The Gene YALI0E20207g from Yarrowia lipolytica Encodes an N-Acetylglucosamine Kinase Implicated in the Regulated Expression of the Genes from the N-Acetylglucosamine Assimilatory Pathway

The non-conventional yeast Yarrowia lipolytica possesses an ORF, YALI0E20207g, which encodes a protein with an amino acid sequence similar to hexokinases from different organisms. We have cloned that gene and determined several enzymatic properties of its encoded protein showing that it is an N-acetylglucosamine (NAGA) kinase. This conclusion was supported by the lack of growth in NAGA of a strain carrying a YALI0E20207g deletion. We named this gene YlNAG5. Expression of YlNAG5 as well as that of the genes encoding the enzymes of the NAGA catabolic pathway—identified by a BLAST search—was induced by this sugar. Deletion of YlNAG5 rendered that expression independent of the presence of NAGA in the medium and reintroduction of the gene restored the inducibility, indicating that YlNag5 participates in the transcriptional regulation of the NAGA assimilatory pathway genes. Expression of YlNAG5 was increased during sporulation and homozygous Ylnag5/Ylnag5 diploid strains sporulated very poorly as compared with a wild type isogenic control strain pointing to a participation of the protein in the process. Overexpression of YlNAG5 allowed growth in glucose of an Ylhxk1glk1 double mutant and produced, in a wild type background, aberrant morphologies in different media. Expression of the gene in a Saccharomyces cerevisiae hxk1 hxk2 glk1 triple mutant restored ability to grow in glucose.


Introduction
Hexose kinases initiate the intracellular sugar metabolism in a variety of organisms. In Saccharomyces cerevisiae three glucose phosphorylating enzymes encoded by the genes HXK1, HXK2 and GLK1 have been characterized. The first two encode typical hexokinases while the third one encodes a glucokinase [1,2]. The lack of growth in glucose of S. cerevisiae triple mutants hxk1 hxk2 glk1 indicates that this organism does not possess other enzymes that phosphorylate this sugar [3]. In S. cerevisiae the gene HXK2 is expressed at high levels during growth in glucose while HXK1 and GLK1 are repressed making Hxk2 the important enzyme for glucose metabolism [4]. Hxk2 also acts as a moonlighting protein regulating the expression of some genes subjected to catabolite repression [5,6]. Hexokinases and glucokinases have also been shown to participate in signalling pathways in a variety of organisms [7,8].
Different yeast species exhibit diverse glucose phosphorylating equipments: in Kluyveromyyces lactis an hexokinase [9] and a low activity glucokinase are present [10], in Schizosaccharomyces pombe there are only two hexokinases [11] while Hansenula polymorpha [12] or Yarrowia lipolytica [13,14] have both an hexokinase and a glucokinase. However, in Y. lipolytica the glucokinase activity accounts for about 80% of the glucose phosphorylating activity during growth in this sugar [14]. Y. lipolytica is a strictly aerobic, dimorphic yeast that separated early from the common yeast evolutionary trunk and is distantly related to other ascomycetous yeasts [15,16]. It is receiving increased attention both in basic and applied research due to a series of particular properties. From a basic point of view it has been used to study protein secretion [17], peroxisome biogenesis [18], dimorphism [19] and mitochondrial complexes [20]. Important differences with the model yeast S. cerevisiae have been shown in some regulatory properties of glycolytic enzymes [21,22], or in the transcription of certain glucose repressed genes [23,24]. Also telomeric proteins present in other yeast species are absent in Y. lipolytica [25]. From a biotechnological point of view this yeast is important in the production of heterologous proteins [26] organic acids [27] or novel biofuels [28,29].
During a study of the Y. lipolytica hexose kinases, we found in a comparative BLAST analysis that Y. lipolytica possesses a putative protein with sequence similarity with a plethora of hexokinases from different origins. The gene encoding it is YALI0E20207g and it appeared of interest to elucidate its function as it could reveal the existence of a kinase missed in conventional tests as it occurred for the glucokinase of K. lactis that allows growth of this yeast in glucose with a doubling time of 30 hours [10]. We have cloned the gene YALI0E20207g and biochemically characterized its encoded protein. In this work we present biochemical and genetic evidence showing that the gene encodes an N-acetylglucosamine (NAGA) kinase whose sequence does not show marked similarity with NAGA kinases from other organisms. Expression of the gene under the control of the YlTEF1 promoter allowed growth in glucose of a Ylhxk1glk1 double mutant of Y. lipolytica. We also present results showing that disruption of YALI0E20207g abolishes growth in NAGA, hinders sporulation, and causes derepression of the genes encoding the enzymes of the NAGA assimilatory pathway while its overexpression affects morphology in different media.

Yeast strains and culture conditions
The Y. lipolytica and S. cerevisiae strains used in this work are shown in Table 1 and S1 Table respectively. Yeasts were cultured at 30°C in a synthetic medium with 0.17% yeast nitrogen base (Difco, Detroit, MI) 0.5% ammonium sulphate and glucose, fructose, mannose, NAGA, or ethanol at 2% or glycerol at 3% as carbon sources. Liquid cultures were shaken in a girotory shaker at 180 rpm. For plates 2% agar was added. Auxotrophic requirements were added at a final concentration of 20 μg/ml. Transformation was done using the lithium acetate method as described by Barth and Gaillardin [30] for Y. lipolytica and by Ito et al. [31] for S. cerevisiae. Growth was followed measuring optical density at 660 nm. Mating, sporulation of diploid strains and spore staining by malachite green were carried out as in Flores et al. [14].

Nucleic acid manipulations, plasmid constructions and RT-qPCR
Recombinant DNA manipulations were done by standard techniques. Genomic DNA was obtained as described in Hoffman and Winston [32]. Total RNA from the different strains of Y. lipolytica was obtained using the Speedtools total RNA extraction kit from Biotools B&M Labs S.A.(Spain) The transcription initiation site of YALI0E20207g was determined using a RLM-RACE reaction with the First Choice RLM-RACE kit from Ambion (Life Technologies). To express YALI0E20207g in yeasts a 1400 bp piece of DNA comprising 3 bp upstream of the first ATG and 8 bp after the TAG termination codon was isolated by PCR using Y. lipolytica DNA as template and oligonucleotides 5´-ATATGTCCATGGGAGATGACG and 5´-TATCTACTGT-GAAAGCTGGCT. This fragment and all subsequent PCR products were sequenced from both strands. The YALI0E20207g DNA fragment was cloned into plasmid pGEM-T Easy (Promega Biotech Iberica, Spain) to give plasmid pCL148, excised with NotI and ligated into plasmids pCL49 [23] or pCL49L [14] both carrying the YlTEF1 promoter, the YlXPR2 terminator, and URA3 or LEU2 respectively as markers for expression in Y. lipolytica, or in plasmid pDB20 [33] for expression in S. cerevisiae. The resulting plasmids were named pCL149, pCL149L and pCL150 respectively.
For RT-qPCR the quality of RNA was checked using the Agilent 2100 Bioanalyzer. Only RNAs with a RIN > 9 were used. Non characterized putative genes of the NAGA utilization pathway were identified from the Génolevures database (http://cbi.labri.fr/Genolevures/index. php, [34]) using a BLAST search. The primers used for RT-qPCR are shown in S2 Table. All of them were checked for specificity using the Primer-BLAST from NCBI against the Y. lipolytica CLIB122 genomic sequence. Total RNA was reverse-transcribed into cDNA using the High Capacity RNA-to-cDNA Kit (Applied Biosystems). The cDNA levels were then determined using a LightCycler 480 from Roche and the LightCycler 480 SYBR Green I Master mix (Roche) with each primer at 250 nM. Each sample was tested in triplicate. After completion of the RT-qPCR melting-curve data were collected to verify PCR specificity, the absence of contamination and primer dimers. Transcripts of the gene YALI0F27533g (ARP4) were used to normalize the data [14].

Disruption of YALI0E20207g
Plasmid pCL148 was cut with StuI and XhoI and blunt ended. This digestion eliminates a fragment of 367 bp internal to the YALI0E20207g ORF starting 364 bp after the initial ATG. Plasmid pINA62 [35] was cut with NcoI and a fragment of 2.1kb containing the YlLEU2 gene was excised, blunt ended and ligated with the previous digestion product. The 3.1kb NotI fragment of the resulting construct was used to disrupt the chromosomal copy of YALI0E20207g in strain PO1a. Disruption was checked by PCR and Southern analysis.

Purification of YlNAGA-kinase expressed in Escherichia coli
Plasmid pCL148 was cut with EcoRI and the 1400 bp fragment containing gene YALI0E20207g was ligated into plasmid pGEX-4T-2 (GE Healthcare Life Sciences) cut with the same enzyme. E. coli BL21(DE3) transformed with the resulting plasmid, pCL151, was grown in LB to an OD of 0.5 at 660 nm and then induced with 1mM IPTG for 2 hours and cells were harvested by centrifugation at 12000 x g. Extracts were done by suspending cells in 50 mM phosphate buffer, 1 mM EDTA, 1 mM PMSF, 1mM DTT and 0.1 mM NAGA pH 7.6 and 0.12 mg/ml lysozyme. The suspension was incubated at 37°C for 10 min and centrifuged for 20 min at 27200 x g at 4°C. The supernatant was bound to a Glutathione Sepharose 4B column (GE-Healthcare) and the recombinant protein was eluted with 50 mM glutathione in the buffer described above.

Enzymatic assays
All assays were carried out at 30°C. Phosphorylation of NAGA, glucose, fructose or mannose by purified protein preparations was assayed spectrophotometrically at 340 nm coupling the production of ADP to the oxidation of NADH in 0.1 M Tris-HCl pH 7.5, 2 mM ATP-Mg, 0.3 mM NADH, 5 mM PEP, and excess pyruvate kinase and lactate dehydrogenase. When Km values were determined the concentration of sugars was varied as needed. When yeast extracts were used, the phosphorylation of glucose, fructose or mannose was assayed by a spectrophotometric assay with glucose-6-P dehydrogenase, phosphoglucose isomerase and phosphomannose isomerase as required [1].Yeast cell free extracts were prepared by breaking the yeast with glass beads in 50 mM phosphate buffer, 1 mM EDTA, 1mM PMSF, 1 mM DTT and 0.1 mM NAGA pH 7.6 in five cycles of 1 min of vortexing and 1 min on ice. The extract was centrifuged at 4°C for 15 min at 20000 x g and the supernatant used for determination of enzyme activities. Apparent Km values were obtained using the Eadie Hofstee or the Lineweaver-Burk plots. Data for all substrates were obtained using at least five different substrate concentrations. Protein was assayed with the commercial BCA protein assay kit (Pierce).

Measurement of glucose utilization
Cells from a glycerol grown culture were resuspended at 1.5 mg dry weigth / ml in 0.17% YNB without nitrogen source with 25mM glucose and shaken at 30°C. Samples were taken along time; centrifuged one minute at 20000 x g and the supernatant was used for glucose determination using a spectrophotometric assay with NADP, hexokinase and glucose-6-P dehydrogenase.

Calcofluor staining
Yeasts were collected, washed with water, suspended in calcofluor (1mg/ml) and left in the dark for 10 minutes. Then they were washed twice with water and used for microscopic examination.

Phylogenetic analysis
Phylogenetic analysis was done using the MEGA6 program [36,37]. Homologous protein sequences were obtained using NCBI protein blast (http://blast.ncbi.nlm.nih.gov/Blast.cgi). The query coverage was established at > 70% and the E value cut off at 4e -25. Sequences were aligned using MUSCLE [38]. Selection of the best-fit model of amino acid replacement for the phylogenetic analysis was done using ProtTest [39]. The LG+G+I+F model [40] was selected by this software as the best model based on the Akaike information criterion. The maximum likelihood tree was estimated using the MEGA6 software with the LG+G+I+F model. The evolutionary analysis involved 36 amino acid sequences. All positions with less than 95% site coverage were eliminated. There were a total of 406 positions in the final dataset. A bootstrap analysis with 1000 replicates was carried out.

Results
Expression of YALI0E20207g in a Y. lipolytica mutant lacking glucose kinases allows growth in glucose and mannose Disruption of the genes encoding glucokinase and hexokinase in Y. lipolytica results in inability to grow in glucose, fructose or mannose (Fig 1). This result indicates that in this yeast these enzymes are the only ones that allow a significant growth in those sugars. However a BLAST search of the genomic sequence of Y. lipolytica revealed the existence of a gene, YALI0E20207g that could encode a protein with amino acid sequence similarity to yeast hexokinases and glucokinases. To explore the function of this protein, DNA corresponding to its coding sequence was obtained by PCR, cloned, and expressed in an Ylhxk1 glk1 double mutant of Y. lipolytica. Since YALI0E20207g presents in its 5´region several putative initiation codons we performed before the PCR a RLM-RACE reaction to determine the initiation site of transcription that allowed us to infer the N-terminal sequence of the protein as MSMGDDDRHYHHQMS. When YALI0E20207g was expressed under the control of the YlTEF1 promoter in an Ylhxk1 glk1 mutant, growth on glucose or mannose but not on fructose was observed (Fig 1). A similar restoration of growth capacity was seen when YALI0E20207g was expressed in a hxk1 hxk2 glk1 triple mutant of S. cerevisiae unable to grow in those sugars [3] (S1 Fig). Since yeasts glucokinases exhibit activity on glucose and mannose but not on fructose [13,41], the growth phenotypes indicate that YALI0E20207g encodes an enzyme with a sugar specificity similar to that of a glucokinase, that is non operative during growth in glucose.

YALI0E20207g encodes a N-acetyl glucosamine kinase
The double mutant Ylglk1 hxk1 overexpressing YALI0E20207g grew in glucose slower than a Y. lipolytica wild type strain and had a glucose utilization rate about 3 times lower than that of a wild type (Table 2). However, determination of its glucose phosphorylating capacity showed no striking difference with that of a wild type (Table 2). These results suggest that the glucokinase activity of YALI0E20207p in vivo is low, even when the gene is expressed under the control of a strong promoter, likely due to a low affinity for glucose or ATP. Disruption of YALI0E20207g in a wild type strain did not influence its duplication time in glucose or its rate of glucose utilization (Table 2), results consistent with the idea that the main hexose phosphorylating activities responsible for growth in glucose in Y. lipolytica are YlGlk1 and YlHxk1 and that the activity of YALI0E20207p during growth in glucose is low. Determination of the Km for glucose and mannose of the protein YALI0E20207p yielded values of 135 and 67 mM respectively (Table 3), both much higher than those of 0.17 mM for glucose and 1 mM for mannose reported for Y. lipolytica glucokinase [13] consistent with the previous results. No activity could be detected towards fructose even at high concentrations in the assay, a result in accordance with the observed growth phenotype of the strains expressing only YALI0E20207g. The elevated Km values for the sugars assayed suggested that the glucokinase-like activity observed could be due to a marginal activity of a sugar kinase with another physiological function. In some animal tissues, enzymes reported to be glucokinases with a high Km towards glucose turned out to be NAGA kinases with a marginal activity on glucose [42][43][44]. We found that YALI0E20207p showed high activity towards NAGA with a Km of 0.5 mM (Table 3) suggesting that the protein is a NAGA kinase. Since Y. lipolytica grows in NAGA it is possible to obtain physiological support for the conclusion reached by the kinetic studies by checking the growth on NAGA of a strain disrupted in YALI0E202027g. This disruption abolished growth in NAGA and reintroduction of the gene in the disruptant restored the capacity to grow in this sugar (Fig 2). All these results demonstrate that YALI0E20207p is a NAGA kinase. To avoid multiplication of names for genes encoding proteins with the same biochemical function we will name YALI0E20207g as YlNAG5, following the designation given to the NAGA kinase encoding gene in Candida albicans by Yamada-Okabe et al. [45]. An Ylhxk1glk1 mutant grew in NAGA showing that the corresponding proteins do not participate in the utilization of this sugar.
The amino acid sequence of the protein encoded by YlNAG5 differs significantly from that of other NAGA kinases Once established by function analysis that the protein encoded by YlNAG5 is a NAGA kinase it appeared interesting to explore why no significant similarity in amino acid sequence with other NAGA kinases was found in the initial BLAST searches. A comparison of the amino acid sequence of YlNag5 against non-redundant protein sequences databases using a Delta-Blast algorithm produced basically hexokinases or uncharacterized proteins from different origins with about 30% identity and sequence coverages of about 70%. In YlNag5 the hexokinase pfam03727 domain is present. In this domain we found two amino acid stretches, GTGIN and NCEASLF that were conserved in alignments with other hexokinases. The sequence of the first stretch is comprised in the phosphate 2 region of the ATPase domain of sugar kinases described by Bork et al. [46]. Considering that the second stretch is very close to the first one in the primary structure of the protein it is likely that it forms part of this domain too. We introduced changes in these sequences to ascertain their functional importance in YlNag5. Expression in a Y. lipolytica strain nag5::LEU2 of plasmids with the mutated variants GT 266 AGIN or NCE 294 QASLF (confirmed by RT-qPCR) did not restore the ability to grow on NAGA to the mutant strain showing the importance of these regions in YlNag5.
We also performed a sequence similarity search using the FASTA algorithm against the UNIPROTKB/Swiss Prot database but no NAGA kinase was found. When the search was done against the UNIPROT database again the overwhelming majority of the proteins found were hexokinases. Only a sequence annotated as a putative NAGA kinase from the yeast Dekkera bruxellensis [47], also named Brettanomyces bruxellensis, [48] with a 25.6% amino acid identity appeared in the search. We constructed a phylogenetic tree using homologous sequences of proteins from selected organisms from the Pezizomycotina and Saccharomycotina taxa limiting the proteins to those which showed a query coverage >70% and a cut off E value of 4e -25 in a BLAST search (Fig 3 and S2 Fig). The tree showed that YlNag5 appeared close but separated from a group containing other proteins annotated as NAGA kinases (of which only one has been biochemically characterized). Also many proteins annotated as hexokinases or unnamed proteins appeared near YlNag5. For these proteins we have marked their accession number since they have not been functionally characterized as hexokinases. Since organisms from the Pezizomycotina and Saccharomycotina taxa exhibited proteins similar to hexokinase and NAGA-kinase it is suggested that the appearance of the NAGA-kinases preceded the separation of these taxa.
Expression of genes encoding the NAGA assimilatory enzymes is induced by NAGA and becomes constitutive in a Ylnag5 mutant A possible explanation for the lack of growth in glucose of a double Ylglk1 hxk1 mutant in spite of the presence of the chromosomal copy of YlNAG5 could be that the expression of this gene is negligible during growth in this sugar. Therefore we examined the levels of expression of this gene and that of the other genes encoding the enzymes of the pathway of NAGA utilization (Fig 4) during growth in glucose and in NAGA. In addition we determined those levels for the genes encoding the enzymes leading from fructose-6-phosphate to chitin since the important intermediate UDP-NAGA is formed also during catabolism of other sugars. The corresponding genes were identified in the genome of Y. lipolytica by sequence homology using the Génolevures database [34]. As shown in Fig 5 all the genes implicated in the utilization of NAGA were expressed at a very low level during growth in glucose while their expression increased between 20 to 40 times in NAGA grown cultures. A similar behaviour has been reported for the genes NAG1, NAG2/DAC2 and NAG5 in C. albicans [45,49]. The genes encoding proteins of the pathway from fructose-6P to chitin (Fig 5) were expressed at similar levels in glucose or NAGA grown cultures suggesting a comparable need for those enzymes in different culture conditions. We found that a strain with a disrupted YlNAG5 gene grown in glucose showed an expression of all the genes encoding the enzymes for NAGA utilization similar to those found in the wild type grown in NAGA (Fig 5). Expression of YlNAG5 itself followed the same pattern as shown by the values of β-galactosidase expressed from a fusion of the promoter of YlNAG5 to E.coli lacZ measured in a wild type and a disrupted strain (Wild type grown in glucose or ethanol <2 mU/mg protein, grown in NAGA 31 mU/mg protein, ΔYlNAG5 grown in glucose 28, grown in ethanol or acetate 32 mU/mg protein). Reintroduction of YlNAG5 in a Ylnag5 background restored the repression by glucose to the genes of the assimilatory pathway (Fig 5) were obtained from the GenBank (NCBI) database and the names of the genera in the tree from top to bottom are: P, Pyrenophora; N, Neurospora; Y, Yarrowia; Sch, Schizosaccharomyces; T, Tilletaria; Cr, Cryptococcus; B, Bretanomyces (Dekkera); C, Candida; S, Saccharomyces; M, Magnaporthe; A, Aspergillus; K, Kuraishia; D, Dactylellina; Ar, Arthrobotrys.  suggesting that the protein YlNag5 participates in the control of the expression of the genes implicated in the NAGA assimilatory pathway. In C. albicans it has been reported that expression of the genes NGT1 and NAG1 encoding NAGA transport and NAGA deacetylase respectively was higher in a double mutant hxk1/hxk1 than in a wild type grown in glucose or glycerol (NAG5 is referred to as HXK1 in that article) [51]. Disruption of YlNAG5 did not affect the expression of the genes of the pathway from fructose-6P to chitin (Fig 5) indicating that the effect of YlNag5 is restricted to the NAGA utilization pathway. Overexpression of YlNAG5 in a wild type background did not influence repression by glucose of the genes of the NAGA assimilatory pathway but it decreased the levels of expression of those genes on NAGA.
Other effects produced by the disruption or overexpression of YlNAG5 in Y. lipolytica A diploid strain carrying disruption of both copies of YlNAG5 sporulated very poorly as compared with a wild type isogenic diploid (Fig 6). Expression of YlNAG5 in a wild type diploid, assayed by RT-qPCR, showed an increase after transfer of the strain to sporulation medium (1.5 times after 8 days and 2.5 after 14 days) thus suggesting a role for YlNag5 in the sporulation process.
A strain overexpressing YlNAG5 exhibited a longer latency to start growth in NAGA when inoculated from a glucose culture as compared with a wild type (400 vs. 200 minutes). This increased lag may be due to an initial overflow of the pathway leading to a transitory depletion of ATP. Another feature of the Y. lipolytica strain overexpressing YlNAG5 was its altered morphology. Y. lipolytica grows in several media as a mixture of yeast-like cells and short hyphae; however the strain overexpressing YlNAG5 produced cells with elongated or abnormal  morphology in different carbon sources (Fig 7). Calcofluor staining showed no striking differences in chitin accumulation between wild type cells and those overexpressing YlNAG5 (Fig 8). However, changes in the cell wall properties might have occurred in the strain overexpressing YlNAG5 as shown by the more voluminous, less compact sediment produced by this strain as compared with a wild type grown in similar conditions (S3 Fig) Expression of YlNAG5 in S. cerevisiae does not produce catabolite repression of the SUC2 gene Catabolite repression of the gene SUC2 encoding invertase in S. cerevisiae requires the presence of the Hxk2 protein [6]. Since expression of YlNAG5 in a S. cerevisiae hxk1 hxk2 glk1 mutant restored growth in glucose we tested if it also could substitute for Hxk2 in the glucose repression of SUC2. No restoration of catabolite repression of the gene SUC2 was observed in the strain tested. Also growth on glucose of that strain did not occur in the presence of the respiratory inhibitor Antimycin A indicating a decreased glycolytic flux insufficient to sustain fermentation and making respiration necessary for the use of glucose.

Discussion
Evidence from enzymatic and genetic tests showed unequivocally that the gene YALI0E20207g from Y. lipolytica encodes the unique N-acetylglucosamine kinase of this yeast. The Km values for glucose and ATP are in the same range as those reported for NAGA kinases from diverse origins [45,[52][53][54]. The low affinity for glucose of the Y. lipolytica enzyme is also characteristic of mammalian NAGA kinases that were initially described as glucokinases with low glucose affinity [42,43]. The only enzymes described with a similar affinity and V max for NAGA and glucose are the RokA hexokinase from Bacteroides fragilis [55] and the hexokinase from the   archeon Sulfobolus tokadai [56]. No activity on fructose has been reported for NAGA kinases and this was also the case for the protein of Y. lipolytica. The abolition of growth in NAGA in a mutant disrupted in that gene supports the conclusion of the enzymatic tests. We have named the gene YALI0E20207g NAG5 following the nomenclature of Yamada-Okabe et al. [45] for the C. albicans gene and not HXK1 as used in the Candida Genome Database to avoid confusion with the name usually employed to designate hexokinases in different organisms and because HXK1 is already used in Y. lipolytica [13].
It is interesting to notice that the sequences of NAGA kinases from different organisms biochemically characterized as such often fail to show extensive similarity among them [53,54,57,58]. This is also the case of the NAGA kinase of Y. lipolytica that showed more sequence similarity with hexo-or glucokinases than with NAGA kinases of other origins. Omelchenko et al. [59] have proposed the denomination of non-homologous isofunctional enzymes for enzymes that catalyze the same reaction but that do not show detectable sequence similarity; many NAGA kinases appear to fit in this category. From these considerations and the situation in the phylogenetic tree it could be speculated that several proteins that have not been functionally characterized and appear annotated in databases as related to or similar to glucokinase or hexokinase would turn out to be NAGA kinases. Likely evolution from an ancestral, not very specific, sugar kinase originated the branches leading to hexo-gluco kinases and to NAGA kinases. Among the differences between Y. lipolytica and other yeasts is the fact that many proteins from this yeast are more similar to proteins from organisms belonging to Pezizomycotina than to those from other Saccharomycotina [60]. Our results with the sequence of its NAGA kinase agree with this observation.
NAGA is a component of several abundant polysaccharides such as chitin, murein or hyaluronic acid from which it can be derived by hydrolytic enzymes of different organisms. However, the use of NAGA as carbon source is not widespread among yeasts [15]. Alvarez and Konopka [61] reported that the ability to use NAGA as carbon source has been lost in several yeast lineages due to loss of different enzymes of the assimilatory pathway. Expression of the corresponding missing heterologous genes renders S. cerevisiae able to use NAGA [62][63][64].
NAGA kinase is the first intracellular enzyme of NAGA metabolism in Y. lipoytica and also in C. albicans [45] and humans [53]. This contrasts with the situation in E. coli in which the sugar is phosphorylated by the PTS system during transport and where the NAGA kinase function appears restricted to the utilization of internally produced NAGA from the degradation of murein [54]. A recycling role for NAGA from lysosomal degraded glycoproteins or glycolipids appears also as the main function of NAGA kinase in mammalian cells [53].
One of the intermediates of the metabolic pathway of NAGA is glucosamine-6P. In S. cerevisiae this compound is formed from the non-metabolizable glucose analog glucosamine, that enters the cell via the glucose transporters and is phosphorylated by hexokinase [65][66][67] but cannot be further metabolized due to the lack of glucosamine-6-P deaminase. It could be expected that Y. lipolytica would grow in glucosamine since the glucose transport and the hexokinase are present and the glucosamine-6-P deaminase is functional as shown by the growth in NAGA. However glucosamine does not support growth of Y. lipolytica [15]. Different explanations may be offered for this behaviour the most plausible ones being a low level of hexokinase [14], a low level of glucosamine-6-P deaminase in the absence of NAGA (this work) or a strong dependence of the deaminase on its allosteric activator NAGA-6P as is the case in E. coli and other organisms [68,69].
The genes encoding NAG5, NAG1 and NAG2/DAC1, in C. albicans are located contiguously in a cluster in chromosome 6 [45]. Moreover, NAG1 and NAG5 share a common bidirectional promoter [45,49]. Looking for possible orthologous clusters in the genome of the yeast Dekkera bruxellensis Curtin et al. [47] found that in this yeast genes putatively encoding the enzymes Nag5, Nag1 and Nag2, of the NAGA assimilatory pathway were also located nearby in the same chromosome. We have identified in Y. lipolytica the genes of the NAGA assimilatory pathway by sequence homology and found that YlNAG5 and YALI0E20163g (NAG2) are located nearby in chromosome E, separated by YALI0E20185g encoding a putative protein of the glycoside hydrolase 3 family. The other genes encoding proteins of NAGA metabolism were scattered in the Y. lipolytica genome. Taking into account the position of Y. lipolytica in the Hemiascomycetous yeasts lineage [16] and that pairs of genes sharing a promoter are not easily dissociated by recombination events [70] it could be speculated that NAG2 and NAG5 were originally located nearby in the same chromosome and that further rearrangements during evolution have placed NAG1 in association with them in other species.
Our results show that the genes of the NAGA assimilatory pathway in Y. lipolytica are induced by NAGA paralleling the situation in C. albicans [45,49]. The low level of YlNAG5 in the absence of NAGA also explains why a double mutant Ylglk1 hxk1 does not grow in glucose in spite of possessing a NAGA kinase able to phosphorylate glucose. The finding that the disruption of the YlNAG5 gene abolished the need of NAGA in the medium to induce the expression of the genes encoding all enzymes of the NAGA utilization pathway and that expression of YlNAG5 restored their inducibility by NAGA indicates that YlNag5 participates in the control of the expression of the genes of the pathway although the detailed molecular mechanism (s) of its action remains to be elucidated. There are several known instances of enzymatic proteins that inhibit the transcription of their own encoding gene. In S. cerevisiae and K. lactis pyruvate decarboxylase inhibits transcription of the PDC promoters when pyruvate decarboxylase reaches a certain level [71,72]. A case in which a protein acts also as a controller of the expression of the other genes of the pathway is that of the galactokinase from K. lactis. The genes encoding the enzymes of the Leloir pathway are induced by galactose and the galactokinase protein is needed to relieve the inhibitory action of the protein Gal80 thus allowing the function of the activator protein Gal4/Lac9 for transcription to proceed [73]. The behaviour of YlNag5 suggests that it might also behave as a moonlighting protein that participates in the repression of the synthesis of the enzymes of the NAGA utilization pathway. The fact that in C. albicans the expression of the genes NGT1 and NAG1 is increased in the absence of NAGA kinase [51] suggests an important role for Nag5 in the control of the NAGA utilization pathway that has been conserved in distantly related yeast species.
The possible moonlighting role of YlNag5 in Y. lipolytica may be a way to regulate the fate of NAGA-6P an intermediate that arises both in the catabolic pathway of NAGA and in that of UDP-NAGA biosynthesis. Simultaneous functioning of the corresponding acetylation/deacetylation reactions and of deamination/amination (Fig 4) could originate futile cycles with detrimental effects to the cell.
The marked negative effect of the disruption of YlNAG5 on sporulation suggests a role for the protein on the process, an idea supported by the increase in expression of YlNAG5 when a wild type diploid is placed in sporulation medium. We do not have data yet to hypothesize on the mode of action of YlNag5.
The increase in the lag phase of growth of the strain overexpressing YlNAG5 when switched from glucose to NAGA is likely caused by an increased phosphorylation rate that cannot be matched by subsequent reactions to regenerate ATP leading to an initial transitory ATP depletion. In mammals this situation is observed upon a fructose load to the liver; an initial precipitous drop in ATP concentration is followed by a slow phase of recovery that lasts for several hours [74]. Also in S. cerevisiae the loss of the hexokinase inhibition by trehalose-6-phosphate produces a similar effect [75]. The growth inhibition caused by NAGA in different carbon sources in E. coli or C. albicans mutants devoid of NAGA-6P deacetylase or of glucosamine-6P deaminase [76,77] is likely due to the ATP sink effect of NAGA-6P besides other possible effects of this compound in metabolism.
In addition to its utilization as a nutrient NAGA plays a role in cell signalling in different organisms by various mechanisms (for a review see [78]). NAGA has been used as an external trigger of morphological differentiation in dimorphic yeasts [19,78,79]. In the opportunistic pathogenic yeast C. albicans NAGA induces filamentous growth, a process that appears to have drastic consequences for the invasivity of that organism [80,81]. The differentiation process is a complex one and elements from different kinase cascades participate in its regulation [82] although with different roles depending on the organism [19]. Rao et al. [51] found that homozygous hxk1/hxk1 mutants of C. albicans (NAG5 referred to as HXK1) presented filamentous growth in media in which a wild type did not form filaments. Alvarez and Konopka [61] reported that a C. albicans mutant with a deleted NGT1 gene, that encodes a NAGA transporter, could form hyphae when exposed at very elevated NAGA concentrations suggesting the need for internalization of the sugar to exert its signalling effect(s). Naseem et al. [77] using mutants lacking the NAGA catabolic enzymes showed that NAGA induction of morphogenesis is not dependent on its metabolism suggesting that the sugar by itself initiates the signalling pathway (s). The altered morphology of Y. lipolytica strains overexpressing YlNAG5 in different media indicates that additional factors different from NAGA play important roles in morphogenesis. In this context it is worth noting that overexpression of NAGA kinase in rat hippocampal neurons upregulated the number of dendrites and increased dendritic branching [83] independently of its enzymatic activity [84] strongly indicating a moonlighting activity of this protein. The alignment generated by MEGA was converted to PIR format using the Format Converter v2.3.5 from the HIV Sequence Database (http://www.hiv.lanl.gov/content/sequence/FORMAT_ CONVERSION/form.html) and coloured with the ClustalX color code using Jalview (http:// www.jalview.org/). (TIF) S3 Fig. Fluffiness of the sediment of a Y. lipolytica strain overexpressing YlNAG5. Y. lipolytica strains CJM660 (PO1a/pCL49L, void plasmid) and CJM762 (PO1a/pCL149L-YlNAG5) were grown to an optical density of 10 in minimal medium glucose; 5 ml were transferred to test tubes and photographed after 5h and 24h standing at room temperature. A, CJM660; B, CJM762. (TIF) S1 Table. Saccharomyces cerevisiae strains used in this work. a) Strain CJM 864 was originated as follows: Strains with individual disruptions were crossed, the diploids sporulated and spores were isolated by micromanipulation. Double disruptants of adequate mating type were crossed to select strain CJM 864. This strain was then transformed with the indicated plasmids. b) Plasmid p381 is centromeric and expresses YALI0E20207g under the control of the ScMET25 promoter. (DOC) S2 Table. Primers used for the RT-qPCR assays. (DOC)