Cloned Mouse Ribonucleotide Reductase Subunit M1 cDNA Reveals Amino Acid Sequence Homology with Escherichia coli and Herpesvirus Ribonucleotide Reductases*

We have isolated and sequenced overlapping cDNA clones containing the entire coding region of mouse ribonucleotide reductase subunit M1. The coding region comprises 2.4 kilobases and predicts a polypeptide of 792 amino acids (Mr 90,234) which shows striking homology with ribonucleotide reductases from Esche- richia coli and the herpesviruses, Epstein-Barr virus and herpes simplex virus. The homologies reveal three domains: an N-terminal domain common to the mammalian and bacterial enzymes, a C-terminal domain common to the mammalian and viral ribonucleotide reductases, and a central domain common to all three. We speculate on the functional basis of this conserva- tion.

Ribonucleotide reductase is a crucial cell enzyme, providing the only route for de novo synthesis of deoxynucleotide substrates for DNA replication via direct reduction of the corresponding ribonucleoside diphosphates. Mammalian ribonucleotide reductase is composed of two nonidentical subunits, M1 and M2, with molecular weights of 90,000 and approximately 55,000, respectively (Thelander and Reichard, 1979). The enzyme is under complex allosteric control mediated by the binding of deoxynucleoside triphosphates and ATP to binding sites on the M1 subunit. Two distinct regulatory sites have been defined the specificity site, which controls substrate specificity, and the activity site, which regulates overall catalytic activity (Thelander and Reichard, 1979;Eriksson et al., 1981a). Photoaffinity labeling experiments have located a third site on M1, a substrate-binding catalytic site, formed only in the presence of the second subunit, M2 (Caras et al., 1983). We have previously characterized mutants of ribonucleotide reductase subunit M1 that carry alterations in the allosteric sites that specifically disrupt the binding of one or more nucleotide effectors (Eriksson et al., 1981a). One of these mutants (an activity site mutant) is resistant to feedback inhibition by dATP, producing elevated deoxynucleotide pools and a mutator phenotype in cells .
In addition to complex structure-function relationships, ribonucleotide reductase presents a number of interesting problems in terms of growth regulation and human disease. The level of enzyme activity is closely correlated with the growth rate of a cell (Elford et al., 1970) and appears to vary within the cell cycle Eriksson et al., 1984). Ribonucleotide reductase is also thought to mediate the pathogenesis of the immunodeficiency that results from an inherited deficiency of adenosine deaminase or purine nucleoside phosphorylase (Martin and Gelfand, 1981). The deoxynucleotides that accumulate in the lymphoid cells of these patients are thought to feedback-inhibit ribonucleotide reductase, preventing DNA replication and cell proliferation.
In this paper, we report the isolation of three overlapping cDNA clones, spanning 2.9 kilobases, which include the entire coding region of mouse ribonucleotide reductase subunit MI. The mRNA used to generate these clones was from a mutant T-lymphoma (S49) cell line which expresses a dATP feedback-resistant mutant form of M1. We have compared the deduced primary amino acid sequence of the M, 90,000 subunit M1 with the recently published sequences of the analogous Escherichia coli subunit B1 (Carlson et al., 1984) and the ribonucleotide reductases of the herpesviruses, Epstein-Barr virus (EBV') (Gibson et al., 1984) and herpes simplex virus (HSV) (McLauchlan and Clements, 1983;Dutia, 1983;Bacchetti et al., 1984). We report here a striking similarity in specific regions of the sequence and propose that this homology may be functionally relevant.

EXPERIMENTAL PROCEDURES
Cell Lines and Antiserum-Mouse T-lymphoma cells, S49 (Horibata and Harris, 1970) or WEHI-7 (Harris et al., 19731, were grown in Dulbecco's modified Eagle's medium with 10% heat-inactivated horse serum. The isolation and characterization of the S49 cell mutant, dGuo-200-1-A, will be described elsewhere.' A polyclonal anti-Ml rabbit serum was raised against S49 cell subunit M1 purified to homogeneity by affinity chromatography on dextran blue-Sepharose and dATP-Sepharose (Eriksson et al., 1981b). IgG was isolated on DEAE-Affi-Gel Blue (Bio-Rad) using conditions recommended by the supplier.
Polysome Purification and RNA Isolation-Polysomes were prepared from exponentially growing ,949 or WEHI-7 cells as described by Goddard et al. (1983). Immunoadsorption of polysomes with anti-M1 antibody, isolation of the antibody-antigen complex by protein A-Sepharose chromatography, and extraction of the poly(A)+ RNA were as described by Kraus and Rosenburg (1982). Total cytoplasmic poly(A)+ RNA from S49 or WEHI-7 cells was prepared as described (Goddard et al., 1983).
In  (1982). except that only one round of immunoprecipitation was performed and formalin-fixed Staphylococcus aureus A cells (Pansorbin) were purchased from Calbiochem-Behring. Chromatography of the reaction products on dextran blue-Sepharose and/or dATP-Sepharose was carried out as previously described (Eriksson et al., 1981b) using 0.1-ml columns. Fractions were concentrated by precipitation with trichloroacetic acid in the presence of bovine serum albumin and analyzed by sodium dodecyl sulfate-10% polyacrylamide gel electrophoresis (Laemmli, 1970). Construction and Screening of cDNA Libraries-Two cDNA libraries were constructed. Initially, double-stranded cDNA was prepared from 100 ng of M1-enriched S49 cell mRNA primed with oligo(dT) as described by Goddard et al. (1983). The double-stranded cDNA FIG. 1. Identification of MI-specific mRNA by in vitro translation followed by affinity chromatography or immunoprecipitation. Poly(A)+ RNA from S49 cells was translated in a reticulocyte cell-free system containing [35S]methionine and treated as follows. a, 5% of the 35S-labeled translation products was loaded directly onto a sodium dodecyl sulfate-10% polyacrylamide gel  was inserted into the PstI site of pBR322 by dC/dG homopolymeric tailing and used to transform E. coli strain K12 MC1061 (Goddard et al., 1983). The resultant clones were screened by colony hybridization (Gergen et al., 1979) using 32P-labeled single-stranded cDNA probes reverse-transcribed from enriched or unfractionated WEHI-7 poly(A)+ RNA.
For the second library, 0.8 pg of a specific 18-mer DNA primer was mixed with 10 pg of unfractionated polysomal poly(A)+ RNA, heated to 70 "C for 3 min, and then chilled on dry ice. The reaction mixture was adjusted to 20 mM Tris-HCI (pH 8.3), 20 mM KCI, 8 mM MgCl2, 25 mM dithiothreitol, 0.5 mM of each deoxynucleoside triphosphate, 1 pCi/pl [w3'P]dCTP, 4 mM sodium pyrophosphate, and 2 units/pl RNasin (Promega Biotec) in a final volume of 100 pl. cDNA synthesis was initiated by the addition of 50 units of reverse transcriptase and incubation a t 42 "C for 20 min. The cDNA was made double-stranded by standard procedures (Pennica et al., 1983;Goeddel, 1980) and electrophoresedon a 6% polyacrylamide gel. Two-hundred nanograms of double-stranded cDNA (greater than 600 base pairs in length) was recovered by electroelution, ligated to a synthetic EcoRI adapter (Wood et al., 1984), and inserted into the unique EcoRI site of the X g t l O phage vector (Hyunh et al., 1984). The DNA was packaged in vitro using the Packagene extract from Promega Biotec and plated onto Y1073, an Hfl derivative of C600 (Young and Davis, 1983). The recombinant plaques were screened as described by Maniatis et al. (1982) using a 32P-labeled nick-translated cDNA fragment of p201 (see "Results").
Identification of Clones by Hybrid Selection-Positive hybrid selection was performed essentially as described by Maniatis et al. (1982). Recombinant plasmid DNA (6 pg) bound to diazobenzyloxymethyl paper (S and S Transa-Bind, aminophenylthioether form) was hybridized with 75 pg of poly(A)+ RNA. The hybridized mRNA was FIG. 3. Restriction map and sequencing strategy. The overlapping M1 cDNA clones are shown below a restriction map of the composite sequence. The protein coding region is indicated by the hatched bar. The arrows beneath each cDNA indicate the direction and extent of sequence determined for each fragment analyzed (dideoxy terminator method (Smith, 1980)). The position of the 18-mer DNA primer used in the cloning is indicated below the restriction map. eluted in two changes of 200 pl of 1 mM EDTA, 5 pg of tRNA at 80 "C for 1 min. The eluted mRNA was ethanol-precipitated and assayed by in vitro translation and immunoprecipitation.
Sequence Analysis-DNA nucleotide sequences were determined by the dideoxynucleotide chain termination method (Smith, 1980) after subcloning of appropriate restriction fragments into derivatives of bacteriophage M13 (Messing et al., 1981). A partial N-terminal sequence of purified subunit M1 (Eriksson et al., 1981b) was determined by the Edman method (Edman and Begg, 1967) using a modified Beckman microsequenator.
Other Methods-Plasmid and bacteriophage DNAs were prepared essentially as described by Maniatis et al. (1982).
Double-stranded DNA was labeled by nick translation (Rigby et al., 1977) to a specific activity of 107-108 cpmlpg.
The oligonucleotide primer was synthesized by the solid-phase phosphotriester method (Crea and Horn, 1980).

RESULTS
The poly(A)+ RNA used to generate the cDNA clones described below was from a mutant S49 cell line, dGuo-200-1-A.2 These cells have the following properties: 1) they are resistant to deoxyadenosine toxicity; 2) they contain altered ribonucleotide reductase subunit M1 molecules which show reduced affinity for dATP at the activity site and do not respond to normal feedback regulation by dATP; and 3) they appear to express predominantly one protein M1 allele, the mutant allele.
Cloning Strategy-Protein M1 represents less than 0.05% of the total cell protein in S49 cells.3 Assuming the mRNA to be of similarly low abundance, we enriched the mRNA approximately 100-fold by immunoadsorbing polysomes with anti-M1 antibody. The enriched mRNA was used to construct a cDNA library, and potential M1 clones were selected by a differential screening procedure (see below). M1 clones were then identified by positive hybrid selection. Since the initial clones lacked the full coding region, a second cDNA library was constructed by specific priming followed by cloning into a X g t l O vector.
Detection and Enrichment of MI mRNA-To follow the enrichment of M1 mRNA and to facilitate identification of M1 clones by hybrid selection, we developed two independent assays for M1-specific RNA sequences. After in uitro translation of S49 cell poly ( immobilized on diazobenzyloxymethyl paper filters was hybridized with 75 pg of poly(A)+ mRNA isolated from either S49 cells (a and b) or WEHI-7 cells (c). The selected mRNA was eluted and translated in a reticulocyte lysate system with [35S]methionine. Two per cent of the translation products from each sample was loaded directly onto a sodium dodecyl sulfate-polyacrylamide gel (a). The remainder was immunoprecipitated with anti-M1 antibody (b and c). Lanes are as follows: translation products of 0.1 pg of S49 or WEHI-7 poly(A)+ RNA not subjected to hybrid selection (lanes 1 and 2); immunoprecipitation of the nonselected S49 mRNA translation products with preimmune serum (lane 8 ) or anti-M1 antibody (lane 9); immunoprecipitation of the nonselected WEHI-7 mRNA translation products with preimmune serum (lane 16) or anti-Ml antibody (lanes 10 and 17); translation products of mRNAs selected by the indicated recombinant clones either before (lanes 3-6) or after (lanes 11-14 and 18-21) immunoprecipitation with anti-MI antibody; and translation products in the absence of exogenous RNA (lanes 7, 15, and 22). The arrows indicate the position of M1 determined by co-electrophoresis with [32P]TTP photoaffinity-labeled protein M1 (Eriksson et al., 1982) (lanes M). 7) or subjected to affinity chromatography on dextran blue-Sepharose followed by dATP-Sepharose (lane 5 ) and analyzed on sodium dodecyl sulfate-polyacrylamide gels. Both procedures lead to the substantial enrichment of a 90-kDa protein which co-migrates with authentic M1 (data not shown) and has identical chromatographic properties. A control with preimmune serum indicated that M1 immunoprecipitation was specific (compare lanes 6 and 7). Because S49 cells contained a mouse mammary tumor virus protein that was weakly precipitated by the anti-M1 antibody (protein V, lane 7), we used mRNA from WEHI-7 cells (which do not express mouse mammary tumor virus (Stallcup et al., 1978)) to generate cDNA probes used in the screening of cDNA libraries as described below.

Molecular Cloning
To prepare M1-enriched mRNA, we used polysomes (S49 or WEHI-7) purified by immunoadsorption with anti-Ml antibody followed by chromatography on protein A-Sepharose. In oitro translation of the enriched mRNA produced one major 90-kDa protein (Fig. 2, lane I ) which was immunoprecipitable with anti-M1 antibody as expected (lane 3) and also bound to dATP-Sepharose (lane 5), providing independent evidence that the immunoadsorbed polysomal mRNA was enriched for M1-specific mRNA. Densitometric scanning of the bands in lane 1 suggested that MI-specific mRNA com-prised approximately 5% of the total polysome-enriched mRNA preparation, representing a 100-fold enrichment.
Molecular Cloning of Ribonucleotide Reductase Subunit MI cDNA-The initial library, prepared from M1-enriched mRNA cloned into the PstI site of pBR322, contained approximately 2000 tetracycline-resistant, ampicillin-sensitive transformants. Potential ribonucleotide reductase subunit M1 clones were identified by differential colony screening using [32P]cDNA probes transcribed from poly(A)+ RNA from either unfractionated or immunopurified WEHI-7 polysomes. M1 clones were not expected to give a signal with the unfractionated cDNA probe in which M1 sequences represented 0.05% of the total. We therefore selected clones which hybridized only to cDNA made from the M1-enriched mRNA. Approximately 50 colonies (2.5%) exhibited the required pat- High molecular weight DNA (approximately 6 pg, except the yeast lane which contained approximately 0.6 pg) was digested to completion with BarnHI, fractionated, and transferred to nitrocellulose by the Southern procedure (Southern, 1975). DNA bands homologous to M1 sequences were visualized by hybridization to nick-translated restriction fragments from p247, p201, and MA-1 spanning the sequence shown in Fig. 3. Fragment sizes are given in kilobases and were determined by reference to a Hind111 digest of bacteriophage A DNA.
tern of hybridization. Plasmid DNAs from these colonies were grouped on the basis of common DdeI restriction fragments (data not shown), and representatives containing the longest cDNA inserts were examined by hybrid selection as described below. We identified two recombinant M1 plasmids, designated p201 and p247, which contained cDNA inserts of 1.6 and 1.2 kb, respectively. Restriction analysis showed that they contain a 700-base pair overlap region and together span 2.1 kb (Fig. 3). To obtain clones extending in the 5' direction, we used a specific 18-mer (dTGGTTAGTCTCCACTCGC, complementary to a region of M1 mRNA 300 base pairs from the 5' end of p201) (Fig. 3) to prime the synthesis of cDNA which was cloned into the unique EcoRI site of AgtlO. Sixty nanograms of double-stranded cDNA gave a library of 7 x lo5 recombinant clones. Subunit M1 clones (-0.3% of the total) were selected by specific hybridization with a 32P-labeled 180base pair restriction fragment of p201 lying 20 base pairs upstream from the primer. One of these clones, designated MA-1, contained a cDNA insert of 1120 base pairs and was chosen for sequence analysis as described below.
Identification of Ribonucleotide Reductase Subunit MI cDNA Clones by Hybrid Selection-In order to determine if the cDNA clones isolated by differential screening contained M1 cDNA sequences, hybrid selection analysis was performed. Potential recombinant M1 plasmids were bound to diazobenzyloxymethyl paper filters and incubated with poly(A)+ RNA from either S49 or WEHI-7 cells. The hybridized mRNAs were eluted and translated in vitro. Two per cent of the translation products was analyzed directly by sodium dodecyl sulfate gel electrophoresis (Fig. 4a), while the remainder was subjected to immunoprecipitation prior to electrophoresis (Fig. 4, b and c). Lanes 11, 19, and 21 of Fig. 4 show that p201 and p247 specifically selected an mRNA encoding a 90-kDa protein immunoprecipitable with anti-M1 antibody. This mRNA was selected from both S49 and WEHI-7 poly(A)+ RNA.
DNA Sequence Analysis and Proof of the Identity of MI cDNA Clones- Fig.  3 shows the strategy used for sequence analysis of the cDNA inserts of p247, p201, and MA-1. A stretch of at least 30 adenosine residues, presumably corresponding to the poly(A) tail of the mRNA, preceded by the eukaryotic polyadenylation signal, AATAAA (Proudfoot and Brownlee, 1976), was present at one end of p247. This allowed us to orient the sequences of the cDNA clones with that of the mRNA. These contiguous clones contained a single open reading frame encoding a 90-kDa protein (Fig. 5). To provide definitive proof that this sequence encodes M1,15 of the first 18 amino acids of purified subunit M1 were determined. As shown in Fig. 5, these matched exactly with the N terminus predicted by the cDNA.
The mRNA used to generate the cDNA clones described above was isolated from mutant S49 cells that appear to express predominantly a dATP feedback-resistant form of M1.' To determine whether the sequence shown in Fig. 5 represents dATP-resistant or wild-type M1, we isolated and sequenced an M1 cDNA from mutant S49 cells that express both wild-type and dGTP/dTTP-resistant M1  (data not shown). This sequence was identical to that shown in Fig. 5, suggesting that it represents wild-type M1.
Southern Blot Analysis of Genomic DNAs from Various Species-To determine whether the mouse M1 gene showed homology with the gene in other species, we probed Southern blots (Southern, 1975) with restriction fragments of p247, p201, and MA-1 spanning the M1 sequence shown in Fig. 3. S49 cell DNA showed a pattern of four bands of sizes 5.2,5.8, 6.6, and -18 kb (Fig. 6). This pattern was identical to that seen with DNA from the BALB/c mouse (from which S49 cells are derived), indicating that no obvious rearrangements of the gene have taken place during 15 years in culture. Under stringent conditions, the mouse probe also hybridized to bands from a number of different species including human, pig, cat, horse, chicken, and yeast, suggesting that M1 is highly conserved.
Inspection of these homologies reveals four distinct domains: an N-terminal domain common to M1 and B1 but largely deleted in EBV, a C-terminal domain common to M1 and the EBV and HSV proteins but not B1, and two central domains present in all three proteins (residues 304-371: 38% homology between all three proteins, 24% exact matches; residues 501-548: 40% homology between the three proteins, 18% exact matches). These central domains were separated by a region of good homology (21%, 9% exact matches) to form a central, highly conserved core region (residues 304-548, 29% homology, 15% exact matches). Three conserved cysteine residues (at positions 218, 411, and 429) are located within or near this central region of homology. This homologous domain was flanked by regions of lower homology (13%, exact matches 4-5%).

DISCUSSION
We have isolated and characterized overlapping cDNAs encoding mouse ribonucleotide reductase subunit M1. The 2.9-kb composite sequence contains a single open reading frame of 2,376 nucleotides, predicting a protein of M, 90,234, which compares well with the estimated size of subunit M1 (Eriksson et al., 1981b). Definitive proof of the identity of these cDNAs was obtained by determining 15 of the first 18 residues of subunit M1 which exactly match the N-terminal sequence predicted by the cDNA (Fig. 5).
The nucleotide sequence contains a 446-base pair 3'-untranslated region with a polyadenylation signal, AATAAA (Proudfoot and Brownlee, 1976), 15 nucleotides from the poly(A) tail. Analysis of M1 mRNA on denaturing gels showed a single message of approximately 4.4 kb in S49, WEHI-7, and mouse L-cells (data not shown). The 3"untranslated region and the coding region together cover 2822 nucleotides, leaving -1580 nucleotides to be accounted for by the 5'untranslated region and the poly(A) tail. Analysis of 50 clones from the second library indicated that apart from a few short clones, the majority terminated approximately 80 nucleotides upstream from the initiator AUG, suggesting that this region may contain a structural block preventing reverse transcription through to the 5' end.
Subunit M1 of ribonucleotide reductase contains three distinct nucleotide-binding sites: the catalytic site and two allosteric sites. Regulatory mutants containing alterations in each of the allosteric sites have been well-characterized (Eriksson et al., 1981a). In this study, we isolated M1 cDNA clones using mRNA from two mutant S49 cell lines, one which appears to express predominantly dATP feedback-resistant (activity site mutant) M1* and one which contains both wildtype and dGTP/dTTP-resistant (specificity site mutant) M1 . A comparison of one complete sequence from each mutant line revealed no differences, suggesting that this sequence represents wild-type M1. To determine the sequence of the mutant forms of M1, we are now analyzing other cDNAs from each mutant line. This will allow us to probe the relationship of the enzyme structure to its function and lead to a deeper understanding of the molecular nature of the nucleotide binding and allosteric regulation.
There is great similarity at a functional level between ribonucleotide reductases isolated from species as widely separated as E. coli and mammals (Thelander and Reichard, 1979). The following two pieces of molecular data support this observation. Genomic DNA from a number of different species (mammalian, avian, and yeast) hybridized to mouse M1 sequences under conditions of high stringency (Fig. 7), suggesting that there is strong conservation at the DNA sequence level. We have also detected extensive amino acid sequence homology between mouse M1 and the recently published sequences of the analogous E. coli B1 (Carlson et al., 1984) and the ribonucleotide reductases of the herpesviruses, EBV (Gibson et al., 1984) and HSV (McLauchlan and Clements, 1983). Although conserved residues extend throughout the length of the polypeptide, we identified a highly homologous central domain, common to all three of the sequences compared (M1 residues 304-548, 28% conserved). Such strong sequence conservation among proteins from widely unrelated species may be taken to reflect functional constraints on the encoded product. For example, in a comparison of the amino acid sequences of bacterial and mammalian dihydrofolate reductases, approximately 70% of the conserved residues were found in the regions which form the hydrophobic binding site of the enzyme (Simonsen et al., 1983). We therefore propose that the conserved central domain in ribonucleotide reductase subunit M1 may comprise part of the catalytic site. In support of this, active dithiols have been implicated in the catalytic reduction by ribonucleotide reductase and are known to be located on subunit B1 (Thelander and Reichard, 1979). We note that this highly conserved region contains 2 conserved cysteine residues (Cys 411 and Cys 429). A third conserved cysteine (Cys 218) lies outside, but close to the region of homology. The homologies revealed two further domains, an N-terminal domain common to M1 and B1 but largely deleted in EBV, and a C-terminal domain containing sequences strongly conserved in M1 and the EBV and HSV reductases, but not in B1. Since the cellular enzymes are both subject to complex and similar allosteric regulation by deoxynucleoside triphosphates and ATP, whereas the viral enzymes are not (Langelier and Buttin, 1981;Lankinen et al., 1982), it is tempting to speculate that this N-terminal region is involved in the regulation. The significance of the C-terminal domain is presently unclear, but might reflect an interaction with a mammalian cellular component such as the second subunit of ribonucleotide reductase.
The homologies described above are consistent with the assigned map locations of the ribonucleotide reductase gene in HSV (Dutia, 1983) and EBV (Gibson et al., 1984). Our data suggest that the EBV 93-kDa reading frame and the HSV 140-kDa gene correspond to the large subunit of mammalian or E. coli ribonucleotide reductase. Although the subunit structure of the virally encoded enzymes has not been defined, an intriguing possibility is that the EBV 34-kDa reading frame (Gibson et al., 1984) and the HSV 38-kDA gene (Bacchetti et al., 1984) encode the viral equivalent of the small subunit (55 kDa) of the cellular enzyme.
The availability of an M1 cDNA should allow identification of sequences that mediate the cell cycle and/or growth regulation of this important cell enzyme. To this end, genomic sequences of mouse M1 have been isolated and are under investigation.