key: cord-0760594-uwbekm5o
authors: KOTTIER, SANNEKE A.; CAVANAGH, DAVID; BRITTON, PAUL
title: Experimental Evidence of Recombination in Coronavirus Infectious Bronchitis Virus
date: 1995-11-30
journal: Virology
DOI: 10.1006/viro.1995.0029
sha: e3941873bc1dd80698b47010d1f666b2167fe4d1
doc_id: 760594
cord_uid: uwbekm5o

Abstract Embryonated eggs were coinfected with two strains of the coronavirus avian infectious bronchitis virus (IBV), IBV-Beaudette and IBV-M41, to investigate whether recombination between the two strains would occur. Virions were isolated from the allantoic fluid of the coinfected eggs and putative hybrid RNAs were detected by polymerase chain reaction (PCR), using strain-specific oligonucleotides. PCR products, of the expected sizes, were obtained as predicted from potential recombination events between the nucleoprotein (N) gene and the 3′-untranslated region of the two IBV genomes. Sequencing confirmed that they corresponded to hybrid RNAs. Virus produced as a result of the mixed infection was treated with an M41-specific neutralizing monoclonal antibody and passaged in Vero cells, in which IBV-Beaudette, but not IBV-M41, replicated. Hybrid RNA was still detectable after three serial passages. Since no IBV-M41 was detectable this confirmed that infectious recombinant genomes had been produced in the embryonated eggs. These findings not only support the circumstantial evidence, from sequencing studies of IBV field strains, that recombination occurs during replication of IBV and contributes to the diversity of IBV, but also show that coronavirus RNA recombination is not limited to mouse hepatitis virus.

A recombination map, using temperature-sensitive MHV mutants, revealed the frequency of recombination to be Infectious bronchitis virus (IBV) is a member of the approximately 1% per 1300 nucleotides. Assuming that Coronaviridae (for reviews see recombination occurs randomly, the frequency of recomal., 1994) and contains a single-stranded positive-sense bination in the entire genome was estimated to be RNA genome of 28 kb (Boursnell et al., 1987) . It is a around 25% (Baric et al., 1990) . highly infectious and contagious pathogen of chickens Although several sequencing studies on field isolates which replicates mainly in the respiratory tract but also of IBV have provided circumstantial evidence that recomin some epithelial cells of the gut, kidney, and oviduct bination had occurred in the field (Cavanagh and Davis, (King and Cavanagh, 1991; Lambrechts et al., 1993) . In 1988; Kusters et al., 1989 Kusters et al., , 1990 Cavanagh et al., 1992 ; order to study the basis of IBV pathogenesis, infectious Wang et al., 1993 Wang et al., , 1994 Jia et al., 1995) , there has been cDNA clones allowing site-directed mutagenesis would no direct experimental evidence for recombination. To be desirable. Infectious cDNA clones have been made investigate recombination in IBV experimentally, we for several RNA viruses (for review see Boyer and chose two strains of IBV, Beaudette and M41. Sequence Heanni, 1994), but the large size of the IBV genome curanalysis had indicated that both strains had a high serently precludes their production. However, site-directed quence identity that may be expected to contribute to mutagenesis in the coronavirus mouse hepatitis virus the production of recombinants as previously indicated (MHV) has recently been made possible using recombiby Kirkegaard and Baltimore (1986) for poliovirus. Comnation (Koetzner et al., 1992; van der Most et al., 1992;  parison of the M41 nucleoprotein gene and the adjacent Masters et al., 1994) . untranslated region (UTR) sequence with the corre-Recombination in coronaviruses has been demonsponding region in Beaudette identified a 184-nucleotide strated experimentally using MHV (Lai et al., 1985) and deletion in M41 (Boursnell et al., 1985) . The presence of has been suggested to occur during both positive-and this deletion permitted synthesis of a Beaudette-specific negative-strand RNA synthesis (Liao and Lai, 1992) . A oligonucleotide. Synthesis of this and additional oligonuhigh frequency of recombination was found not only in cleotides, with differences in their 3 ends, made possivitro but also in vivo (Keck et al., ble a polymerase chain reaction (PCR) method for the 1988). Recombination sites were detected over almost discrimination of recombinant and parental RNAs. the entire genome and many recombinants were found MATERIALS AND METHODS with multiple crossover sites (as reviewed in Lai, 1992) .

Viruses and cells IBV-M41 has been adapted to grow in chick kidney adapted strain which, after passage in CK cells, has been tides B7 or M8 and negative-sense oligonucleotides B5 or M6, respectively (Table 1) . Reactions were performed passaged in Vero cells . Both strains were passaged once in 11-day-old specified at 94Њ for 1 min, 42Њ for 2 min, and 72Њ for 3 min for 25 cycles, with a final elongation step for 9 min at 72Њ, on pathogen-free Rhode Island Red (RIR) embryos for use in this work.

a Hybaid Omnigene heating block. The positions of the oligonucleotides used for RT-PCR on the IBV genomes Vero cells were obtained from Flow Laboratories (ATCC, No. CCL81) and were grown in Eagle's MEM are shown in Fig. 1 . (Flow Laboratories) supplemented with 0.29% tryptose Cloning and sequencing phosphate broth, 0.2% bovine serum albumin, 20 mM N,N-bis[2-hydroxyethyl]-2-aminoethanesulphonic acid, PCR products, derived from putative hybrid RNAs, 0.2% NaHCO 3 , 10 mM L-glutamine and antibiotics.

were ligated into 50 ng pCR-II (Invitrogen TA cloning vector) according to the manufacturer's protocol and se-Coinfection of embryonated eggs with IBV M41 and quenced on a 373A DNA Sequencer (Applied Biosystems Beaudette

Inc.) using the PRISM ready reaction DyeDeoxy terminator cycle sequencing kit according to the manufacturer's Eleven-day-old RIR embryonated eggs were inoculated in quintuplicate in the allantoic cavity, with 7 1 10 6 instructions (Applied Biosystems Inc.). The oligonucleotides M13n, M13p, 107, and 108 used for sequencing are egg infectious dose 50 (EID 50 ) IBV-Beaudette, or with 3 1 10 7 EID 50 IBV-M41, or coinfected with 7 1 10 6 EID 50 IBV-shown in Table 1 . Beaudette and 3 1 10 7 EID 50 IBV-M41 (X1), or with 7 1

Neutralization of IBV M41 with MAb A38 10 6 EID 50 IBV-Beaudette and 3 1 10 6 EID 50 IBV-M41 (X2), or with 7 1 10 6 EID 50 IBV-Beaudette and 3 1 10 5 EID 50 Monoclonal antibody (MAb) A38 (Mockett et al., 1984) , M41 (X3), or inoculated with PBSa. Eggs were sealed produced against the spike (S) protein, was used to spewith collodium and incubated at 37Њ for 24 hr. After overcifically neutralize IBV M41. Allantoic fluid or medium night at 4Њ the allantoic fluid was harvested.

containing M41 at a titer of 1 1 10 6 pfu/ml was incubated with 0.2 vol of MAb A38 containing hybridoma culture Extraction of RNA fluid for 1 hr at room temperature. A 780-fold dilution of the MAb A38-containing hybridoma culture fluid com-Allantoic fluid from embryonated eggs infected with IBV was clarified by centrifugation in a Sorval RC-5B pletely neutralized 10 6 pfu/ml IBV M41. centrifuge with a Sorval SS-34 rotor for 30 min (700g)

Passage of IBV in Vero cells and the virions were pelleted by centrifugation in a Sorval ultracentrifuge using a Sorval AH629 rotor for 2 hr Confluent cultures of Vero cells, in 25-cm 2 flasks, were (30,000g). Virus RNA was extracted from the virions, or washed three times with medium and inoculated with total cellular RNA from IBV-infected Vero cells, using the 0.5 ml undiluted allantoic fluid from embryonated eggs guanidinium isothiocyanate method of Chomczynski and infected with either Beaudette or M41 or coinfected with Sacchi (1987) and dissolved in 50 ml of nuclease-free both IBV strains. After 1 hr incubation at 37Њ cells were water (Sigma) for virion extracts or 150 ml for cell extracts.

washed three times in medium and incubated further at 37Њ in 5 ml medium for 16-40 hr until 50% of cells showed Reverse transcription (RT) cytopathic effects (CPE). The medium was collected and used undiluted for further passage. cDNA syntheses were carried out using 3 ml of either the virion or the total cellular RNA extracts following de-Hybridization of 32 P-labeled oligonucleotide to PCR naturation at 60Њ for 10 min in the presence of 0.21 nmol products of oligonucleotide 100, complementary to the 3 end of the IBV-genome (Table 1 ). The mixture was immediately Agarose gels, containing PCR products, were partially cooled on ice and incubated at 42Њ for 2 hr in a 35dried for 2 hr at 55Њ and directly hybridized with 4.2 nmol ml reaction mixture containing 250 U of M-MLV reverse of 32 P-labeled oligonucleotide 104 (Table 1) for 5 hr at 37Њ transcriptase (GIBCO BRL), 50 mM Tris-HCl (pH 8.3), 75 in a Hybaid hybridization oven (Hybaid Ltd.) according to mM KCl, 3 mM MgCl 2 , 7 mM DTT, and 1.5 mM dNTPs. the method of Meinkoth and Wahl (1984) . Oligonucleotide 104 was labeled using 1.5 MBq [g-32 P]ATP (Ç111 TBq/ Polymerase chain reaction mmol, NEG002A, DuPont) and 8 U T4 polynucleotide kinase (GIBCO-BRL) for 30 min at 37Њ and heated at 65Њ PCR amplifications were carried out using 5 ml of the cDNA reactions in a total volume of 50 ml containing 1.5 before use. Following hybridization, the gel was washed twice at 37Њ in wash buffer (0.18 M NaCl, 10 mM NaPO 4 , U Taq DNA polymerase (Promega), 50 mM KCl, 10 mM Tris-HCl (pH 9.0), 0.1% Triton X-100, 1.5 mM MgCl 2 , 0.2 pH 7.7, 1 mM EDTA, 0.2% SDS) and placed against Xray film. mM dNTPs, and 0.3 nmol of positive-sense oligonucleo- (Boursnell et al., 1987) . The nucleotide positions for the M41-specific oligonucleotides M6 and M8 refer to the corresponding positions in the Beaudette sequence. All oligonucleotide sequences are in the 5 r 3 direction.

genomes but contain three of four nucleotides mismatching at their 3 ends. Oligonucleotide M8 has two Design of strain-specific oligonucleotides adjacent nucleotide mismatches at the 3 end when compared to the same sequence in the Beaudette genome. In order to detect putative hybrid IBV RNAs it was first necessary to design oligonucleotides that could distin-

The Beaudette-specific and M41-specific oligonucleotides were spaced 1210 and 970 nucleotides apart, re-guish between M41 and Beaudette using RT-PCR. For this, Beaudette and M41 sequences were analyzed for spectively. Embryonated eggs were infected with either M41 or regions with the highest number of contiguous mismatches. Apart from the Beaudette insert, in the 3-UTR, Beaudette and virion RNA, extracted from virions isolated from the allantoic fluid, was used for the synthesis of no differences were found of more than two consecutive nucleotides. Previously, Kwok et al. (1990) had shown cDNA using oligonucleotide 100, complementary to a sequence present within the 3-UTR in both strains of that oligonucleotides with only two mismatches can be specific, provided that the mismatches are at the 3 end IBV (Fig. 1) . PCR amplifications were carried out on either the Beaudette-or the M41-derived cDNAs using either of the oligonucleotide. Figure 1 shows the oligonucleotides that were designed for use in PCR amplifications the Beaudette-specific oligonucleotides B5 and B7 or the M41-specific oligonucleotides M6 and M8. As can be for the synthesis of products specific for either of the two IBV strains. Oligonucleotide B7 corresponds to a seen from Fig. 2A , lane 3 (Beaudette cDNA using oligonucleotides B5 and B7) and Fig. 2B , lane 2 (M41 cDNA sequence found only in Beaudette; oligonucleotides B5 and M6 correspond to the same position in the two IBV using oligonucleotides M6 and M8) the strain-specific oligonucleotides produced the expected 1210-and 970bp PCR products, respectively. In contrast, the strainspecific oligonucleotides did not produce any product when used in conjunction with the heterologous strain ( Fig. 2A , lane 2, M41 cDNA using oligonucleotides B5 and B7; Fig. 2B , lane 3, Beaudette cDNA using oligonucleotides M6 and M8). These results showed that the strain-specific oligonucleotide pairs were unable to generate DNA from the heterologous strain and were suitable for detecting hybrid RNAs derived from recombination events between the 3 ends of the genomes of the two IBV strains. M41. IBV virions were isolated from the allantoic fluid RT-PCR amplifications were also carried out on extracts from allantoic fluid from mock-infected embryonated eggs, to check whether any PCR products could have been produced from cellular RNA.

PCR amplifications were carried out, on oligonucleotide 100-derived cDNA samples, using the Beaudettespecific oligonucleotides B5 and B7 ( Fig. 2A) , the M41specific oligonucleotides M6 and M8 (Fig. 2B) , and the oligonucleotide combinations B5 and M8 (Fig. 2C ) and M6 and B7 (Fig. 2D ) for the detection of putative hybrid RNAs. The Beaudette-specific oligonucleotides produced the 1210-bp Beaudette-derived PCR product from embryonated eggs coinfected with both IBV strains ( Fig . These results indicated 1570 bp (Beaudette) and 1380 bp (M41), produced from oligonucleotides B5 or M6 and 100, respectively, in addi-that mispriming of the strain-specific oligonucleotides did not occur and that no RT-PCR artefact products were tion to the 1210-bp (Beaudette) and 970-bp (M41) PCR products derived from the strain-specific oligonucleo-produced as a result of template switching between the two RNA or cDNA templates. tides. The presence of excess oligonucleotide 100 in PCRs containing mixed oligonucleotide pairs preferen-Mixed oligonucleotide combinations produced RT-PCR products of the expected sizes, 970 bp for oligonu-tially resulted in PCR products of 1570 or 1380 bp, produced from oligonucleotides 100 and B5 or M7, from the cleotide combination B5 / M8 and 1210 bp for oligonucleotide combination M6 / B7, from coinfection X1 (Figs. parental RNAs templates. No putative hybrid RNAs were detected using the mixed oligonucleotide pairs in the 2C and 2D, lane 4). These results indicated that virions isolated from the allantoic fluid of coinfected em-presence of oligonucleotide 100, presumably due to the low amounts of any putative hybrid RNA compared to bryonated eggs contained hybrid RNAs. No RT-PCR products were detected from allantoic fluid extracts produced the amounts of parental RNAs. Therefore, the amount of oligonucleotide 100 required in the RT reactions, for from mock-infected embryonated eggs using either the strain-specific oligonucleotide pairs or the mixed oligo-cDNA synthesis, which did not produce oligonucleotide 100-derived PCR products was determined prior to nucleotide combinations. This confirmed that the RT-PCR products observed above were not derived from any cel-screening for putative hybrid RNAs (data not shown).

Reverse transcriptase and Taq polymerase have been lular RNAs present in the allantoic fluid. All results were confirmed by three independent experiments. shown to produce hybrid products due to template switching (Luo and Taylor, 1990; Meyerhans et al., 1990) .

Although putative hybrid RNAs were detected in one of the coinfections, X1, they were not detected in the two To check whether potential recombinant PCR products could result from RT-PCR artefacts, virion RNAs isolated other coinfections, X2 and X3. The coinfections differed in the ratios of the amounts of Beaudette and M41 virus from the allantoic fluid of embryonated eggs infected with only Beaudette or M41 were mixed and used as a control.

used to infect the embryonated eggs. The coinfection in which potential recombinant RNAs were detected, X1, strain. For example, clone D1 had the first 161 nucleotides characteristic of M41 with a potential crossover had a ratio of Beaudette:M41 of approximately 1:5 EID 50 . We consistently observed that if more Beaudette than region of 26 nucleotides and thereafter the nucleotides were characteristic of Beaudette. None of the eight se-M41 was used (i.e., ratios of Beaudette:M41 of Ç2:1 (X2) or Ç20:1 (X3)) putative hybrid RNAs were not detected quences had the same crossover site, indicating that the RT-PCR products were derived from hybrid RNAs by the above method. This observation might result from the fact that IBV Beaudette has been adapted to grow in resulting from several recombination events. It should be noted that at this stage we could not say whether any embryonated eggs and appears to grow much faster in embryonated eggs than M41. The combination of more of these hybrid RNAs resulted in recombination events leading to infectious genomic RNA. Beaudette than M41 and the previous observation that Beaudette grew quicker in embryonated eggs than M41

Although the eight sequences consisted of both M41and Beaudette-like sequences, some nucleotide differ-might result in Beaudette preferentially infecting more cells. Alternatively, the higher amount of Beaudette might ences were observed between the M41 and Beaudette sequences present in the PCR products, as shown in either preclude M41 from efficiently infecting the eggs or that it out grew any potential recombinant virus to a Fig. 3 , and those in the published sequences (Boursnell et al., 1985) . This could partly have been due to RT-point that such putative hybrid RNAs could not be detected by strain-specific PCR. These results indicated PCR errors. However, in some cases the published M41 sequence differed from the sequences determined for that it was important to use an appropriate ratio of viruses in a coinfection for detection of recombinants. It should several of the hybrid RNAs, as illustrated in Fig. 3 by arrows. These observations indicated a change in the be noted that the ratio of the control mixed RNAs, derived from virion RNA isolated from embryonated eggs infected parental M41 sequence, of the particular M41 strain used in this work, from that of the published M41 sequence. with only one or the other strain of IBV, were as close as possible to the ratio of input viruses used for the X1

The nucleotide differences observed between the previously published sequence of M41 and the sequence coinfection.

Overall, these observations supported the proposal of the M41 virus used in these experiments might have resulted from adaptation of the M41 to CK cells. that the RT-PCR products had been derived from hybrid RNAs resulting from recombination events between the two IBV strains during mixed infection.

Passage of IBV in Vero cells following neutralization of IBV-M41 with MAb A38 Cloning and sequencing of PCR products derived

To determine whether any of the hybrid RNAs were from recombinant RNA genomic RNA that could replicate, we passaged virus isolated from the allantoic fluid of coinfected em-To verify that the PCR products, derived from hybrid RNAs using either oligonucleotide pairs B5 / M8 or M6 bryonated eggs, shown to contain virion-derived hybrid RNAs. To reduce the possibility of the two parental vi-/ B7, consisted of both M41 and Beaudette sequences, they were cloned and sequenced. Eight clones were ex-ruses outgrowing recombinant virus and to further reduce the possibility that the hybrid RNA was a result of amined, four from each of the oligonucleotide combinations, containing PCR products of 1210 bp (clones C1 to some RT-PCR artefact due to the presence of the two parental genomes, we decided to remove one of the two C4) and 970 bp (clones D1 to D4), respectively. Figure 3 shows the alignment of the eight sequences, parental strains. One potential method comprised the passage of the C1 to C4 and D1 to D4, derived from the putative hybrid RNAs together with the published sequences of the cor-viruses in a cell line, Vero, previously demonstrated to sustain the growth of Beaudette but not M41. The Vero responding genome regions from Beaudette and M41 (Boursnell et al., 1985) . Each of the PCR product se-cells should therefore act as a selective cell allowing the growth of Beaudette but not M41 from a mixed popula-quences contained nucleotides characteristic of both the Beaudette and the M41 sequences, indicating that the tion. Previous studies (unpublished data) had shown that no CPE was observed following the infection of Vero PCR products were derived from hybrid RNAs resulting from recombination events between the two IBV strains. cells with M41. However, we did not know if IBV RNA was produced in M41-infected Vero cells or the effect of As can be seen from Fig. 3 , Beaudette and M41 have only a few nucleotide differences within the genome re-M41 in Vero cells in the presence of Beaudette. Allantoic fluid either containing only one parental strain, Beaudette gion being examined; therefore, the exact crossover sites of the hybrid RNAs could not be determined. The cross-or M41, or containing potential recombinants in the presence of both parental viruses was passaged in Vero cells over sites between the two parental IBV sequences, shown as boxes in Fig. 3 , are defined as lying within (P 1 ) followed by passaging progeny virus an additional two times in Vero cells (P 2 and P 3 ). RNA was extracted regions where the sequence characteristic for one strain changed to the sequence characteristic of the other from each passage, cDNA was synthesized using oligo-nucleotide 100, and the strain-specific oligonucleotide pairs B5 / B7 or M6 / M8 were used to detect either Beaudette-or M41-derived cDNA by PCR. The RT-PCR products obtained from RNA extracts of passages P 1 , P 2 , and P 3 are shown in Fig. 4 . The Beaudette-specific oligonucleotide combination B5 / B7 iden- 1, 4, and 7) or Beaudette (lanes 2, 5, and 8)

The M41-specific oligonucleotide combination M6 / M8 or coinfected with M41 and Beaudette (lanes 3, 6, and 9 ). The three detected the 970-bp PCR product in Vero cells, infected passages were; lanes 1 to 3, P 1 ; lanes 4 to 6, P 2 ; lanes 7 to 9, P 3 . The with allantoic fluid from embryonated eggs infected only fragments were electrophoresed in 1% agarose gels and the DNA was with M41, only in P 1 (Fig. 4B, lane 1) . No M41-derived PCR detected using ethidium bromide staining. The sizes of the Beaudette (1210 bp) and M41 (970 bp) PCR products are shown. The DNA size products were detected in the subsequent passages P 2 standards (M) used was the 1 kb ladder obtained from GIBCO BRL. and P 3 in Vero cells (Fig. 4B, lanes 4 and 7) . No M41derived PCR products were detected in Vero cells only infected with Beaudette (Fig. 4B, lanes 2, 5, and 8) . These hybridoma culture fluid completely neutralized 10 6 pfu/ observations indicated that although the M41 present in ml of M41. The titers of M41 in allantoic fluid obtained allantoic fluid was initially able to infect Vero cells it from M41-infected embryonated eggs were observed to subsequently failed to grow. This indicated that either be no more than 10 6 pfu/ml. A 20-fold dilution of the A38 no viable virus was produced in Vero cells at P 1 or that hybridoma culture medium was subsequently used to virus, if produced, was not released or that too little virus neutralize the M41 in either allantoic fluid from coinfected was released to successfully initiate infection. However, eggs or potentially in the medium from Vero cells. This analysis of RNA extracts from Vero cells infected with dilution of MAb A38 was shown to have no effect on allantoic fluid isolated from coinfected embryonated Beaudette. eggs using the M41-specific oligonucleotide pair identi-Ideally at least two such antibodies should be used fied the 970-bp M41-derived PCR product in all three in order to eliminate the possibility of selecting escape passages (Fig. 4B, lanes 3, 6, and 9 ). This result indicated mutants. However, because we only had one MAb capaeither that the Beaudette virus had been able to rescue ble of neutralizing M41 but not Beaudette, we decide the M41 genome for subsequent passage through Vero to examine the elimination of M41 virus from a mixed cells or that the M41 PCR product had been derived from population using a combination of virus neutralization hybrid Beaudette viruses containing M41 sequences with MAb A38 followed by passage in Vero cells. Results with the crossover sites upstream of the oligonucleotide of the experiments described above indicated that no B5/M6 sites. As it was not possible to differentiate be-M41 was detectable, following neutralization of M41 in tween these two possibilities, a different approach to allantoic fluid, in RNA extracts following subsequent pasremove M41 was examined.

sage on Vero cells. This indicated that if any MAb A38-A second method for removing one virus in a mixed resistant M41 viruses were selected from the allantoic population is by the use of specific neutralizing MAbs. fluid they would be present in such low amounts that it For such a purpose we used one such MAb, A38, which would be unlikely there would be subsequent rescue of has been demonstrated to selectively neutralize M41 but M41 genomes by Beaudette. We proposed that neutralnot Beaudette (Mockett et al., 1984) . To determine the ization of M41 with A38 followed by subsequent passage amount of MAb A38 required various dilutions of a MAb on Vero cells would eliminate M41. Therefore, the detec-A38-containing hybridoma, culture fluid was used to neutralize M41. We determined that a 780-fold dilution of the tion of hybrid RNAs, following neutralization of M41 with Boursnell et al., 1985) . Sequences C1 to C4 and D1 to D4 were produced using oligonucleotides B5 / M8 and M6 / B7, respectively. The nucleotides of the PCR product sequences are only presented where differences were observed between the aligned sequences. Crossover sites are indicated as boxed areas and are defined as those regions in which the sequence of the hybrid RNA changes from one parental sequence to the other parental sequence. Areas that were not sequenced are indicated with dots; positions where the nucleotides were not certain are indicated by an N. Nucleotides that were identified as different between the PCR products and the published M41 sequence (Boursnell et al., 1985) are indicated with an arrow. The numbers above the sequences refer to the nucleotide positions for the complete sequence of the Beaudette genome. produce any M41-derived PCR products in any Vero cell extract following infection with any of the allantoic fluid samples (Fig. 5B) . To discount the possibility that some PCR products may have been produced from the Vero cell extracts but were not detectable under the conditions used, the gels were hybridized with radiolabeled oligonucleotide 104 (Table 1) . No M41-derived PCR products were detected (data not shown). These results confirmed that the combination of the neutralizing MAb A38 followed by selective passage on Vero cells eliminated M41 even in the presence of Beaudette. These observations supported our previous results that PCR-derived products derived from hybrid RNAs were not a result of RT-PCR artefacts due to the presence of the two virus ge- 1, 4, and 7) . However, this oligonucleotide the Vero cells were M41 (lanes 1, 4, and 7) or Beaudette (lanes 2, 5, combination did produce a 970-bp PCR product from and 8) or they were coinfected with M41 and Beaudette (lanes 3, 6, and 9). The three passages were: lanes 1 to 3, P 1 ; lanes 4 to 6, P 2 ; Vero cell extract P 3 from cells infected with allantoic lanes 7 to 9, P 3 . The fragments were electrophoresed in 1% agarose fluid derived from eggs coinfected with both IBV gels and the DNA was detected using ethidium bromide staining. The strains (Fig. 5C, lane 9 ). This result indicated that a to contain hybrid RNAs, in the absence of one of the parental strains, M41. The oligonucleotide combination M6 / B7 did not produce any hybrid-derived PCR MAb A38 followed by selective passage of resulting virus products in any Vero cell passage using any of the on Vero cells, would indicate that recombinant viruses allantoic fluid samples (Fig. 5D ). This indicated that existed in the presence of Beaudette and could not arise no hybrid RNAs were detected as a result of potential as a result of some RT-PCR artefact due to the presence recombinants consisting of predominantly M41 seof the two virus genomes. quence with Beaudette sequence at the 3 end. Poten-Samples of allantoic fluid, previously used for the analtial viruses of this nature would be presumably be ysis of growth of M41 in Vero cells, were treated with like the parental M41 strain and have been eliminated MAb A38 to neutralize M41. The A38-treated allantoic by passage in Vero cells following neutralization with fluid samples were used to infect Vero cells (P 1 ) and

MAb A38. progeny virus subsequently passaged, following treatment with MAb A38, twice in Vero cells (P 2 and P 3 ). RNA Cloning and sequencing of the PCR products from was extracted from the P 1 to P 3 Vero cells, cDNA was potential recombinants detected after passage in synthesized using oligonucleotide 100, and the presence Vero cells of parental-or hybrid-derived cDNAs was analyzed using the strain-specific oligonucleotides.

The 970-bp PCR product derived from the putative The Beaudette-specific oligonucleotide combination hybrid RNA detected in Vero cell P 3 , following selec-B5 / B7 produced Beaudette-derived PCR products in tive passage of virions from the allantoic fluid of emall three Vero cell passages, following infection with albryonated eggs coinfected with the two IBV strains, lantoic fluid derived from embryonated eggs infected with was cloned and partially sequenced. Analysis of the either Beaudette (Fig. 5A, lanes 2, 5, and 8) or Beaudette 5 end of the PCR product from four clones (VPCR1 to and M41 (Fig. 5A, lanes 3, 6, and 9) . No Beaudette-de-VPCR4) showed it to contain M41-specific nucleotides rived PCR products were detected in Vero cell extracts preceded by a short region of Beaudette-specific sefollowing infection with allantoic fluid from eggs infected quence (Fig. 6A) . The G residue at position 26,061 with only M41 (Fig. 5A, lanes 1, 4, and 7) . The M41on the PCR product was derived from the Beaudette sequence, although the published sequence for M41 specific oligonucleotide combination M6 / M8 did not FIG. 6 . Alignment of the DNA sequences of the PCR product derived from the hybrid RNA detected in Vero cells following selective passage. The sequences consist of (A) the 5 and (B) the 3 ends of four independent clones (VPCR 1 to 4). The published sequences for the corresponding regions of Beaudette (top) and M41 (bottom) are included. Asterisks above the sequence show the positions where the two sequences differ. Nucleotide differences observed between the published M41 sequence and those observed from the M41-CK virus used in this work, as determined for the sequences in Fig. 3 , are indicated with arrows. The potential crossover site, defined as the region in which the sequence derived from the PCR products changes from one parental strain to the other, is shown as a boxed region. Regions in which the sequence was not determined are indicated as dots; positions where the nucleotides were not certain are indicated by an N. The numbers above the sequences refer to the nucleotide positions for the complete sequence of the Beaudette genome. The lines at the 5 end of (A) and 3 end of (B) indicate the sequences of oligonucleotides B5 and M8, respectively, used to produce the PCR products. also has a G at this position our previous sequence the two IBV sequences occurred over a five-nucleotide region corresponding to nucleotides 26,068 -data showed that the CK adapted M41 (M41-CK) virus, used in this work, had an A residue at this position 26,072 on the Beaudette genome. (Fig. 6A ). The C residue at position 26,062 on the PCR product could have been derived either from the DISCUSSION Beaudette sequence or the M41-CK virus. However, the C at position 26,067 on the PCR product corre-Our experiments have shown that recombination between two strains of IBV, leading to the production of sponded to the Beaudette sequence because although the published M41 sequence also contained infectious recombinant virions, occurred during the mixed infection of embryonated eggs. This is the first experimen-a C residue, we found that the M41-CK sequence consisted of a T residue at this position. The remainder tal demonstration of recombination in IBV. Previous sequencing studies had provided only circumstantial evi-of the residues at the 5 end of the PCR product that differed between Beaudette and M41 or M41-CK cor-dence that recombination occurs in field isolates of IBV (Cavanagh and Davis, 1988; Kusters et al., 1989 Kusters et al., , 1990 ; responded to the M41-CK sequence as did the 3 end of the PCR product (Fig. 6B) . Interestingly, all four Cavanagh et al., 1992; Wang et al., 1993 Wang et al., , 1994 Jia et al., 1995) . For example, the S gene sequences of the Dutch clones contained the same crossover site, indicating that there was either a bias in the PCR population or isolate D207/78 and the UK isolate UK/6/82 are almost identical, ú98%, until nucleotide 3315 after which the iden-that there was only a small population of replicating recombinant viruses detectable within the region of tity falls to only 57%. However, after nucleotide 3315 the nucleotide identity between the UK/6/82 isolate and an-the IBV genome under investigation. Analysis of the sequences showed that the crossover site between other Dutch isolate, D1466/78, is 97%, whereas the pre-ceding nucleotides in the S sequence show an identity of recombination events had occurred. These recombination events seemed to have occurred randomly between 73%. These observations suggested that the evolution of the UK/6/82 isolate involved a recombination event, near the selected primers. Previous work, on MHV, had suggested that there might be favored recombination sites the end of the S gene, between strains resembling the Dutch isolates D207 and D1466 (Kusters et al., 1990) . How-in coronavirus genomes , although it is now believed that the apparent clustering of recombi-ever, there has previously been no direct experimental evidence that recombination can occur in IBV.

nation sites resulted from the use of selection pressure in the experiments (Banner and Lai, 1991) . To detect The technique of strain-specific PCR has been shown to be an extremely powerful method for distinguishing hybrid IBV RNAs, following as little selection pressure as possible, we used RT-PCR on virion RNAs produced between closely related virus strains (Banner and Lai, 1991; Jarvis and Kirkegaard, 1992) and opens the possi-from the mixed infections without passage. This method also allowed the detection of potential hybrid RNAs de-bility of detecting potential recombinants in the absence of selection. Our oligonucleotides were designed for their rived from recombinants with selective disadvantages that otherwise would be lost following any selection pro-ability to distinguish between two IBV strains and could therefore be used for the detection of potential recombi-cedure(s). An alternative explanation for the detection of hybrid nants using strain-specific PCR. Although oligonucleotide M8 had only two nucleotide mismatches in relation RNA molecules in this study is that they were not derived from genomic RNAs but from subgenomic RNAs, as small to the corresponding Beaudette sequence, the oligonucleotide was shown to be specific for M41. This is in amounts of subgenomic mRNAs were detected in IBV preparations (Zhao et al., 1993) . However, such mole-agreement with the observation that an oligonucleotide with three nucleotide mismatches at the 3 end was cules would still have been generated by recombination. Definitive proof that IBV can undergo recombination, re-shown to be strain-specific for MHV using PCR (Baker and Lai, 1990) . Previous studies on the effects of primer-sulting in the production of hybrid genomes, would require the demonstration that hybrid RNAs can be pas-template mismatches in PCR revealed that double nucleotide mismatches at the 3 end of an oligonucleotide saged. It was very unlikely that we would have been able to isolate a recombinant virus because the recombina-drastically reduced the PCR product yield (Kwok et al., 1990; Jarvis and Kirkegaard, 1992) , indicating that PCR tion event studied is unlikely to have a selective advantage over the parental strains and potential recombinant can differentiate between sequences with no more than two consecutive base mismatches.

viruses could be out competed. We currently have no tools, e.g., temperature-sensitive (ts) mutants and/or ap-Our oligonucleotides for distinguishing between two IBV strains gave rise to products of 970 and 1210 bp.

propriate monoclonal antibodies with which to select our recombinant viruses. Liao and Lai (1992) had previously shown that for the coronavirus MHV a region of 300 nucleotides was suffi-In order to demonstrate that the hybrid RNAs were able to replicate, we used a selective passage method cient for recombination events, indicating that a distance of 970 or 1210 nucleotides in IBV should be sufficient for (neutralization with a MAb, Vero cells) followed by strainspecific PCR. The procedure resulted in no IBV-M41 ge-the detection of recombination events.

One of the disadvantages of using PCR for detection nomes being present in any of the cell extracts following passage in Vero cells of the progeny virus following of potential recombinants is the fact that both reverse transcriptase and Taq polymerase are capable of produc-mixed infections of embryonated eggs. This had two advantages. Firstly, removal of one parental strain would ing recombinant RT-PCR products by switching between templates. One of our control RT-PCRs, using a simple have reduced the chance of the parental viruses outgrowing recombinant viruses. Second, this removed the mixture of M41 and Beaudette RNA, indicated that the hybrid RNAs produced following mixed infection had not possibility that the hybrid RNA could have been produced as a result of some RT-PCR artefact. We demonstrated arisen from template switching by either of these two enzymes. The result supported the proposal that the that a hybrid RNA, consisting of the Beaudette genome with M41 sequence at the 3 end, downstream of the RT-PCR products derived from virion RNA isolated from viruses in the allantoic fluid of coinfected eggs resulted Beaudette-specific oligonucleotide B5 site, was passaged from viruses present in the allantoic fluid of coin-from hybrid RNAs produced by the coronavirus polymerase.

fected embryonated eggs. This result demonstrated that a replicating hybrid RNA was produced as a result of the Sequence analysis of cloned RT-PCR products, synthesized from virion RNA isolated from viruses in the allan-coinfection of embryonated eggs and confirmed, experimentally, that recombination can occur in IBV. toic fluid of coinfected eggs, confirmed that they were derived from hybrid IBV RNAs. All the cloned PCR prod-This demonstration shows that RNA recombination can occur in coronaviruses in addition to that previously ucts sequenced consisted of regions derived from Beaudette-and M41-specific sequences and showed dif-published for MHV. The observation of recombination with IBV and MHV supports the notion that RNA recombi-ferent crossover sites, demonstrating that a number of

Infectious transcripts and cDNA nation may be a common phenomenon within the Coroclones of RNA viruses

Evolution of avian coronavirus and divergence of this group of viruses. IBV: Sequence of the matrix glycoprotein gene and intergenic region Unlike poliovirus, for which Kirkegaard and Baltimore of several serotypes

Infectious bronchitis virus: Evidence for recombination within the Massachusetts serobination was template switching, the mechanism for retype

Boursnell, such great detail

Arg-Arg at the cleavage site of the spike precursor propolypeptide of IBV strains Beaudette and M41

Revision of tion. We could not find a link between any potential the taxonomy of the Coronavirus, Torovirus and Arterivirus genera

Single-step method of RNA identified in our recombinants. Nevertheless, recombinaisolation by acid guanidinium thiocyanate-phenol-chloroform extion resulting from the nonprocessive nature of the corotraction

navirus polymerase is the likely reason for the high fre

Poliovirus RNA recombination: quency in coronaviruses

3135-Our demonstration of recombination in IBV opens new 3145

A novel variant of possibilities for the study of IBV. Recombinants could be avian infectious bronchitis virus resulting from recombination among used to study the virus on a molecular level as previously three different strains

In vivo RNA-RNA binants for studying the function of the virus gene prodrecombination of coronavirus in mouse brain

The mechanism of RNA recoming defective RNA of IBV-Beaudette

and are currently investigating the possibility of using

Repair and mutagenesis of the genome of a deletion mutant of the coronavirus mouse hepatitis virus by targeted RNA of recombinants to study the pathogenicity of IBV. recombination

Effects of primer-template mismatches

An in vitro system for the leaderon the polymerase chain reaction: Human immunodeficiency virus primed transcription of coronavirus mRNA

Coronavirus: Organization, replication and expres-RNA recombination in the absence of selection pressure. Virology sion of genome

RNA recombination in animal and plant viruses

Analysis of intracellular small RNAs of mouse hepatitis virus

449-ing a genetic recombination map for murine coronavirus strain A59 456. complementation groups

Sequences of the nucleocapsid genes from two strains of kidney level by vaccine strains and Belgian nephropathogenic isoavian infectious bronchitis virus

Completion of the sequence of the rus: Recombination between viral genomic RNA and transfected RNA genome of the coronavirus avian infectious bronchitis virus

Template switching by reverse tran-Penzes

Highfective RNA of avian coronavirus infectious bronchitis virus. Virology frequency RNA recombination of murine coronaviruses

Homologous RNA recombination allows efficient introduction tion of targeted RNA recombination and mapping of a novel nucleoof site-specific mutations into the genome of coronavirus MHV-A59 capsid gene mutation in the coronavirus mouse hepatitis virus. J. via synthetic co replicating RNAs

Evidence of natural

Hybridization of nucleic acids immobirecombination within the S1 gene of infectious bronchitis virus. Virollized on solid supports

combination during PCR

Presence of subgenomic bronchitis coronavirus strain Massachusetts M41

ACKNOWLEDGMENTS van der Zeijst, B. A. M. (1989). Phylogeny of antigenic variants of avian coronavirus IBV. Virology 169, 217-221. We thank Miss Karen Mawditt for synthesizing the oligonucleotides.Kusters, J. G., Jager, E. J., Niesters, H. G. M., and van der Zeijst, This work was supported by the Ministry of Agriculture, Fisheries and B. A. M. (1990). Sequence evidence for RNA recombination in field Food, UK.isolates of avian coronavirus infectious bronchitis virus. Vaccine 8, 605-608.