To test whether any of the nucleoside kinases proposed to phosphorylate tiazofurin are uniquely or collectively responsible for utilization of nicotinamide riboside, we prepared a qns1 deletion strain that was additionally deleted for all of the candidate genes for which yeast homologs exist, namely adenosine kinase ado1 (Lecoq et al., 2001), uridine/cytidine kinase urk1 (Kern 1990, Kurtz et al. 1999), and ribokinase rbk1 (Thierry et al., 1990). As shown in Figure 2B, despite these deletions, the strain retained the ability to utilize nicotinamide riboside in an anabolic pathway independent of NAD+ synthetase. Given that mammalian pharmacology provided no useful clue to the identity of a putative fungal Nrk, we considered whether the gene might have been conserved with the Nrk of H. influenza. The Nrk domain of H. influenza is encoded by amino acids 225 to 421 of the NadR gene product (the amino terminus of which is NMN adenylyltransferase). Though this domain is structurally similar to yeast thymidylate kinase (Singh et al., 2002), sensitive sequence searches (not shown) revealed that bacterial Nrk has no ortholog in yeast. Indeed, genomic searches with the Nrk domain of H. influenza NadR have identified a growing list of bacterial genomes predicted to utilize nicotinamide riboside as an NAD+ precursor (Kurnasov et al., 2002). Thus, had fungi possessed NadR Nrk-homologous domains, comparative genomics would have already predicted that yeast can salvage nicotinamide riboside.
To identify the Nrk of S. cerevisiae, we established an HPLC assay for the enzymatic activity and utilized a biochemical genomics approach to screen for the gene encoding this activity (Martzen et al., 1999). Sixty-four pools of 90–96 S. cerevisiae open reading frames fused to glutathione S-transferase (GST), expressed in S. cerevisiae, were purified as GST fusions and screened for the ability to convert nicotinamide riboside plus ATP to NMN plus ADP. As shown in Figure 4A, whereas most pools contained activities that consumed some of the input ATP, only pool 37 consumed nicotinamide riboside and produced NMN. Examination of the 94 open reading frames that were used to generate pool 37 revealed that YNL129W encodes a predicted 240 amino acid polypeptide with a 187 amino acid segment containing 23% identity with the 501 amino acid yeast uridine/cytidine kinase Urk1 and remote similarity with a segment of E. coli pantothenate kinase panK (Yun et al., 2000) (Figure 4B). Cloning of YNL129W into a bacterial expression vector allowed us to test the hypothesis that this homolog of metabolite kinases is the eukaryotic Nrk. Strikingly, the specific activity of purified YNL129W was ∼100 times that of pool 37, consistent with the idea that all the Nrk activity of pool 37 was encoded by this open reading frame. To test genetically whether this gene product phosphorylates nicotinamide riboside in vivo, we created a deletion of YNL129W in the qns1 background and found that nicotinamide riboside rescue of the qns1 deletion strain is entirely dependent on this gene product (Figure 4C). Having shown biochemically and genetically that YNL129W encodes an authentic Nrk activity, we named this gene NRK1.