Advertisement
Research Article

Genome and Transcriptome Analysis of the Food-Yeast Candida utilis

  • Yasuyuki Tomita,

    Affiliations: Central Laboratories for Frontier Technology, KIRIN Holdings Company, Ltd., Yokohama, Kanagawa, Japan, Center for Information Biology and DNA Data Bank of Japan, National Institute of Genetics, Mishima, Shizuoka, Japan

    X
  • Kazuho Ikeo,

    Affiliation: Center for Information Biology and DNA Data Bank of Japan, National Institute of Genetics, Mishima, Shizuoka, Japan

    X
  • Hideyuki Tamakawa,

    Affiliation: Central Laboratories for Frontier Technology, KIRIN Holdings Company, Ltd., Yokohama, Kanagawa, Japan

    X
  • Takashi Gojobori,

    Affiliation: Center for Information Biology and DNA Data Bank of Japan, National Institute of Genetics, Mishima, Shizuoka, Japan

    X
  • Shigehito Ikushima mail

    Shigehito_Ikushima@kirin.co.jp

    Affiliation: Central Laboratories for Frontier Technology, KIRIN Holdings Company, Ltd., Yokohama, Kanagawa, Japan

    X
  • Published: May 18, 2012
  • DOI: 10.1371/journal.pone.0037226

Abstract

The industrially important food-yeast Candida utilis is a Crabtree effect-negative yeast used to produce valuable chemicals and recombinant proteins. In the present study, we conducted whole genome sequencing and phylogenetic analysis of C. utilis, which showed that this yeast diverged long before the formation of the CUG and Saccharomyces/Kluyveromyces clades. In addition, we performed comparative genome and transcriptome analyses using next-generation sequencing, which resulted in the identification of genes important for characteristic phenotypes of C. utilis such as those involved in nitrate assimilation, in addition to the gene encoding the functional hexose transporter. We also found that an antisense transcript of the alcohol dehydrogenase gene, which in silico analysis did not predict to be a functional gene, was transcribed in the stationary-phase, suggesting a novel system of repression of ethanol production. These findings should facilitate the development of more sophisticated systems for the production of useful reagents using C. utilis.

Introduction

Candida utilis (Lindnera jadinii) is a Crabtree effect-negative yeast that is currently used to produce several industrially important compounds, such as glutathione and RNA [1][3]. C. utilis can grow on inexpensive substrates, such as pulping-waste liquors from the paper industry [4], and high cell density culture and large-scale production are possible under efficient continuous-culture conditions [1]. Furthermore, C. utilis is able to assimilate nitrate, a naturally occurring mineral source of nitrogen found as potassium nitrate. Along with Saccharomyces cerevisiae and Kluyveromyces fragilis, dried C. utilis cells have been approved for use as a food additive by the Food and Drug Administration [1]. In addition, since efficient methods of C. utilis transformation have been developed [5][7], the yeast has been used for the heterologous production of monellin, α-amylase, carotenoids, and organic acids such as L-lactic acid [8][12].

In a previous study, C. utilis NBRC0988 (ATCC 9950, CBS 5609) was suggested to be tetraploid based on multiple gene disruption using the Cre-loxP system and flow-cytometric analysis of nuclear DNA content [7]. Tetraploidy is relatively uncommon, as most yeast do not maintain a stable ploidy level of greater than two [13]. The recent completion of the C. utilis genome by massively parallel sequencing has led to the identification of 37 potential secretome proteins and the conclusion that C. utilis is not a member of the CUG clade, which comprises species such as Candida albicans, Debaromyces hansenii, and Pichia stipitis [14]. To date, phylogenic analyses have been conducted based on ribosomal DNA sequences [14][16], and not on whole genome data. In addition, the whole transcriptome analysis of C. utilis has not yet been reported.

We first sequenced the entire genome of the C. utilis NBRC0988 strain and analyzed its transcriptome. Using these data, we then determined both its phylogenic position, and the relationship between its phenotype and its genetic background. In addition, we identify a putative novel genetic mechanism of regulation of ethanol production based on comparative genome and transcriptome analyses using the next-generation sequencing.

Results

Genome sequencing, assembly and accuracy

We first subjected Candida utilis NBRC0988 to whole genome sequence analysis. employed a hybrid approach to integrate Sanger and 454/Roche sequencing (GS-FLX) data, leading to the production of 975 supercontigs (Materials and Methods). supercontigs out of 975 were larger than 40-kb (kilobases) in length, and we defined the 13 supercontigs as chromosomes with a total estimated size of 12.8 Mb., the other supercontigs were organized into a single hypothetical chromosome referred to as unmapped chromosome (Table 1) (estimated total genome size: 14.6 Mb) (accession numbers BAEL01000001-BAEL01001163 and DG000065-DG000077). The nuclear DNA content of C. utilis NBRC0988 was 3- to 5-fold greater than that of a S. cerevisiae haploid strain (12.3 Mb), supporting our previous suggestion that the former is tetraploid [7].

thumbnail

Table 1. Genome sequencing overview.

doi:10.1371/journal.pone.0037226.t001

The sequencing reads produced by the GS-FLX pyrosequencers are prone to nucleotide over-calls and under-calls that eventually manifest as insertion and deletion errors particularly within a homopolymer region (for example, poly-A) [17], [18]. calling errors may lead to two types of annotation errors: interrupted open reading frames (ORFs), with less than a 600-bp (base pair) gap between two ORFs, or overlapping ORFs. The interrupted and overlapping ORFs represented 6.3% (296) and 1.1% (52) of the total 4,706 annotated genes with at least one homolog in reference database mentioned below, respectively (frame-shift errors). Any doubtful assembly was considered to arise due to frame-shifts errors that resulted in two ORFs being annotated with the same name without having a truly orthologous relationship with each other (BLASTP e-value of more than 1e-5).

Annotation of the C. utilis genome

Protein coding genes were predicted using GlimmerM [19] and GeneLook [20] for the 975 supercontigs (Materials and Methods; Figure S1). This resulted in the identification of a total of 8,646 protein coding genes of which 4,041 (46.7%) had at least one homolog (with a BLASTP e-value of 1e-10 and the prescribed annotation accuracy in Figure S2) in the Saccharomyces genome database (SGD), Candida Genome Database (CGD), and/or SwissProt release 55.6. Thus, the protein coding genes occupy 59.2% of the genome sequence (Table 1). In addition, each putative 8,864 genes in C. utilis was compared with S. cerevisiae S288C mitochondrial genome (85,779 bp) to reveal that nine orthologous genes in the unmapped chromosome were identified as mitochondrial genes (e-values were each less than 1e-10). Although Buerth et al. [14] predicted only a putative 6,417 ORFs, our RNA-Seq data shown below supported the existence of approximately 90% of our 8,864 predicted genes based on the detection at least once in RNA-Seq tags described in fourth section.

Protein functions were assigned to 40.2% (3,476 genes) of the predicted 8,646 genes according to the Eukaryotic orthologous groups of proteins (KOG) classification [21] (Figure S3). secretion signal peptide was predicted in approximately 9% (770 genes) of the total number of genes with PrediSi [22], and 4,048 of the predicted proteins contained Pfam domains. Comparison of protein domains with those present in nine other yeasts species (Saccharomyces cerevisiae [23], Schizosaccharomyces pombe [24], Pichia stipitis [25], Yarrowia lipolytica [26], Ashbya gossypii [27], Kluyveromyces lactis [26], Candida albicans [28], Candida glabrata [26], and Debaryomyces hansenii [26]), identified 23 functional domains in 24 genes that are specific to C. utillis (Table S1).

The phylogenic position of C. utilis

We have first reported the phylogenic position of C. utilis using whole genome sequencing data. The phylogenic tree (Materials and Methods; Figure 1) shows that C. utilis diverged long before the formation of the CUG or Saccharomyces/Kluyveromyces clades. The codon usage of C. utilis differs significantly from that of other yeasts such as S. cerevisiae. For example, the relative frequencies of the usage of the two codons for Phe (UUU and UUC) and Tyr (UAU and UAC) are reversed in C. utilis relative to their frequencies of usage in S. cerevisiae (Figure S4).

thumbnail

Figure 1. Phylogenetic tree of single copy orthologous genes in 14 species.

This tree was built on the concatenated sequence of 32 single copy orthologous genes common to all 14 species by the maximum likelihood (ML) method using RAxML 7.2.8 with a JTT + Γ (gamma) model [58]. Individual orthologous gene families were aligned with the MUSCLE program [56], [57] and concatenated together to yield an alignment of 19,371 amino acid sites with gaps. One thousand bootstrap replicates were performed resulting in a fully resolved tree with all but one node having bootstrap values of 100.

doi:10.1371/journal.pone.0037226.g001

Analysis of the C. utilis transcriptome

A summary of the RNA-Seq data obtained using the next-generation sequencer is shown in Table 2. We obtained a total of 117,764,576 RNA-Seq tags from a log-phase sample in the presence of D-glucose and a total of 152,414,537 tags from a stationary-phase sample following depletion of D-glucose (accession number DRA000498). A majority of the expressed RNAs from each sample (approximately 60%) were successfully aligned to C. utilis genome (Table 2; Materials and Methods), resulting in the detection of more than 90% of the predicted genes based on detection of their transcription under the given conditions. In addition, more than 70% of the 3,186 genes that did not correspond to any protein in the reference database (predicted protein in Figure S2) were identified at least once in RNA prepared from cells at either phase. In this study, the RNA-Seq data were normalized based on upper-quartile (uq; defined by Bullard et al. [29]) and the gene whose uq value was more than 10 (approximately top 2 percent) was considered as “high expression gene (HEG)”. We then identified the pathways in which more than 10% of enzymatic genes were classified into HEGs either in log- (Table S3) or stationary-phase (Table S4). The numbers of the HEGs at log-phase, such as aminoacyl-tRNA and steroid biosyntheses genes, were notably more than that of stationary-phase, whereas citrate cycle and glyoxylate and dicarboxylate metabolism were more activated at stationary-phase. Besides, the HEGs at both phases were identified in oxidative phosphorylation pathway.

thumbnail

Table 2. RNA-seq overview.

doi:10.1371/journal.pone.0037226.t002

RNA-Seq analyses also revealed that approximately 97% of the genome covering both strands was expressed as RNA-Seq tags within each sample (theoretical maximum was 200%). Transcripts that were mapped to locations distinct from predicted exons may represent extensions of existing transcripts, non-coding RNA (ncRNA), antisense transcripts, or simply biological or experimental background errors. particular, we searched for transcripts within intergenic regions and antisense regions of coding sequences by identifying more than 180-bp contigs with a significant level of expression. Of 4,031 expressed regions identified at log-phase, 1,825 and 2,206 contigs were expressed in intergenic or antisense regions, respectively. Similarly, 1,750 and 2,341 contigs were found within intergenic or antisense regions during the stationary-phase (Table 2).

Construction and comparison of protein families

Comparison of the amino acid sequences of C. utilis ORFs with those of the aforementioned nine yeasts and Pichia pastoris [30] was conducted using a tBLASTN search (e-value 1e-10). Thirty-five annotated proteins lacked homologous sequences (defined as “unique”) within these genomes and three proteins also lacked annotation in the other yeast genomes (defined as “specific”) (Table S2).

Next, the proteins were subsequently classified into families, using the protein sequences from the 10 yeasts (Materials and Methods). 36.4% (3,144) of the predicted coding genes of C. utilis belonged to a protein family common to at least one other yeast. We show venn diagram groupings of protein families in C. utilis, S. cerevisiae, and C. albicans in Figure S5, as the two yeast genomes were used for the annotation of C. utilis genes. a result, we identified 4,894 families that were unique to C. utilis. is, the 4,894 families lacked homologous sequences in S. cerevisiae and C. albicans. In addition, we identified 40 pathways in which either S. cerevisiae or C. albicans, or both, lack one or more of the genes in C. utilis defined in the KEGG database (Table S5).

For example, the C. utilis nitrate reductase (EC 1.7.1.1/2/3) (CuNR) does not belong to a protein family common to the other two yeasts, and a nitrite reductase (EC 1.7.1.4) was found in C. utilis (CuNIR) but not in the others. Taken together with the existence of a nitrate/nitrite transporter (CuNRT) specific to C. utilis (Table S2), these results are consistent with the existence of a nitrate/nitrite assimilation phenotype [31] that is unique to C. utilis among the yeasts analyzed.

A cluster of nitrate assimilation genes in C. utilis

The cluster of nitrate/nitrite assimilation genes was identified in various organisms which can assimilate nitrate and nitrite [32]. Our C. utilis genome sequence revealed an existence of contiguous nitrate/nitrite assimilation gene cluster located on chromosome 2. The gene cluster is also considered to be responsible for in the nitrate assimilation phenotype in C. utilis. Since some reports have suggested that this gene cluster was transferred horizontally from a Basidiomycete (mushrooms and smuts) to an ancestor of the Ascomycetous mold Trichoderma reesei [32], we performed concatenated phylogenetic analyses of homologs of all three genes (CuNRT, CuNR, and CuNIR encoding putative proteins of 541, 865, and 1,077 residues, respectively) in Ascomycota and Basidiomycota from the subkingdom Dikarya (Materials and Methods) as shown in Figure 2. The phylogenic relationship of the gene cluster in the fungi is similar to the organismal phylogenic position except for the Ascomycete Trichoderma reesei, which indeed suggests horizontal gene transfer [32]. Because the three genes of the nitrate/nitrite assimilation pathway of C. utilis were strongly suggested to be clustered in Saccharomycetales, we presume that they mediate this assimilation activity in C. utilis as well as in P. angusta. Although we have yet to perform knock-out studies to confirm this, deletion study of nitrate/nitrite transporters in P. angusta was performed and the strain was found not to grow in nitrate media [33]. Comparison of the nitrate/nitrite transporters, nitrate reductases, and nitrite reductases from C. utilis and P. angusta revealed sequence identifies of 48% (251/522), 58% (492/840) and 61% (637/1043), respectively. On the other hand, the sequential order of the nitrate and nitrite reductase genes in C. utilis differs from that in P. angusta [33] and the nitrate/nitrite transporter of C. utilis is transcribed in the opposite direction relative to that in P. angusta [33]. We next constructed a phylogenic tree including P. angusta was constructed using 18 single copy orthologous genes in 12 sequenced yeast genomes and P. angusta was found to be a part of distinct clade from C. utilis (Figure S6).

thumbnail

Figure 2. Phylogenetic tree of nitrate assimilation related genes.

This tree was built on concatenated homologous sequences of all three genes (nitrate/nitrite transporter, nitrate reductases, and nitrite reductase) in Ascomycota and Basidiomycota from the subkingdom Dikarya (Materials and Methods) by the ML method.

doi:10.1371/journal.pone.0037226.g002

We next examined the transcription of the three genes of the nitrate/nitrite assimilation pathway based on the RNA-seq data, which was subjected to upper-quartile (uq) normalization [29] (Materials and Methods). The expression of each of the genes, annotated as CuNRT, CuNR, and CuNIR, was greater during log-phase (uq = 0.485, 1.270, and 4.300) than during stationary-phase (uq = 0.059, 0.020, and 0.216), respectively. The expression of CuNR, and CuNIR was particularly high during the log-phase grown in the nutrient rich media.

Hexose transporters in C. utilis

A phylogenic tree was constructed based on 114 transporters in C. utilis, S. cerevisiae, S. pombe, P. stipitis, A. gossypii, K. lactis, C. albicans, C. glabrata, D. hansenii, and P. pastoris. Base on this tree, thirty-one proteins were assumed to be hexose transporters. Next, the phylogenic tree was reconstructed using the 31 transporters and is shown in Figure 3. In contrast to S. cerevisiae, which possesses 20 isogenes encoding low- and high-affinity hexose transporters, only two putative hexose transporters are present in the C. utilis genome (cut01g0001347 and cut01g0020536). RNA-Seq analysis revealed that cut01g0001347, annotated as CuHXT6, was more highly expressed than cut01g0020536 (CuHXT10) (Table 3), suggesting that CuHXT6 functions as a pivotal hexose transporter under both high and low D-glucose concentrations.

thumbnail

Figure 3. Phylogenetic tree of 31 hexose transporters by the ML method.

These were selected from 114 transporters in C. utilis, S. cerevisiae, S. pombe, P. stipitis, A. gossypii, K. lactis, C. albicans, C. glabrata, D. hansenii, and P. pastoris.

doi:10.1371/journal.pone.0037226.g003
thumbnail

Table 3. Transcription results of 2 putative hexose transporters of C. utilis.

doi:10.1371/journal.pone.0037226.t003

Annotated genes involved in ethanol production and assimilation in C. utilis

Analysis of the genome sequence revealed that C. utilis has six and seven orthologous genes encoding alcohol dehydrogenase (ADH) and aldehyde dehydrogenase (ALD), respectively. We next performed gene expression analysis by RNA-Seq in order to examine which ADH and ALD genes were transcribed (Table 4). the ethanol concentrations and the transcription of ADH genes (Table 4), we presume that cut01g0000110 [named CuADH1-F (“F” stands for “forward.”)] functions in the reduction of acetaldehyde to ethanol because the gene was strongly expressed during the log-phase. In contrast, cut01g0050119 (CuADH2) and cut01g0021005 (CuADH3), which were both more highly expressed during the stationary-phase than during the log-phase, probably play a role in the conversion of ethanol to acetaldehyde. Based on the level of their expression during the log-phase, we presume that the products of these two genes act to assimilate ethanol even in the presence of D-glucose. Furthermore, the product of the ALD gene (CuALD6), which functions downstream of ethanol assimilation (the conversion of ethanol to acetaldehyde) through CuAdh2/3 was also highly expressed during both phases (Table 4). We also predicted the cellular localization of these genes using WoLF PSORT [34] as shown in Table 4. Ethanol assimilation in both the cytoplasm and mitochondria is thought to occur during both phases of growth in C. utilis, and is believed to contribute to the alleviation of the D-glucose overflow metabolism that is characteristic of Crabtree effect-negative yeast such as C. utilis.

In our genome sequencing, C. utilis was found to lack the conserved components required for RNA interference (RNAi), including Dicer-like RNase and Argonaute which were found in several budding yeast, such as Saccharomyces castellii and Candida albicans [35], [36]. However, some budding yeasts including S. cerevisiae, have been reported to have genes that are regulated by non-coding RNAs [37]. In the present study, antisense transcripts in C. utilis were analyzed using the RNA-seq data, which revealed a novel antisense transcript complementary to the entire CuADH1-F region, shown in Table 4 and Figure 4. Although the expression of CuADH1-F during the stationary-phase was one-sixth lower than during log-phase, the expression of the antisense transcript [CuADH1-R (“R” stands for “reverse.”)] during the stationary-phase was 93.5-fold greater than during the log-phase. In addition, the number of tags aligned at CuADH1-R (226,551) was 1.6-fold greater than those at CuADH1-F (145,562) during the stationary-phase. The inverse correlation between the expression of these two overlapping transcripts strongly suggests a possible mechanism whereby the expression of the antisense transcript acts to repress the expression of CuADH1-F.

Discussion

In the present study, we have conducted genome sequencing and whole genome transcriptome analysis of the C. utilis NBRC0988 strain. Although the phylogenic position has been already reported based on analysis of ribosomal DNA [14][16], we constructed a phylogenic tree, to clarify this phylogenic position using whole genome sequence data. We conclude that C. utilis represents a distinct clade among the 13 yeasts analyzed and that it diverged before the Saccharomyces/Kluyveromyces clade and long before the formation of the CUG clade.

thumbnail

Table 4. C. utilis ADH and ALD genes, and transcription results.

doi:10.1371/journal.pone.0037226.t004

We obtained a new finding about nitrate assimilation related genes in C. utilis and its phylogenic position among other microbes. As a noteworthy characteristic, C. utilis has been reported to assimilate nitrate and then to convert nitrite to ammonia as a nitrogen source [38]. This metabolic pathway is industrially useful since nitrate is a naturally occurring mineral source of nitrogen that can be found in the form of potassium nitrate. Thus, C. utilis can survive some poor conditions where other yeasts including S. cerevisiae can not survive. The amino acid sequences of C. utilis nitrate reductase, nitrite reductase, and nitrate/nitrite transporter are all homologous to the correspding enzymes in P. angusta and the corresponding genes are clustered in Saccharomycetales (Figure 2). However, the putative order of the nitrate and nitrite reductase genes differs between the two yeasts, and the nitrate/nitrite transporter gene in C. utilis is located in the opposite direction compared to that in P. angusata [33]. In addition, C. utilis and P. angusta are classified within distinct clades of the phylogenetic tree (Figure S6). In conclusion, C. utilis has maintained the capacity for nitrate assimilation and it is likely that the loss of nitrate assimilation genes in other yeasts analyzed clustered in Saccharomycetales occurred in both the clades containing C. utilis and P. angusata, respectively.

thumbnail

Figure 4. Antisense transcripts of C. utilis ADH1 (cut01g0000110) differentially expressed between log- and stationary-phases.

Expressions at the ADH1 locus was visualized using the Genaris integrated Next-Generation Sequencing Data Analysis Platform (GiNeS) (Genaris, Inc., Kanagawa, Japan).

doi:10.1371/journal.pone.0037226.g004

C. utilis is Crabtree effect-negative which is partially derived from limitation of sugar uptake to avoid overflow in metabolism [39]. Hence, we focused on hexose transporters in C. utilis. As shown in Figure 3, the phylogenic analysis revealed that C. utilis had fewer hexose transporters than S. cerevisiae. In addition, the gene cut01g0001347, annotated as CuHxt6, is assumed to be a pivotal hexose transporter at both high and low concentrations of D-glucose. The protein encoded by CuHxt6 exhibits high sequence similarity (more than 63% identity) to the Crabtree effect-negative yeast K. lactis hexose transporter 2 (Kht2), which has an intermediate affinity for D-glucose (Km of 3.7 mM) [40]. Although K. lactis Hgt1 is a high-affinity D-glucose transporter (Km of 1 mM) [41] and homologous (27.8% identity) to C. utilis Hgt1 (cut01g0021027), it is poorly transcribed during both phases (uq = 0.088 at log-phase and uq = 0.0066 at stationary-phase). On the other hand, no low-affinity hexose transporter was predicted in the genome of C. utilis. These data suggest that C. utilis limits its uptake of D-glucose, and this is consistent with previous reports indicating that C. utilis only uses a high affinity D-glucose uptake system [39] and that Crabtree-effect negative yeasts generally exhibit Km values for D-glucose in the micromolar range [42].

The C. utilis transcriptome during the log- and stationary-phases included more than four thousand transcripts within either intergenic or antisense regions (Table 2), which had not been predicted by any in silico approach. There have been previous reports of antisense transcriptional regulation in yeasts [43][45] which may play a different role in acting post-transcriptionally by RNAi mechanisms. In S. cerevisiae, zinc deficiency has been shown to induce the expression of the transcription factor Zap1, which in turn induces expression of intergenic RNA and repression of ADH1 gene expression [46]. However, we did not identify any Zap1 orthologs in C. utilis. Since the antisense transcript CuADH1-R is expressed during the stationary-phase but not the log-phase, we propose an alternative hypothesis that this transcript is involved in repression of ethanol production during the stationary-phase. Our results appear to differ from a previous report [46], because we believe that CuADH1-F may be repressed by its antisense transcript during the stationary-phase. Suppression mechanism of antisense transcripts has been reported to be enforced by two distinct mechanisms in S. cerevisiae: Rpd3S-mediated deacetylation that prevents antisense initiation [47], and Nrd1–Nab3–Sen1 complex-mediated termination [48]. Since C. utilis lacks the conserved components required for RNA interference (RNAi), including Dicer-like RNase and Argonaute as is the same with S. cerevisiae, we paid attention to the four homologs in C. utilis genome to reveal that all the four were actually transcribed. We detected the RNA-Seq tags in CuRPD3: (11,002 at log-phase and 12,059 at stationary-phase), CuNRD1: (20,513 and 44,159), CuNAB3 (: 17,071 and 35,917), and CuSEN1: (10,541 and 13,082). Herein, there might be a possibility that the expression of CuADH1-R might be controlled by either of the two mechanisms in C. utilis.

RNA-Seq is not limited to detecting transcripts that correspond to existing exon sequences, but also reveals the precise genome locations of transcripts corresponding to non-coding regions by comparison with genome reference information. A second advantage of RNA-Seq relative to microarray analysis is that it gives rise to very low background signals because the DNA sequences can be unambiguously mapped to unique regions of the genome. In addition, it can detect a large dynamic range of expression [49]. These advantages have led here to the identification of genes that play central roles in hexose transport and of a novel antisense transcript (CuADH-R). We believe that our RNA-Seq data has provided tremendous insight into and should help inform the design of follow-up experiments that aim to clone novel genes and useful promoters, quantitative trait locus (eQTL) mapping, and so on.

The wealth of information provided by complete genome sequence and transcriptome analysis will enable us to develop more sophisticated systems for the production of heterologous proteins and novel compounds. The genome data of C. utilis reported in the present study are now publicly available, and we expect this information to generate novel findings and developments in the food and pharmaceutical industries.

Materials and Methods

DNA preparation

C. utilis NBRC0988 (ATCC 9950, CBS 5609) was obtained from the National Institute of Technology and Evaluation (NITE) Biological Resource Center. This particular yeast strain was cultured at 30°C in YPD medium (1% yeast extract, 2% peptone, and 2% dextrose). Genomic DNA (>300 μg) was prepared according to a published protocol (Molecular Cloning 3rd edition). A C. utilis DNA library for sequencing was prepared by random cleavage of its DNA using a Hydroshear Device (GeneMachines, San Carlos, CA). Plasmid and fosmid DNA libraries comprised fragments of approximately 3 to 4-kb and 40-kb, respectively.

Genome sequencing and assembly

Shotgun sequencing (Sanger).

DNA templates were sequenced in a 384-well format, with the forward and reverse reactions (paired ends) being performed on the same plate to maximize the paired end pass rate. BigDyeTM Terminator version 3.1 (Applied Biosystems, Foster City, CA) reactions were used in preparation for sequencing on the ABI PRISM 3730xl. Thermal cycling was performed using 384-well Thermocyclers (Applied Biosystems). Sequencing reactions were purified using the CleanSeq dye-terminator removal kit (Agencourt, Danvers, MA). A passing read was defined as an average high quality PHRED score of 20 or greater for at least 100 bases, producing 57.9 Mb from plasmid and 18.5 Mb from fosmid libraries (read length: 500–600 bp).

Pyrosequencing and assembly using a hybrid approach.

We also used the GS-FLX (Roche) to produce 211.7 Mb and generate 1,017-K pair-end reads with 250 bp (averaged depth 20), to enable assembly with the Newbler assembler version 1.1.03.24 that is a part of the software package distributed with the sequencing machines. Subsequently, we obtained 975 supercontigs (total supercontig length: 14.6 Mb) through both the GS-FLX and the Sanger sequencing data with the Arachne assembler version 2.0 [50], respectively.

Codon usage.

Nucleotide sequences of the predicted 8,646 ORFs in C. utilis and the 5,865 ORFs in S. cerevisiae S288C strain were analyzed with ANACONDA 2.0 [51] to calculate codon usage.

Genome sequence annotation

Protein coding genes were predicted using GlimmerM [19] and GeneLook [20] from 975 supercontigs (Figure S1). The Saccharomyces genome database (SGD) (www.yeastgenome.org) was used as training data for GlimmerM, resulting in the prediction of 10,560 genes. Meanwhile, 8,927 genes were predicted using the ab initio gene finder, GeneLook. Among the genes designated by the two methods, we first selected those that were assumed to contain less than 10 consecutive nucleotide gaps and to comprise 60 or more amino acids. Among the latter, we subsequently selected common 4,536 genes in terms of strand direction, transcriptional initiation, and termination sites. In addition, another 598 and 1,757 genes were predicted individually (Figure S1). Those genes predicted by one method that overlapped partially with genes predicted by another method, but with less annotation accuracy (Figure S2) were not analyzed further. In sum, a total of 8,646 protein-coding genes were annotated using the SGD (6,717 ORFs), the Candida Genome Database (CGD) (www.candidagenome.org) (6,107 ORFs), and SwissProt release 55.6 (788,247 proteins). These were subsequently classified into four categories (conserved, slightly conserved, hypothetical, and predicted protein) according to their annotation accuracy (Figure S2).

Phylogenic tree construction

A phylogenic tree was constructed from the alignment of 32 proteins in Candida utilis NBRC0988, Saccharomyces cerevisiae S288C [23], Schizosaccharomyces pombe 972h- [24], Pichia stipitis CBS6054 [25], Yarrowia lipolytica CLIB122 [26], Ashbya gossypii ATCC10895 [27], Kluyveromyces lactis NRRL Y-1140 [26], Candida albicans SC5314 [28], Candida glabrata CBS138 [26], Debaryomyces hansenii CBS767 [26], Pichia pastoris GS115 [30], Kluyveromyces thermotolerans CBS6340 [52], Saccharomyces kluyveri CBS3082 [52], and Zygosaccharomyces rouxii CBS732 [52], the whole genomes of which have been completely sequenced (Figure 1). Thirty-two proteins were selected from universal single-copy orthologous genes exhibiting synteny in Yeast Gene Order Browser YGOB [53][55]. Also, eighteen proteins were selected in 12 yeast genomes including P. angusta under the same condition (Figure S6). The amino acid sequences of members of individual orthologous gene families were aligned using the MUSCLE 3.6 program [56], [57] and these sequences were concatenated together so as to yield an alignment of 19,371 amino acid including gaps. A phylogenic tree (Figure 1 and Figure S6) was constructed on the concatenated sequence by the maximum likelihood (ML) method using RAxML 7.2.8 with a JTT + Γ (gamma) model [58]. Gaps were considered to be complete deletions, and 1,000 bootstrap replicates were performed resulting in a fully resolved tree with all but one node having bootstrap values of 100. Similarly, multiple sequence alignments of the sequences of the genes involved in nitrate assimilation and the 31 hexose transporters were performed using the MUSCLE program [56], [57], and a phylogenic tree was also constructed using the ML method using RAxML 7.2.8 with a JTT + Γ (gamma) model [58] (Figure 2 and Figure 3). In order to identify orthologous genes involved in nitrate assimilation in Ascomycota and Basidiomycota, we performed a BLASTP search comparing each of the three genes involved in this pathway in C. utilis against the NCBI protein database (e-values were each less than 1e-40) and constructed a phylogenic tree (Figure 2).

Comparative analysis of protein families

In order to classify genes encoding proteins falling within specific families (orthologous and paralogous genes), we compared the amino acid sequence similarity of all proteins among 10 yeasts (C. utilis, S. cerevisiae, S. pombe, P. stipitis, Y. lipolytica, A. gossypii, K. lactis, C. albicans, C. glabrata, and D. hansenii) under two conditions: All-against-all BLASTP with an e-value of 1e-20 and a cover rate calculated by a total length of high-scoring segment pairs of more than 50%. We subsequently formed a triangle cluster when an orthologous relationship (e.g., protein A to protein C, protein B to protein C) was detected. In those cases where two triangle clusters shared 2 ORFs, we combined the two clusters and this procedure was continued iteratively so as to construct the final protein families [59].

SOLiD sequencing and mapping

To characterize the C. utillis transcriptome, total RNA was extracted from the NBRC0988 strain under aerobic conditions at 25°C in YPD medium (1% yeast extract, 2% peptone, and 5% D-glucose) during both log-phase with remaining D-glucose (10.5 h) and during stationary-phase when the D-glucose was depleted (24 h) using the RNeasy Mini Kit (QIAGEN GmbH., Hilden, Germany). Poly (A) RNA was isolated from the total RNA by using Poly (A) PuristTM mRNA Purification Kit (Ambion). The poly (A)-selected RNA was fragmented by incubation with RNase III (Ambion), and 100 to 200-bp fragments were separated by gel electrophoresis. According to the standard protocol described in the SOLiD Library Preparation Guide, cDNA was amplified onto beads and was subjected to emulsion PCR. Each library was sequenced using SOLiDTM 3 to generate 50-bp long reads according to the supplied protocol.

Sequence data were mapped and aligned to the reference genome using BioScopeTM software 1.0. The obtained sequences comprised 50-bp reads and they were divided into the former 25-bp (5′) and latter 25-bp (3′) and were required to map uniquely to the C. utilis genome, allowing up to two mismatches at the former 25-bp and no insertions or deletions. There are three parallel mappings: mapping to filter sequences [filtering out ribosomal RNA (about 1.5% and 2.1 % in the mapped tags in log- and stationary-phase samples, respectively)], mapping to the reference genome (exon sequences defined by the genome annotation), and mapping to splice junctions (defined by the genome annotation). The number of tags mapping to unique regions of the genome were used as an estimation of quantitative expression of the particular genes.

Determination of normalized gene expression by RNA-Seq

The expression levels of genes determined by RNA-Seq were subjected to upper-quartile (uq) normalization [29] for comparison of expression between log- and stationary-phases. The upper quartile (75th percentile) was computed for each sample ( = each column), after excluding genes with zero tags across all the columns. The normalized count equals the raw count divided by the upper quartile. We also applied this normalization method to the analysis of antisense transcript expression separately from the aligned tags at ORFs.

Data access

The C. utilis genomic sequence has been deposited in the DDBJ/EMBL/GenBank under accession numbers BAEL01000001-BAEL01001163 and DG000065-DG000077. RNA-Seq data have been deposited in the DDBJ Sequence Read Archive (DRA) under accession number DRA000498.

Supporting Information

Figure S1.

Summary of integrated gene prediction platform. Annotation accuracy is indicated in Figure S2.

doi:10.1371/journal.pone.0037226.s001

(TIFF)

Figure S2.

Summary of gene sequence annotation.

doi:10.1371/journal.pone.0037226.s002

(TIFF)

Figure S3.

Functional comparison of (A) C. utilis and (B) S. cerevisiae according to KOG categories. The rate of genes categorized to the following 3 KOGs in C. utillis is more than 1% compared to that of S. cerevisiae: *amino acid transport and metabolism (279, 216), **lipid transport and metabolism (226, 173), and ***secondary metabolites biosynthesis, transport and catabolism (135, 80). The figures in the brackets represent the number of genes of C. utillis and S. cerevisiae, respectively.

doi:10.1371/journal.pone.0037226.s003

(TIFF)

Figure S4.

(A) C. utilis and (B) S. cerevisiae codon usage.

doi:10.1371/journal.pone.0037226.s004

(TIFF)

Figure S5.

Venn diagram grouping protein families in C. utilis (A), S. cerevisiae (B), and C. albicans (C). The 2,397 of the protein families in C. utilis belonged to families that were present in either S. cerevisiae or C. albicans, or in both shown in ABC, ABc, and AbC.” The 4,894 families were found to lack homologous sequences in S. cerevisiae and C. albicans. Each lower-case character (a, b, and c) represents a group of complementary families in each yeast.

doi:10.1371/journal.pone.0037226.s005

(TIFF)

Figure S6.

Phylogenetic tree of single copy orthologous genes in 12 species. This tree was built on concatenated sequence of 18 single copy orthologous genes in 12 species by the ML method using RAxML 7.2.8 with a JTT + Γ (gamma) model [58].

doi:10.1371/journal.pone.0037226.s006

(TIFF)

Table S1.

Overview of Pfam domains specific to C. utilis compared to nine other yeasts (S. cerevisiae, S. pombe, P. stipitis, Y. lipolytica, A. gossypii, K. lactis, C. albicans, C. glabrata, and D. hansenii).

doi:10.1371/journal.pone.0037226.s007

(PDF)

Table S2.

Annotated proteins of C. utilis without homologous sequences in ten other yeast genomes (S. cerevisiae, S. pombe, P. stipitis, Y. lipolytica, A. gossypii, K. lactis, C. albicans, C. glabrata, D. hansenii, and P. pastoris) detected with a tBLASTN search (e-value 1e-10).

* Proteins without any annotations in other yeast genomes.

doi:10.1371/journal.pone.0037226.s008

(PDF)

Table S3.

KEGG pathways in which more than 10% of enzymatic genes were classified into high expression genes (uq>10) at log-phase.

doi:10.1371/journal.pone.0037226.s009

(PDF)

Table S4.

KEGG pathways in which more than 10% of enzymatic genes were classified into high expression genes (uq>10) at stationary-phase.

doi:10.1371/journal.pone.0037226.s010

(PDF)

Table S5.

Metabolic pathways defined in the KEGG database in which either S. cerevisiae or C. albicans, or both, lack one or more of the genes in C. utilis within the pathway.

doi:10.1371/journal.pone.0037226.s011

(PDF)

Acknowledgments

We are grateful to members of Central Laboratories for Frontier Technology, especially to Dr. Satoru Mizutani, Dr. Aruto Yoshida, Dr. Satoshi Yoshida, and Ms. Yuki Konoeda for their helpful discussion and to Dr. So Nakagawa, Dr. Yuki Kobayashi, and Dr. Naobumi V. Sasaki (National Institute of Genetics) for their technical help for phylogenic analyses.

Author Contributions

Conceived and designed the experiments: SI. Performed the experiments: YT SI. Analyzed the data: YT HT SI. Contributed reagents/materials/analysis tools: KI TG. Wrote the paper: YT TG SI.

References

  1. 1. Boze H, Moulin G, Galzy P (1992) Production of food and fodder yeasts. Crit Rev Biotechnol 12: 65–86.
  2. 2. Ichii T, Takehara S, Konno H, Ishida T, Sato H, et al. (1993) Development of a new commercial-scale airlift fermentor for rapid growth of yeast. J Ferment Bioeng 75: 375–379.
  3. 3. Li Y, Wei G, Chen J (2004) Glutathione: a review on biotechnological production. Appl Microbiol Biotechnol 66: 233–242.
  4. 4. Kurtzman CP, Fell JW (1998) “The Yeast, A Taxonomic Study” Fourth edition, Elsevier Science B. V., Amsterdam.
  5. 5. Kondo K, Saito T, Kajiwara S, Takagi M, Misawa N (1995) A transformation system for the yeast Candida utilis: use of a modified endogenous ribosomal protein gene as a drug-resistant marker and ribosomal DNA as an integration target for vector DNA. J Bacteriol 177: 7171–7177.
  6. 6. Ikushima S, Minato T, Kondo K (2009) Identification and application of novel autonomously replicating sequences (ARSs) for promoter-cloning and co-transformation in Candida utilis. Biosci Biotechnol Biochem 73: 152–159.
  7. 7. Ikushima S, Fujii T, Kobayashi O (2009) Efficient gene disruption in the high-ploidy yeast Candida utilis using the Cre-loxP system. Biosci Biotechnol Biochem 73: 879–884.
  8. 8. Kondo K, Miura Y, Sone H, Kobayashi K, Iijima H (1997) High-level expression of a sweet protein, monellin, in the food yeast Candida utilis. Nat Biotechnol 15: 453–457.
  9. 9. Miura Y, Kettoku M, Kato M, Kobayashi K, Kondo K (1999) High level production of thermostable alpha-amylase from Sulfolobus solfataricus in high-cell density culture of the food yeast Candida utilis. J Mol Microbiol Biotechnol 1: 129–134.
  10. 10. Miura Y, Kondo K, Shimada H, Saito T, Nakamura K, et al. (1998) Production of lycopene by the food yeast, Candida utilis that does not naturally synthesize carotenoid. Biotechnol Bioeng 58: 306–308.
  11. 11. Ikushima S, Fujii T, Kobayashi O, Yoshida S, Yoshida A (2009) Genetic engineering of Candida utilis yeast for efficient production of L-lactic acid. Biosci Biotechnol Biochem 73: 1818–1824.
  12. 12. Tamakawa H, Ikushima S, Yoshida S (2012) Efficient production of L-lactic acid from xylose by a recombinant Candida utilis strain. J Biosci Bioeng 113: 73–75.
  13. 13. Mable BK, Otto SP (2001) Masking and purging mutations following EMS treatment in haploid, diploid and tetraploid yeast (Saccharomyces cerevisiae). Genet Res 77: 9–26.
  14. 14. Buerth C, Heilmann CJ, Klis FM, de Koster CG, Ernst JF, et al. (2011) Growth-dependent secretome of Candida utilis. Microbiology 157: 2493–2503.
  15. 15. Yamada Y, Matsuda M, Mikata K (1995) The phylogenetic relationships of Pichia jadinii, formerly classified in the genus Hansenula, and related species based on the partial sequences of 18S and 26S ribosomal RNAs (Saccharomycetaceae). Biosci Biotechnol Biochem 59: 518–520.
  16. 16. Yamada Y, Yano J, Matsuda M, Higashi T, Mikata K (1995) The phylogeny of Williopsis salicorniae Hinzelin, Kurtzman et Smith based on the partial sequences of 18S and 26S ribosomal RNAs (Saccharomycetaceae). Biosci Biotechnol Biochem 59: 945–948.
  17. 17. Huse SM, Huber JA, Morrison HG, Sogin ML, Welch DM (2007) Accuracy and quality of massively parallel DNA pyrosequencing. Genome Biol 8: R143.
  18. 18. Quinlan AR, Stewart DA, Stromberg MP, Marth GT (2008) Pyrobayes: an improved base caller for SNP discovery in pyrosequences. Nat Methods 5: 179–181.
  19. 19. Salzberg SL, Pertea M, Delcher AL, Gardner MJ, Tettelin H (1999) Interpolated Markov models for eukaryotic gene finding. Genomics 59: 24–31.
  20. 20. Nishi T, Ikemura T, Kanaya S (2005) GeneLook: a novel ab initio gene identification system suitable for automated annotation of prokaryotic sequences. Gene 346: 115–125.
  21. 21. Tatusov RL, Fedorova ND, Jackson JD, Jacobs AR, Kiryutin B, et al. (2003) The COG database: an updated version includes eukaryotes. BMC Bioinformatics 4: 41.
  22. 22. Hiller K, Grote A, Scheer M, Munch R, Jahn D (2004) PrediSi: prediction of signal peptides and their cleavage positions. Nucleic Acids Res 32: W375–W379.
  23. 23. Goffeau A, Barrell BG, Bussey H, Davis RW, Dujon B, et al. (1996) Life with 6000 genes. Science 274: 546, 563–567:
  24. 24. Wood V, Gwilliam R, Rajandream MA, Lyne M, Lyne R, et al. (2002) The genome sequence of Schizosaccharomyces pombe. Nature 415: 871–880.
  25. 25. Jeffries TW, Grigoriev IV, Grimwood J, Laplaza JM, Aerts A, et al. (2007) Genome sequence of the lignocellulose-bioconverting and xylose-fermenting yeast Pichia stipitis. Nat Biotechnol 25: 319–326.
  26. 26. Dujon B, Sherman D, Fischer G, Durrens P, Casaregola S, et al. (2004) Genome evolution in yeasts. Nature 430: 35–44.
  27. 27. Dietrich FS, Voegeli S, Brachat S, Lerch A, Gates K, et al. (2004) The Ashbya gossypii genome as a tool for mapping the ancient Saccharomyces cerevisiae genome. Science 304: 304–307.
  28. 28. Jones T, Federspiel NA, Chibana H, Dungan J, Kalman S, et al. (2004) The diploid genome sequence of Candida albicans. Proc Natl Acad Sci U S A 101: 7329–7334.
  29. 29. Bullard JH, Purdom E, Hansen KD, Dudoit S (2010) Evaluation of statistical methods for normalization and differential expression in mRNA-Seq experiments. BMC Bioinformatics 11: 94.
  30. 30. De Schutter K, Lin YC, Tiels P, Van Hecke A, Glinka S, et al. (2009) Genome sequence of the recombinant protein production host Pichia pastoris. Nat Biotechnol 27: 561–566.
  31. 31. Choudary PV, Rao GR (1982) Energy-dependence of nitrate reductase induction in Candida utilis. Biochem Biophys Res Commun 108: 1293–1299.
  32. 32. Slot JC, Hibbett DS (2007) Horizontal transfer of a nitrate assimilation gene cluster and ecological transitions in fungi: a phylogenetic study. PLoS One 2: e1097.
  33. 33. Perez MD, Gonzalez C, Avila J, Brito N, Siverio JM (1997) The YNT1 gene encoding the nitrate transporter in the yeast Hansenula polymorpha is clustered with genes YNI1 and YNR1 encoding nitrite reductase and nitrate reductase, and its disruption causes inability to grow in nitrate. Biochem J 321 (Pt 2): 397–403.
  34. 34. Horton P, Park KJ, Obayashi T, Fujita N, Harada H, et al. (2007) WoLF PSORT: protein localization predictor. Nucleic Acids Res 35: W585–W587.
  35. 35. Drinnenberg IA, Weinberg DE, Xie KT, Mower JP, Wolfe KH, et al. (2009) RNAi in budding yeast. Science 326: 544–550.
  36. 36. Weinberg DE, Nakanishi K, Patel DJ, Bartel DP (2011) The inside-out mechanism of Dicers from budding yeasts. Cell 146: 262–276.
  37. 37. Harrison BR, Yazgan O, Krebs JE (2009) Life without RNAi: noncoding RNAs and their functions in Saccharomyces cerevisiae. Biochem Cell Biol 87: 767–779.
  38. 38. Bruinenberg PM, van Dijken JP, Scheffers WA (1983) An enzymic analysis of NADPH production and consumption in Candida utilis. J Gen Microbiol 129: 965–971.
  39. 39. van Urk H, Postma E, Scheffers WA, van Dijken JP (1989) Glucose transport in crabtree-positive and crabtree-negative yeasts. J Gen Microbiol 135: 2399–2406.
  40. 40. Milkowski C, Krampe S, Weirich J, Hasse V, Boles E, et al. (2001) Feedback regulation of glucose transporter gene transcription in Kluyveromyces lactis by glucose uptake. J Bacteriol 183: 5223–5229.
  41. 41. Billard P, Menart S, Blaisonneau J, Bolotin-Fukuhara M, Fukuhara H, et al. (1996) Glucose uptake in Kluyveromyces lactis: role of the HGT1 gene in glucose transport. J Bacteriol 178: 5860–5866.
  42. 42. Boles E, Hollenberg CP (1997) The molecular genetics of hexose transport in yeasts. FEMS Microbiol Rev 21: 85–111.
  43. 43. Prescott EM, Proudfoot NJ (2002) Transcriptional collision between convergent genes in budding yeast. Proc Natl Acad Sci U S A 99: 8796–8801.
  44. 44. Hongay CF, Grisafi PL, Galitski T, Fink GR (2006) Antisense transcription controls cell fate in Saccharomyces cerevisiae. Cell 127: 735–745.
  45. 45. Tuch BB, Mitrovich QM, Homann OR, Hernday AD, Monighetti CK, et al. (2010) The transcriptomes of two heritable cell types illuminate the circuit governing their differentiation. PLoS Genet 6: e1001070.
  46. 46. Bird AJ, Gordon M, Eide DJ, Winge DR (2006) Repression of ADH1 and ADH3 during zinc deficiency by Zap1-induced intergenic RNA transcripts. EMBO J 25: 5726–5734.
  47. 47. Churchman LS, Weissman JS (2011) Nascent transcript sequencing visualizes transcription at nucleotide resolution. Nature 469: 368–373.
  48. 48. Arigo JT, Eyler DE, Carroll KL, Corden JL (2006) Termination of cryptic unstable transcripts is directed by yeast RNA-binding proteins Nrd1 and Nab3. Mol Cell 23: 841–851.
  49. 49. Wang Z, Gerstein M, Snyder M (2009) RNA-Seq: a revolutionary tool for transcriptomics. Nat Rev Genet 10: 57–63.
  50. 50. Batzoglou S, Jaffe DB, Stanley K, Butler J, Gnerre S, et al. (2002) ARACHNE: a whole-genome shotgun assembler. Genome Res 12: 177–189.
  51. 51. Pinheiro M, Afreixo V, Moura G, Freitas A, Santos MA, et al. (2006) Statistical, computational and visualization methodologies to unveil gene primary structure features. Methods Inf Med 45: 163–168.
  52. 52. Souciet JL, Dujon B, Gaillardin C, Johnston M, Baret PV, et al. (2009) Comparative genomics of protoploid Saccharomycetaceae. Genome Res 19: 1696–1709.
  53. 53. Fitzpatrick DA, Logue ME, Stajich JE, Butler G (2006) A fungal phylogeny based on 42 complete genomes derived from supertree and combined gene analysis. BMC Evol Biol 6: 99.
  54. 54. Byrne KP, Wolfe KH (2005) The Yeast Gene Order Browser: combining curated homology and syntenic context reveals gene fate in polyploid species. Genome Res 15: 1456–1461.
  55. 55. Byrne KP, Wolfe KH (2006) Visualizing syntenic relationships among the hemiascomycetes with the Yeast Gene Order Browser. Nucleic Acids Res 34: D452–D455.
  56. 56. Edgar RC (2004) MUSCLE: a multiple sequence alignment method with reduced time and space complexity. BMC Bioinformatics 5: 113.
  57. 57. Edgar RC (2004) MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res 32: 1792–1797.
  58. 58. Stamatakis A (2006) RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics 22: 2688–2690.
  59. 59. Tatusov RL, Koonin EV, Lipman DJ (1997) A genomic perspective on protein families. Science 278: 631–637.