesis in Saccharomyces cerevisiae is controlled by three non-essential genes that make up the inositol regulon: ScINO2 and ScINO4, which together encode a heterodimeric transcriptional activator, and ScOPI1, which encodes a transcriptional repressor. ScOpi1p inhibits the ScIno2-ScIno4p activator in response to extracellular inositol levels. An important gene controlled by the inositol regulon is ScINO1, which encodes inositol-3-phosphate synthase, a key enzyme in inositol biosynthesis. In the pathogenic yeast Candida albicans, homologues of the S. cerevisiae inositol regulon genes are ‘transcriptionally rewired’. Instead of regulating the CaINO1 gene, CaINO2 and CaINO4 regulate ribosomal genes. Another Candida species that is a prevalent cause of infections is Candida glabrata; however, C. glabrata is phylogenetically more closely related to S. cerevisiae than C. albicans. Experiments were designed to determine if C. glabrata homologues of the inositol regulon genes function similarly to S. cerevisiae or are transcriptionally rewired. CgINO2, CgINO4 and CgOPI1 regulate CgINO1 in a manner similar to that observed in S. cerevisiae. However, unlike in S. cerevisiae, CgOPI1 is essential. Genetic data indicate that CgOPI1 is a repressor that affects viability by regulating activation of a target of the inositol regulon.

INTRODUCTION

Fungi of the genus Candida are the most common cause of human fungal infections and can lead to both mucosal and systemic infections (Calderone, 2002). Candida albicans is the most common cause of these infections, but non-albicans Candida species are increasingly associated with disease (Coleman et al., 1998). One of these species, Candida glabrata, is now the second most common cause of both mucosal and systemic Candida infections (Kaur et al., 2005).

Phylogenetically, C. glabrata is more closely related to the non-pathogenic yeast Saccharomyces cerevisiae than most of the other common species of Candida associated with human disease (Kaur et al., 2005). C. glabrata lacks a number of the virulence factors associated with Candida pathogens, such as secreted hydrolases and hyphal growth (Kaur et al., 2005). Despite this, C. glabrata is a growing challenge in clinical settings where it causes mucosal infections and is associated with approximately 15 % of all Candida-related systemic bloodstream infections (Pfaller & Diekema, 2004). These observations are interesting in light of the fact that C. glabrata is more closely related to S. cerevisiae, but is still a pathogen, whereas S. cerevisiae is only very rarely associated with infection (Piarroux et al., 1999). A clearer understanding of how S. cerevisiae and C. glabrata differ from one another may help shed light on why one is a significant human pathogen and the other is not.

A number of recent studies have shown that C. albicans is transcriptionally rewired compared to S. cerevisiae. For example, the genes encoding enzymes of the Leloir pathway for galactose catabolism in S. cerevisiae are regulated by the ScGal4p transcription factor. However, in C. albicans these genes are regulated by the transcription factor Cph1p, while the C. albicans CaGal4p homologue regulates TCA cycle genes such as CaLAT1 (Martchenko et al., 2007). Other examples of transcriptional rewiring between C. albicans and S. cerevisiae include regulatory systems controlling mating type (Tsong et al., 2003), mitochondrial ribosomal genes (Ihmels et al., 2005) and de novo myo-inositol biosynthesis genes (Hoppen et al., 2007; Y. L. Chen & T. B. Reynolds, unpublished). Myo-inositol will be referred to as inositol throughout the rest of this article.

Since there are several examples of transcriptional rewiring between these two more distantly related yeasts (S. cerevisiae and C. albicans), it was of interest to determine if similar rewiring is present between the more closely related yeasts C. glabrata and S. cerevisiae. The Leloir enzymes for galactose metabolism are not present in C. glabrata (Kaur et al., 2005), and this yeast is a galactose auxotroph (Kreger-vav Rij, 1984), so this pathway is unavailable for comparison. Mating has never been described for C. glabrata, thus this pathway is not useful for study either. In contrast to these pathways, the inositol regulon appears to be an excellent pathway to compare between these two yeasts. The inositol regulon is a very well-studied transcriptional regulon in S. cerevisiae (reviewed by Chen et al., 2007; Greenberg & Lopes, 1996), and there are C. glabrata orthologues for both the transcription factors and targets of the S. cerevisiae inositol regulon (see Results and Discussion).

The inositol regulon in S. cerevisiae has been well-described (Chen et al., 2007; Greenberg & Lopes, 1996), and consists of three main transcription factors that regulate target gene expression in response to extracellular inositol levels (Fig. 1). The roles of each of these transcription factors in controlling this regulon are described in more detail below. Transcriptional targets of the inositol regulon include a number of phospholipid biosynthetic genes, but the most highly expressed and well-characterized of these targets is the S. cerevisiae INO1 (ScINO1) gene. ScINO1 encodes an enzyme that occupies the rate-limiting step in de novo inositol biosynthesis. Inositol is essential and is required for the synthesis of phosphatidylinositol (PI) which is a precursor for several essential lipids, including inositol-phosphate signalling lipids, glycosylphosphatidylinositol (GPI) anchors and sphingolipids (Dickson & Lester, 1999; Michell, 2008; Strahl & Thorner, 2007).

Fig. 1.

The inositol regulon in S. cerevisiae controls transcription of phospholipid biosynthetic genes like ScINO1 in response to the level of extracellular inositol. (a) The ScIno2p-ScIno4p heterodimer activates transcription of ScINO1 and other target genes in the absence of extracellular inositol. ScIno2p-ScIno4p binds UAS_INO present in the promoters of genes like ScINO1. (b) In the presence of extracellular inositol, ScOpi1p binds to ScIno2p and prevents it from activating transcription of ScINO1 and other targets.

ScIno1p is an inositol-3-phosphate synthase which converts glucose 6-phosphate to inositol 3-phosphate (Majumder et al., 1997). Inositol 3-phosphate is dephosphorylated by the inositol monophosphatases ScInm1p or ScInm2p to create inositol (Lopez et al., 1999). An Scino1Δ mutant cannot make inositol de novo and is an inositol auxotroph. In the absence of extracellular inositol (or at low concentrations like 10 μM) ScINO1 is expressed, and in the presence of higher concentrations of extracellular inositol ScINO1 is repressed (Graves & Henry, 2000) (Fig. 1).

The inositol regulon transcription factors ScIno2p and ScIno4p form a heterodimeric transcriptional activator that binds to the upstream activator sequence (UAS_INO) in the promoters of target genes such as ScINO1 (Ambroziak & Henry, 1994; Bachhawat et al., 1995; Loewen et al., 2003; Lopes & Henry, 1991; Nikoloff & Henry, 1994; Schwank et al., 1995). Both ScIno2p and ScIno4p are absolutely required for transcription of ScINO1, so Scino2Δ and Scino4Δ mutants are inositol auxotrophs.

The regulation of ScINO1 in response to extracellular inositol is dependent on the repressor protein ScOpi1p (Heyken et al., 2005; Jiranek et al., 1998; Loewen et al., 2003, 2004; Loewen & Levine, 2005). ScOpi1p senses the level of extracellular inositol indirectly based on the level of the PI precursor lipid phosphatidic acid (PA). Inositol is the rate-limiting metabolite in PI synthesis, and when there is abundant extracellular inositol, PI synthesis is maximal, and PA levels are lower because PA is consumed during synthesis of PI. In these conditions ScOpi1p binds to ScIno2p and represses transcription of ScINO1. When extracellular inositol is not available (or greatly decreased), PI synthesis slows, and PA levels increase in the endoplasmic reticulum (ER). ScOpi1p, which binds to PA, is recruited to the ER where it also binds Scs2p. Thus, the ScIno2p-ScIno4p heterodimer is freed to transcribe ScINO1. A mutation that deletes ScOPI1 results in constitutive overexpression of ScINO1 and other genes carrying the UAS_INO sequence in their promoters.

The purpose of this study was to determine if C. glabrata carried an inositol regulon that is similar to that in S. cerevisiae, or if these yeasts are transcriptionally rewired for inositol regulation. In order to do this, C. glabrata homologues of the S. cerevisiae inositol regulon proteins were identified and disrupted. Analysis of these mutants revealed the surprising finding that CgOPI1 is essential. CgIno2p and CgIno4p (which are not essential) are activators of CgINO1, and CgOpi1p is a transcriptional repressor of CgINO1. This is similar to the inositol regulon in S. cerevisiae. However, unlike in S. cerevisiae, CgOPI1 is required for viability. This difference from S. cerevisiae may indicate that the inositol regulon in C. glabrata has some additional or different targets than in bakers' yeast, or C. glabrata may be metabolically rewired compared to S. cerevisiae.

METHODS

Strains and media.

Strains are listed in Table 1. Integrations and manipulations were confirmed by PCR (Table 2) in most cases and by Southern blotting with the tetO : : HOP1 : : CgOPI1 strains. Strain BG14, which is a uracil auxotroph (Table 1), was a gift from Brendan Cormack and was used to generate all of the strains reported in this study. Disruption of CgINO2, CgINO4 and CgINO1 was performed in BG14 (Table 1) utilizing a two-step gene disruption strategy (Cormack & Falkow, 1999; Cormack et al., 1999; Rothstein, 1991) with disruption constructs built in the pRS306 vector (Sikorski & Hieter, 1989; Table 3) carrying the ScURA3 marker that can be selected for on media lacking uracil and counterselected against using media containing 5-fluororotic acid (5-FOA). Reconstitution of gene disruption mutants was done by transforming the strains with ScURA3-marked integrating plasmids carrying the wild-type gene (Table 3).

Table 1.

Strains used in this study

Strain	Genotype	Reference
BG14	Cgura3 : : Tn903neoR	Cormack & Falkow (1999)
BG2	CgURA3, parent of BG14	Cormack & Falkow (1999)
Cg11	Cgura3 : : Tn903neoR pCgOPI1	This study
Cg12	Cgura3 : : Tn903neoR pGRB2.1	This study
Cg14	Cgura3 : : Tn903neoR Cgopi1Δ pCgOPI1	This study
Cg5	Cgura3 : : Tn903neoR Cgino2Δ	This study
EBCg019	Cgura3 : : Tn903neoR Cgino2Δ : : CgINO2	This study
Cg23	Cgura3 : : Tn903neoR Cgino2Δ pCgOPI1	This study
Cg18	Cgura3 : : Tn903neoR Cgino2Δ Cgopi1Δ pCgOPI1	This study
EBCg005	Cgura3 : : Tn903neoR Cgino4Δ	This study
EBCg008	Cgura3 : : Tn903neoR Cgino4Δ : : CgINO4	This study
EBCg014	Cgura3 : : Tn903neoR Cgino1Δ	This study
EBCg017	Cgura3 : : Tn903neoR Cgino1Δ : : CgINO1	This study
Cg33	Cgura3 : : Tn903neoR Cghis3Δ	This study
EBCg046	Cgura3 : : Tn903neoR Cghis3Δ Cgtrp1Δ	This study
EBCg048	Cgura3 : : Tn903neoR Cghis3Δ Cgtrp1Δ pINTG4 pEB48	This study
EBCg049	Cgura3 : : Tn903neoR Cghis3Δ Cgtrp1Δ pINTG4 pEB48	This study

Table 2.

Primers used in this study

Primer	Sequence (5′–3′)*	Restriction site	Purpose†
TRO605	AAAAAAGCTTTGCCTCCTTATCGGTAACAA	HindIII	Forward for 5′ NCR of CgOPI1l
TRO606	AAAAGAATTCTCCAGCAGCACAGTTTATTCA	EcoRI	Reverse for 5′ NCR of CgOPI1l
TRO607	AAAAGAATTCCTCTTCAAATTGAAAACGTTACGAC	EcoRI	Forward for 3′ NCR of CgOPI1l
TRO608	AAAAACTAGTGATTTCTGTTTGACTATTGGTTCTC	SpeI	Reverse for 3′ NCR of CgOPI1l
TRO652	AAAAGAATTCCAGCTGAAGCTTCGTACGC	EcoRI	Forward for NAT1 gene
TRO653	AAAAGAATTCGCATAGGCCACTAGTGGATCTG	EcoRI	Reverse for NAT1 gene
VSO1	AAAACTCGAGCCTTTACACCTGTGAACTACAGTCA	XhoI	Forward for 5′ NCR CgINO4
VSO2	AAAAGAATTCTCCGACTTTTGAAATGGGGT	EcoRI	Reverse for 5′ NCR CgINO4
VSO3	AAAAGAATTCTGCCTGTAATTGAGAAATCCTATTG	EcoRI	Forward for 3′ NCR CgINO4
VSO4	AAAAGGATCCTTGAAAAGAGAAGTTAAAACAGAGG	BamHI	Reverse for 3′ NCR CgINO4
MHO3	AAAACTCGAGTGTCCCCTTTTTTTTTGCC	XhoI	Forward for 5′ NCR CgINO1
MHO4	AAAACCCGGGTCGTGTGGTAAGTGTAGTTGGTCA	SmaI	Reverse for 5′ NCR CgINO1
MHO1	AAAACCCGGGCAGACTTCCCAATGAGGGAAA	SmaI	Forward for 3′ NCR CgINO1
MHO2	AAAACCGCGGCGTTCGTTGGCGAAACTTTT	SacII	Reverse for 3′ NCR CgINO1
BCO3	AAAAAAGCTTTCGCCCGTCTGAAAAAAA	HindIII	Forward for 5′ NCR of CgINO2
BCO4	AAAAGAATTCGGTTCGTGTATTAAATTAAGCACTC	EcoRI	Reverse for 5′ NCR of CgINO2
BCO1	AAAAGAATTCCTACTGACTGTATGTTAGGCTGCAA	EcoRI	Forward for 3′ NCR of CgINO2
BCO2	AAAAGGATCCCAAACTCTTCTTTGAATGACTTTG	BamHI	Reverse for 3′ NCR of CgINO2
TRO665	AAAATCTAGAATTCCCCCATGTACCACAGTC	XbaI	Forward for 5′ NCR of CgHIS3
TRO667	AAAAGGATCCTTGCTCGATGCTTCTCTTTG	BamHI	Reverse for 5′ NCR of CgHIS3
EBO24	GTTGCTGTTAAGTATGTTTGA		Check insertion of pEB13 (ino4Δ)
EBO27	TGGCAACTAGAATTTTTCACATGC		Check insertion of pEB19 (ino1Δ)
TRO668	TACGTTGTTACCCACACGATT		Check insertion of pRS306-his3Δ
TRO623	ACAGTCATCCAAAGGTGACTCTCAT		Check insertion of pRS306-ino2Δ
TRO536	CCAGGGTTTTCCCAGTCA		Reverse for primers EBO24, EBO27 and TRO623
TRO614	GAAGGAATCTCAAAAGTGCGGA		Check insertion of pCgopi1Δ
TRO461	GTGCGCAGAAAGTAATATC		Reverse primer for TRO614
EBO64	AAAAGAATTCACCATGGACACAAGGCGTG	EcoRI	Forward for CgOPI1tetO promoter
EBO65	AAAAGGGCCCGTAGATGTAGGTTCTCCTTTTCATTA	ApaI	Reverse for CgOPI1tetO promoter
EBO40	ATGACTGTGAATAAAGGTATTAGCATTC		Forward for CgINO1 probe
EBO41	CTATTTCAATCTTTCTTCGAATCTCAG		Reverse for CgINO1 probe
TRO656	GCCGGTTTCGCCGGTGACG		Forward for CgACT1 probe
TRO636	CCAAAGCGACGTAACATAGCTTT		Reverse for CgACT1 probe
EBO12	GACACAAGGCGTGGGTG		Forward for CgOPI1 probe
EBO11	TCATACCTTTTGTAAATGCATA		Reverse for CgOPI1 probe

*Underline shows the restriction site.

†NCR, Non-coding region that is either 5′ or 3′ to the designated ORF.

Table 3.

Plasmids used in this study

Plasmid	Description	Type	Restriction site	Reference
pRS306-ino2Δ	CgINO2 knockout	Integrating	SnaBI	This study
pRS306-CgINO2	CgINO2 reintegration	Integrating	SnaBI	This study
pEB19	CgINO1 knockout	Integrating	NruI	This study
pEB21	CgINO1 reintegration	Integrating	NruI	This study
pEB13	CgINO4 knockout	Integrating	PmlI	This study
pEB17	CgINO4 reintegration	Integrating	PmlI	This study
pCgOPI1	Contains CgOPI1	Episomal		This study
pCgopi1Δ	Cgopi1 : : NAT1	PCR template		This study
pRS306-his3Δ	CgHIS3 knockout	Integrating	BglII	This study
pGRB2.1	ScURA3 vector	Episomal		Frieman et al. (2002)
pAG25	NAT1 cassette	PCR template		Goldstein & McCusker (1999)
pEB48	tetO : : HOP1 : : CgOPI1	Integrating		This study
pINTG4	tetR : : GAL4AD	Integrating		Nakayama et al. (1998)
pRS306	Shuttle vector	Integrating		Sikorski & Hieter (1989)

The Cgopi1Δ mutant (Cg14, Table 1) was generated in BG14 by transforming it with the pCgOPI1 episomal plasmid (ScURA3 marker, Table 3), and then disrupting the CgOPI1 gene on the chromosome with the Cgopi1 : : NAT1 construct amplified from plasmid pCgopi1Δ (Table 3). The Cgopi1Δ Cgino2Δ double mutant was made in the same manner as the Cgopi1Δ mutant except it was done in strain Cg5 (Cgino2Δ).

The tetO : : HOP1 : : CgOPI1 strains EBCg048 and EBCg049 (Table 1) were generated by first making BG14 a histidine and tryptophan auxotroph (Cghis3Δ Cgtrp1Δ). This was done by disrupting CgHIS3 in BG14 as described above for the Cgino2Δ gene to create strain Cg33. The CgTRP1 gene was disrupted in Cg33 using a plasmid from Karl Kuchler's lab to create strain EBCg046. EBCg046 was used to create the tetO : : HOPI1 : : CgOPI1 strains EBCg048 and EBCg049 by first integrating plasmid pINTG4 carrying the tetR : : GAL4AD repressor-activator into its genome as described by Nakayama et al. (1998). Then the tetO : : HOP1 : : CgOPI1 construct was PCR-amplified from plasmid pEB48 (Table 3) and was used to replace the CgOPI1 locus on the chromosome.

Media used in this study included 2 % agar plates or liquid media made with YPD, YNB, or inositol-free media (Styles, 2002) that included supplements of amino acids, nucleotides, inositol, doxycycline, 5-FOA, etc., as described in the text. For inositol-free agar plates, Bactoagar was used because it does not contain trace amounts of inositol.

Plasmids and constructs.

The gene disruption plasmids (pRS306-ino2Δ, pRS306-ino4Δ and pRS306-ino1Δ) were made by PCR-amplifying DNA corresponding to approximately 500 bp of non-coding DNA that flanked the 5′ and 3′ edges of each ORF (5′ or 3′ NCRs), and then cloning them into pRS306 adjacent to one another to create a disrupted allele (primers and restriction sites used for cloning are listed in Table 2). The pRS306-his3Δ disruption plasmid was created in a similar manner by subcloning the CgHIS3 5′ NCR into plasmid pGRB2.1 (Frieman et al., 2002) upstream of the 3′ NCR of CgHIS3 contained in this plasmid. The whole 5′ and 3′ NCR disruption cassette was then subcloned into the pRS306 integrating vector (Sikorski & Hieter, 1989) as an XbaI–KpnI fragment to create pRS306-his3Δ. The disrupted alleles from all of the pRS306 disruption cassette plasmids were used to replace the respective wild-type alleles on the chromosome of BG14 by the two-step deletion strategy (Cormack & Falkow, 1999; Cormack et al., 1999; Rothstein, 1991). For example, the pRS306-ino1Δ cassette was cut with NruI in the 5′ NCR to linearize it and target it to recombine upstream of CgINO1. PCR was used to confirm correct integrations and disruptions. In a similar manner, plasmids pRS306-ino4Δ, pRS306-ino1Δ and pRS306-his3Δ were cut with PmlI, SnaBI and BglII, respectively. Following replacement of the wild-type genes with the disrupted alleles, the mutant alleles were confirmed based on phenotype and PCR. Reintegration constructs were generated for each of the above disruptants in the vector pRS306 by PCR-amplifying the corresponding ORFs plus ∼500 bp 5′ and 3′ NCRs with the primers BCO3 and BCO2 for CgINO2, USO1 and USO4 for CgINO4, and MHO3 and MHO2 for CgINO1. Cut sites for subcloning are listed in Table 2. Plasmids were linearized with the enzymes mentioned above, and transformed into their respective disruptant strains. Correct integrations were confirmed by PCR with primers listed in Table 2.

The episomal plasmid pCgOPI1 carrying CgOPI1 plus 215 bp of upstream DNA and 439 bp of downstream DNA was generated by amplifying the CgOPI1 ORF and flanking sequences from purified BG14 DNA using primers TRO605 and TRO608, and cutting the PCR product with BamHI and SpeI enzymes. The SpeI site was introduced by primer TRO608, and the BamHI site was internal to the amplified DNA fragment. This PCR product was cloned into the ScURA3-bearing C. glabrata CEN/ARS plasmid pGRB2.1 (Frieman et al., 2002) using SpeI and BamHI. Plasmid pCgopi1Δ was generated by subcloning the NAT1 cassette from pAG25 (Goldstein & McCusker, 1999) with primers TRO652 and TRO653 into an EcoRI site between a 469 bp fragment containing the 5′ NCR of CgOPI1 (cloned by primers TRO605 and TRO606) and a 439 bp fragment containing the 3′ NCR of CgOPI1 (cloned by primers TRO607 and TRO608), both carried on the pRS306 vector. The pINT4 vector was a gift from Hironobu Nakayama (Nakayama et al., 1998) and was integrated into the EBCg046 genome after cutting it with EcoRV (Nakayama et al., 1998). Plasmid pEB48 was generated by subcloning an ∼500 bp 5′ NCR of the CgOPI1 ORF into p97CGH upstream (5′) of the HIS3 tetO-HOP1 cassette using primers EBO76 and EBO77. The CgOPI1 ORF was then subcloned 3′ to the HIS3 tetO-HOP1 cassette using a PCR product made by primers EBO64 and EBO65. This whole segment from pEB48, containing 5′-CgOPI1, NCR-HIS3 and the tetO-HOP1-CgOPI1 ORF was then transformed into strain EBCg046 carrying pINTG4 to replace the CgOPI1 promoter with the tetO-HOP1 cassette.

Northern blotting.

Northern blotting was performed essentially as described by Reynolds (2006) using probes generated from primers EBO40 and EBO41 for CgINO1, EBO11 and EBO12 for CgOPI1, TRO656 and TRO636 for CgACT1, TRO480 and TRO648 for ScACT1, EBO56 and EBO57 for CgCHO1, EBO54 and EBO55 for CgOPI3, and EBO58 and EBO59 for CgCHO2. Probes specific for ScACT1 or CgACT1 were used to reprobe Northerns for CgACT1 expression in C. glabrata for normalization, and both probes appeared to work well.

RESULTS

CgINO2 and CgINO4 encode transcriptional activators of the CgINO1 gene and control de novo inositol biosynthesis

A homologue of ScINO1 was identified in C. glabrata by blast analysis of the ScINO1 translated protein sequence against the C. glabrata genome at the Genolevures website (www.genolevures.org). The blast search revealed that the gene CAGL0I06050g encoded a protein that was 73.9 % identical to ScIno1p. We refer to CAGL0I06050g as CgINO1 based on the experiments described below.

CgINO1 expression was examined in media containing or lacking inositol by Northern blotting which revealed that CgINO1 was highly expressed in medium lacking inositol, but it was poorly expressed in medium containing 75 μM inositol (Fig. 2). This is similar to what has been observed for ScINO1 in S. cerevisiae (Graves & Henry, 2000). In order to determine if CgINO1 was required for de novo inositol biosynthesis, the CgINO1 ORF was disrupted by homologous recombination using a two-step gene deletion strategy (Cormack & Falkow, 1999; Cormack et al., 1999). The resulting Cgino1Δ mutant was unable to grow on inositol-free medium (Fig. 3). However, when the CgINO1 gene was reintegrated into the genome at the CgINO1 locus, the reconstituted strain (Cgino1Δ : : CgINO1) could grow in medium lacking inositol (Fig. 3).

Fig. 2.

CgINO2 and CgINO4 are required to express CgINO1 in the absence of exogenous inositol. (a) Northern blotting was used to assess the expression of CgINO1 in wild-type (WT), mutant and reconstituted strains in the presence (+) or absence (−) of exogenous inositol. CgACT1 was used as a loading control. Strains were grown overnight at 30 °C in inositol-free medium supplemented with 75 μM inositol, washed with water, and resuspended in inositol-free media containing either 0 or 75 μM inositol, after which the cultures were incubated with shaking for 6 h before collecting total RNA for Northern blotting. (b) A Storm Phosphorimager was used to quantify expression of CgINO1 in each strain relative to the wild-type strain grown in inositol-free medium, which was arbitrarily set at 100 %. Expression was normalized to CgACT1 expression. These results are representative of four different experiments.

Fig. 3.

The CgINO1, CgINO2 and CgINO4 genes are all required for de novo inositol biosynthesis, like their S. cerevisiae homologues. The Cgino1Δ, Cgino2Δ and Cgino4Δ mutants and their respective reconstituted strains (along with the wild-type control) were streaked onto inositol-free media supplemented with 0 (−) or 75 μM inositol (+), and grown at 30 °C for 3 days.

The CgINO1 gene is regulated in a similar manner as the ScINO1 gene in synthetic medium containing or lacking inositol, which suggests that the inositol regulon that controls ScINO1 in S. cerevisiae might be conserved in C. glabrata. In order to test this, homologues of the ScINO2 and ScINO4 transcriptional activator genes were identified by blast searches, querying the protein sequences of ScIno2p and ScIno4p against the C. glabrata genome at the Genolevures website. The blast searches revealed only one strong homologue for each protein, and these were encoded by the genes CAGL0B01947g for ScIno2p (35.6 % identity) and CAGL0I07359g for ScIno4p (44.5 % identity). These genes, CAGL0B01947g and CAGL0I07359g, were named CgINO2 and CgINO4, respectively. CgINO2 and CgINO4 were disrupted using the two-step gene disruption method which completely removed the ORF of each gene. The resulting Cgino2Δ and Cgino4Δ mutants were tested to determine if they could grow in the absence of inositol in the medium. As seen for orthologous Saccharomyces mutants, the Cgino2Δ and Cgino4Δ mutants were unable to grow on inositol-free medium (Fig. 3). These data suggest that CgINO2 and CgINO4 control the expression of the CgINO1 gene. Northern blotting revealed that Cgino2Δ and Cgino4Δ mutants showed a complete lack of CgINO1 expression even in inositol-free medium (Fig. 2). Reconstituted Cgino2Δ : : CgINO2 and Cgino4Δ : : CgINO4 strains, conversely, grew well on medium lacking inositol (Fig. 3) and regulated CgINO1 much like the wild-type strain (Fig. 2).

CgOPI1 is an essential gene in C. glabrata

In S. cerevisiae, the ScOPI1 gene encodes the main regulator of de novo inositol biosynthesis, and its homologue was identified in a blast search against the C. glabrata genome at Genolevures as described above. The protein encoded by CAGL0K03267/g was found to be 52.3 % identical to ScOpi1p. An attempt was made to disrupt the CgOPI1 gene by the two-step gene disruption strategy; however, this method continuously yielded strains carrying a wild-type ORF of CgOPI1.

The above results suggested that CgOPI1 might be essential, and this was shown to be the case using a counterselectable episomal plasmid expressing CgOPI1. CgOPI1 was cloned, along with non-coding DNA flanking both 5′ and 3′ of the ORF (including the transcriptional promoter and terminator, respectively), into the CEN/ARS vector pGRB2.1 (Frieman et al., 2002) to create plasmid pCgOPI1. Plasmid pCgOPI1, which carries the S. cerevisiae URA3 gene, was transformed into the wild-type strain, and the chromosomal CgOPI1 ORF was disrupted by homologous recombination using a construct in which the CgOPI1 ORF is replaced with the nourseothricin resistance marker cassette NAT1 (Goldstein & McCusker, 1999). The resulting pCgOPI1 Cgopi1Δ strain was then streaked onto medium containing 5-FOA. Processing of 5-FOA by the URA3 gene product from S. cerevisiae or C. glabrata leads to production of 5-fluorouracil which is toxic to C. glabrata (Boeke et al., 1987; Cormack & Falkow, 1999).

If the Cgopi1Δ mutation is lethal, then no Cgopi1Δ pCgOPI1 colonies should grow on 5-FOA medium because the 5-FOA would select against the cells carrying pCgOPI1. In Fig. 4 it is clear that the Cgopi1Δ pCgOPI1 strain cannot grow on 5-FOA. In contrast, colonies from the parental strain carrying pCgOPI1 or the empty vector grew well on this medium, indicating that they had lost the plasmid but were still viable.

Fig. 4.

The CgOPI1 gene is essential. Wild-type (WT) and Cgopi1Δ strains carrying CgOPI1 on a URA3 plasmid (pCgOPI1) were grown for 3 days on medium with (+) or without (−) 5-FOA at 30 °C.

These experiments were performed in the BG2 strain background (Cormack & Falkow, 1999). To be sure that this phenotype was not strain-specific, the above strategy was used to test the essentiality of CgOPI1 in the ATCC 2001 strain background (a gift from Karl Kuchler). It was found that CgOPI1 was essential in strain ATCC 2001 as well, which suggested that the essentiality of CgOPI1 is not just a BG2 strain-specific phenomenon (data not shown).

The regulation of viability by CgOPI1 is dependent on the CgINO2 transcription factor

Based on the model from S. cerevisiae (Fig. 1), it was hypothesized that disruption of CgOPI1 causes overexpression of a downstream target of the inositol regulon that then results in a loss of viability. In S. cerevisiae, overexpression of ScINO1 in the Scopi1Δ mutant is due to unrepressed transcriptional activation by ScIno2p (Fig. 1). The ScINO1 overexpression phenotype of the Scopi1Δ mutant can be blocked by a Scino2Δ mutation. The Scopi1Δ Scino2Δ double mutant acts like the Scino2Δ single mutant and fails to express ScINO1 because the Scino2Δ mutation is epistatic to the Scopi1Δ mutation (Graves & Henry, 2000).

If the Cgopi1Δ mutant compromises viability due to overexpression of a downstream target of CgIno2p, then a Cgino2Δ mutation should restore the viability of a Cgopi1Δ mutant by blocking expression of this putative target. In order to test this hypothesis, the Cgino2Δ mutant was transformed with plasmid pCgOPI1, and the chromosomal ORF of CgOPI1 was disrupted as described above. The resulting Cgopi1Δ Cgino2Δ pCgOPI1 strain was streaked onto 5-FOA medium, and it was found to grow like the wild-type strain carrying pCgOPI1 or empty vector (Fig. 5). The Cgino2Δ strain also grew like the wild-type strain (Fig. 5).

Fig. 5.

The viability defect of the Cgopi1Δ mutant is dependent on the CgINO2 transcriptional activator. The wild-type (WT), Cgopi1Δ, Cgino2Δ and Cgopi1Δ Cgino2Δ strains containing plasmid pCgOPI1 (URA3-based) were streaked onto synthetic medium plates with (+) or without (−) 5-FOA. The WT strain containing the empty vector was included as a control.

The CgOPI1 gene product represses expression of CgINO1

The results above suggest that CgOpi1p can act as a repressor of CgIno2p targets such as CgINO1. To determine if CgOpi1p could repress CgINO1, CgOPI1 was placed under the control of a doxycycline-repressible promoter (Nakayama et al., 1998). C. glabrata strain BG14 (Cgura3Δ) was modified by disrupting both the CgHIS3 gene (making it a histidine auxotroph) and the CgTRP1 gene (making it a tryptophan auxotroph). Using the resulting triple auxotroph (Cgura3Δ Cghis3Δ Cgtrp1Δ), the promoter of the CgOPI1 gene was replaced on the chromosome by homologous recombination with the tetO-HOP1 promoter construct derived from plasmid p97CGH (CgHIS3) (Nakayama et al., 1998). This strain was further modified by integration of plasmid pINTG4 (CgTRP1) carrying the tetR : : GAL4AD-encoded repressor-activator that activates the tetO-HOP1 chimeric promoter in the absence of doxycycline, but represses it in the presence of doxycycline.

The resulting strain was then tested for growth in the presence and absence of doxycycline, and it was found that this strain showed very poor growth in 10 μg doxycycline ml⁻¹, but grew quite well in the absence of doxycycline (Fig. 6b). These experiments were performed on synthetic minimal medium, but similar results were seen in YPD (rich) medium (E. K. Bethea & T. B. Reynolds, unpublished). When CgOPI1 levels were assessed in this strain by Northern blotting, it was found that CgOPI1 was expressed in the absence of doxycycline, but was not expressed in the presence of the drug (Fig. 6a).

Fig. 6.

CgOpi1p is a transcriptional repressor of CgINO1. The promoter of the chromosomal copy of CgOPI1 was replaced with the doxycycline-repressible tetO : : HOP1 promoter. (a) Expression of CgOPI1, CgINO1 and CgACT1 (loading control) were tested by Northern blotting in the presence (+) and absence (−) of doxycycline. Cells were grown to saturation overnight in synthetic medium (containing ∼11 μM inositol) lacking doxycycline, washed with water, and were then resuspended in fresh synthetic medium containing 0 or 10 μg doxycycline ml⁻¹ and grown for ∼6 h at 30 °C with shaking at which time samples were taken for Northern blotting. (b) Cells were grown on plates containing 0 (−) or 10 μg doxycycline ml⁻¹ (+) to confirm that loss of CgOPI1 expression decreased viability. Two different transformants were tested for the doxycycline repressible promoter.

CgOpi1p is a repressor of CgINO1. When CgOPI1 is repressed by doxycycline, CgINO1 expression increases substantially, but when CgOPI1 is expressed in the absence of doxycycline, then CgINO1 mRNA is very low or undetectable (Fig. 6a). This indicates that CgOpi1p acts as a repressor of CgINO1. We were also able to demonstrate the repression of CgINO1 expression by CgOpi1p using the C. glabrata copper-inducible MTII promoter (El Barkani et al., 2000) as well (E. K. Bethea & T. B. Reynolds, unpublished).

Overexpression of CgINO1 is not responsible for the loss of viability in the Cgopi1Δ mutant

Since CgINO1 is overexpressed in the absence of CgOPI1, it seemed possible that CgINO1 overexpression is responsible for the loss of viability. This was tested by creating a Cgopi1Δ Cgino1Δ pCgOPI1 double mutant. However, when this double mutant is streaked onto 5-FOA it fails to grow (E. K. Bethea & T. B. Reynolds, unpublished) indicating that the disruption of CgINO1 is not sufficient to restore viability as seen with CgINO2 (Fig. 5).

DISCUSSION

Transcriptional rewiring appears to exist between the transcription factor homologues of the inositol regulons of S. cerevisiae and C. albicans (Hoppen et al., 2007; Y. L. Chen & T. B. Reynolds, unpublished). C. glabrata is much more closely related to S. cerevisiae than C. albicans based on phylogenetic trees comparing 18S ribosomal sequences (Kaur et al., 2005). Our analysis suggests that the C. glabrata inositol regulon is not transcriptionally rewired compared to S. cerevisiae, at least for CgINO1 regulation. However, a major difference between the two species is that the OPI1 homologue in C. glabrata appears to be essential (Figs 4–6), whereas it is not in S. cerevisiae.

It is not clear why the Cgopi1Δ mutation is lethal, but our data suggest that the Cgopi1Δ mutation causes overexpression of CgIno2p-CgIno4p target genes and one of these targets causes a loss in viability when overexpressed. In support of this hypothesis, disruption of CgINO2 in the Cgopi1Δ strain rescues growth of the Cgopi1Δ mutant, presumably because the downstream target is no longer overexpressed (Fig. 5).

One possible target for compromising viability appeared to be CgINO1. However, disruption of CgINO1 in the Cgopi1Δ strain did not rescue viability on 5-FOA plates (E. K. Bethea & T. B. Reynolds, unpublished). This indicates that CgINO1 overexpression is not toxic.

There are two main models to explain the loss of viability of the Cgopi1Δ mutant. (1) A direct downstream target gene involved in phospholipid biosynthesis is overexpressed, and C. glabrata is particularly sensitive to this imbalance in lipid biosynthesis and loses viability. (2) Expression of a direct target gene not involved in lipid biosynthesis is affected by Cgopi1Δ and results in a loss of viability.

In the first model, there are several phospholipid biosynthetic genes that may be targets of the C. glabrata inositol regulon based on sequence similarity to homologues in S. cerevisiae. Direct downstream targets of the S. cerevisiae inositol regulon have been identified by the presence of the UAS_INO consensus sequence CATGTGAAAT in their promoters and their misregulation in Scino2Δ, Scino4Δ and Scopi1Δ mutants (Bailis et al., 1992; Graves & Henry, 2000; Jackson & Lopes, 1996; Lai & McGraw, 1994; Li & Brendel, 1993). Using these genes as a guide, there are five genes in C. glabrata that contain the sequence CATGTG (the most important part of the UAS_INO consensus sequence; Greenberg & Lopes, 1996) in their promoters. These genes include CgINO1, CgOPI3, CgCHO1, CgCHO2 and CgITR2. In addition to these five genes, two other genes, CgERG20 and CgCKI1, contain the sequence CATGTT, which differs by only one base and could possibly also respond to the inositol regulon.

In the second model, the target may be unrelated to phospholipid biosynthesis and/or may not have a homologue that is regulated by the inositol regulon in S. cerevisiae. Such a gene might also not be the direct cause of the loss of viability, but might itself regulate a downstream effector and cause the loss of viability. For example, if CgOPI1 were to repress a transcriptional repressor of an essential gene, then loss of CgOPI1 could result in loss of expression of the essential gene and compromise viability.

The essential nature of Opi1p in C. glabrata appears to be unique among the few characterized members of the growing Opi1p family of proteins, a family that is specific to fungi (Hirakawa et al., 2009). Opi1p homologues in S. cerevisiae, C. albicans, and Yarrowia lipolytica (Yas3p) are not essential (Hirakawa et al., 2009; Jiranek et al., 1998) (Y. L. Chen & T. B. Reynolds, unpublished). It is of interest to note that in S. cerevisiae and C. glabrata, two closely related yeasts, the Opi1p proteins both regulate inositol biosynthesis, whereas in the more distantly related yeasts C. albicans and Y. lipolytica, their Opi1p family members, CaOpi1p (C. albicans) and Yas3p (Y. lipolytica) do not regulate inositol biosynthesis (Endoh-Yamagami et al., 2007; Hirakawa et al., 2009; Yamagami et al., 2004) (Y. L. Chen & T. B. Reynolds, unpublished). Thus, the Opi1p family of proteins may have a number of diverse functions.

We are currently investigating the mechanism by which CgOPI1 controls viability in C. glabrata. Considering that the Opi1p family of proteins is unique to fungi, CgOpi1p may be a possible future drug target for treating C. glabrata infections.

Abbreviations

• 5-FOA, 5-fluororotic acid

• NCR, non-coding region, UAS_INO, upstream activator sequence for inositol regulation