key: cord-253024-b393ea2u
authors: Fu, Kaisong; Baric, Ralph S.
title: Evidence for variable rates of recombination in the MHV genome
date: 1992-07-31
journal: Virology
DOI: 10.1016/0042-6822(92)90684-h
sha: 
doc_id: 253024
cord_uid: b393ea2u

Abstract Mouse hepatitis virus has been shown to undergo RNA recombination at high frequency during mixed infection. Temperature-sensitive mutants were isolated using 5-fluorouracil and 5-azacytidine as mutagen. Six RNA+ mutants that reside within a single complementation group mapping within the S glycoprotein gene of MHV-A59 were isolated which did not cause syncytium at the restrictive temperature. Using standard genetic techniques, a recombination map was established that indicated that these mutants mapped into two distinct domains designated F1 and F2. These genetic domains may correspond to mutations mapping within the S1 and S2 glycoproteins, respectively, and suggest that both the S1 and S2 domains are important in eliciting the fusogenic activity of the S glycoprotein gene. In addition, assuming that most distal is alleles map roughly 4.0 kb apart, a recombination frequency of 1 % per 575–676 by was predicted through the S glycoprotein gene. Interestingly, this represents a threefold increase in the recombination frequency as compared to rates predicted through the polymerase region. The increase in the recombination rate was probably not due to recombination events resulting in large deletions or insertions (>50 bp), but rather was probably due to a combination of homologous and nonhomologous recombination. A variety of explanations could account for the increased rates of recombination in the S gene.

Mouse hepatitis virus (MHV), a member of Coronaviridae, contains a single-stranded nonsegmented plus-polarity RNA of about 32 kb in length (Lee et a/., 1991; Pachuk et a/., 1989) . The genomic RNA is arranged into seven or eight coding regions and is encapsidated within multiple copies of a 50-kDa nucleocapsid protein (N) (Lai, 1990; Siddell, 1983) . The N protein forms a helical nucleocapsid structure that is probably associated with the transcriptional complex Compton et a/., 1987; Stohlman and Lai, 1979; Stohlman eta/., 1988; Sturman eta/., 1980) . The nucleocapsid structure is surrounded by a lipid envelope and contains two or three virus-specific glycoproteins. The S glycoprotein has a molecular weight of 180 kDa and is frequently cleaved into two 90-kDa glycoproteins designated Sl and S2 (Frana et a/., 1985; Sturman et a/., 1980 . The N-terminal signal sequence is contained within the Sl domain while the C-terminus is contained within the S2 region (Boireau et al., 1990; Luytjes eta/., 1987; Rasschaert et al., 1990; Schmidt et al., 1987) . The Sl domain can undergo significant amino acid alteration without losing its function while the S2 domain is more highly conserved and contains a hydrophobic heptad repeat element that probably forms a complex ' To whom correspondence should be addressed.

coiled coil (Luytjes et a/., 1987; Schmidt et a/., 1987) . The S glycoprotein is responsible for virus binding to the specific cellular receptor, induction of cell fusion, and elicitation of neutralizing antibody and cell-mediated immunity (Collins et al., 1982; Fleming et a/., 1986; Sturman et a/., 1980 . In addition to S, a 23-kDa M glycoprotein is present in the virion that probably functions in assembly and release (Sturman et a/., 1980 . In some strains of MHV, a 65-kDa hemagglutinin (HE) that shares considerable sequence homology with the influenza C virus hemagglutinin is also present (Luytjes et al., 1988; Shieh et al., 1989; Yokomori et al., 1989) . Upon entry into susceptible cells, the genomic RNA is translated into one or more polyproteins that act as an RNA dependent RNA polymerase (Mahy et a/., 1983; Brayton et al., 1982 Brayton et al., , 1984 . While the exact mechanism for MHV transcription is still under study, the preponderance of data suggest that the genome is initially transcribed into a full-length minus strand RNA, which acts as template for the synthesis of a genomelength RNA and six subgenomic mRNAs by "leaderprimed" transcription (Baker et a/., 1990; Baric et al., 1983 Baric et al., , 1985 Makino et a/., 1986a Makino et a/., , 1991 . In turn, the subgenomic mRNA then act as template for the synthesis of subgenomic minus strands, which participate in successive rounds of mRNA amplification (Sawicki and Sawicki, 1990; Sethna et a/., 1989 Sethna et a/., , 1991 .

Homologous recombination among viruses with nonsegmented RNA genomes has been reported among picornaviruses (Cooper, 1968 (Cooper, , 1977 Cooper et al., 1975; King et al., 1982 King et al., , 1985 King et al., , 1987 Lake et al., 1975) , brome mosaic virus (Bujarski and Kaesberg, 1986) alphaviruses (Hahn et al., 1988; Weiss and Schlesinger, 1991) , cowpea chlorotic mottle virus 3a (Allison et al., 1990) , and coronaviruses (Keck et al., 1988a,b; Lai et a/., 1985; Makino et al., 1986b) . During MHV replication, the RNA recombination frequency is unusually high and approaches 25% or more during mixed infection . Recently, biochemical analysis has revealed extensive polymorphisms and deletions in the sequence of the S glycoprotein gene of different MHV variants. The variant viruses were probably derived from nonhomologous recombination because no consensus or conserved sequences flanked the deletions (Banner et al., 1990; Parker et a/., 1990) . Sequence analysis of RNA recombinant viruses also suggests the presence of a recombination "hot spot" in the MHV S glycoprotein gene (Banner el al., 1990) . However, in the absence of selection pressure, RNA recombination sites in this region are random, suggesting that this preferred site reflects the selection for biologically more efficient recombinants .

Recombination analysis provides a powerful tool for mapping genetic loci and determining the recombination frequency between individual temperature sensitive (ts) mutations Cooper, 1968 Cooper, , 1977 Lake et al., 1975) . In this study, several group F RNA+ ts mutants were used to establish a genetic recombination map in the S glycoprotein gene of MHV-A59.

Temperature-sensitive mutants representing the group F RNA+ mutants (LA7, LA1 2, NC6, NC1 4, NC1 6, NC1 7) and group E RNA-mutants (LA1 8) of MHVA59 were used throughout the course of this study. All virus stocks were propagated at 32" in 1 50-cm2 flasks containing DBT cells as described previously . Plaque assays were performed at either the permissive (32") or nonpermissive (39.5") temperatures in DBT cells in Dulbecco's modified essential medium (DMEM) (Sigma) containing 10% Nu-serum (Collaborative Research, Inc.), 1% gentamycin/kanamycin (GIBCO), and 0.8% agarose (BRL). All plaque assays were stained 28-36 hr postinfection by the addition of neutral red for 2 hr. Cloned 17CI-1 cells were kindly provided by Dr. Sue Baker (Loyola University, Chicago) and maintained at 37" in DMEM containing 7% newborn calf serum (GIBCO), 1% gentamycin/kanamycin, 2% glucose (Sigma), and 5% ttyptose-phosphate broth (Sigma) .

Most of the mutants used in this study have been characterized previously (LA1 8, LA7, LA1 2, NC6) . To increase the accuracy of our recombination mapping data, additional mutants residing within complementation group F were isolated. Briefly, cultures of MHV-A59-infected cells were mutagenized with either 20 pug/ml 5-azacytidine or 100 pg/ ml 5-fluorouracil for 12 hr. Supernatant fluids containing mutagenized virus were plaqued at permissive temperature and individual plaques were isolated. Each plaque was diluted into 0.5 ml ice-cold PBS, screened at 32" and 39.5" by plaque assay, and isolates displaying ts phenotypes were repurified two to three times at 32". Virus stocks were assayed at the permissive and restrictive temperatures and only those isolates displaying reversion frequencies of less than 1 OM3 were used for future studies. Complementation and RNA phenotype analysis were performed as described previously (Leibowitz et al., 1982; Martin et al., 1988; , and three additional group F RNA+ mutants were isolated and designated NC14, NC1 6, and NC1 7. NC1 4 and NC1 6 were isolated using 5-fluorouracil, while NC 17 was isolated using 5-azacytidine.

To select for revertants from the group F RNA+ ts mutants, virus stocks were plaque assayed at 39.5", and individual plaques were isolated and repurified by plague assay at the restrictive temperature. Individual stocks of revertant virus were grown in 1 50-cm2 flasks at 37" and the reversion frequency of each was determined by plague assay at 32 and 39.5". At least three independent revertants from each RNA+ mutant were isolated. To isolate RNA recombinant virus, DBT cells were coinfected with two different ts mutants at an m.o.i. of 10 each at 32". The medium was harvested at 16 hr postinfection and stored at -70". Recombinant viruses were isolated at the restrictive temperature, repurified at 39.5", and grown in DBT cells at 37".

Culture of 17CL-1 cells (1 X 106) in 25-mm six-well dishes (Costar) were infected with various group F ts mutants, revertants, or recombinant viruses at a multiplicity of Infection of 10 for 1 hr at room temperature. Following adsorption, the inoculum was removed, and the cultures were rinsed with PBS and incubated in complete DMEM for 4 hr at 32". Actinomycin D (2 PgI ml) was added for 1 hr and one-half the cultures shifted to the restrictive temperature at 5 hr postinfection by the addition of prewarmed media. Viral progeny were harvested at varying intervals and analyzed by plaque assay.

To determine whether these mutants could transcribe viral mRNA, cultures of cells were infected with various ts mutants and maintained at the permissive or restrictive temperature for 8 hr. The media were removed and the cultures washed with 2 ml of ISO-TKM (10 m/W Tris-HCI, pH 7.5, 150 mNI KCI, and 1.5 mM MgCI,). The RNA was isolated as described by Sawicki and Sawicki, 1990 , and extensively purified by successive rounds of phenol (U.S. Biochemical Company), phenol/chloroform/isoamy alcohol, and chloroform extraction. The RNA was bound to nitrocellulose filters and probed with radiolabeled RNA probes specific for the MHV N gene . The blots were washed and exposed to XAR-5 film for 24 hr.

Recombination analysis and establishing a genetic recombination map of the group F RNA+ mutants Recombination analyses were performed as previously described and the recombination frequencies calculated as the percentage of ts+ virus present in the progeny from the following formula: RF = VW,,., -tA + B)3~.s x , ooo,o, . . -* bw32 (AB),,., was the titer of the cross at nonpermissive temperature while (AB),, represented the titer of the same cross at permissive temperature. (A + B)39.5 was the sum of the revertants of each parent strain assayed at the nonpermissive temperature. The formula only measured single or odd-number cross-over events resulting in the ts+ phenotype and did not account for recombination events that resulted in the doublets mutant phenotype. All recombination frequencies were standardized to a standard cross [LA7 X NC1 6; recombination frequency 3.0 f 1.21 as previously described . The mutants were aligned according to their standardized recombination frequencies and positioned from LA7, LA1 8, and NC1 6.

Cultures of DBT cells in lOO-mm2 dishes were infected at a m.o.i. of 5 with different ts mutants, revertant viruses, or recombinant viruses, and intercellular RNA was harvested at 8-l 2 hr postinfection. The precipitate was resuspended in TE buffer (10 mNI Tris-HCI, pH 7.2, 1 mM EDTA) and used as template for reverse transcription and TAQ amplification reactions.

cDNA synthesis was carried out in a 20-~1 reaction mixture containing 1 pg intercellular RNA, 5 mM MgCI,, 10 m/l/l Tris-HCI pH 8.4, 50 mM KCI, 0.1% Triton X-l 00 (Sigma), 1 mM dNTP each (Promega), 0.5 pg random hexamer (Perkin Elmer Cetus), 1 U/PI RNasin (Promega), 15 U AMV reverse transcriptase (Promega) and incubated at 42" for 1 hr. The products were extracted and precipitated in ethanol and resuspended in 50 ~1 H,O prior to successive rounds of polymerase chain amplification (PCR).

One-tenth of the cDNA was amplified in a 50 ~1 reaction mixture containing 10 mM Tris-HCI (pH 9.0) 50 mM KCI, 3 mM MgCI,, 0.1 O/O Triton X-l 00, 1.25 U TAQ DNA polymerase (Promega), 0.2 mM dNTP mixture and the appropriate primer pairs (100 ng of each). PCR amplification was performed for 30 successive cycles at 1.5 min at 94" to denature the DNA, 2 min at 55" for primer annealing, and 4 min at 72" for primer extension. The products were loaded onto 1% agarose gels in TAE (40 mltlTris, 9 mM NaCI, 1 mM EDTA, (pH 8.0)) separated electrophoretically, stained with ethidium bromide, and visualized under UV light. The DNA products were transferred to nitrocellulose filters, and confirmed by hybridization with 32P-labeled oligomer probes specific for internal sequences in each gene fragment amplified.

Three sets of overlapping primer pairs were obtained from highly conserved sequences in the MHV-A59 and JHM S glycoprotein genes (Banner et a/., 1990; Luytjes et al., 1987; Parker et a/., 1990) . Set A was derived from sequences spanning nucleotides -1 to 15 (CAT GCT GGT CGT GTT T) and 662-677 (AAC GTA GTA GCG GAG G), respectively, and should result in a DNA fragment of 0.67 kb in length. Set B was derived from nucleotides 642-662 (GCG TAC TAT TCG GAT AAA CC) and 2 103-2 115 (CCC ACG ACC GAA TAC G) and were used to amplify a -1.5-kb fragment encompassing the hypervariable region in the Sl domain (Banner et a/., 1990; Parker et a/., 1990) . Set C was derived from nucleotides 2067-2083 (ACG GAT GAG GCG CTT C) and 4055-4070 (GTC llT CCA GGA GAG G) and were used to amplify a -2.0-kb fragment spanning the C terminus of the S glycoprotein gene. Internal oligodeoxynucleotides were used as probe to demonstrate the specificity of the PCR reaction.

Six complementation group F RNA+ mutants and one group E RNA-mutant were used during the course of this study. Three of the RNA' mutants (LA7, LA1 2, 

3.0 x lo7 1.0 x 103 3.3 x 1o-4 1.1 x lo* 1.0 x lo2 1.0 x 10-6 5.5 x 107

1.3 x lo* 2.4 X 1O-6 5.5 x lo7 9.5 x lo4 1.7 x 10-3 3.5 x 10' 4.3 x lo3 1.2 x 1o-4 7.8 x 10' 1.4 x lo4 1.8 x 1O-4 6.9 X 10' 8.7 x 10' 1.2 x loo 7.3 x lo7 7.2 X 10' 1.0 x loo 9.1 x 10' 9.5 x lo7 1.0 x loo 2.1 x lo7 2.0 x 10' 9.5 x 10-l 8.5 X lo* 5.5 x 10" 6.4 X lo-' 9.1 x lo7 7.2 X 10" 7.9 x 10-Z 2.2 x 10' 2.4 X lo7 1.1 x loo 4.0 x 10' 2.0 x lo6 5.0 x 10-2 8.6 x 10' 8.4 X 10' 9.7 x 10-l 1.2 x lo8 1.2 x lo8 1.0 x lo0 1.2 x lo8 1.2 x 108 1.0 x loo 9.4 x 10' 1.2 x 108 1.2 x loo 8.2 X lo7 9.2 x lo7 1.1 x lo0 1.0 x 108 9.0 x lo7 9.0 x 10-l 9.2 x 10' 8.5 x 10' 9.2 x 10-l 2.8 X 10' 3.2 X 10' 1.1 x loo 5.2 x lo7 4.5 x 10' 8.6 x 10-J 4.7 x 10' 4.6 X 10' 9.7 x 10-l 5.9 x 10' 4.8 X 10' 8.1 x 10-l + + + + + + + + "Calculated as 39.5132" NC6) and one RNA-mutant (LA1 8) have been characterized previously (Schaad el al., 1990) . To increase the accuracy of the recombination map and obtain more representative mutants in the S glycoprotein gene, three additional RNA+ mutants were isolated by mutagenesis as previously described . Two mutants were isolated following mutagenesis with 100 pg/ml 5-fluorouracil (NC14, NC16) and one mutant with 20 pg/ml 5-azacytidine (NC17). All mutants had reversion frequencies of less than 10e3, and did not complement each others defect at the restrictive temperature (Table 1 , data not shown). All of the RNA+ mutants were capable of transcribing viral mRNA when maintained at the restrictive temperature (data not shown, Table 1 ). While NC14 produced a small amount syncytium, the remainder of the mutants were not capable of giant cell formation at the restrictive temperature (Fig. 1 ).

To assist in precisely defining the location of these mutations and examining the mechanism for MHV RNA recombination within the S glycoprotein gene, a panel of revertants were obtained from six of the RNA+ mutants (Table 1) . Revertant viruses had similar titers, produced syncytium, and were of the RNA+ phenotype when assayed at both permissive and restrictive temperatures. To conclusively document the revertant phenotype and determine if growth characteristics were similar to parental controls, growth curves were compared at the permissive and restrictive temperatures. Cultures of cells were infected with NC1 6 or its revertant, NC1 6-Rl, at 32" and shifted to 39.5" at 5.0 hr postinfection. Viral progeny were harvested at different times after infection and assayed by plaque assay at the permissive temperature.

Similar growth curves were evident between ts NC1 6 and its revertant NC1 6-Rl in cultures maintained at the permissive temperature (Fig. 2) . However, following temperature shift, growth of ts NC16 was inhibited significantly under conditions in which the revertant replicated normally. Similar results have also been obtained with other mutants and their revertants (data not shown).

To assess the ability of these mutants to be used in recombination studies, we performed a series of crosses between different group F ts mutants and examined the differences between the recombination and reversion frequencies. Cultures of cells were infected at a m.o.i. of 10 each with two different ts mutants and maintained at 32" for 14 hr postinfection. The progeny were harvested and titered at the permissive and restrictive temperatures. Recombination frequencies ranged from as little as 0.1% to >4.00/0 suggesting that the mutants map in different locations in the gene. The ratio of recombinants to revertants ranged from 9 to 3 140 times higher than the sum of the spontaneous reversion frequencies of each individual ts mutant used in the cross and indicated that recombination mapping was feasible throughout this region ( Table 2 ). The extent of the difference in the ratio of recombinants to revertants probably reflected the stability of the mutants used in the cross and the distance between the individual ts lesions.

Establishing a recombination map for the group F RNA+ mutants In contrast to poliovirus and apthovirus ts mutants, MHVA59 ts mutants were amenable to complementa- tion analysis, providing strong evidence that all of the RNA+ ts mutants used in this study contain defects in a single gene or a noncomplementable function (Koolen et al., 1983; Leibowitz et al., 1982; Martin et al., 1988; Schaad et al., 1990) . All mutants were originally crossed three to five times with the reference mutants LA7, NC16, or a group E RNA-mutant LA18 and standardized to the reference cross LA7 X NC16. To obtain a more detailed map of the group F mutants, crosses were also performed among each of the viruses used in this study and standardized to the reference cross. The results of these experiments are shown in Table 3 . The distances between different ts mutants were within statistical limits and permitted the construction of a genetic map (Fig. 3) . On the basis of their genetic recombination frequencies, the group F RNA+ ts mutants appeared to map into two distinct domains, designated Fl (LA7, LA12, NC6) and F2 (NC14, NC16, NC17, LA1 2). Ts LA1 2 mapped closely to representative mu-tants within each subgroup suggesting that it may be a double mutant.

Assuming that crossover events occur in both directions and that NC6/LA7 and NC 14/NC 16 map at different ends of the 4.0-kb S glycoprotein gene, a 1% recombination frequency occurred over 575-667 nucleotide pairs of RNA. Since these mutants do not produce syncytium at the restrictive temperature ( Fig.  1) and sequence analysis of RNA recombinant viruses suggested that the defect in LA7 mapped within the first 1 .l kb of the S glycoprotein gene (Banner et al., 1990; Keck eta/., 1987 Keck eta/., , 1988a Makino eta/., 1986b Makino eta/., , 1987 , these data provided an anchor for mapping the location of the remaining group F RNA+ mutants. These data predicted that the Fl mutants map in the Sl glycoprotein while the F2 mutants map within the S2 glycoprotein sequences (Fig. 4) .

Comparisons between the recombination frequencies in the polymerase and S glycoprotein genes

Using well characterized ts mutants of MHV-A59, we have predicted that the recombination frequency between the group F RNA' mutant LA7, and the group A RNA-mutants LA6/LA3 to be approximately l%/ 1300-l 400 nucleotides of dsRNA ( Fig. 5) . To determine if recombination frequencies vary in different portions of the genome, we compared the maximum recombination frequencies between mutants spanning the polymerase genes of MHV (Group A (LA3/LA6) X Group E (LA1 8)) across gene B encoding the p30 and HE genes (Group E (LA18) X Group F (LA7/LA12)) or through the S glycoprotein gene (Group F (LA7 X NC16INC14)). All mutants were crossed three to five times and average recombination frequencies calculated as previously described (Fig. 5) . Interestingly, the maximum recombination frequency predicted between the two most distant ts mutants in complementation groups A (LA3, LA6) and E (LA18) in the 22-kb polymerase region ranged from about 1%/l 800 (LA1 8 X LA3) to 1 %I2500 (LA1 8 X LA6) nucleotides of ds RNA. The maximum recombination frequency predicted between the group E and F mutants that spanned the p30 and HE genes ranged from about 1%/l 100 (LA1 2 X LA1 8) to 1 %/1400 (LA7 X LA18) nucleotides of dsRNA. Since the maximum predicted recombination frequency across the 4.0-kb S glycoprotein gene ranged from about 1 o/0/575 (LA7 x NC 14) to 1 O/o/667 (LA7 X NC 16) nucleotides of dsRNA, these data suggested that variable rates of recombination occurwithin different portions of the MHV genome ( Figure 5 ). Are deletions contributing to the increased rates of recombination among the group F RNA+ mutants?

Extensive amounts of polymorphism and deletion have been detected in both the HE and the S glycoprotein gene sequences of different MHV strains and in the sequence of other coronaviruses (Banner et al., 1990; Boireau et a/., 1990 Parker et al., 1990; Rasschaert et a/., 1990) . It is possible that the increased rate of recombination within the S glycoprotein gene may represent the sum of homologous recombination and recombination (homologous or nonhomologous) resulting in deletions. To test this possibility, cultures of cells were infected with various combinations of ts mutants as shown in Fig. 4 and RNA recombinant viruses isolated 14 hr postinfection. Since the ratio of recombinants/revertants ranged from 9 to 3 140 in these crosses (Table 2) these data suggested that the majority of ts+ isolates represented true RNA recombinants and not revertants. Based on the recombination rate in the S glycoprotein gene, the predicted distance between the ts alleles used in these crosses ranged from -100 bp to >l 000 nucleotides (Fig. 4) .

To determine if deletions were present within the S gene of RNA recombinant viruses, a series of overlapping primers were synthesized that span the entire S glycoprotein gene of MHV-A59. The location of these primers as well as the size of predicted PCR products are shown in Fig. 4 . Cultures of cells were infected with various RNA recombinant viruses and intercellular RNA was harvested at 8-12 hr postinfection. Following cDNA synthesis and 30 rounds of amplification with the TAQ polymerase and specific oligonucleotide primer pairs, the DNA products were separated on 1% agarose gels and visualized by UV light or southern blotting techniques (Fig. 6) . In Fig. 6a , PCR products spanning the hypervariable region of MHV-A59, JHM, and MHV2 clearly demonstrate the 89 amino acid deletion that is present in the JHM S glycoprotein gene (Banner eta/., 1990; Parker eta/., 1990) . Southern blotting techniques using an internal oligomer probe demonstrate that the -1.5-kb fragment is specific for the MHV S glycoprotein sequences spanning nucleotides 642-21 15 and is not present in uninfected cells (Fig.  6b) . No obvious deletions were detected through the hypervariable region (Fig. 6c) , N-terminus, or C terminus in the S glycoprotein gene of over 20 different RNA recombinant viruses tested (Fig. 6d) . Since the distance between the group Fl and F2 mutants would have required sizable deletions (>lOO bp) to result in ts+ virus (Fig. 4) these data suggested that large deletions were rare events in these crosses, and did not contribute to an increase in the recombination frequency within the S glycoprotein gene.

These studies cannot rule out the possibility of small deletions in one ts parent contributing to an increase in the ts+ progeny. However, no obvious deletions were detected in the S glycoprotein gene of 19 different revertant viruses. In Fig. 6e , PCR products spanning the hypervariable region of NC17 and its four revertants are presented. While these data cannot conclusively rule out the possibility of small deletions (~50 bp) that would not be detected under these conditions, the data suggested that most revertants probably contain a single nucleotide reversion.

Although homologous recombination occurs during the replication strategy of several nonsegmented RNA viruses (Bujarski and Kaesberg, 1986; Cooper, 1968; Hahn et al., 1988; King el al., 1987; Weiss and Schlesinger, 1991) , high frequency RNA recombination is probably a unique phenomenon associated with MHV and perhaps the replication strategies of other coronavirus Keck et al., 1987 Keck et al., , 1988a Makino et a/., 1986b Makino et a/., , 1987 Lai et a/., 1985) . Using highly characterized ts mutants of MHV-A59, we established a genetic recombination map in the polymer- Predrcted nucleotide domarns of the group F RNA' mutants. Assumrng a recombination frequency of lo/01575 bp of dsRNA and that LA7 maps roughly 0.5 kb from the 5' end of the S glycoprotern gene, the tentatrve nucleotide domains of the mutants mapping in the Fl and F2 subgroups were predrcted. The hatched boxes represent the locatron of the region of polymorphism and putative hot spot of recombination in the MHV genome (Banner et a/., 1990; Parker et al., 1989) . Arrows represent the approxrmate locatron of neutralizing epitopes In the S glycoprotein (Routledge et al., 1991; Weissmiller et a/., 1990) . To determine if recombinatron events resulting in deletions were contributing to the increased recombinatron rate in the S glycoprotern gene, a series of crosses were obtained between different RNA+ mutants as shown at the top of the figure. PCR products spannrng different portions of the MHV S glycoprotein gene were obtained using specific oligomer products and the TAQ polymerase. The locatron of these primer pairs and their predicted products are shown in the bottom of the figure. To predict the maximum recombination frequency at the 5'end of the genome, we calculated the recombination frequency between the three most distant RNA-mutants (Group A RNA-mutants: LA3. LA6; Group F RNA-mutant: LA1 8) assuming that these mutants map at the most distant ends of the 22.kb polymerase region (-21 kb apart). To map the maximum recombination rate across the HE and p30 genes, recombination rates were determined between mutants mapping approximately 3.5 kb apart at the 3' end of the polymerase gene (LA1 8) and the 5' end of the S glycoprotein gene (Group F RNA+ mutants: LA7, LA1 2). Recombination rates were calculated through the S glycoprotein gene, assuming the group F RNA+ mutants LA7 and NC1 4/NC16 map roughly 4.0 kb apart. ase genes of MHV. Assuming that the recombination frequency was equivalent throughout the entire genome, these data suggested that the recombination frequency approached 25% or more . In this study, we have demonstrated that the ts mutants from a RNA+ complementation group of MHV-A59 can also be arranged into an additive, linear, genetic map.

Several lines of evidence strongly suggest that the group F RNA+ mutants map in the S glycoprotein gene of MHV-A59. First, the group F mutants do not produce syncythium when maintained at the restrictive temperature. Induction of cell fusion has been mapped to either the S2 domain of the bovine coronavirus S glycoprotein gene (Yoo eta/., 199 1) or the MHV Sl and S2 glycoprotein domains (Gallagher et al., 199 1; Routledge et al., 1991; Weismiller eta/., 1990) . Second, Tl fingerprint analysis of RNA recombinant viruses derived from crosses between ts LA7 or LA1 2 and MHV-JHM demonstrated that the MHV-A59 Sl domain was always replaced by heterologous MHV-JHM sequences. Since no other region in the MHV-A59 genome was uniformly replaced in recombinant viruses, these data strongly suggested that the mutations in these ts viruses reside within the Sl coding sequences (Keck et al., 1987 (Keck et al., , 1988a Makino et al., 1986b Makino et al., , 1987 . Finally, sequence analysis of RNA recombinant viruses derived from these mutants place the ts allele in LA7 within the 5' most 1.1 kb of the Sl glycoprotein gene (Banner et a/., 1990) . While these data do not definitively prove that the Group F mutants map in the S glycoprotein gene sequence, these data have strongly supported the localization of these mutations within this coding region.

On the basis of recombination mapping data, the group F RNA+ mutants appear to map into two discrete domains in the S glycoprotein gene sequence (Fig. 4) . Data from our laboratory and others suggest that the Fl mutants probably map within the Sl glycoprotein while the F2 mutants map within the 52 glycoprotein sequences (Banner et a/., 1990; Keck et al., 1987 Keck et al., , 1988a Makino et al,, 1987 Makino et al,, , 1988 . Sequence analysis of the group F RNA+ mutants and revertant viruses will be required to definitively map the location of these alleles. However, since mutants from both subgroups are incapable of producing syncytium at the restrictive temperature, these data suggest that both domains are important in eliciting cell fusion. Currently, it is unclear whether the alterations in the Fl and F2 conditional lethal mutants result in a temperature-sensitive fusogenic domain or alter the synthesis, transport, and surface expression of the S glycoprotein gene.

The recombination frequency for the entire MHV genome has been estimated to approach 25% or more . These estimates were based on the assumption that the recombination frequency was uniform and approached 10/0/l 300-1400 nucleotides of dsRNA throughout the entire 32-kb MHV genome. However, recombination rates measured within the polymerase region and S glycoprotein gene were estimated to occur at frequencies of about 1%/l 800-2500 and 1%/575-667 bp of dsRNA, respectively (Fig.  5) . Recombination rates across the p30/HE genes were estimated to occur at about 19/o/l 100-l 400 bp of dsRNA. Currently, our best estimate for an MHV recombination frequency was measured within the S glycoprotein gene because these mutations must have resided within a 4.0-kb stretch of RNA. Moreover, the frequency of recombination was internally consistent between several independent crosses within a single complementation group (Table 3) . While definitive proof will require the identification of the exact location of the ts alleles used in these studies, these data suggest that the recombination frequency varies in different portions of the genome and is roughly threefold higher in the S glycoprotein gene.

Several explanations could account for an increase in the recombination frequency within the S glycoprotein gene. First, the recombination rate through the FIG. 6. PCR ampliflcatron and size analysrs of the S glycoprotein gene of wildtype, Ts, recombmant, and revertant viruses. Cultures of cells were infected with different strains of MHV ts mutants, recombinant viruses, or revertant viruses. The Intercellular RNA was extracted and used as template for reverse transcription and 30 rounds of PCR amplification usrng the TAQ polymerase and different primer parrs. (A) PCR products spanning the hypervariable region of MHVA59, JHM, and MHV-2 (642-2155) (lanes 2-4, respectively; lane 5, uninfected control); (B) Southern blot analysis of the PCR products shown in (A) using an Internal ollgomer probe; (C) PCR products spanning the hypervariable region In the S glycoprotein gene of MHV-A59 (lane 2) and RNA recombinant viruses derived from NC1 7 x NC1 6 [lanes 3-5) NC6 x LA7 (lanes 6-8) and NC1 6 X LA7 (lanes 1 O-l 2); (D) PCR products spanning the C (lanes 2-1 1) and N (lanes 12-21) terminl of MHV-A59 (lanes 2, 12) and RNA recombinant viruses derived from NC1 7 X NC1 6 (lanes 3-5 and 13-l 5). NC6 X LA7 (lanes 6-8 and 16-l 8) and NC1 6 X LA7 (lanes 9-l 1 and 19921); (E) PCR products spanning the hypervariable region of NC 17 (lane 2) and four different revertant viruses (lanes 3-6). Lane 1 contains DNA markers in all panels. polymerase region of MHV-A59 may be grossly under-LA6) was unclear. This seems unlikely because a uniestimated because the exact location of the ts allele in form recombination frequency of lo/o/575 nucleotides the group E mutant (LA1 8) and group A mutants (LA3/ of dsRNA throughout the genome would: (1) predict the recombination frequency for the MI-IV genome to approach -56% or more;

(2) place RNA-complementation groups A, B, and C in ORF 1 b and RNA-complementation groups D and E in the p30 /HE nonstructural proteins, and (3) suggest that conditional lethal mutants were not isolated within ORFla at the 5'most three-fifths (-14 kb) of the genome. This is highly unlikely since several proteolytic, hydrophobic membrane-anchoring and cysteine-rich domains have been identified in ORF 1 a, which should be amenable to mutagenesis (Baker et a/., 1989; Lee et a/., 1991) . More importantly, Tl fingerprint analysis of RNA recombinant viruses derived from ts mutants in complementation groups C, D, and E clearly place each genetic function in the polymerase gene (Keck et a/., 1987; Lai et al., 1985) . Since RNA-complementation groups C, D, and E have also been demonstrated to function in mRNA synthesis and map within ORF 1 b, which contains polymerase, helicase, and metal-binding sequence motifs, it is extremely unlikely that these mutants map in the HE or p30 subgenomic ORFs Bredenbeck et a/., 1990; Lee et al., 1991; Schaad et al., 1990) . In contrast, the recombination rate predicted through the polymerase region was not inordinately high (1%/1800-2500 bp of dsRNA), but rather closely approximated the recombination frequencies estimated to occur in apthovirus and poliovirus infections (Cooper, 1977; King, 1987) . Thus, it seems likely that the frequencies measured between the polymerase and S glycoprotein genes actually represent true differences in the recombination rate.

Since the enhanced recombination frequencies in MHV were observed within a physically smaller (4.0-kb) region as compared to the 22-kb polymerase region, it is possible that a mechanism similar to high negative interference in DNA viruses may account for the increased rates of recombination (Chase and Doermann, 1958) . This seems unlikely since high negative interference is probably mediated by DNA polymerase repair mechanisms, which have not been demonstrated in the MHV RNA polymerase (Glickman and Radman, 1980) . The increase in the recombination rate in the S glycoprotein gene could also not be attributed to a combination of true homologous recombination and recombination resulting in large deletions. Nonhomologous recombination probably explains the appearance of defective interfering RNAs of Sindbis virus, vesicular stomatitis virus, and MHV (Holland, 1987; Makino et a/., 1988 Monroe and Schlesinger, 1983) . Since deletions have been demonstrated in the MHV S and HE glycoprotein genes as well as in the nonstructural genes encoded in mRNA 4 and 5 and in the TGEV S glycoprotein gene and other subgenomic ORFs (Ban-ner el al., 1990; La Monica et al., 1991; Luytjes et al., 1987; Parker et al., 1990; Rasschaert et al., I 990; Schwartz et al., 1990; Yokomori and Lai, 1991) , these data have suggested that the subgenomic ORFs may be very amenable to frequent deletions and insertions. Insertions in the MHV S glycoprotein gene have also been reported previously (Taguchi et al., 1987) . Analysis of 20 RNA recombinant viruses derived from different crosses between the group F mutants suggested that large deletions or insertions were rare and occurred at <&the rate of true homologous recombination. While these data could not conclusively rule out the presence of very small deletions, this seems unlikely since the predicted distances between the ts alleles used in these crosses should have resulted in deletions ranging from -100 bp to > 1000 bp in length. It is also unlikely that the ts+ virus isolated from these crosses were revertants containing small deletions (~50 bp) in a singlets parent, since the ratio of recombinants: revertants was at least 9: 1 or greater in each of the crosses and none of the revertants analyzed from each parental ts mutant had evidence of deletions. Since revertants of SB ts mutants usually contain single nucleotide reversions at the site of mutation (Hahn et al., 1989a,b) , the most likely interpretation from these data is that the increase in the rate of recombination in the S glycoprotein gene probably reflects an increase in the rate of true homologous RNA recombination. If the mechanism for homologous recombination and recombinations resulting in deletions are similar as suggested by Banner et al., 1990 , these data suggest that the frequency of these two recombination events is very different. Alternatively, the regions flanked by the ts mutations used in these crosses may be critical for S glycoprotein gene function or plaque formation and could not be deleted.

Two possible mechanisms could explain variable rates of recombination in the MHV genome. The first possibility is a preferred recombination site in the MHV S glycoprotein gene. A clustering of RNA recombination sites adjacent to the hypervariable region in the S glycoprotein region suggest the presence of a preferred site of recombination (Banner eta/., 1990) . However, in the absence of selection, crossover sites spanning the hypervariable region of the MHV S glycoprotein gene were random, suggesting that the preferred site of recombination reflects in vitro selection for certain types of recombinants .

Although the majority of data support non-site-specific homologous recombination throughout the entire poliovirus genome (Kirkegaard and Baltimore, 1986; Sarnow et al,, 1990) Brian, 1989, and Sawickr and Sawicki, 1990 , have clearly demonstrated the presence of subgenomic minus strands In MHV and TGEV infected cells. If the subgenomic minus strands partrcipate In template switching with full-length genomrc RNAs, recombrnation rates should increase from the 5' to the 3' end of the genome. Panel A demonstrates how template switching could occur during negative-strand synthesis between a full-length and subgenomic-length plus strand RNA and result In a full-length negative-stranded RNA recombinant molecule. Panel B demonstrates how template switching between full-length and subgenomic negative strands during positrve-strand synthesis could result in recombinant genome-length molecules, switching occur after the synthesis of UU in sites chosen to minimize the adverse free energy change involved in switching to a heterotypic template (King, 1988) . In addition, a nonrandom distribution of recombination sites has also been reported among intertypic poliovirus recombinants (Tolskaya et al., 1987) . If MHV RNA recombination is mediated by freely segregating RNA segments that are generated by transcriptional pausing during RNA synthesis, preferred sites of recombination may also exist in AU-rich regions and/or in regions of secondary structure in the MHV genome (Baric et al., 1987) . However, it is difficult to envision how these types of preferred sites of recombination would result in higher intratypic recombination frequencies between MHV-A59 ts mutants since extensive secondary structure is found throughout the MHV genome and the G:C:A:U and AA/UU dimer ratios are roughly equivalent in the polymerase and S glycoprotein regions (Banner et al., 199 1; Baric et al., 1987; Soe et al., 1988; Fu et al., unpublished) .

A more likely mechanism to explain the increase in the recombination rate in the S glycoprotein gene is based on the basic replication strategy of coronaviruses (Fig. 7) . Recently, subgenomic minus-strand and replicative intermediate

RNAs have been demonstrated during TGEV and MHV infection (Sethna et a/., 1989 (Sethna et a/., , 1991 Sawicki and Sawicki, 1990) . These data indicate that the amount of negative-strand template RNA is unequal and increases from the 5' to 3' end of the genome. Because of the availability of more negative strand template, recombination rates should increase proportionately from the 5' to 3' end of the genome and be highest in the N gene coding region contained within mRNA 7. For example, recombination events in the polymerase region can only involve template switching between full-length negative-or positive-strand RNA templates. However, recombination events in the S glycoprotein gene or other subgenomic ORFs could not only occur between full-length RNA templates but also involve subgenomic mRNA and subgenomic negative-strand RNA templates as well.

Several findings support this hypothesis. First, recombination frequencies between the polymerase gene and S glycoprotein genes of MHV may vary threefold. Second, in vitro transcribed subgenomic-length DI RNAs rapidly undergo RNA recombination with fulllength (or subgenomic) RNAs supporting the notion that recombination events can occur between different-sized template RNAs . Third, analysis of sepharose 2B-Cl column purified full-length RI RNA has demonstrated the presence of mRNA 7 subgenomic nascent-plus strands bound to the fulllength negative-strand RNA (Baric et al., 1983) . While the original interpretation of these data supported "leader-primed" transcription, an alternative explanation for these findings is that these subgenomic nascent-plus strands have disassociated from subgenomic RI RNA template and recombined with the fulllength negative-strand RNA. Finally, a large number of RNA recombinant viruses selected with markers in the S glycoprotein gene of MHV contain additional crossover sites in the M and N structural genes encoded at the 3' end of the genome (Keck et al., 1988b) . These findings are consistent with the idea that subgenomic minus strands function in RNA recombination and suggest that recombination events may increase in frequency toward the 3' end of the genome.

High-frequency RNA recombination is a unique property associated with MHV replication. Our data suggest that the frequency of RNA recombination is mediated in part by the large size of the MHV genome (32 kb) and its novel mechanism for RNA synthesis involving "leader primed" discontinuous transcription and mRNA replication via subgenomic minus strands (Baric et al., 1983 (Baric et al., , 1987 Sethna et a/,, 1989 Sethna et a/,, , 1991 Sawicki and Sawicki, 1990) . The possibility that the MHV recombination rate increases proportionately from the 5' to 3' ends of the genome may also contribute to the evolution, genetic diversity, and organization of coronavirus genomes. For instance, the gene order of IBV is different from other coronaviruses and it contains an additional gene between the M and N genes at the 3'end of the genome (Cavanagh eta/., 1991) . TGEV also contains an additional gene, designated X, which is located at the 3'end of the genome (Kapke and Brian, 1986) . More remarkably, so,me coronaviruses, but not all, contain a p30 nonstructural protein and an HE glycoprotein that is related to the influenza C virus HE (Luytjes et al., 1988) . Coronaviruses and toroviruses may also have diverged by nonhomologous recombination events in their polymerase and envelope proteins (Snijder et al., 1991) . These findings are consistent with the fact that the coronavirus genome has a remarkable ability to evolve by homologous and nonhomologous recombination, especially in the genes encoded by subgenomic mRNA.

Regeneration of a functional RNA virus genome by recombination between deletion mutants and requirement for cowpea chlorotic mottle virus 3a and coat genes for systemic infection

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

An in vitro system for the leader-primed transcription of coronavirus mRNAs. f/&Y0

A clustering of RNA recombination sites adjacent to a hypervariable region of the peplomer gene of murine coronavirus

Random nature of coronavirus RNA recombination the absence of selection pressure

Characterization of replicative intermediate RNA of mouse hepatitis virus: Presence of leader RNA sequences on nascent chains

Characterization of leader-related small RNAs in coronavirus-infected cells: Further evidence for leader-primed mechanism of transcription

Analysis of intercellular small RNAs of mouse hepatitis virus: Evidence for discontinuous transcription

Interactions between coronavirus nucleocapsid protein and viral RNAs: Implications for viral transcription

Establishing a genetic recombination map for MHV-A59 complementation groups

Nucleotide sequence of the glycoprotein S gene of bovine enteric coronavirus and comparison with the S proteins of two mouse hepatitis virus strains

Characterization of two RNA polymerase activities induced by mouse hepatitis virus

Further characterization of mouse hepatitis virus RNA-dependent RNA polymerases

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

Genetic recombination between RNA components of a multipartite plant virus

Molecular basis of the variation exhibited by avian infectious bronchitis coronavirus (IBV)

High negative interference over short segments of the genetic structure of bacteriophage T4

Monoclonal antibodies to murine hepatitis virus-4 (strain JHM) define the viral glycoprotein responsible for attachment and cell-cell fusion

ln vitro replication of mouse hepatitis virus strain A59

A genetic map of poliovirus temperature-sensitive mutants

Attempts to extend the genetic map of poliovirus temperature-sensitive mutants

Genetics of picornaviruses

Pathogenicity of antigenic variants of murine coronavirus JHM selected with monoclonal antibodies

Proteolytic cleavage of the E2 glycoprotein of murine coronavirus: Host dependent differences in proteolytic cleavage and cell fusion. 1. Viral

Alteratlon of the pH dependence of coronavirus-induced cell fusion: Effect of mutations in the spike glycoprotein. 1. Viral

Escherichia co/i mutator mutants deficient in methylation-Instructed DNA mismatch correction

Western equine encephalitis virus is a recombinant virus

Mapping RNA-temperature sensitive mutants of slndbls virus: Complementation group F mutants have lesions in NSP4

Mapping of RNA-temperature sensitive mutants of Sindbis virus: Assignment of complementatlon groups A, B, and G to nonstructural proteins

Defective lnterfenng rhabdoviruses

Sequence analysis of the parclne transmissible gastroenteritis coronavirus nucleocapsid protein gene

Multiple recomblnatlon sites at the 5' end of the murine coronavirus RNA

In viva RNA-RNA recomblnatlon of coronavlrus In mouse brain

RNA recombination of munne coronaviruses: Recomblnatlon between fusion-positive mouse hepatitis virus A59 and fusion-negatjve mouse hepatitis virus 2

Multiple sites of recombination within the RNA genome of foot-and-mouth disease virus

Genetic recombination In RNA viruses

Preferred sites of recombination in pollovirus RNA: An analysis of 40 intertypic cross-over sequences

The mechanism of RNA recombination In pollovirus

Temperature-sensitive mutants of mouse hepatitis virus strain A59: Isolation, characterlzatlon and neuropathogenlc properties

Coronavlrus: Organization, replication and expression of genome

Recombination between nonsegmented RNA genomes of murine coronavirus

A genetic recomblnatlon map of foot-and-mouth disease virus. I. Gem Viral

Locallzatlon of extensive deletions in the structural genes of two neurotroplc variants of murine coronavirus JHM

The complete sequence (22 kilobases) of murine coronavirus gene 1 encoding the putative proteases and RNA polymerase

Genetic analysis of munne hepatltls virus strain JHM

Primary structure of the glycoproteln E2 of coronavlrus MHV-A59 and ldentlfication of the trypsln cleavage site

Sequence of mouse hepatitis virus A59 mRNA2: Indications for RNA-recombination between coronavlruses and Influenza C virus

RNA dependent RNA polymerase activity in munne coronavirus infected cells

Leader sequences of murine coronavlrus mRNAs can be freely reassorted. Evidence for the role of free leader RNA In transcription

High frequency RNA recombination of munne coronaviruses

RNA recombination of coronaviruses; localization of neutralizing epitopes and neuropathogenlc determinants on the carboxy terminus of peplomers

Prlmary structure and translation of a defective Interfering RNA of murlne coronavtrus

High frequency leader sequence switching during coronavirus defective RNA replication. I Viroi

A system for study of coronavlrus mRNA synthesis: a regulated, expressed subgenomlc defective interfering RNA results from lntergenlc Insertion

Temperature sensitive mutants of mouse hepatitis virus type 3(MHV-3): Isolation, biochemical and genetic characterization

RNAs from two Independently Isolated defective interfering parttcles of Sinbdis virus contain a cellular RNA sequence at their 5'ends

Molecular cloning of the gene encodlng the putative polymerase of mouse hepatltls coronavirus, strain A59

Sequence analysis reveals extensive polymorphism and evidence of deletions wlthin the E2 glycoprotetn gene of several strains of munne hepatitis virus Mroiogy

Porcine respiratory coronavirus differs from transmissible gastroenteritis virus by a few genomic deletions

Analysis of murine coronavirus surface glycoproteins functions by using monoclonal antibodies. 1. Viral

Poliovirus genetics

Coronavirus transcription: subgenomic mouse hepatitis virus replicative intermediates function in RNA synthesis

Genetics of murine coronavirus transcription: Identification of a cistron required for MHV negative strand synthesis

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

Murine coronavirus nonstructural protein ns2 is not essential forvirus replication in transformed cells

Coronavirus subgenomic minus-strand RNAs and the potential for mRNA replicons

Minus-strand copies of replicating coronavirus mRNAs contain antileaders

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

Coronavirus JHM: Coding assignments of subgenomic mRNAs

Comparison of the genome organization of Toro-and Coronaviruses: Evidence for two nonhomologous RNA recombination events during Berne virus Evolution

Sequence and translation of the murine coronavirus 5'.end genomic RNA reveals the N-terminal structure of the putative polymerase

Specific interaction between the coronavirus leader RNA and nucleocapsid protein

Phosphoproteins of murine hepatitis virus

Isolation of coronavirus envelope glycoproteins and interaction with the viral nucleocapsid. 1. Viral

The novel glycoproteins of coronaviruses

Proteolytic cleavage of the E2 glycoprotein of murine coronaviruses: Activation of cell fusing activity of virions by trypsin and separation of two different 90 K cleavage fragments

Characterization of JHMV variants isolated from rat brain and cultured neural cells after wild type JHMV infection

Studies on the recombination between RNA genomes of poliovirus: The primary structure and nonrandom distribution of crossover regions in the genomes of intertypic poliovirus recombinants

Monoclonal antibodies to the peplomer glycoprotein of coronavirus mouse hepatitis virus identify two subunits and detect a conformational change in the subunit released under mild alkaline conditions

Recombination between sindbis virus RNAs

Biosynthesis, structure, and biological activities of envelope protein gp65 of murine coronavirus

Mouse hepatitis virus S RNA sequence revealed that nonstructural proteins ns4 and ns5a are not essential for murine coronavirus replication. 1. Viral

The S2 subunit of the spike glycoprotein of bovine coronavirus mediates membrane fusion in insect cells

We thank Sheila Peel, Lorraine Alexander and Mary Schaad for helpful comments and criticisms. This work was supported by American Heart Association Grant (AHA 87-l 135, AHA 90-4635) and a grant from the National Institutes of Health (A123946). This work was done during the tenure of an Established Investigator of the American Heart Association (AHA 89-0913) (R.S.B.).