Coordinate expression of NADPH-dependent flavin reductase, FRE-1, and Hint-related 7meGMP-directed hydrolase, DCS-1

A novel human cytosolic flavin reductase, Nr1, was recently described that contains FMN, FAD, and NADPH cofactors. Though the targets of the related NADPH-dependent flavoprotein reductases, cytochrome P450 reductase, methionine synthase reductase, and nitric oxide synthase, are known, the cellular function of Nr1 is not clear. To explore expression and regulation of Nr1, we cloned fre-1, the Caenorhabditis elegans ortholog of Nr1, and discovered that it is transcribed as a bicistronic pre-mRNA together with dcs-1, the ortholog of the recently described scavenger mRNA decapping enzyme. We used the novel substrate, 7meGpppBODIPY, to demonstrate that DCS-1 has low micromolar specificity for guanine ribonucleotides with the 7me modification, whereas trimethylated G substrates are poor competitors. Contrary to earlier classification, DCS-1 is not a pyrophosphatase but a distant member of the Hint branch of the histidine triad superfamily of nucleotide hydrolases and transferases. These observations are consistent with the hypothesis that DCS-1 homologs may function in the metabolism of capped oligonucleotides generated following exosome-dependent degradation of short-lived mRNA transcripts. We find that fre-1 and dcs-1 are coordinately expressed through worm development, are induced by heat shock, and have a nearly identical expression profile in human tissues. Furthermore, immunocytochemical analysis of the endogenous proteins in COS cells indicates that both are present in the nucleus and concentrated in a distinct perinuclear structure. Though no connection between these enzymes had been anticipated, our data and data from global expression and protein association studies suggest that the two enzymes jointly participate in responses to DNA damage, heat shock, and other stresses.

Determination of the cellular and biochemical specificities of thousands of newly described enzymes is a major problem of post-genomic biology. In humans, the flavincontaining NADPH cytochrome P450 reductases (CPRs) 1 were described more than 25 years ago (1,2), while related nitric oxide synthases (NOSs) (3,4) and methionine synthase reductases (MSRs) (5) were described ten and five years ago respectively. CPRs are endoplasmic reticulum-localized enzymes that serve to maintain cytochrome P450s and heme oxygenases in the reduced active state (6). NOSs mediate the synthesis of NO from arginine in the immune, vascular and nervous systems (7) while MSRs are required to maintain the methylcobalamindependent enzyme methionine synthase in a reduced state for sulfur amino acid biosynthesis, and are mutated in children with homocystinuria and other birth defects (5,8). A novel cytoplasmic reductase named Nr1 was recently described that binds FMN, FAD and NADPH cofactors and is highly expressed in human cancer cells (9). The substrates and biological pathways involving this enzyme are unknown.
Histidine triad (HIT) enzymes are a superfamily of nucleoside monophosphate hydrolases and nucleoside monophosphate transferases named for an active site sequence related to His-φ-His-φ-His-φ-φ (φ is a hydrophobic amino acid) in which the nucleoside monophosphate portion of substrates becomes covalently linked to the second His in the first step of the reaction (10).
Branches 1 and 2 of the HIT superfamily are hydrolases. Branch 3 enzymes include transferases such as galactose-1-phosphate uridylyltransferase. The most well understood branch 1 enzymes are rabbit Hint and baker's yeast Hnt1 which hydrolyze the adenosine 5' monophosphoramide substrate, AMP-lysine (11). Losses in Hnt1 lead to yeast cells that are hypersensitive to p. 4 mutations in Kin28 and other components of general transcription factor TFIIH (11), which has lead to the hypothesis that a post-translationally modified form of Cdk7/Kin28 may be a lysyladenylylated protein substrate of Hint (10). In birds, Hint and an apparent Hint inhibitor, Asw, are encoded on the Z and W sex chromosomes, respectively, suggesting that Asw-mediated inhibition of Hint may play a role in female avian development (12). Branch 2 enzymes consist of the human tumor suppressor Fhit (13)(14)(15) and its fungal (16) and animal homologs (17), which hydrolyze dinucleoside polyphosphates such as ApppA and AppppA to AMP plus the other mononucleotide (10).
Human Dcps (scavenger decapping enzyme) was recently purified and cloned as an enzyme that hydrolyzes compounds such as 7meGpppG and small capped oligoribonucleotides, and was proposed to function in the hydrolysis of short oligomers that remain after 3' to 5' exonucleolytic degradation of mRNA (18). Deletion of DCS1 from baker's yeast showed that the enzyme has 7meGpppG-hydrolase activity in vivo (18). In fission yeast, the homologous enzyme (Nhm1) was co-purified with the mRNA cap-binding complex eIF-4F (19).
Here we identify fre-1, the C. elegans homolog of the human NR1 NADPH-dependent flavin reductase, as the second gene in a two-gene operon with C. elegans dcs-1. Biochemical analysis with a newly synthesized 7meGpppBODIPY substrate and structure-based sequence analysis of DCS-1 indicates that DCS-1 homologs are a new sub-branch of HIT hydrolases (Hint branch) with a 7meGMP-binding pocket in place of the AMP-binding pocket of Hint and Fhit.
In dcs-1:fre-1-green fluorescent protein (GFP) transgenic worms the dcs-1:fre-1 promoter directs expression to neurons and pharyngeal muscle and is induced by heat shock. Reverse transcription-polymerase chain reaction (RT-PCR) analysis of the two genes in worms and humans indicates that they are co-regulated in time and tissue across phyla. p. 5 Immunocytochemistry indicates that these two proteins are found together in the nucleus and in a distinct perinuclear structure in COS cells. Expression data suggest that the two enzymes jointly participate in responses to DNA damage, heat shock and other stresses.

EXPERIMENTAL PROCEDURES
Cloning and Characterization of Worm Transcripts-Cultures of C. elegans (Bristol N2) were grown and maintained as described (20). Total RNA, prepared from a mixed population of wild type hermaphrodites, was reverse-transcribed into cDNA with SuperScript Reverse Transcriptase (Invitrogen) using oligo dT primers followed by PCR using gene-specific oligonucleotides 1 and 2 for fre-1, and 3 and 4 for dcs-1. Full-length cDNAs were cloned into pCR2.1 (Invitrogen) and sequenced. Trans-splicing specificity was assessed by performing RT-PCR with SL1 and SL2 as the 5' primers and primers 2 and 4 as the 3' primers as described (21).
To examine developmental expression patterns, RNA was isolated from synchronized N2 nematodes cultured at 20 0 C and RT-PCR was performed using the primers for fre-1 and dcs-1, along with primers 5 and 6 for the S-adenosyl L-homocysteine hydrolase (AHH) gene, whose mRNA abundance is unchanged throughout development (C. Thacker, unpublished results). For expression analysis by RT-PCR, 1 µg of total RNA was subjected to 35 cycles of amplification with an annealing temperature of 59 0 C. The same primers were used for RT-PCR analysis of worms subjected to a two hour 33 0 C heat shock. Primer sequences are provided in Supplemental Table 1.
Construction and Characterization of Transgenic Worms-A PCR product was obtained consisting of 1.8kbp of DNA upstream of the dcs-1 initiator codon, the entire dcs-1coding region, and the first 38 bp of the fre-1 open reading frame using primers 7 and 8. This fragment, p. 6 digested with XbaI and SmaI (sites engineered in the primers), was introduced into GFP expression vector pPD95.70 (23). The resulting plasmid was injected at 10 µg/µl along with plin-15 containing the wild type lin-15 gene at 50 ng/µl and pBlueScript KS at 50 ng/µl into the gonad arms of lin-15 (n765ts) adult hermaphrodites as described (24). Transgenic animals were identified by rescue of the lin-15(n765ts) multi-vulval phenotype at 20 0 C and stable lines were isolated. Cell identification was based on the characteristic morphology and position of GFPpositive nuclei viewed by simultaneous fluorescence and differential interference contrast microscopy. Antibodies and Immunocytochemistry-Antibodies were raised in rabbits against synthetic peptides derived from human Nr1 (aa 197-207) and Dcps (aa 181-195), conjugated to KLH, and affinity purified using the peptide antigens. In western blots of protein from COS and HEK cells, or mouse brain, the antibodies recognized bands of the predicted size for Nr1 (68 kD) and Dcps (38 kD). For immunocytochemistry, COS cells were grown on coverslips, fixed in 4% paraformaldehyde, and incubated with the antibodies at a final concentration of 2 µg/ml.

Cloning and Characterization of Human
Following washing in PBS, the cells were incubated in Cy3-conjugated donkey anti-rabbit antisera (Jackson Labs.), counterstained with DAPI and examined by fluorescence microscopy.

Homologs of Human NADPH-Dependent Flavin Reductase and Scavenger Decapping
Enzyme are Encoded on a Two Gene Operon in Worms-We examined the inventory of predicted flavoprotein reductases in the worm genome (30) and discovered homologs of CPR (accession number NP_498103), MSR (accession number NP_495978), and Nr1 (accession number 6425246), the latter of which we named fre-1. As shown in Figure 1, the predicted 585 amino acid fre-1 polypeptide conserves sequences defined to bind FMN, FAD and NADPH but lacks the N-terminal hydrophobic sequences that anchor CPR to the endoplasmic reticulum, the p. 9 N-terminal heme-binding oxidase domain found in NOS, and the central sequences that are diagnostic for MSR.
Just 180 nucleotides upstream of the fre-1 initiator codon, we located the stop codon of the ortholog of human scavenger decapping enzyme Dcps, which we name dcs-1.
Approximately 15% of worm genes occur in operons of two to eight genes (31). In these cases, pre-mRNAs are formed containing multiple cistrons that are cleaved and polyadenylylated. In all known cases in which there is an intercistronic length of greater than a few nucleotides, the downstream genes are trans-spliced with the downstream gene-specific SL2 RNA (32). To examine whether the potential downstream gene fre-1 is, in fact, synthesized as part of an operon, we performed RT-PCR with an SL2 primer and fre-1-specific primer 2 (21) and demonstrated amplification of fre-1 ( Figure 2). Moving upstream to dcs-1, we showed that dcs-1 cDNA can be amplified with a primer to the SL1 trans-spliced leader but not the SL2 leader, demonstrating that dcs-1 is the 5' cistron within the dcs-1:fre-1 operon. Because the next downstream and upstream genes are more than 10 kbp away from dcs-1 and fre-1 exceeding the maximally observed intercistronic distance by more than 20-fold (31), we conclude that dcs-1 and fre-1 constitute a two-gene operon in C. elegans. A recent high-throughput study of SL2transpliced transcripts predicted that Y113G7A.9 (dcs-1) and Y113G7A.8 (fre-1) are in a twogene operon (31). It should be noted that because the mRNAs for both dcs-1 and fre-1 are transspliced to SL sequences, these transcripts, are expected to contain the trimethylguanosine (TMG) cap from the nuclear SL RNA molecules, like those of other small nuclear RNAs (33).

DCS-1 is a Member of the Hint Branch of the HIT Hydrolase Superfamily with Specificity
for 7meGMP in the Primary Nucleotide Binding Site-Human scavenger decapping enzyme was recently reported to hydrolyze 7meGpppG to 7meGMP plus GDP (18). Sequences of Dcps p. 10 and Dcs orthologs (hereafter, these enzymes are termed Dcs to distinguish them from unrelated Dcp mRNA decapping enzymes (34)) were described as containing a His-φ-His-φ-His-φ-φ motif in which the second His is required for activity but with no other similarity to HIT hydrolases (18). The orthologous enzyme from fission yeast, named Nhm1p, was described as having low level similarity to Fhit but only similarity with the orthologous enzymes was shown (19). Based on our crystal structures of nucleotide-bound forms of Hint (35) and Fhit (15), we performed a structure-based alignment of three Hint orthologs, three Aprataxin orthologs, three Fhit orthologs and three Dcs orthologs. As shown in Figure 3, though the level of sequence identity between Dcs sequences and the three other enzymes is low, there is significant conservation that begins one residue before the first α-helix of the Hint structure through one residue after the fifth and final β-strand that is conserved in all HIT hydrolases (this strand is numbered 7 in Fhit because two short helices replace the first helix in Hint (15,36)). There is no similarity outside this core structure, suggesting that Dcs homologs are unique outside of the dimeric 10-stranded GalT halfbarrel that has been found in all HIT enzymes (10). Though human Dcs has a Thr in the seventh position of the His-φ-His-φ-His-φ-φ motif, the alignment shows that this β-branched and moderately hydrophobic residue does not violate the rule as Dcs homologs have β-branched hydrophobic residues Ile and Val, which along with Leu, are the consensus. More surprising than Thr in the seventh position of the HIT motif is the lack of a His that aligns with Hint residue 51 and Fhit residue 35. This residue is in a hydrogen bonding network interacting with the final His of the HIT motif that has been proposed to be involved in leaving group leaving and water attack (10). Dcs orthologs have a polar residue at this position. Beyond conserved N-terminal and C-terminal extensions that are diagnostic for Dcs sequences, the Dcs enzymes have a short, acidic loop insertion between β-strands 1 and 2 and a distinctive loop between β-strands 2 and 3 p. 11 containing the sequence L-P-D-φ-K-W-D-G. On the basis of the crystal structure of Hint bound to GMP (35), this loop is positioned to make contact with 7me group of 7meG, which we predict accounts for the 7meGMP-specificity of DCS-1 and its homologs.
Specificity of DCS-1 Defined with 7meGpppBODIPY -We expressed His-tagged worm DCS-1 in E. coli and purified it to homogeneity. To characterize the specificity of this enzyme, we synthesized a novel fluorescent analog, 7meGpppBODIPY. As the specificity of human Dcps was reported to be for analogs of capped RNAs such as 7meGpppG, 7meGpppA or 7meGppp(Np) n , where n<10 (18), we reasoned that the lack of specificity for substrate features beyond 7meGTP would allow us to substitute a BODIPY reporter group for a nucleoside, nucleotide or short oligonucleotide. In the past, GpppBODIPY was synthesized from GTPγS by conjugation of an iodoacetamide-linked BODIPY probe (29). Here, we used nucleoside diphosphate kinase to transfer the γ phosphorothio group of ATPγS to 7meGDP. The resulting 7meGTPγS was then converted to 7meGpppBODIPY by established methods (29). As shown in Table 1, this turned out to be a nearly essential modification for DCS-1 characterization. While Fhit substrates ApppBODIPY and GpppBODIPY were cleaved by DCS-1 with k cat values of less than 0.01 s -1 and 0.035 s -1 , the 7me modification of GpppBODIPY increased k cat to > 0.17 s -1 .
Taking k cat / K m as the specificity constant, DCS-1 prefers 7meGpppBODIPY to ApppBODIPY and GpppBODIPY by > 35-fold and 75-fold, respectively. In the context of the BODIPY nucleotide substrates, the 7me modification to G accounts for a 7-fold discrimination in K m and a 5-fold discrimination in the k cat term. The 1.21 µM K m measured for DCS-1 hydrolysis of 7meGpppBODIPY compares favorably with good substrates for other histidine triad enzymes: Fhit cleaves GpppBODIPY with a 1.4 µM K m , though the higher k cat gives an overall higher specificity constant (29).

p. 12
Having demonstrated that 7meGpppBODIPY is a useful probe for DCS-1, we sought to further dissect substrate recognition by this enzyme. By titrating nonlabeled nucleotides into 7meGpppBODIPY assays and measuring the concentration-dependent reduction in k cat / K m (apparent) (29), we determined the K i of 12 additional nucleotides (Table 1). This analysis revealed that most but not all of the binding specificity is for the 7meG mononucleotide group as we measured Ki values for 7meGDP, 7meGTP and 7meGpppG in the 2.2 to 3.5 µM range.
There is a hydrophobic binding site on the other side of the scissile bond, however, because GpppBODIPY competes for the DCS-1 active site better than GpppA, which, in turn, competes better than GTP. Similarly, binding specificity for a hydrophobic leaving group can be illustrated in the series of ApppBODIPY, AMP-pNA, ApppA and AMP-NH2, which exhibited increasingly poor binding. As a consequence of the lack of a 7me modification and the lack of a hydrophobic leaving group, the abundant guanine nucleotides such as GDP and GTP are not DCS-1 competitors even at 10 mM. The 2, 3, 7-trimethylated GpppG analog had a K i value eight-fold higher than that of 7meGpppG. Thus, the in vitro specificity of the enzyme does not support the idea that small nuclear RNAs with TMG caps (37,38) might be in vivo substrates for DCS-1 and its orthologs. Earlier, human Dcps was termed a pyrophosphatase (18). In contrast, our work indicates that DCS-1 enzymes are a distinct sub-branch of the Hint branch of HIT hydrolases (10) with specificity for 7meGMP in the primary nucleoside monophosphate binding site.

Evolutionary Origin of DCS-1 and FRE-1 Orthologs-Earlier we have argued that Hint,
a prototypical AMP-lysine hydrolase, was the evolutionary precursor of Fhit homologs (10,35).
While the specificity of DCS-1 is somewhat closer to Fhit (29) than Hint (11), i.e., DCS-1 prefers a phosphorylated leaving group such as ppBODIPY or GDP to an amine leaving group, p. 13 structure-based sequence analysis shown in Figure 2 indicates greater similarity to Hint than to Fhit. Moreover, sequence searches indicate that DCS-1 orthologs are found in animals including mammals, other vertebrates, and invertebrates, fungi including S. cerevisiae and S. pombe, and microsporidia including E. coniculi (our unpublished work). As we have shown earlier, Fhit homologs are found additionally in green plants and mycetozoa such as D. discoidium (10).
Thus, to argue that DCS-1 orthologs evolved from a Fhit orthologs would imply loss of DCS-1 from significant branches of the evolutionary tree. In contrast, independent descent of DCS-1 and Fhit from Hint requires only the assumption that DCS-1 is a more recently evolved hydrolase, i.e., after the rift of plants from animals and fungi (39), that specializes in 7meGMP hydrolysis from small nucleotide or oligonucleotide substrates. A similar analysis of FRE-1 orthologs indicates that they are a somewhat older group of enzymes, having homologs on either side of the plant/animal rift (our unpublished work). Thus, in the first fungal-animal precursor cell in which a DCS-1 hydrolase appeared, a FRE-1-type reductase is expected to have already been part of the protein complement. Examination of transgenic animals carrying this construct showed GFP expression throughout development in the nuclei of neurons and the pharyngeal muscle, with heat shock enhancing p. 14 GFP expression ( Figure 4A). RT-PCR analysis of dcs-1 and fre-1 indicates that both mRNAs are expressed in all developmental stages, that L3 and adults have higher expression levels, and that both messages are present in dauer animals and similarly induced by heat shock ( Figure 4B).

Coordinate Regulation of dcs-1 and fre-1 in Worms and
The RT-PCR assay was applied to human tissues and revealed that brain, liver, kidney, pancreas and testes were positive for both messages. Heart, lung, small intestine, skeletal muscle, colon, leukocytes, spleen, ovaries, and thymus were negative for both messages. Only prostate was positive for DCPS and negative for NR1 and no characterized human tissue was positive for NR1 and negative for DCPS. Thus, though DCPS and NR1 are not neighboring genes in humans, their expression profile is highly correlated ( Figure 4C).
Endogenous Nr1 and Dcps were detected in COS cells by immunocytochemistry ( Figure   5). Staining for both proteins was detected in the nucleus, but they were concentrated in a distincet perinuclear structure in these cells.

DISCUSSION
Consistent with our observations that dcs-1 and fre-1 are heat shock-induced genes in worm, the baker's yeast DCS1 and DCS2 genes are induced by heat shock, mutagens, and are repressed by cAMP withdrawal. Dcs1 was reported to have a two-hybrid interaction with neutral trehalase (41) and to be found in a large complex with the stress-induced protein kinase Mpk1 by systematic mass spectrometry (42). These interactions were validated by showing that the genes are coordinately regulated at the level of mRNA accumulation and show similar heat shocksensitive and caffeine-sensitive phenotypes (43). Though cell cycle initiation is highly sensitive to reduction of eIF-4E levels, translation of heat shock proteins such as polyubiquitin is insensitive (44) so it would appear unlikely that scavenger decapping activity is required to allow p. 15 immediate-early stress-induced changes in gene expression. Resumption of the cell cycle after stress may entail an increased requirement for capping or cap-recognition, either of which could be inhibited by accumulation of short capped RNAs.
The precise nature of the DCS-1 substrate remains an interesting question as it has little activity on long capped RNAs (18). It is interesting to note that the class of genes most frequently transcribed in operons in C. elegans encodes proteins involved in RNA degradation (31). mRNA decapping is mediated by the Dcp1 and Dcp2 Nudix-type hydrolases (45), which are located in distinct cytoplasmic foci (46)(47)(48). The human Hint-related Dcps enzyme was purified with the thought that its activity might be associated with a late stage of mRNA degradation (18), however, the enzyme appears to be nuclear in fission yeast (19). Thus DCS-1 may function in exosome-mediated mRNA degradation in the cytoplasm and nucleus. We note that the stresses that induce dcs-1 gene expression may result in production of aborted mRNA transcripts. As capping and methylation of messages occurs early in the elongation phase of transcription (49), conditions that lead to the early dissociation of the pol II machinery would be expected to lead to production of short capped RNAs with no useful function. Given the lack of specificity for trimethylated nuclear caps, we suggest that capped, monomethylated, aborted transcripts may be relevant DCS-1 substrates. Discovery of dcs-1 and fre-1 as coordinately expressed genes will allow testing of the hypothesis that scavenger decapping activity and a specific NADPH-dependent flavin reductase activity are required for responses to heat shock and genotoxic stress in animals.

p. 19
Legends to Figures   Fig. 1 (35). The sequences of DCS-1 orthologs are slightly closer to those of Hint orthologs than to other HIT hydrolases. As discussed in the text, the L-P-D-φ-K-W-D-G loop between β strands 2 and 3 is predicted to account for 7meGspecificity in the primary nucleoside monophosphate-binding site. Note that fre-1 is detected in all stages except the embryo and dcs-1 was detected at all stages, although egg and larval stages L1 and L2 had a very low expression. When mixed stages were compared before and after heat treatment both genes showed a higher expression following the heat treatment. D. Tissue expression patterns of Nr1 and Dcs in human tissues. MTC panel cDNAs were used in PCR reactions with the gene specific primers, and both transcripts were expressed in liver, brain, kidney and testis.

A S V T T S A A A S K S N Q S F I D R M K N E N R Q V L I M Y G S Q T G T A E E M S G R L A K D L T R Y T K K A V V V D C P R 5 8 Q T L T S S V R E S ----S F V E K M K K T G R N I I V F Y G S Q T G T A E E F A N R L S K D A H R Y G M R G M S A D N R 1 1 ---------------------M P S P Q L L V L F G S Q T G T A Q D V S E R L G R E A R R R R L G C R V Q A Y 1 1 3 G A . 8 1 ------------------------M S I A I L Y G S E T G T A Q D I A E S L R R E A Q Q R H L Q A R V H E
1 5 5 G L Y P P P P G L T E I P P G V P L P S K F T L L F -L Q E A P S T G S E G Q R V A H P G S Q E P P S E S K P F L A P M Y 1 1 3 G 7 A .