Genes encoding ribonucleoside hydrolase 1 and 2 from Corynebacterium ammoniagenes

The N-ribosidic cleaving reactions of nucleosides occur through actions of nucleoside phosphorylases (EC 2.4.2.) and nucleoside hydrolases (NHs; EC 3.2.2.) (West, 1988; Camici et al., 1990). Nucleoside phosphorylases catalyse the phosphorolytic cleavage of nucleosides and have a reversible ribosyl transferase activity. These enzymes have been well studied and used as catalysts for the synthesis of nucleoside analogues through base-exchange reactions (Krenitsky et al., 1981; Ling et al., 1990; Hamamoto et al., 1996). In contrast, the NHs mediating the irreversible hydrolysis of nucleosides to ribose and free nucleic bases are widely distributed in bacteria, fungi, protozoa, insects, fish, amphibians and plants, but so far not in mammals. They have been studied extensively in pathogenic protozoa as potential targets for chemotherapy, since pathogenic protozoa lack de novo purine biosynthetic pathways and rely on exogenous purines salvaged from the hosts (Degano et al., 1996; Miles et al., 1999; Shi et al., 1999).

Enzymes responsible for nucleoside salvage are quite different depending on the organism. In Escherichia coli, exogenous ribonucleosides are predominantly metabolized by nucleoside phosphorylases encoded by deoD, udp and xapA, while NHs encoded by rihA, rihB and rihC have been reported and play a minor role (Koszalka et al., 1988; Petersen & Moller, 2001). In the case of the genus Bacillus, the metabolic pathways mediated by only nucleoside phosphorylases have been reported and characterized (Hamamoto et al., 1996; Rocchietti et al., 2004), whereas in the yeast Saccharomyces cerevisiae, purine ribonucleosides, inosine and guanosine, and pyrimidine ribonucleosides, cytidine and uridine, were salvaged by purine nucleoside phosphorylase (encoded by pnp1) and uridine ribohydrolase (encoded by urh1), respectively (Desgranges et al., 2001; Kurtz et al., 2002; Mitterbauer et al., 2002). In addition, NHs, but not nucleoside phosphorylases, are crucial enzymes in purine-pyrimidine salvage in protozoan parasites (Parkin et al., 1991): an inosine-uridine NH and a guanosine-inosine NH from Crithidia fasciculata (Estupinan & Schramm, 1994; Degano et al., 1996), a purine-specific inosine-adenosine-guanosine NH from Trypanosoma brucei subsp. brucei (Parkin, 1996) and a non-specific NH from Leishmania major (Shi et al., 1999).

Corynebacterium ammoniagenes is a Gram-positive micro-organism that has been used for the industrial production of flavour-enhancing purine nucleotides and other compounds (Chung et al., 1996; Koizumi et al., 2000; Noguchi et al., 2003). The biosynthetic pathways leading to nucleotides/nucleosides have been reported by Chung et al. (1996) and Noguchi et al. (2003); however, very little is known about the N-ribosidic cleavage reactions of the salvage pathways in this organism. Although several enzymes of other coryneform bacteria were predicted to be putative NHs during genome sequencing projects, to date their functions have not been verified experimentally.

In this work, we cloned and identified two major genes involved in the salvage of ribonucleosides from C. ammoniagenes ATCC 6872, and designated rih1 and rih2. To investigate the characteristics and physiological function of the NHs, we have determined their substrate specificity and characterized their functions using deletion mutants and gene-bearing plasmids. This is the first report of the identification and characterization of putative NHs in Corynebacterium species.

Bacterial strains and plasmids.
The bacterial strains and plasmids used in this study are listed in Table 1. Wild-type C. ammoniagenes ATCC 6872 was used to provide the DNA template for cloning of ribonucleoside-cleaving genes. E. coli CGSC 6885 and DH5α were employed for the selection of positive clones and general DNA manipulation, respectively. An E. coli/Corynebacterium glutamicum shuttle plasmid, pECCG117, was used for molecular biological work in C. ammoniagenes (Chung et al., 1996).

Table 1. Bacterial strains and plasmids used in this study

Media and culture conditions.
C. ammoniagenes was routinely grown in YPB medium (10 g yeast extract, 10 g peptone, 10 g beef extract, 2·5 g NaCl per litre water at pH 7·2) (Chung et al., 1996). E. coli was cultured in LuriaBertani (LB) medium (Sambrook & Russell, 2001). Nucleoside-cleaving genes were screened on MacConkey agar medium (Difco) supplemented with 2·0 % inosine or uridine. Minimal (M) medium [1 g (NH₄)₂SO₄, 1 g KH₂PO₄, 1 g K₂HPO₄, 0·3 g MgSO₄.7H₂O, 10 mg CaCl₂.2H₂O, 10 mg FeSO₄.7H₂O, 1 mg ZnSO₄.7H₂O, 3·6 mg MnCl₂.4H₂O, 20 mg L-cysteine, 5 mg thiamine-HCl, 10 mg Ca-D-pantothenate, 30 µg biotin per litre water at pH 7·2) containing 10 mM nucleoside as a carbon source was used to analyse the growth patterns of several mutants. If necessary, kanamycin and gentamicin were used at final concentrations of 10 and 0·4 µg ml¹, respectively, in C. ammoniagenes. For E. coli, 100 µg ampicillin ml¹, 50 µg kanamycin ml¹ or 10 µg gentamicin ml¹ was added to the medium. E. coli was grown at 37 °C and C. ammoniagenes was cultured at 32 °C in a 500 ml baffled flask containing 50 ml medium with an agitation speed of 120 r.p.m.

Cloning and sequencing of nucleoside-cleaving genes
Primary screen of nucleoside cleaving genes.
Genomic DNA of C. ammoniagenes ATCC 6872 was isolated by using a genomic DNA purification kit (Qiagen) and partially digested with Sau3AI. DNA fragments ranging from 2 to 7 kb were eluted from an agarose gel and cloned into the BamHI site of pECCG117. These constructs were transformed into E. coli CGSC 6885 by electroporation (Dunican & Shivan, 1989). The deoD- and gsk-defective strain, E. coli CGSC 6885, unable to utilize purine nucleoside as a carbon source, did not show any colour change around the colonies on a MacConkey agar plate supplemented with 2·0 % inosine or uridine, the representative purine and pyrimidine ribonucleoside (Degano et al., 1996; Kurtz et al., 2002). Transformants were incubated overnight at 37 °C on MacConkey agar plates containing 2·0 % inosine or uridine, respectively. The positive clones showed a faint or deep red colour around colonies due to the decrease in pH caused by inosine or uridine metabolism (Peist et al., 1997).

Sequencing and sequence analysis.
Nucleotide sequences of clones were determined by using the dideoxy chain-termination method using double-strained DNA as a template with either universal or synthetic primers (Sambrook & Russell, 2001). The nucleotide and amino acid sequences of the putative ORFs were analysed using the BLAST server of the National Center for Biotechnology Information (). The multiple sequence alignments of related or retrieval enzymes were performed with the CLUSTAL W program ().

Construction of expression and integration plasmids.
To overexpress rih1 and rih2 the genes were subcloned in an E. coli expression vector, pTrc99A. For amplification of rih1, primers A (5'-CATGCCATGGATGAACGCAGATTCCACCTCACC-3') and B (5'-CCCAAGCTTTTAAAAGTTGGTGTTGCC-3') were used, whereas for rih2 primers C (5'-CATGCCATGGATGAAGATGATTTTAGATCTAGA-3') and D (5'-CCCAAGCTTTTACTGGTTTTGGACCAA-3') were chosen. The amplified fragments were double digested with NcoI and HindIII and ligated with NcoI/HindIII-cleaved pTrc99A. The resulting plasmids were designated pTNI12 and pTNU23, respectively.

Plasmid pBSG102 harbouring the sacB gene was used for constructing deletion mutants based on allelic replacement. pBSG102 was constructed as follows. A vector harbouring a gentamicin marker, pBG101 (2·04 kb), was constructed by digesting pBR322 with AhdI and NdeI (1·07 kb fragment), followed by Klenow treatment and ligation with 0·88 kb of Klenow-treated pBRINT-Gm/EagI fragment, which contained gentamicin acetyltransferase (aacC1) and multiple cloning sites derived from pBluescript II KS+ (Balbas et al., 1996). To construct a plasmid containing levansucrase, a DNA fragment of sacB (1·9 kb) was amplified by PCR using Bacillus subtilis Marburg 168 as template and primers E (5'-GATCCTTTTTAACCCATCACATAT-3') and F (5'-TCGTGATGGCAGGTTGGGCGTCGC-3') (Ohshima et al., 2002). pBG101 was digested with SapI, followed by filling with Klenow fragment and ligated to the sacB fragment, yielding plasmid pBSG102.

Construction of mutant strains.
The deletion mutants were constructed by the method of allelic replacement, based on the selection of a chromosomal deletion by two recombination events (Stibitz, 1994; Ikeda & Katsumata, 1998). To construct an rih1 deletion mutant, two regions in the rih1 gene (0·54 and 0·58 kb), were amplified by PCR using primer set G-H (G, 5'-CGCGGATCCATCGACCACGGTTTCTTCAC-3'; H, 5'-AGCATGGCAATGGCATCGTC-3') and primer set I-J (I, 5'-GACGATGCCATTGCCATGCTTCCATCTATCCTGCGAACTG-3'; J, 5'-CTAGTCTAGCGACGATATCGGAAAGACGCG-3'), and the resulting two PCR products were combined and used for another round of PCR using primers G and J. Subsequently, the fusion fragment was digested with BamHI and XbaI, ligated with BamHI/XbaI-digested pBSG102 vector and transformed into E. coli DH5α. The transformants were selected on gentamicin plates at 37 °C. To disrupt rih1, C. ammoniagenes ATCC 6872 was transformed by electroporation with the non-replicative plasmid pK-Rih1 purified from E. coli GM2929. Gentamicin-resistant clones contained chromosomally integrated plasmids. Subsequently, the gentamicin marker was eliminated by negative selection with the sacB gene in 10 % sucrose (Stibitz, 1994). rih2 was disrupted by the same method, except that the following primer sets were used: K-L (K, 5'-TTAGATCTAGACACCGGTATCGATG-3'; L, 5'-ATCATCGCCGTATTTCGCCG-3') and M-N (M, 5'-CGGCGAAATACGGCGATGATGCTTTATGACTGCAGTTCGC-3'; N, 5'-CGCGGATCCACGGTCGACGTTGTTAGCAGC-3'). Primers K and N, and two PCR fragments (0·35 and 0·33 kb) were used for another round of PCR as described above. Double deletion mutant Δrih1Δrih2 was generated starting from the parent strain, Δrih1, according to the same method described above. All the deletion mutants resulting from a double chromosomal recombination event were confirmed by PCR analysis (data not shown).

Enzyme assay.
After induction with 0·1 mM IPTG in LB medium, samples were taken from cultures of E. coli CGSC 6885(pTNI12) and 6885(pTNU23), respectively. Cells were harvested by centrifugation at 5000 g for 10 min, washed twice with 10 mM Tris/HCl (pH 7·5) and then resuspended in the same buffer. Cells were disrupted by sonication and cell debris was removed by centrifugation (10 000 g, 15 min) giving crude extracts which were used immediately or stored at 20 °C. To measure the NH enzyme activity, a 1 ml reaction mixture was used that contained 100 mM Tris/HCl (pH 7·5), 25 mM nucleoside and 1·0 mg crude extract. All of the reactions were started by adding 0·2 ml crude cell extract, incubated at 30 °C for 30 min and terminated by adding 0·1 ml 0·35 % (w/v) trichloroacetic acid (Parkin et al., 1991). The enzyme activity of nucleoside phosphorylase and uridine phosphorylase was determined spectrophotometrically using inosine and uridine as substrates, and 10 mM sodium arsenate was added as phosphate analogue (Krenitsky et al., 1976). Inosine kinase activity was assayed by measuring the formation of IMP from inosine. The 1 ml reaction mixture consisted of 100 mM HEPES (N-2-hydroxyethylpiperazine-N-2-ethanesulfonic acid) (pH 7·2), 10 mM MgSO₄, 300 mM KCl, 2 mM ATP, 5 mM inosine and 1·0 mg crude extract (Mori et al., 1995). One unit (U) of enzyme activity was defined as the amount of enzyme required to convert 1 µmol substrate into the product per minute per mg protein. Each measurement was performed in triplicate.

Recombinant DNA techniques.
DNA manipulations and PCR were carried out by standard procedures (Sambrook & Russell, 2001). Chromosomal DNA and PCR fragments were purified by using a Wizard genomic DNA purification kit (Promega) and a QIAquick gel extraction kit (Qiagen), respectively. Genes were amplified by PCR with Pfu DNA polymerase (Stratagene). All constructs involving PCR were verified by DNA sequencing according to standard methods (Sambrook & Russell, 2001).

Analyses.
Cell growth was measured at 562 nm using a spectrophotometer (Beckman DU650). HPLC (Kromasil C18 column; Waters Co.) was used to separate nucleotides, nucleosides and bases, which were detected at 254 nm. A 5 µl sample was injected after equilibration with mixed solvent (solvent A : solvent B=97 : 3) at a flow rate of 1 ml min¹. Solvent A consisted of 0·02 % tetrabutylammonium dihydrogenphosphate and 0·2 % ammonium dihydrogenphosphate, adjusted to pH 2·4 with phosphoric acid, while solvent B consisted of acetonitrile. The amount of protein was determined by the dye-binding method of Bradford (Bio-Rad) with bovine serum albumin as a standard.

Cloning of nucleoside-cleaving genes from C. ammoniagenes
After ligation of partially digested genomic DNA with the cloning vector pECCG117, the genomic DNA library was transformed into the deoD- and gsk-defective E. coli strain CGSC 6885 on MacConkey agar supplemented with 2·0 % inosine and uridine, respectively. From 20 transformants that gave rise to an apparent red colour around the colony on MacConkey agar containing inosine, genes encoding an inosine-cleaving protein were obtained. All of the plasmid DNAs isolated from the transformants displayed an approximately 1·3 kb fragment with an identical restriction enzyme pattern when digested with NheI and SalI, and had the same nucleotide sequence. Of these samples, one plasmid containing the longest DNA insert was selected and named pNI13. Genes encoding a uridine-cleaving protein were isolated from 25 clones that showed red colonies on the screening medium containing uridine as a pyrimidine nucleoside substrate. Restriction analysis of extracted plasmid DNAs from these 25 clones revealed two types of restriction pattern: one pattern contained a fragment of 0·6 kb when digested with NcoI; the other contained a fragment of 1·3 kb when double-digested with NheI and SalI. Sequence analysis of the 0·6 kb NcoI and 1·3 kb NheI/SalI fragments revealed that they contained different nucleotide sequences. However, the 1·3 kb NheI/SalI fragments in plasmids obtained from both inosine- and uridine-supplemented MacConkey agar plates had the same sequence. Therefore, one clone was selected based on the insert size and the degree of red colour around the colony from clones exhibiting a 0·6 kb fragment after digestion with NcoI and designated pNU7. The isolated plasmids pNI13 and pNU7 were retransformed into E. coli CGSC 6885 and all of the independent clones gave rise to the same red colour, as expected. Therefore, it is believed that the products of the putative genes in plasmids pNI13 and pNU7 are involved in nucleoside-cleaving reactions.

Analysis of nucleoside-cleaving genes and homology search
The nucleotide sequences of a 2·8 kb fragment of pNI13 and a 2·3 kb of pNU7 were determined. Analysis of the insert fragment of pNI13 indicated that it contained an ORF, ORF1, encoding a product of 308 aa with a calculated molecular mass of 32 310 Da. The ORF identified in pNU7, ORF2, consisted of 1011 nt, corresponding to a 337 aa product with a predicted molecular mass of 35 892 Da. To confirm the molecular mass of the putative proteins encoded by ORF1 and ORF2, crude extracts of E. coli EC100, EC202 and EC303 were prepared and subjected to SDS-PAGE analysis. As shown in Fig. 1, distinct bands appeared corresponding to molecular masses of about 33 and 36 kDa, respectively, and this result correlated with the predicted molecular mass of the proteins encoded by ORF1 and ORF2.

(79K): Fig. 1. Expression of Rih1 and Rih2 in E. coli CGSC 6885. Proteins were separated by 12 % SDS-PAGE. The arrows indicate the protein bands corresponding to ORF1 (NH 2, dotted arrow) and ORF2 (NH 1, solid arrow). Lanes: 1, 6885(pTrc99A); 2, 6885(pTNI12); 3, 6885(pTNU23); M, molecular size marker.

The deduced amino acid sequences of ORF1 and ORF2 were compared to the NCBI database using the BLAST program to search for homologous proteins. The deduced amino acid sequence encoded by ORF1 displayed high sequence similarity to the sequences of the putative NHs or inosine-uridine NHs (IunH) of Corynebacterium diphtheriae NCTC 13129 (NP_939817·1, 53 %), Corynebacterium efficiens YS-314 [NP_738480.1 (CE1870), 52 %], C. glutamicum ATCC 13032 [NP_601183.1 (NCgl1902), 52 %], Pediococcus pentosaceus ATCC 25745 (ZP_00323009.1, 48 %), Lactobacillus plantarum WCFS1 (NP_786011.1, 46 %) and Lactobacillus reuteri (AAX82611.1, 46 %). The protein encoded by ORF2 also showed high levels of sequence similarity to putative NHs or IunHs of Corynebacterium jeikeium K411 (CAI38260.1, 69 %), Photorhabdus luminescens subsp. laumondii TTO1 (NP_931465.1, 52 %), C. efficiens YS-314 [NP_738092.1 (CE1482), 51 %], Pasteurella multocida subsp. multocida Pm70 [NP_246706.1, 51 %], C. glutamicum ATCC 13032 [NP_600581.1 (NCgl1309), 50 %], C. diphtheriae NCTC 13129 (NP_940091.1, 50 %) and Crithidia fasciculata (Q27546, 49 %) (see Fig. 2). All of these enzymes, except IunH of C. fasciculata, were predicted as putative NHs or IunHs from whole-genome sequencing projects and are still uncharacterized. This comparison of protein identity suggested that the deduced polypeptides of ORF1 and ORF2 are putative NHs related to the salvage pathways of nucleoside cleavage. However, the identity (26 %) and similarity (44 %) between the two proteins were very low.

(66K): Fig. 2. Comparison of Rih1 and Rih2 with characterized NHs from different organisms. The alignment was performed with the CLUSTAL W program. Based on the characterized structure of IunH in C. fasciculata, conserved amino acid residues which occur at the same position in all sequences are indicated by a black background. The symbols #, + and @ indicate involvement of hydrogen bonds with the ribose hydroxyl groups, Ca2+ binding and a proton donor for the leaving group, respectively. Rih1 and Rih2, NHs of C. ammoniagenes; RihA, B and C, NHs of E. coli; Urh1, uridine ribohydrolase of S.cerevisiae; IunH, inosine/uridine-preferring NH of C. fasciculata.

Substrate specificity of the two NHs
To characterize the function and substrate specificity of the nucleoside-cleaving enzymes, crude extracts prepared from IPTG-induced E. coli EC202 [6885(pTNI12)] and EC303 [6885(pTNU23)], respectively, were used to measure the enzyme activity of NHs toward different purine and pyrimidine nucleosides. As shown in Table 2, the deduced enzyme 1 encoded by ORF1 in pTNI12 can hydrolyse both purine and pyrimidine ribonucleosides with the following activity order: inosine>adenosine>uridine>guanosine>xanthosine>cytidine. The deduced enzyme 2 encoded by ORF2 in pTNU23 had stronger activity toward uridine than cytidine, but no detectable activity toward purine ribonucleosides, except for inosine where a little activity was measured, indicating that it had an essentially pyrimidine-specific activity. No enzyme activity was detected toward deoxynucleosides in crude extracts of EC202 and EC303. Furthermore, the enzyme activities of nucleoside phosphorylase, uridine kinase and inosine-guanosine kinase were not detected at the same level in EC202 and EC303 compared to EC100 (data not shown). Thus, according to earlier reports regarding the classification of characterized NHs based on their substrate specificity (Petersen & Moller, 2001; Versees & Steyaert, 2003), we concluded that the deduced protein 1 belongs to a class of non-specific NHs and was designated Rih2. The deduced protein 2 had NH activity for pyrimidine ribonucleosides and was named Rih1.

Table 2. Substrate specificity of NHs The crude extracts prepared from IPTG-induced E. coli EC101 [6885(pTrc99A)], EC202 [6885(pTNI12)] and EC303 [6885(pTNU23)] were used to measure enzyme activities. , No detectable activity. E. coli EC101 [6885(pTrc99A)] showed no detectable activity with any of the substrates.

Effect of the rih1 and rih2 in C. ammoniagenes
To investigate the physiological role of rih1 and rih2 in C. ammoniagenes, deletion mutants CA313 (Δrih1), CA315 (Δrih2) and CA322 (Δrih1Δrih2), and rih1- and rih2-expressing strains CA313(pNU7) and CA315(pNI13) were constructed. Based on the analysis of the substrate specificity of Rih1 and Rih2 in the Table 2, the cell density of each strain was measured on M medium supplemented with different nucleosides or glucose, respectively.

The Δrih1 strain, CA313, exhibited a severe decrease in cell growth on uridine and cytidine compared with the wild-type strain, but growth recovered substantially by introducing pNU7 (Table 3). As described above, since Rih1 showed only pyrimidine-specific activity, we expected that an rih1 deletion mutant would grow well in M medium supplemented with purine nucleosides as carbon source. Contrary to our expectation, CA313 also exhibited decreased growth on purine nucleosides compared with the wild-type strain. However, unlike the growing pattern on pyrimidine medium, CA313(pNU7) harbouring rih1 exhibited a similar level of cell growth compared to CA313, indicating that the rih1 gene could not complement cell growth of Δrih1 on purine ribonucleosides. These results suggested that Rih1 in C. ammoniagenes has strong activity with respect to the salvage of pyrimidine nucleosides. The Δrih2 strain displayed very low levels of growth on M medium with purine nucleosides, especially medium containing inosine or xanthosine. In addition, the expression of Rih2 substantially recovered growth of strain CA315. The observed growth of strain CA315 (Δrih2) was also slower than that of the wild-type strain on pyrimidine nucleosides, but the introduction of Rih2 complemented its growth. Therefore, it is likely that Rih2 in C. ammoniagenes has a major function in the salvage of both purine and pyrimidine nucleosides. The double deletion mutant, CA322, was unable to grow on M medium with both purine and pyrimidine nucleosides. These results indicated that rih1 and rih2 play major roles in the salvage pathways of nucleosides in this micro-organism.

Table 3. Growth comparison of wild-type and deletion mutants of rih1 and/or rih2 A culture grown in YPB medium overnight was washed with saline and 1 % broth was inoculated into M medium supplemented with glucose or a nucleoside as sole carbon source. The final cell density shown was measured after 48 h of culture. The relative growth of mutants compared to wild-type is shown in parentheses. Each experiment was performed with three independent cultures.

C. ammoniagenes can grow on M medium supplemented with purine or pyrimidine ribonucleosides as sole carbon source. Auling & Moss (1984) identified a uridine phosphorylase and also suggested that there may be two different pyrimidine NHs in C. ammoniagenes. In the present study, when we attempted to identify all of the genes responsible for the nucleoside salvaging pathway by complementation in E. coli; contrary to our expectations, two genes, rih1 and rih2, annotated as putative NHs or IunHs, instead of uridine phosphorylase were identified.

Previously, the classification of NHs was based on substrate specificity and they were divided into four different classes (Versees & Steyaert, 2003). According to our assay of NH enzyme activities with different purine and pyrimidine nucleosides (Table 2) and the complementation test using knockout mutants and plasmids containing each gene (Table 3), we identified that the Rih1 enzyme in C. ammoniagenes belongs to the third class of pyrimidine-specific NHs and had a similar function to RihA and RihB in E. coli (Petersen & Moller, 2001) and URH1 in S. cerevisiae (Mitterbauer et al., 2002). Rih2 was closely related to the first class of non-specific NHs and representative examples of this group include IunH of C. fasciculata and the RihCs of E. coli and Salmonella enterica serovar Typhimurium (Hansen & Dandanell, 2005). Considering the sequence homology among these enzymes, pyrimidine-specific Rih1 was much more homologous to the non-specific IunH of C. fasciculata (49 % sequence identity) and the pyrimidine-specific RihA of E. coli (43 % identity), whereas Rih2 displayed a very low level of sequence identity with the enzymes described above. Thus, there was no high correlation between function and the sequence homology of these proteins.

According to the crystal structure of the IunH protein in C. fasciculata, Asp10, Asp14, Asp15, Asn39, Asn160, Glu166, Asn168 and Asp242 are required for hydrogen bond formation with the ribose of the substrate, and three Asp residues (positions 10, 15, 242) and one Thr residue (position 126) are involved in the binding of a Ca²⁺ ion at the active centre (Petersen & Moller, 2001; Versees & Steyaert, 2003). The highly conserved His241 residue has been shown to be a proton donor for the leaving group in the catalytic mechanism of IunH of C. fasciculata (Kurtz et al., 2002). In addition, the characteristic sequence of NHs is a recurring N-terminal DXDXXXDD motif (Versees & Steyaert, 2003). The sequence alignment between Rih1 and Rih2 of C. ammoniagenes and the functionally identified NHs demonstrated that all of these residues involved in substrate and Ca²⁺ ion binding, the proton donor and the N-terminal motif of NH activity were well conserved in Rih1 and Rih2 (Fig. 2). Even though NH 1 and 2 had the same conserved amino acid residues, they displayed different substrate specificity. These discordances might be derived from a small number of amino acid residues that restrict the access of nucleosides to the active site, as shown in a study on RihA, RihB and RihC of E. coli (Petersen & Moller, 2001).

Although several putative NHs have been annotated in the genomes of Corynebacterium species from genome sequencing projects (Ikeda & Nakagawa, 2003; Kalinowski et al., 2003), none of these has been characterized. We searched for ORFs in the complete genome sequence of Corynebacterium species, including C. glutamicum, C. efficiens, C. diphtheriae and C. jeikeium, that would encode putative nucleoside salvage enzymes. As a result of searching for NHs using the BLAST program, three putative NHs or inosine-uridine nucleoside N-ribohydrolases (NP_600581.1, NP_601183.1 and NP_602026.1) were predicted in C. glutamicum ATCC 13032. C. efficiens YS-314 had two putative NHs (NP_738092.1 and NP_738480.1) and C. diphtheriae NCTC 13129 contained three putative NHs (NP_939817.1, NP_940091.1 and NP_940641.1). Finally, two IunHs (CAI36385.1 and CAI38260.1) were annotated in C. jeikeium K411. Structurally, Rih1 exhibited more than 50 % sequence identity to NP_600581.1 of C. glutamicum, NP_738092.1 of C. efficiens, NP_940091.1 of C. diphtheriae and CAI38260.1 of C. jeikeium, whereas Rih2 was much more homologous to NP_601183.1 of C. glutamicum, NP_738480.1 of C. efficiens and NP_939817.1 of C. diphtheriae. Thus, based on quite high homology relationships, we suppose that enzymes of the first group in Corynebacterium, CAI38260.1, NP_600581.1, NP_940091.1 and NP_940641.1, may be assigned as pyrimidine-specific ribonucleoside hydrolases, whereas the proteins in the second group, NP_601183.1, NP_738480.1 and NP_939817.1, may be annotated as non-specific ribonucleoside hydrolases.

The complementation tests confirmed that the rih1 and rih2 deletion mutants displayed a decrease in cell growth on pyrimidine and purine/pyrimidine nucleosides, respectively, compared with the wild-type strain (Table 3). Also, growth of each mutant was substantially complemented by introducing rih1 and rih2, respectively. Furthermore, surprisingly, cell growth was completely abolished in CA322 (Δrih1Δrih2) on M medium with purine or pyrimidine nucleosides, although Auling & Moss (1984) reported the presence of uridine phosphorylase in C. ammoniagenes. Thus, we suggest that rih1 and rih2 in C. ammoniagenes play a prominent role in the salvage pathways of ribonucleosides on M medium supplemented with different ribonucleosides as sole carbon source, whereas uridine phosphorylase might be induced under different physiological conditions. Considering that Rih1 acts only on pyrimidine nucleosides, growth retardation in the rih1 deletion mutant on purine nucleosides medium was unexpected. In terms of genomic organization (Fig. 3), a putative permease of the major facilitator superfamily is preceded by rih1 followed by a 37 nt intercistronic region and an ORF encoding a putative membrane protein in the reverse direction is located 81 nt upstream of rih1. Therefore, we suppose that the growth inhibition of the Δrih1 strain on purine nucleosides might be related to the deletion of a region in the rih1 ORF which may affect the expression of the putative permease or the putative membrane protein.

(5K): Fig. 3. Diagrammatic representation of the DNA region containing rih1 (shaded grey). The number of amino acids for the annotated ORFs is shown within the arrows. The arrows indicate the direction of transcription. The (putative) functions of the gene products are indicated below the arrow. The numbers of intergenic base pairs are given in parentheses.

We have demonstrated that two genes, rih1 and rih2, play major roles in the salvage pathways of nucleosides in C. ammoniagenes on the basis of enzyme assays and complementation tests. Information on the salvage pathway in this organism is expected to offer more rational strategies for developing a nucleotide or nucleoside overproducing C. ammoniagenes strain with greater potential. This study was supported by grants from The Ministry of Commerce, Industry and Energy (IMT-2000 Grant).

Footnotes

^,†, Jin-Ho Lee² †These authors contributed equally to this work.

The GenBank/EMBL/DDBJ accession numbers for the sequences of rih1 and rih2 reported in this paper are AY603363 and AY603360, respectively.