key: cord-345630-bam3pa70
authors: Lee, Han-Jung; Shieh, Chien-Kou; Gorbalenya, Alexander E.; Koonin, Eugene V.; La Monica, Nicola; Tuler, Jeremy; Bagdzhadzhyan, Anush; Lai, Michael M.C.
title: The complete sequence (22 kilobases) of murine coronavirus gene 1 encoding the putative proteases and RNA polymerase
date: 1991-02-28
journal: Virology
DOI: 10.1016/0042-6822(91)90071-i
sha: 
doc_id: 345630
cord_uid: bam3pa70

Abstract The 5′-most gene, gene 1, of the genome of murine coronavirus, mouse hepatitis virus (MHV), is presumed to encode the viral RNA-dependent RNA polymerase. We have determined the complete sequence of this gene of the JHM strain by cDNA cloning and sequencing. The total length of this gene is 21,798 nucleotides long, which includes two overlapping, large open reading frames. The first open reading frame, ORF 1 a, is 4488 amino acids long. The second open reading frame, ORF 1 b, overlaps ORF 1 a for 75 nucleotides, and is 2731 amino acids long. The overlapping region may fold into a pseudoknot RNA structure, similar to the corresponding region of the RNA of avian coronavirus, infectious bronchitis virus (IBV). The in vitro transcription and translation studies of this region indicated that these two ORFs were most likely translated into one polyprotein by a ribosomal frameshifting mechanism. Thus, the predicted molecular weight of the gene 1 product is more than 800,000 Da. The sequence of ORF 1 b is very similar to the corresponding ORF of IBV. In contrast, the ORF 1 a of these two viruses differ in size and have a high degree of divergence. The amino acid sequence analysis suggested that ORF 1 a contains several functional domains, including two hydrophobic, membrane-anchoring domains, and three cysteine-rich domains. It also contains a picornaviral 3C-like protease domain and two papain-like protease domains. The presence of these protease domains suggests that the polyprotein is most likely processed into multiple protein products. In contrast, the ORF 1b contains polymerase, helicase, and zinc-finger motifs. These sequence studies suggested that the MHV gene 1 product is involved in RNA synthesis, and that this product is processed autoproteolytically after translation. This study completes the sequence of the MHV genome, which is 31 kb long, and constitutes the largest viral RNA known.

Mouse hepatitis virus (MHV), a murine coronavirus, contains a single-stranded, positive-sense RNA genome (Lai and Stohlman, 1978; Wege eta/., 1978) . The genomic organization is well understood Lai, 1990) . It contains 8 genes, each of which is expressed from the 5'-end of a polycrstronic mRNA species. These mRNAs have a 3'-coterminal, nestedset structure (Lai et al., 1981) . Starting from the 5'-end of the genome, the genes are named 1, 2a, 2b, 3, and so on until gene 7 (Cavanagh eta/., 1990) . Genes 2b, 3, 6, and 7 encode the four known viral structural proteins, i.e., HE (hemagglutinin-esterase), S (spike), M (membrane), and N (nucleocapsid) proteins, respectively. The remaining genes presumably encode nonstructural proteins, most of which are yet to be identified in the virus-infected cells. The nucleotide sequences of genes 2 to 7 have been determined for two strains, A59 and JHM, of MHV (Armstrong eta/., 1983 (Armstrong eta/., , 1984 Skinner and Siddell, 1983, Sequence data from this article have been deposlted with the EMBUGenBank under Accession No. M55148 .

' To whom correspondence should be addressed.

1985; Schmidt et a/., 1987; Luytjes et a/., 1987 Luytjes et a/., , 1988 Shieh et al., 1989) . Altogether these seven genes account for roughly 9.5 kb. The remaining gene, gene 1, which is the 5'-most gene, has been estimated to be longer than the size of all of the other genes combined (Pachuk eta/., 1989; Baker et al., 1990) . Only the 5'-terminal 5.3 kb in JHM strain and the 3'-terminal 8.4 kb of this gene in A59 strain have so far been sequenced Baker et al., 1989; Pachuk et al,, 1989; Bredenbeek et al., 1990) . The corresponding gene of an avian coronavirus, infectious bronchitis virus (IBV), has been completely sequenced and shown to be 20 kb long . This IBV gene consists of two open reading frames (ORFs), which can be translated into a polyprotein via a ribosomal frameshifting mechanism (Brierley et a/., 1987 (Brierley et a/., , 1989 . Again, the gene products have yet to be detected in the virus-infected cells. The size of MHV gene 1 has not been determined. From the approximate sizes of the cDNA clones, it has been estimated to be roughly 22-23 kb (Pachuk et al,, 1989; Baker et al,, 1990) . Comparison of the published partial sequences of gene 1 showed that IBV and MHV share sequence similarity in the 3'-terminus of the gene ( , 1990) , and yet their 5'-ends are diverged (Soe eta/., 1987; Baker et al., 1989) . Thus, the evolutionary relationship of these two viruses in gene 1 is not clear.

Several pieces of evidence suggest that gene 1 may encode proteins which are directly involved in viral RNA synthesis: First, since MHV does not contain RNA polymerase (Brayton et al., 1982) , this enzyme has to be synthesized from the incoming virion genomic RNA. This translation is only possible if the gene is located at the 5'-end of the genome. Second, RNA recombination studies using temperature-sensitive (ts) mutants indicated that the ts lesions affecting RNA synthesis are localized within the gene 1 region (Keck et al., 1987) . This conclusion has been confirmed by RNA recombination mapping studies (Baric et a/., 1990) . Third, the 3'-half of the gene 1 sequences of IBV and MHV-A59 contains the sequence motifs for RNA polymerase and helicase, which are the activities expected to be involved in RNA synthesis Gorbalenya et a/., 198913; Bredenbeek et a/., 1990) . However, these postulated functions have not been directly demonstrated. At least one enzymatic activity, i.e., an autoprotease , has been associated with the gene product. The presence of the protease activity suggests that the gene 1 product is likely to be processed into multiple proteins.

The properties of the RNA polymerase of coronavirus are of considerable interest since the coronavirus RNA synthesis utilizes an unusual mechanism of discontinuous transcription, probably involving a free leader RNA species (Lai, 1988) . The understanding of the RNA polymerase should shed further light on the mechanism of RNA synthesis. To this end, we have obtained the complete sequence of gene I of the JHM strain of MHV. This gene is nearly 22,000 nucleotides long and contains two overlapping ORFs, similar to the corresponding IBV gene. Sequence analysis shows that the MHV gene may have undergone extensive divergence from the IBV gene, particularly at its 5'-half. Several functional domains were identified, which may be important for the processing and the enzymatic activities of its gene product.

Virus and cells. The plaque-cloned JHM strain of MHV (Makino et a/., 1984) was used throughout this study. The virus was propagated on DBT cells (Hirano et a/., 1974) at m.o.i. of 1. Virus was harvested and purified from the medium, and viral RNA was prepared as previously described (Makino et a/., 1984) .

CDNA cloning. The cDNA c\ones encompassing ,  I  I  I  I  ,  I  I  I  /  ,  I  I  I  I  I  1  2000  2500  3000  3500   3   ORF la/lb  0   ORF lb   ,  I  I  I  I  ,  I  I  I  I  ,  I  I  I  I  I  '  '   1000   1500  2000  2500 FIG. 2. Hydropathy profiles of the predicted amino acid sequences of ORF 1 a and ORF 1 b. Values above the lrne are hydrophobic and values below the line are hydrophilic. The hydropathicrty was calculated using a moving window of 40 amino acids, with a value plotted every 16 residues (Kyte and Doolittle, 1982) . gene 1 were obtained by using specific synthetic oligo-gonucleotides were derived from RNA sequence analynucleotides as primers and purified virion genomic sis of the RNase Tl-resistant oligonucleotides which RNA as template. Initially, the sequences of these olihad been mapped to either gene 1 or 2 (Shieh et al, , 13, 600 13,650

Nucleotide number 13,700

FIG. 3. Dragram of the codon preference in the region between ORF 1 a and ORF 1 b. The codon usage patterns for the three reading frames of the predicted amino acid sequences at the junction between the ORF 1 a and ORF 1 b are shown. The two stop codons at 13600 (TAG) and 13679 (TAA) are marked. The codon usage table was generated for genes 3, 6, and 7, which encode the viral structural proteins, of MHV-IHM (Schmidt er a/., 1987; Skinner and Siddell, 1983) and used for comparison with ORFs 1 a and 1 b. The parameters used are a window length of 25 and a maximum scale of 1.1 (Gribskov et al., 1984) . l BV-M42 5' 12337 GAUAAGAAUUAUUUAAACGGGUACGGGGUAGCAGUG----AGGCUCGGCUGAUACCCCUUGCUAGUGG 3'   II I lllll  11111111111 llllllI  IIII  I II II 11-11111 III Illll  MHV-JHM 5' 13643 GACACGAAUUUUUUAAACGGGUUCGGGGUACAAGUGUAAAUGCCCGUCUUGUACCCUGUGCCAGUGG   3'   MHV-A59 5' 284   llllllll  11111111111 IIIIIIIIIIIIIIIlTrIIIIIIIIIIIIIIIIIIIIIIIIIIII   GACACGAACUUUUUAAACGGAUUCGGGGUACAAGUGUAAAUGCCCGUCUUGUACCCUGUGCCAGUGG  3 

Comparison of the RNA sequences and the proposed secondary structure of the MHV-JHM, MHV-A59 and IBV RNAs at the junction between ORF 1 a and ORF 1 b. (A) Alignment of nucleotide sequences. The first nucleotides are numbered according to Boursnell era/. (1987) for IBV, and Bredenbeek ef al. (1990) for MHVA59, and termrnation codons are underlrned. (B) Tertiary RNA structure at the region of ribosomal frameshifting. The potential signal for ribosomal frameshifting is boxed, and the stop codon is underlined. Arrows indicate the differences in the RNA sequence of MHV-JHM in comparison with that of IBV (boldfaced) and MHV-A59 (outlined). Soe et al., 1987) . cDNA synthesis was were trimmed with T4 DNA polymerase and ligated to performed by the general method of Gubler and Hoff-pTZ18U (United States Biochemical Corp.) either by man (1983) . The double-stranded cDNA molecules blunt-end ligation or EcoRl linker ligation. The recombi- 4, and 7) . generating a 0.5.kb RNA. Translation was performed in a rabbit reticulocyte lysate system using [35S]methionine. Translation products were analyzed directly (lanes l-3) or after immunoprecipitation using the ORF 1 a-specific antiserum (lanes 4-6) or rabbrt preimmune serum (lanes 7-9). M indicates molecular weight markers in kilodaltons; lanes 3. 6, and 9, translation of pTZ(ORFaug). nant DNAs were transformed into Escherichia co/i strain MVl 190 competent cells (Dagert and Ehrlich, 1979) . Homopolymer dC tailing to the 3'-end of the cDNAs using terminal transferase were also used to anneal to Pstl-linearized pBR322 with oligo(dG) tails and transformed into E. co/i strain MC1 061. Specific cDNA clones were identified using 5'-end-labeled oligonucleotides as probes and confirmed by subsequent hybridization to viral mRNA . Once the sequences of the cDNA clones were obtained, oligonucleotides complementary to the 5'-ends of these clones were synthesized to serve as primers for addi-tional cDNA cloning to obtain overlapping cDNA clones.

DNA sequencing. Sequencing was performed as previously described . Both chemical modification (Maxam and Gilbert, 1980) and dideoxynucleotide chain termination @anger et a/., 1977) methods were used directly on plasmid DNA (Chen and Seeburg, 1985) .

Construction of recombinant plasmids for the frameshifting analysis. Subcloning and mutagenesis of cDNA clone T-l 2 was accomplished using synthetic oligonucleotides and polymerase chain reaction (PCR). Briefly, oligomer #166 (5'-GATCGAATTCCTTTACAT-GGTGAAGGGGTG-3') which extends from nucleotide 13,147 to 13,167 of gene 1 and contains mismatches at both nucleotides 13,154 and 13,156, and oligomer #199

(5'-CATATGACACAGGATCCTTTATGCC-3') which is complementary to nucleotides 13,529 to 13,553 and includes the BamHl site at nucleotide 13,537, were used for DNA amplification by PCR according to the standard procedures (Saiki et al., 1988) . The resulting PCR DNA product encompasses sequences from nucleotide 13,147 to 13,537 with a specific mutation (T to A) at nucleotide 13,154 and another (T to G) at nucleotide 13,156, resulting in the introduction of an ATG codon. The DNAwas then digested with II  I IIll  III  IIII  lllll  Ir III1  IIIII  III  QtS~GVCVVCNSPTI The overall alignment was generated by combining segments aligned by programs OPTAL (Gorbalenya et al., 1989a) and MULTALIN (Corpet, 1988) . It consists of four distinct pieces separated by regions that could not be aligned with certainty. For the latter regions, only the total numbers of amino acid residues are indicated. The amino acid numbers of the first and the last residues of each aligned segment are indicated in parentheses. Two dots, identical residues; single dots, similar residues. Conserved Cys residues are highlighted by boldface. Asterisks, putative catalytic residues of proteases; arrows, putative cleavage sites for BC-like proteases. Box, the putative cleavage site for 3CLpr0 in IBV substituted by a KR dipeptide in MHV-JHM. The IBV sequence was from Boursnell et al. (1987) . MHVA: ORF la of MHV. IBVFl: ORF la of IBV. previously described . The resulting In vitro transcription and translation. Recombinant RNA was translated in the mRNA-dependent rabbit replasmids pTZ(ORFaUg) and pTZ (FrSh) Shin and Morrison (1989) and analyzed by electrophoresis on 7.5 to 15% polyacrylamide gel.

Computer analysis of nucleotide and amino acid sequences. Sequence data were analyzed on a VAX 1852 using the GCG sequence analysis software package developed by Genetics Computer Group of University of Wisconsin. Detailed comparative analyses of coronavirus protein sequences were done by programs MULTALIN (Corpet, 1988 ) OPTAL (Gorbalenya et a/., 1989a), DOTHELIX (Leontovich et al., 1990) , and SITE (Koonin et a/., 1990) . The programs DOTHELIX and SITE are parts of the GENBEE program package for biopolymer sequence analysis.

To clone the gene I region, which represents more than two thirds of the MHV genome, a synthetic oligonucleotide (oligo 30; 5'-CTGAATFTGGGG-GTTGGG-3') was initially used as a primer for cDNA synthesis . The sequence of this oligonucleotide was based on the sequence analysis of the RNase Tl-resistant oligonucleotide No. 30, which had previously been mapped to gene 2 (Makino et al., 1984) . The resulting cDNA clones contained inserts ranging from 0.5 to 3 kb in size. These cDNA clones detected only the genomic RNA on Northern blots of intracellular RNAfrom MHV-infected cells (data not shown). Based on the nested-set structure of MHV 9 . A schematic presentation of the relationship between the ORF 1 a of MHV-JHM and IBV. The two ORF 1 a are shown to scale. The designation of regions, for which specific functional predictions could be made, and of regions of similarity between the two viruses are shown in the bottom of the figure. High similarity, statistical significance over 10 SD (standard deviation), when aligned by the program OPTAL (Gorbalenya eta/., 1989a,b) ; moderate similarity, significance of 3 to 10 SD. The alignments in the regions, with predicted functions, were significant at the level of at least 5 SD. Regrons of similarity between the two viruses are joined. Vertical arrows, putative cleavage sites for 3CLpr0'). Horizontal arrows, putative papain-like proteases (two copies in MHV-JHM, and one copy in IBV). mRNAs (Lai et a/., 1981) , this result indicated that these cDNA clones represent part of gene 1. The 5'ends of these DNAs were sequenced, and synthetic oligonucleotides complementary to these sequences were generated to prime further cDNA synthesis for walking toward the 5'-end of gene 1. In this way, overlapping DNA clones which encompass about 11 kb at the 3'-end of gene 1 were obtained (Fig. 1) . cDNA clones representing the 5'-terminal 6.2 kb of gene 1 were derived as described . The cDNA clones spanning the gap between the two cDNA groups were obtained by using specific primers representing both the sequences downstream and upstream of the gap as primers for first-strand and second-strand cDNA synthesis, respectively. The overlap of these cDNA clones was determined by Southern blotting and confirmed by DNA sequencing. The complete cloning of JHM gene 1 indicated that the size of gene 1 is approximately 22 kb in length (Fig. l) , longer than that of IBV , and agrees with the previous estimate for the gene 1 of the A59 strain of MHV (Pachuk et al., 1989) .

Analysis of the nucleotide sequence and the predicted amino acid sequence. The complete MHV-JHM gene 1 sequence was obtained from the cDNA clones as indicated in Fig. 1 . This sequence has been deposited with GenBank (Accession No. M55148), and will not be duplicated in this publication. The complete sequence of gene 1 contains 21,798 nucleotides preceding the UCUAUAC, which is the transcriptional initiation site for gene 2 . Analysis of the sequence revealed two large, overlapping open reading frames (ORFs), ORF 1 a and ORF 1 b (Fig. 1 a) . ORF 1 a is 4488 amino acids long and has a predicted molecular weight of 499,319, which includes the coding region for p28 protein at its N-terminus . The hydropathy plot (Kyte and Doolittle, 1982) shows that ORF 1 a has several long stretches of hydrophobic regions at the carboxy-terminal region, which indicate potential membrane-spanning domains (Fig. 2) . ORF 1 b, which overlaps ORF 1 a for 75 nucleotides but is located at a different reading frame, is 2731 amino acids long with a predicted molecular weight of 308,483. The ORF 1 b sequence is very similar to that of MHV-A59 in both nucleotide and predicted amino acid sequences (Bredenbeek et al., 1990) . Only minor substitutions were noted between the two strains (data not shown). The ORF 1 b starts with CUG instead of AUG. The first potential initiator codon AUG is located 399 nucleotides downstream of the first amino acid The numbers of amino acid residues to the known or postulated termini of the respective viral 3CLPro and between the aligned segments are indicated. For MHV 3CLpro, the postulated N-terminus is at amino acid residue 3350 ( Fig. 8 Gorbalenya et a/. (1989b) , except BWYV (Veidt et a/., 1988) and SBMV (Wu et al., 1987) . Riore. In the aa position columns, the amino acid positions of the respective Q residues are indicated. The arrows show the predicted cleavage sites, Abbreviations:

MPl , MP2, putative membrane proteins flanking the 3CLpr0 at the N-and C-sides, respectively. POL: polymerase motif. HEL: helicase motif. GFL: growth factor-like domain. The data on IBV was obtained from Gorbalenya ef al. (1989b) . The sequence analysis was performed using the computer program as described under Materials and Methods. codon in ORF 1 b. Nevertheless, the codon preference plot suggests that the 399 nucleotides upstream of the first AUG are most likely translated together with the downstream sequences using the same reading frame (Fig. 3) . In light of the corresponding sequences of IBV and MHV-A59 Bredenbeek et a/., 1990) , this result suggests that this region could be translated via a ribosomal frameshifting mechanism (Brierley et al,, 1989) .

Comparison of tertiary structure of RNA in the frameshift regions. It has been proposed that the nucleotide sequences in the overlapping regions between ORF 1 a and ORF 1 b in IBV and MHV-A59 RNAs are able to fold into a pseudoknot tertiary structure, which is essential for efficient frameshifting and, thus, expression of the downstream ORF 1 b (Brierley et a/., 1989; Bredenbeek et al., 1990) . Comparison of the primary sequence revealed that the corresponding region of MHV-JHM contains a "slippery" sequence, UUUAAAC, similar to that of IBV (Fig. 4A) . The possible folding of RNA in this region into a pseudoknot tertiary structure is similar among IBV, MHV-A59, and MHV-JHM (Fig. 4B) . It is interesting to note that the nucleotide changes between MHV-JHM and IBV in either the stem or loop regions are compensated by mutations at the complementary positions (Fig. 48) . This suggests the significance of the putative tertiary structure in ribosomal frameshifting.

Only two nucleotides differ between MHV-JHM and MHV-A59 in this region; they are located at the regions immediately upstream and downstream of the UUUAAAC sequence.

Ribosomal frameshifting in vitro. To confirm that the ORF 1 a and 1 b of MHV-JHM could be translated into one polypeptide by ribosomal frameshifting, we cloned the region spanning from nucleotide 13,147 to 14,164 of gene 1 into an expression vector under the control of the T7 promoter for in vitro translation studies. Because of the lack of a translational initiation codon, an ATG codon was introduced by PCR-mediated mutagenesis at nucleotide 13,154-l 3,156. If the translation of this transcript terminates at the UAA stop codon in ORF 1 a, a 19-kDa protein will be produced. However, if the -1 translational frameshift occurs, a 37-kDa protein will be synthesized. As shown in Fig. 5 , the in vitro translation of this RNA yielded both proteins (lane 2). The 37-kDa protein was heterogeneous; the smaller proteins may represent aberrant translational initiation or specific processing of the translation products. The addition of protease inhibitors in the rabbit reticulocyte lysates did not alter this translation pattern (data not shown). The antiserum prepared against the amino acid sequence just upstream of the frameshift (unpublished) precipitated both proteins (lane 5). Surprisingly, the major products precipitated by this anti- serum migrated faster than the respective primary translation products, suggesting that protein processing had occurred. None of the proteins was immunoprecipitated by the preimmune serum. As controls, the transcripts containing either the 5'-or the 3'-halves [pTZ(ORF""g)] of the ORF did not yield the 37-kD protein. As predicted, only the products of the 5'-half were precipitated by this antibody (Fig. 5B, lane 4) . These results are in agreement with the results obtained with IBV (Brierly et a/., 1987) and MHV-A59 (Bredenbeek et a/., 1990) .

Analysis of sequence homology among MHV-JHM, MHV-A59, and IBV. The comparison of nucleotide and predicted amino acid sequences between MHV-JHM and IBV revealed considerable similarity between the two. The dot matrix comparison of the amino acid sequences shows that ORF 1 b is very similar between MHV and IBV (Fig. 6) . Overall, there are 47.7% similarity at nucleotide level and 52.8% at amino acid level. Similar to the ORF 1 b of IBV, the MHV ORF contains the polymerase and helicase motifs at the corresponding positions (Gorbalenya et al., 1989b ) (data not shown). The putative zinc-binding domain is also largely conserved between the two viruses. On the other hand, two of the residues implicated in metal binding for IBV (Gorbalenya et al., 198913) are replaced in MHV, suggesting that the specific structures of the putative "fingers" may differ (Fig. 7) . The ORF 1 b of MHV-JHM and MHV-A59 are also very similar (95.9% at nucleotide level, and 94.9% at amino acid level) (data not shown).

In contrast, the ORF 1 a is more diverged (Fig. 8) . The MHV ORF 1 a is longer than the corresponding IBV ORF by 537 amino acids. The C-terminal half of the ORF 1 a is relatively conserved between MHV-JHM and IBV, while the N-terminal half is very diverged (Fig. 6) . The alignment of amino acids in ORF la of MHV-JHM and IBV showed that there are four possible stretches of moderate homology which are separated by highly diverged sequences (Fig. 8) .

Analysis of the functional domains of ORF la. Although ORF la is highly diverged between MHV-JHM and IBV, common functional domains could be identified in this ORF of both viruses by detailed amino acid sequence analysis (see Materials and Methods) (Fig.  9 ). Two hydrophobic, potentially membrane-anchoring regions are present in the C-terminal half. There are three cysteine-rich domains, one of which contains a segment distantly resembling growth factors and their receptors (Gorbalenya et al., 1989b) . In both coronaviruses, homologous domains of about 300 residues each have been identified to be related to the putative 3Clike proteases (3CLpro) of picorna-, coma-, nepo-, poty-, sobemo-and luteoviruses (Gorbalenya et al., 1989b) . The sequences of the putative coronavirus 3C-like proteases possess certain unusual features distinct from that of other viral 3Clike proteases (Fig.  10 , and see Discussion). The search for sequences resembling the cleavage sites for the 3C-like proteases revealed six conserved putative target sites for the MHV and IBV 3C-like proteases (Table 1 ) (see Discussion). These potential cleavage sites are localized in the ORF 1 b and the C-terminal half of the ORF la. Interestingly, the N-terminal one of these cleavage sites marks the N-end of the putative 3Clike protease itself. Finally, there is a region of moderate conservation between MHV and IBV, which contains short segments resembling those around the catalytic Cys and His residues of papain-like proteases (Fig. 11) . This region is duplicated in the MHV genome, but not in IBV, at an upstream site in the ORF la. This upstream papain-like cysteine protease has been identified as the one responsible for the cleavage of p28 from the N-terminus of the gene 1 protein . A domain of considerable conservation between MHV and IBV (X domain in Fig. 9 ) has been found next to the putative coronavirus papain-like proteases. Interestingly, a homologous conservative domain also flanks the putative thiol proteases of alpha-and rubiviruses (A. E. Gorbalenya, unpublished observations) .

The complete sequence of gene 1 of MHV presented in this paper shows that this gene is probably the largest known viral gene among RNA viruses. Evidence was presented suggesting that the two ORFs in this gene may be translated into a large polyprotein. This interpretation is consistent with the lack of the transcriptional initiation signal (UCUAAAC) in the entire gene 1 sequence except at the extreme 5'-end. Although the putative "slippery" sequence (UUUAAAC) between the ORF 1 a and 1 b (Brierly et a/., 1989) is similar to the transcriptional initiation signal, no major subgenomic mRNAs have been detected within this gene. Thus, this gene most likely encodes a single polyprotein of at least 800 kDa. The total size of the RNA genome of MHV is approximately 3 1 kb, which is considerably larger than any of the other known viral RNA. The evolution of the coronavirus RNA genome into such a large RNA may have reflected the unusual mechanism of coronavirus RNA synthesis. The complexity of the discontinuous mode of coronavirus RNA synthesis (Lai, 1988) suggests that the coronavirus RNA polymerase needs a variety of different enzymatic activities.

The amino acid sequence of gene 1 of MHV shows considerable similarity to that of IBV. The ORF 1 b is particularly conserved. Its degree of conservation between MHV and IBV is higher than that for any of the other genes in the coronavirus genomes. The ORF 1 b contains the polymerase, helicase, and metal-binding motifs (Gorbalenya el a/., 1989b) , suggesting that this region may be directly involved in RNA synthesis. These structural features are conserved between these viruses. The proposed pseudoknot structure which is important for the ribosomal frameshifting for cotranslation of ORF 1 a and ORF 1 b (Brierley et a/., 1989) is also highly conserved. This fact has previously been recognized in the partial sequence of gene 1 of MHV-A59 (Bredenbeek eta/., 1990) . The sequence differences between MHV-A59 and MHV-JHM within this junction region are located at the nucleotides which do not affect the putative pseudoknot structure. In contrast, ORF la is much more diverged. It is nearly 2 kb longer than the ORF la of IBV, and contains several stretches of sequence which are not present in the IBV genome. These nonhomologous stretches of sequence are interspersed between the conserved regions. Furthermore, a papain-like protease domain, which is present once in the IBV genome, is duplicated in the 5'-half of the ORF la of MHV. The N-terminal sequence including ~28, which is cleaved by the papain-like protease of MHV , is also highly diverged between MHV and IBV. Thus, it appears that the 5'-end of ORF 1 a has undergone considerable sequence rearrangement and possibly recombination, while the remaining sequences in gene 1 are almost colinear between MHV and IBV.

In contrast to the ORF 1 b which contains sequence motifs related to the synthesis of RNA, the ORF la contains several domains suggestive of other functions. First of all, there are two long stretches of hydrophobic domains, which are conserved between IBV and MHV. The presence of these domains suggests that the gene 1 products may be anchored to the membrane. This possibility is consistent with the finding that MHV RNA synthesis occurs on the membrane fractions in the infected cells (Brayton et al., 1982) . Second, there are three cysteine-rich regions, which are also homologous between MHV and IBV. The function of the Cys-rich domains is still not clear. However, it has been noted with IBV that the C-terminal Cys-rich domain is related to that of the growth factors and their receptors (Gorbalenya et al., 1989b) . Third, there is a 3C-like protease domain (3CLpro) in the 3'-half of ORF 1 a, which is also conserved in IBV. The putative catalytic His and Cys residues previously predicted in IBV have also been observed in MHV (Fig. 10) . However, the putative coronavirus proteases remain unique in that they do not contain a conserved Asp(Glu) residue that could serve as the third catalytic residue as suggested for the other 3C-like proteases (Gorbalenya et al., 1989b) . Furthermore, the unusual substitution of Tyr for Gly in the putative substrate-binding region, described previously in IBV, is also observed in the putative MHV 3CLp" (Fig. 10) . The potential cleavage sites for this 3C-like protease have been identified to be mainly in ORF 1 b and the C-terminus of ORF 1 a (Gorbalenya et al., 1989b) . These sites (QS) are either conserved or converted to QA in MHV (Table 1 ). The potential cleavage at Q/S and Q/A sites by picornavirus 3CLP" has been demonstrated previously (Parks and Palmenberg, 1987) . Two QG dipeptides proposed to be cleaved in IBV were substituted in MHV by QC in one case, and by KR dipeptide in another (Table 1) . Substitution of a C (unlike several other residues) for G in a cleavage site for encephalomyocarditis virus protease did not abolish processing in an in vitro system (Parks et a/., 1989) . Dibasic dipeptides are cleaved in the polyproteins of flaviviruses (Strauss and Strauss, 1988) . Thus, these postulated cleavage sites are potentially cleavable by MHV 3CLpro despite the divergence. These cleavages could separate different functional domains of the gene 1 polyprotein into distinct protein products. Whether these sites are indeed cleaved in MHV-infected cells remains to be studied. Fourthly, the N-terminal portion, which is the most diverged region, contains a papain-like protease domain as pointed out previously for IBV (Gorbalenya et al., 1989b) . The papain protease domain is duplicated in the MHV ORF 1 a (Fig. 1 1) and is homologous with the known proteases (Fig. 12) . This protease is probably involved in the cleavage of the N-terminus of the gene 1 polyprotein (Baker eta/., 1989) , which has been demonstrated in MHV-infected cells (Denison and Perlman, 1987) . Site-specific mutagenesis studies demonstrated that this protease has Cys and His at its active site (unpublished observation).

The possible presence of the protease domains suggests that the gene 1 polyprotein is processed into many proteins. It has been shown that there are at least five to six complementation groups involving MHV RNA synthesis, five of which have been mapped within gene 1 (Leibowitz eta/., 1982; Baric eta/., 1990) . These proteins conceivably participate in various aspects of MHV RNA synthesis. None of the proteins have been detected so far.

Sequence of the nucleocapsid gene from murine coronavirus MHV-A59

Sequence and topology of a model intracellular membrane protein, El glycoprotein, from a coronavirus

Murine coronavirus gene 1 polyprotein contains an autoproteolytic activity

Identification of a domain required for autoproteolytic cleavage of murine coronavirus gene A polyprotein. 1. Viroi

Establishing a genetic recombination map for murine coronavirus strain A59 complementation groups

Completion of the sequence of the genome of the coronavirus avian infectious bronchitis virus

Characterization of two RNA polymerase activities induced by mouse hepatitis virus

The primary structure and expresslon of the second open reading frame of the polymerase gene of the coronavirus MHV-A59; a highly conserved polymerase is expressed by an efficient rlbosomal frameshifting mechanism

An efficient ribosomal frame-shifting signal in the polymerase-encoding region of the coronavirus IBV

Characterization of an efficient coronavirus ribosomal frameshifting signal: Requirement for an RNA pseudoknot

Recommendations of the coronavirus study group for the nomenclature of the structural proteins, mRNAs and genes of coronavirus

Supercoil sequencing: A fast and simple method for sequencing plasmid DNA

Multiple sequence alignment with hierarchical clustering

Prolonged incubation in calcium chloride improves the competence of EscherYchia co/i cells

Identification of putative polymerase gene product in cells infected with murine coronavirus A59

A comprehensive set of sequence analysis programs for the VAX

Purification and amino acid sequence of chicken liver cathepsin L

An NTP-binding motif is the most conserved sequence in a highly diverged monophyletic group of proteins involved in positive strand RNA viral replication

Coronavirus genome: Prediction of putative functional domains in the nonstructural polyprotein by comparative amino acid sequence analysis

The codon preference plot: Graphic analysis of protein coding sequences and prediction of gene expression

A simple and very efficient method for generating cDNA libraries

Replication and plaque formation of mouse hepatitis virus (MHV-2) in mouse cell line DBT culture

Multiple recombination sites at the 5'-end of murine coronavirus RNA

A method for localization of motifs in amino acid sequences. Biopolim. Kletka

A simple method for displaying the pathic character of a protein

Replication of coronavirus RNA

Coronavirus: Organization, replication and expression of genome

Mouse hepatitis virus A59: Messenger RNA structure and genetic localization of the sequence divergence from the hepatotropic strain MHV 3

The RNAof mouse hepatitis virus

Genetic analysis of murine hepatitis virus strain JHM. 1. Viral

A method for generation of complete local similarity maps between two amino acid sequences. DOTHELIX program of the GENBEE package

Primary structure of the glycoprotein E2 of coronavirus MHV-A59 and identification of the trypsin cleavage site

Sequence of mouse hepatitisvirus A59 mRNA 2: Indications for RNA recombination between coronavirus and influenza C virus

Analysis of genomic and intracellular viral RNAs of small plaque mutants of mouse hepatitis virus

Sequencing end-labeled DNA with base-specific chemical cleavages. ln

Evolutionary origin of a calcium-dependent protease by fusion of genes for a thiol protease and a calcium-binding protein

Cloning and characterization of a mouse cysteine proteinase

Molecular cloning of the gene encoding the putative polymerase of mouse hepatitis coronavirus strain A59

Proteolytic cleavage of encephalomyocarditis viral capsid region substrates by precursors to the 3C enzyme

Site-specific mutations at a picornavirus VPB/VPl cleavage site disrupt in vitro processing and assembly of capsid precursors

Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase

DNA sequencing with chain-terminating inhibitors

Nucleotide sequence of the gene encoding the surface projection glycoprotein of coronavirus MHV-JHM

Identification of a new transcriptional initiation site and the corresponding functional gene 2b in the murine coronavirus RNA genome

The 5'.end sequence of the murine coronavirus genome: Implications for multiple fusion sites in leaderpnmed transcription

Production and properties of chimeric antibody molecules

Coronavirus MHV-JHM mRNA 5 has a sequence arrangement which potentially allows translation of a second, downstream open reading frame. 1. Gem Viral

Coronavirus JHM: Nucleotide sequence of the mRNA that encodes nucleocapsid protein

Coding sequence of coronavirus MHV-JHM mRNA 4

Sequence and translation of the murine coronavirus Y-end genomic RNA reveals the N-terminal structure of the putative RNA polymerase. 1. Viral

Coronaviruses: Structure and genome expression. 1. Gem Viral

Replication of the RNAs of alphaviruses and flaviviruses

Nucleotide sequence of beet western yellows RNA

Genomic RNA of the murine coronavirus JHM. 1

Sequence and organization of southern bean mosaic virus RNA

We thank Dr. Susan Baker for advice throughout the course of the study. We also thank Lisa Banner and Daphne Shimoda for editorial assistance. A.E.G. and E.V.K. are most grateful to Professor V. I, Ago1 for constant support, to Dr. L. I. Brodsky for supply of the GEN-BEE program package, and to Dr.