The purine salvage enzyme inosine 5'-monophosphate (IMP)-specific 5'-nucleotidase catalyzes degradation of IMP to inosine. Although this enzymatic activity has been purified and characterized in Saccharomyces cerevisiae, the gene encoding IMP 5'-nucleotidase had not been identified.
Mass spectrometry analysis of several peptides of this enzyme purified from yeast allowed identification of the corresponding gene as YOR155c, an open reading frame of unknown function, renamed ISN1. The deduced Isn1p sequence was clearly not homologous to 5'-nucleotidases from other species. However, significant similarities to Isn1p were found in proteins of unknown function from Neurospora crassa, Plasmodium falciparum and several yeast species. Knock-out of ISN1 resulted in the total loss of IMP-specific 5'-nucleotidase activity, thus confirming that the ISN1 gene indeed encodes the enzymatic activity purified from yeast. In vivo studies revealed that, when IMP is overproduced through constitutive activation of the IMP de novo synthesis pathway, ISN1 is required for excretion of inosine and hypoxanthine in the medium.
We have identified a new yeast gene, ISN1 (YOR155c), as encoding IMP-specific 5'-nucleotidase activity. The ISN1 gene defines a new type of 5'-nucleotidase which was demonstrated to be functional in vivo.
Purine salvage pathway allows interconversion of bases, nucleosides and nucleotides. Importance of this pathway in humans has been clearly established since several defects in purine salvage enzymes have been associated to pathologies such as mental retardation and severe immunodeficiencies [1,2]. In yeast, mutations in several purine salvage genes lead to deregulation of AMP biosynthesis  and purine excretion. Taking advantage of complete genome sequencing of Saccharomyces cerevisiae we were able to identify several new purine salvage genes such as adenosine kinase or purine nucleoside phosphorylase [4,5] which are highly homologous to their mammalian counterpart. Strikingly, no 5'-nucleotidase encoding gene could be inferred from comparison of the completely sequenced yeast genome with 5'-nucleotidase genes from other species, although several lines of evidence argued for the presence of 5'-nucleotidase in yeast. First, an IMP-specific 5'-nucleotidase activity was detected and purified from yeast . Second, deregulation of IMP biosynthesis resulted in increased production of IMP which was excreted in the medium as a mix of inosine and hypoxanthine , thus establishing that, in vivo, IMP could be efficiently degraded to inosine in yeast.
In this paper, we report the identification of the yeast IMP-specific 5'-Nucleotidase (ISN1) gene encoding a new type of 5'-nucleotidase.
Results and discussion
IMP-specific 5'-nucleotidase corresponds to Yor155p
The IMP-specific 5'-nucleotidase purified as described previously  was submitted to EndoLysC proteolytic treatment. The MALDI-MS spectrum of the peptide mixture generated by EndoLysC proteolysis revealed 9 major peaks in the m/z 1000–3000 range, that were used for data base search (Table 1). YOR155c, an open reading frame of unknown function (Swissprot/TrEMBL accession number Q99312) was identified with 8 peptide matches within an 0.1 Da error range (MS-Fit search, see Experimental). Identification was further confirmed by an independent data base search (MS-Tag) with PSD fragment ions obtained from a precursor ion measured at 1845.91 Da (calculated mass 1845.98 Da, matching the 318–334 stretch of the protein, Table 1). The unmatched peptide ion, measured at 1479.73 Da, was submitted to PSD analysis (Figure 1) and assigned to the N-terminal peptide SSRYRVEYHLK having undergone methionine excision and bearing an acetyl blocking group (calculated mass of the modified peptide 1479.77 Da). Based on mass spectrometry identification and further results presented below, YOR155c was renamed ISN1 (
Figure 1. Partial spectrum of PSD fragment ions obtained from m/z 1479.76 precursor ion. Masses of N-terminal ion series of the b and a type  are in agreement with the modified sequence Acetyl-SSRYRVEYHK. His and Tyr immonium ions, as well as C-terminal ions (y type) confirm the sequence of the modified peptide.
Table 1. Assignment of major peptide ions observed in the MALDI-ToF mass spectrum of the EndoLysC protein digest Mass values are monoisotopic for [M+H]+ ions. For the N-terminal peptide, the calculated mass takes into account a modification by methionine excision (- 131.04 Da) and acylation (+ 42.01 Da), the theoretical mass of the unmodified peptide being 1568.80 Da.
ISN1 encodes an atypical 5'-nucleotidase
Search for conserved motif in S. cerevisiae Isn1p revealed a typical leucine zipper sequence (Fig. 2) that could be involved in multimerization of the protein which was first isolated as a tetramer . Indeed, Yor155p was found to interact with itself in a systematic two hybrid analysis . The sequence analysis of the ISN1 gene product revealed that it does not have significant homologies with any 5'-nucleotidase from other organisms neither eucaryotes or prokaryotes nor to the S. cerevisiae pyrimidine 5'-nucleotidase encoded by the SDT1 gene . The only significant similarities were found with proteins of unknown function derived from complete genome sequencing of Schizosaccharomyces pombe, Neurospora crassa and Plasmodium falciparum (Fig. 2) as well as partially sequenced open reading frames of Saccharomyces bayanus, Saccharomyces exiguus, Kluyveromyces lactis, Pichia angusta, Pichia sorbitophila (Table 2).
Figure 2. Sequence comparison of Isn1p and putative proteins from various species. The amino acid sequence deduced from the nucleotide sequence of the YOR155c ORF (Isn1p) from S. cerevisiae (Sc) was compared with putative proteins derived from complete genome sequencing of Plasmidium falciparum (Pf) Schizosaccharomyces pombe (Sp) and Neurospora crassa (Nc). Accession numbers are presented in Table 2. Residues conserved in all four proteins are shown by asterisks. Dashes indicate the gaps created for alignment. The putative leucine zipper sequence in Isn1p is highlighted in grey.
Table 2. Sequence comparison between Isn1p and similar proteins from various species.
It therefore seems that IMP degradation in these species occurs through an atypical enzyme. Strikingly, the other yeast IMP metabolizing enzymes such as IMP dehydrogenase and adenylosuccinate synthetase are highly similar to bacterial and mammalian enzymes [15,16]. Interestingly, hypoxanthine-guanine phosphoribosyl transferase from yeast is also poorly conserved  except among yeast species, while other purine salvage enzymes such as adenosine kinase or purine nucleoside phosphorylase are highly conserved from yeast to mammals [4,5]. It therefore clearly appears that some purine salvage enzymes are derived from a common ancestor with mammalian enzymes, while others have evolved differently in yeast and mammals. This intriguing observation could be useful for future development of specific antifungal drugs.
Activities of 5'-nucleotidase and non-specific phosphatase in yeast cell extracts
To evaluate the role of Isn1p in IMP-specific 5'-nucleotidase activity, the enzymatic activity was measured in extracts from a wild-type strain and an isogenic isn1 knock-out mutant. IMP-hydrolyzing activity in the extract from wild-type strain was ca 17 × 10-3 units/mg protein, while the activity in the extract from the mutant strain was virtually undetectable (Fig. 3). The hydrolyzing-activities with phenylphosphate, which is not a substrate for IMP 5'-nucleotidase , were ca 12 × 10-3 units/mg for the knock-out strain and ca 11 × 10-3 units/mg for the wild-type strain (Fig. 3.). Similar results were obtained in experiments assessing phosphomonoesterase activities with various phosphomonoesters as substrates (Table 3). IMP-hydrolyzing activity in the crude extracts from the isn1 knock-out strain was lower than 3% of that in the wild-type strain. Activities with each tested phosphomonoester other than IMP were almost equal in both strains.
Figure 3. IMP 5'-nucleotidase activity is undetectable in an isn1 mutant. Time course of hydrolysis of IMP (circles) or phenylphosphate (squares) in crude extracts from the isn1 knock-out strain (A) or wild-type strain (B). Reaction mixtures contained 10 mM IMP or phenylphosphate, 25 mM MgCl2, and 0.1% BSA in 100 mM imidazole/HCl (pH 6.5). The amounts of protein in the cell extracts used in these experiments were 270 μg for both isn1 knock-out and wild type strains.
Table 3. Specific activities in cell extracts with various phosphomonoesters Each reaction mixture contained 10 mM substrate, 25 mM MgCl2, 0.1% BSA in 100 mM imidazole/HCl (pH 6.5)
These results clearly show that in the absence of the ISN1 gene, the IMP-specific 5'-nucleotidase activity is absent. Together with identification of the purified protein by mass spectrometry, this result clearly establishes that ISN1 encodes yeast IMP-specific 5'-nucleotidase.
Role of Isn1p in vivo
Because of its high substrate specificity, Isn1p most probably regulates intracellular IMP amounts. To test the role of Isn1p in vivo, we took advantage of mutants in the ADE4 gene increasing the flux in the IMP biosynthesis pathway . In these mutants, hypoxanthine and inosine are excreted in the growth medium, most probably as a result of degradation of overproduced IMP . We therefore used such an IMP overproducing mutant to assay whether IMP degradation would be affected in an isn1 mutant strain. A plasmid (P2047) overexpressing a dominant mutant form of ADE4 leading to purine excretion  was transformed in wild-type and isn1 yeast strains and purine excretion was assayed. Indeed, crossfeeding experiments revealed that hypoxanthine excretion (allowing growth of a halo of hypoxanthine auxotrophic cells) was much lower in an isn1 mutant strain (Fig. 4A). Analysis of the purine compounds excreted in the growth medium by isogenic wild-type and isn1 mutant strains revealed very different excretion patterns (Fig. 4B). While the P2047-transformed wild-type strain excreted inosine, hypoxanthine and also a significant amount of IMP (Fig. 4B) (compared to the non-excreting control, not shown), in the P2047-transformed mutant strain, IMP only was still efficiently excreted, whereas neither inosine nor hypoxanthine were detected in the growth medium (Fig. 4B). This result clearly shows that Isn1p plays a critical role in IMP degradation in vivo.
Figure 4. Excretion of purine compounds in the isn1 mutant strain. (A.) Purine excretion phenotype of the wild-type strain (BY4741) and the isn1 mutant strain transformed with the P2047 plasmid leading to IMP overproduction. Transformed strains were spotted on a lawn of ade1 homozygous diploid cells plated on purine-free SD casa medium. Purine excretion was monitored after 3 days at 30°C. (B.) HPLC analysis of excreted purine compounds. Transformed strains as in (A) were grown in adenine- and uracil-free SD casa medium. Cells were then harvested and the medium was filtered. Separation of purine compounds was achieved by HPLC and monitored by following absorbance at 260 nm. Specific peaks were identified as IMP, hypoxanthine and inosine as described previously . Arrows indicate the identified peaks with the following abbreviations: Hyp, hypoxanthine; Ino, inosine.
Why should yeast have a specific IMP 5'-nucleotidase? One possibility is that Isn1p could be involved in scavenging IMP toxic derivatives or analogs. Our attempts to document such an effect were unsuccessful : the isn1 knock-out mutant, grown in the presence of several purine analogs (8-azaadenine, 6-chloropurine, 8-azaguanine, 6-mercaptopurine, 2,6-diaminopurine...), did not show any growth alteration compared to the isogenic wild-type strain (data not shown). Alternatively, Isn1p could play some role in DNA repair. Indeed, transcriptome analysis revealed that ISN1 transcription increased three folds when cells were treated with methyl methanesulfonate, a DNA damaging agent , and furthermore, Isn1p co-purified with Mlh1p , a protein involved in mismatch repair . It is therefore tempting to propose that Isn1p could for example be involved in removing dIMP residues resulting from deamination of dAMP. However, while the mutagenic effect of dIMP has been documented , this deoxynucleotide is a poor substrate for Isn1p in vitro . Whatever its involvement in DNA damage response, it is clear that Isn1p activity is present in normally growing cells . This result together with the extreme substrate specificity of Isn1p argues for a physiological role of Isn1p in IMP metabolism. Deregulation of IMP production leads to excretion of inosine and hypoxanthine , suggesting that IMP should not normally accumulate, and we have shown (Fig. 4) that Isn1p is clearly required for this process. However, overproduction of IMP in an isn1 mutant strain did not result in any obvious growth phenotype that would help us to further understand why yeast cells have developped such an efficient way of degrading IMP to inosine and hypoxanthine. Therefore, while the enzymatic activity of Isn1p is well characterized, its physiological role still remains unclear.
In this paper we report the identification of a new yeast gene, ISN1 (YOR155c), as encoding IMP-specific 5'-nucleotidase activity. The ISN1 gene is the first member of a new type of 5'-nucleotidase which appears conserved among several species, from yeast to human malaria parasite. IMP-specific 5'-nucleotidase was found to be functional in vivo although no clear physiological role could be associated to Isn1p activity. Beside future physiological work, identification of Isn1p will make possible structural analysis required to further characterize this new type of nucleotidase.
Purification of IMP-specific 5'-nucleotidase for amino acid sequence analyses
Yeast IMP-specific 5'-nucleotidase was purified as described previously  and run on a 15% one-dimensional SDS-PAGE . Fixation and staining of proteins in the gel were achieved by a two hours treatment with 0.003% Amido Black in 45% ethanol, 10% acetic acid. The gel was then washed several times with water to remove staining excess and the band of interest was cut out and washed in a large volume of water to remove acetic acid and ethanol.
In-Gel Protein Digestion
The purified 5'-nucleotidase protein band was washed three times with 250 μl of 50% acetonitrile in 25 mM Tris-HCl (pH 8.6) for 30 min at room temperature with gentle agitation. The band was then sliced into small pieces (~1 × 1 × 0.75 mm) and partially dried under vacuum in a SpeedVac concentrator. The gel was rehydrated in digestion buffer (0.1 M Tris-HCl, pH 8.6, 10% acetonitrile) containing 5 μg/mL of endoproteinase Lys-C (Roche Applied Science). Digestion was carried out for 16 h at 37°C in a minimal volume allowing total immersion of gel pieces and was stopped by addition of 1% (final concentration) aqueous trifluoroacetic acid (TFA). Peptides recovered from the digestion mix and from the gel pieces through three cycles of extraction with 60% acetonitrile, 0.1%TFA in water, were then pooled and concentrated in a Speed-Vac. Finally, a micro-chromatographic separation step of proteolytic digest was performed by means of a C18 ZipTip (Millipore), using a 10 μl sample load. The elution was achieved with 2 μl of 70% acetonitrile, 0.1% TFA in water
The matrix assisted laser desorption-ionization (MALDI) mass spectrometer was a Reflex III (Bruker) instrument equipped with a 337 nm laser source, ion gate and ion reflector suitable for Post Source Decay (PSD) experiments. α-Cyano-4-hydroxy-cinnamic acid (Sigma) was used as a matrix (saturated solution in 50% acetonitrile, 0.1% aqueous TFA). The dried droplet method was used for sample loading on stainless steel targets. External mass calibration was achieved with a mixture of eight peptides (angiotensin, bradikynin, bombesin, ACTH clips and oxidized insulin B chain, all from Sigma) covering the m/z 1000–3500 range. For PSD analysis, the reflector voltage was stepped down in 10 to 12 steps, starting from 30 kV, in order to collect fragment ions from the precursor down to immonium ions. Data base search was effected by means of the Protein Prospector MS-Fit ans MS-Tag search engines http://falcon.ludwig.ucl.ac.uk webcite, using a mass tolerance of 0.2 Da for precursor ions, and 0.5 Da for PSD fragment ions.
Yeast strains and media
Yeast strains used in this study were purchased from Euroscarf.
BY4741: MATa ura3Δ0 leu2Δ0 met15Δ0 his3Δ1
Y02411: MATa ura3Δ0 leu2Δ0 met15Δ0 his3Δ yor155c::KanMX4
SD is 0.17% nitrogen base, 2% glucose and 0.5% ammonium sulfate. SD casa is SD supplemented with 0.2% casamino acids (Difco).
The P2047 ADE4 mutant plasmid was constructed as follows : the mutant ADE4 coding sequence was amplified from the BRA11-2  genomic DNA using oligonucleotides 48 and 429.
48 : 5' CGCGGATCCAAATGTGTGGTATTTTAG 3'
429 : 5' AAACTGCAGTCAATAATCTGCACAATTATATAATC 3'
The resulting PCR fragment was digested with BamHI and PstI and introduced in the pCM189 vector  digested with BamHI and PstI. The resulting plasmid allows expression of the ADE4 mutant gene under control of a tetracyclin regulatable promoter.
Preparation of cell extracts for enzyme assay
Harvested yeast was suspended in ca 3 vol of Tris/HCl buffer (pH 7.4) containing 1 mM EDTA and 10 mM 2-mercaptoethanol (Buffer A). The cells were disrupted by vigorous shaking for 3 min with glass beads (ca 1 mm diameter) in a Pyrex test tube. The resulting solution was centrifuged at 100,000 g for 60 min. Supernatant was thoroughly dialyzed against ca 200 vol of buffer A overnight.
Phosphomonoesterases were assayed colorimetrically by measuring inorganic phosphate (Pi) released from various phosphomonoesters using the methods by Chen et al. . Assays were performed at 37°C. The reaction mixture contained 100 mM imidazole/HCl buffer (pH 6.5), 25 mM MgCl2, 0.1% BSA, 10 mM phosphomonoester substrate and appropriate amount of cell extracts. For IMP-specific 5'-nucleotidase assay IMP was used as a substrate. For assay of non-specific phosphatases, phenylphosphate or 1-glycerophosphate was used as a substrate. One unit of the enzyme activity corresponds to the release of 1 μmol of Pi/min from the substrate. Protein concentration was determined by the method of Lowry et al.  using crystalline BSA as a standard.
HPLC analysis of excreted purine compounds
Wild type and isn1 strains transformed with P2047 were grown in adenine and uracil free SD casa medium. Cells were then harvested and the medium was filtered on 0.22 μm filter units. Separation of purine compounds of the medium was achieved by HPLC on a Supelcosil LC-18 5 μm reversed phase column using a step gradient set up with A buffer (0.01 M KH2PO4) and B buffer (20% A buffer – 80% methanol). The following proportions of A and B buffers, respectively (indicated in parentheses) were used at the indicated run time : 0 min (97/3), 13 min (89/11), 17 min (75/25), 19 min (30/70) and 27 min (97/3) and the flow rate was 1 ml/min. Excreted purine compounds separation was monitored by following absorbance at 260 nm at the column end and the different peaks obtained were identified by comparison with the retention time of known standards.
RI carried out Isn1p purification and enzymatic activities studies. CSM did the in vivo and HPLC studies. SC prepared the sample for mass spectrometry analysis. RK contributed to enzymatic activities studies. JMS did the mass spectrometry analysis and computer work to identify YOR155c. BDF designed the study, did the sequence alignements and organized the manuscript.
We thank F. Borne for construction of the P2047 plasmid, C. Desgranges and I. Belloc for their help with the HPLC experiments, A. Breton and B. Pinson for discussions. This work was supported by grants to CNRS UMR 5095.
Nature 1970, 227:680-685. PubMed Abstract
Biomed Mass Spectrom 1984, 11:601. PubMed Abstract
Eur J Biochem 1996, 239:487-493. PubMed Abstract
Ho Y, Gruhler A, Heilbut A, Bader GD, Moore L, Adams SL, Millar A, Taylor P, Bennett K, Boutilier K, et al.: Systematic identification of protein complexes in Saccharomyces cerevisiae by mass spectrometry.
Mol Cell Biol 1994, 14:407-415. PubMed Abstract