cis-Acting Sequences Required for Coronavirus Infectious Bronchitis Virus Defective-RNA Replication and Packaging

Virus and cells.

IBV Beaudette-US was grown in embryonated eggs (29). African green monkey kidney Vero cells were used for the electroporation step (passage P₀) in most experiments, and primary chick kidney (CK) cells were used in all experiments for passage of the D-RNAs for up to five or six passages (P₁ to P₆) to amplify the D-RNAs, as previously described (24, 25, 28, 29).

Construction of cDNAs encoding D-RNAs.

A plasmid, pCD-91, containing the cDNA of the IBV D-RNA CD-91, derived from IBV Beaudette (24), was used as the starting point for the production of pCD-61, which contained D-RNA CD-61 cDNA (25). Other D-RNAs were produced by deletion mutagenesis of pCD-61 using the restriction enzymes indicated in Fig. 1, 2, 4, and 6. CAT was inserted into pCD-61 at two positions, the SnaBI (nt 13045) (Fig. 2) and PmaCI (nt 26416) (Fig. 1) sites, under the control of IBV transcription-associated sequence 5 (TAS5) (28). These constructs were used as templates for the construction of smaller D-RNAs, as indicated in Fig. 1, 2, 4, and 6.

PCR mutagenesis was used to create the D-RNA clones pCD-38CATstem+ and CD-38CATstem− (Fig. 2). The positive-sense oligonucleotides used were dom1 end (TGAT²⁷²⁶⁹CATTAGTTTGCTTTATCGTAG²⁷²⁹⁰) and stmout (TGATCAG²⁷³⁶²TCTAATCTGTCTACTTAG²⁷³⁸³), respectively; underlined nucleotides show a BclI restriction site, and the superscript numbers denote the IBV genomic position of the nucleotide to the left of the superscript numbers. The negative-sense oligonucleotide was 3′ beau (GCGGCCGCTTTTTTTTTTTTTTTTTTTTTTTTTTTTTGCTCTAACTC), where a NotI site is underlined and GCTCTAACTC are the last 10 nt of the Beaudette genome. 3′ UTRs of the desired length were cloned into CD-61CATSnaBI digested with BclI and NotI.

The restriction enzyme sites XcmI and HindIII were joined onto the 5′ and 3′ ends, respectively, of the IBV TAS5-CAT sequence, using PCR with oligonucleotides XcmICAT+ (CCACTTCTAAGTTGGGTGTTTTACTTAACAAAAACT) and HinCAT− (AAGCTTGGGGGTTACGCCCCGCCCTGC), where the underlined nucleotides indicate the XcmI and HindIII restriction sites. This product was ligated to the 5′-terminal XcmI fragment and 3′-terminal HindIII fragment of the IBV genome, to produce CD-12CAT, modified to contain unique PacI and SacI sites between the XcmI and HindIII sites.

A D-RNA containing all the genes downstream of the replicase gene, CD-84CATXcmI-struct, was generated by insertion of a PacI-SacI fragment, comprising nt 20341 to 27608 of the IBV genome, downstream of the CAT gene in CD-12CAT (Fig. 6). Another D-RNA without any replicase 1b gene and only a small part of the S gene plus part of the N gene and all of the 3′ UTR, CD-30CATXcmI-PSF4 (Fig. 6), was generated from CD-84CATXcmI-struct by deletion mutagenesis using SphI present at position 20743 of the IBV genome and SphI introduced at nt 25983 by PCR.

Sequence analysis of cloned D-RNAs was performed to check the integrity of the D-RNAs at the positions where deletions, insertions, or other mutations had been made.

RNA electroporation of Vero cells.

T7-derived RNA transcripts corresponding to the various D-RNAs were synthesized in vitro from 2 μg of the corresponding NotI-linearized D-RNA-containing plasmids (25). Vero cells (P₀) were grown to 80 to 90% confluence in 25-cm² tissue culture flasks (Falcon) and infected with 0.5 ml of Beaudette helper virus in allantoic fluid. At 8 h postinfection (p.i.), the cells were electroporated with the transcription reaction mixtures (29). Following incubation of the electroporated cells for 16 h, virus (V₁) in 1 ml of culture medium was used to infect CK cells (P₁) and after 20 to 24 h p.i. virus (V₂) in culture medium was passaged on CK cells (P₂) for up to P₅ or P₆. Experiments showed that optimal rescue was obtained when the electroporation stage (P₀) was done with Vero cells and subsequent passages (P₁ to P₆) were done with CK cells.

Analysis of IBV-derived RNAs.

Northern blot analysis on total cellular RNA was performed as described in reference 29. Two probes were used: (i) a 590-bp IBV 3′ probe, to detect all IBV-derived RNAs, corresponding to nt 27017 to 27607 at the 3′ end of the IBV genome, and (ii) an IBV 5′ probe, minus the leader sequence, to detect IBV genomic RNA and D-RNAs, consisting of an 1,120-bp AgeI-SphI fragment (nt 340 to 1460). All the probes were labeled with [³²P]dCTP (29).

Analysis of CAT reporter gene.

The CAT reporter gene has been described previously (28). Briefly, cells were lysed, serial dilutions were made, and CAT protein was detected by enzyme-linked immunosorbent assay (ELISA) (Boehringer Mannheim; product no. 1363727), the amount of CAT protein present being determined by comparison to standard amounts of CAT protein.

Nucleotide sequence accession numbers.

The sequences of regions I and II of the 3′ UTRs of five strains of IBV sequenced for this report have been submitted to the EMBL database and have been assigned accession numbers as follows: strain H120, AJ278336; D207, AJ278335; HV10, AJ278337; HVI-140, AJ278338; and UK/918/68, AJ278334. The accession number for the complete sequence of the IBV Beaudette genome is M95169.

Extent of domain I (5′ end) required for D-RNA replication.

Domain I of CD-61 comprises the 528 nt of the 5′ UTR plus the first 608 nt of open reading frame (ORF) 1a. A number of deletion mutants were created with progressively fewer nucleotides at the 3′ end of domain I (Fig. 1). D-RNAs CD-61CATPmaCI and CD-61, with and without CAT, respectively, were used as positive controls. Examples of CAT ELISA values are shown in Table 1. The number in the name of each D-RNA refers to the size, in kilobases, of the D-RNA moiety, excluding the reporter gene, when present.

Constructs CD-53CAT and CD-37CAT (Fig. 1) produced CAT protein, showing that the 5′ first 804 and 544 nt, respectively, were sufficient for replication. CD-53 was also detected in infected cells by Northern blot analysis after serial passage. Constructs CD-35CAT (comprising 338 nt of the 5′ UTR) and CD-42CAT (missing nt 338 to 544) (Fig. 1) did not produce any CAT in P₀. This indicated that the 5′-terminal 339 or more nt, up to 544 nt, were required for replication.

No part of domain II (replicase gene) is specifically required for D-RNA replication.

Domain II of CD-61 comprised discontinuous parts of ORF 1b (6,322 nt) (Fig. 1). A construct, CD-12CAT (Fig. 1), consisting of the 5′ 804 nt and 3′ 400 nt from CD-61 plus a CAT gene was created. CD-12CAT produced CAT protein at P₀, with amounts of CAT similar to those for CD-61CAT, showing that the 5′-terminal 804 nt and 3′-terminal 400 nt were sufficient for replication; no part of ORF 1b of the replicase gene or of the N gene was required.

Extent of domain III (3′ end) required for D-RNA replication.

Domain III (1,626 nt in total) of CD-61 comprised part of the N gene and the whole of the 3′ UTR. A series of deletion mutants were created within domains II and III of CD-61. Defective RNAs CD-51 and CD-31 retained the 3′-terminal 775 and 400 nt, respectively, and lacked the remainder of domain III and part of domain II (Fig. 2). They were rescued, as shown by detection following Northern blot analysis after serial passage in CK cells, confirming that no part of the N gene was required in cis for replication or packaging.

Williams et al. (33) compared the sequences of the 3′ UTRs of six IBV strains (Beaudette, M41, Gray, Ark99, KB8523, and H52), isolated over a period of several decades and showed that they could be divided into two regions. Region I, adjacent to the N gene (Fig. 2), was hypervariable (53.2 to 92.8% nucleotide identity), including large deletions. In contrast, the 3′-most region II (Fig. 2) was highly conserved (94.3 to 97.8% identity). We have sequenced the 3′ UTR of the H120 strain (closely related to strain H52) and of four additional European isolates (D207, HV10, HVI-140, and 918/68). Sapats et al. (27) and Breslin et al. (3) sequenced the 3′ UTRs of eight Australian IBV and three turkey coronavirus (TCoV) isolates, respectively. Taken together, the data confirm that region I is highly variable (comprising 212 nt for strain Beaudette) and that region II is relatively conserved (comprising 293 nt for strain Beaudette).

CD-38CATstem+ was designed to lack most of region I of the UTR and to retain the last 338 nt of the genome, i.e., it retained region II (Fig. 2). This D-RNA was replicated and packaged, as observed by detection of CAT protein from P₀ to P₅.

Construct CD-38CATstem− was similar to CD-38CATstem+ except for the deletion of a further 93 nt from the 3′ UTR, corresponding to the rest of region I and the 5′ end of region II (Fig. 2). This construct was not replicated by helper virus; P₀ and subsequent passages were negative for CAT protein, and the D-RNA was not detectable by reverse transcription-PCR (RT-PCR) using oligonucleotides 93/117 and Beau3′. Thus, the first 57 nt of region II of the 3′ UTR were essential for replication.

Predicted stem-loop.

Secondary structure analysis of the entire 3′ UTRs of the 19 U.S., European, and Australian IBV strains referred to above, plus 3 strains of TCoV (2, 11), using the software package RNAdraw (23), predicted a conserved stem-loop structure of 42 nt situated from nt 27312 to 27353 in the Beaudette genome. Figure 3 shows the predicted stem-loop structure for IBV Beaudette and the nucleotide substitutions identified for the U.S., European, and Australian IBV strains. The nucleotide differences were predicted not to affect the stem-loop structure. Either the base changes in one side of the stem were covariant, or a single base change did not lead to loss of base pairing and alteration of the predicted structure. These changes strengthened the likelihood that the predicted stem-loop structure did exist. The Australian N1-88 and V18-91 strains and the American Gray strain showed the most sequence differences, including transitions, transversions, and deletions, from the Beaudette-U.S. sequence. The deletions occurred exclusively in the predicted loop region (N1-88, Gray, and V18-91). The pseudoknot predicted by Williams et al. (35) for BCoV commenced 9 nt downstream from the 3′ end of the predicted stem-loop structure and was retained in CD-38CATstem−. Thus, if the pseudoknot is present in IBV at the analogous location it alone is not sufficient for replication. Other predicted structures in the 3′ UTRs differed from strain to strain. CD-38CATstem+, which retained the nucleotides comprising the predicted stem-loop, was replicated, and CD-38CATstem−, which lacked the stem-loop, was not replicated, as reported above (Fig. 2). This indicated that nt 183 to 276 of the 3′ UTR (genome nt 27291 to 27384), inclusive of the predicted stem-loop structure, were essential for replication of the D-RNA.

Sequences required for packaging of IBV D-RNA.

Several constructs which each lacked part of the ORF 1b replicase sequence were made (Fig. 4). All of them were replicated and packaged, including CD-44CAT. A previous version of CD-44 without CAT (25) had not been rescued; we assume that a mutation during the production of CD-44 was responsible for that. The rescue of several D-RNAs that did not contain the CAT gene is shown in Fig. 5. The results showed that no specific region of 1b was required for replication or packaging.

CD-22 and CD-19 (Fig. 4) and CD-12 and CD-12CAT (same deletions as each other) (Fig. 1) contained the 5′- and 3′-terminal regions of the genome that had been shown to be the only specific sequences required for replication of IBV D-RNA. However, none of these four D-RNAs were detected by Northern blot analysis after rescue attempts, following several passages in CK cells, although CD-12 was detectable by RT-PCR in P₂ and P₃ extracts. CD-12CAT was replicated in P₀ cells, as evidenced by CAT production, and on subsequent passage (P₁ to P₃) cells were still positive for CAT though to a lesser extent than in P₀ and less than control CD-61CAT in P₁ to P₃ cells. Thus, although the small D-RNAs were replicated, the rescue was poor.

D-RNAs containing structural protein genes were rescued poorly.

The small size of CD-12CAT might have contributed to its poor rescue compared with CD-61CAT. To investigate this, CD-84CATXcmI-struct was constructed. This comprised the 5′ end of CD-12CAT, corresponding to the first 804 nt of CD-61 with the CAT gene inserted at the XcmI site; the 3′ part of the replicase gene, starting at position 20341 (a PacI site); and all the remainder of the genome, i.e., including all the structural protein genes, total length, including the CAT gene, being 9,140 nt (Fig. 6). The amount of CAT expressed at P₀ and at P₁ to P₅ produced from this construct was compared with that from CD-61CATPmaCI in two experiments (Table 1). Although the absolute CAT values varied between the two experiments, the trends in the two experiments were the same. Thus, whereas passage of CD-61CATPmaCI resulted in increasing amounts of CAT, the amount of CAT protein produced by CD-84CATXcmI-struct declined after P₀.

D-RNA CD-30CATXcmI-PSF4 (3.7 kb) was derived from CD-84CATXcmI-struct by removal of the sequence corresponding to genome positions 20740 to 25983 (Fig. 6). Thus, this D-RNA had approximately 400 nt corresponding to the beginning of the S protein gene and most of the N gene. This construct produced amounts of CAT protein similar to those for CD-61CATPmaCI in P₀, after which CAT expression declined rapidly on serial passage, as it did for CD-84CATXcmI-struct (Table 1). The small D-RNA CD-12CAT behaved similarly to CD-84CATXcmI-struct and CD-30CATXcmI-PSF4 (Table 1; in other experiments, the amounts of CAT protein expressed by CD-12CAT in P₀ were the same as or higher than those expressed by CD-61CATPmaCI). In case the expression of CAT was affected by the position of the CAT reporter gene, i.e., at the XcmI site at position 804 rather than further downstream, as in CD-61CATSnaBI and CD-61CATPmaCI, D-RNA CD-53CATXcmI was made. In this D-RNA, the CAT gene was at position 804 (Fig. 6) but most of domain II from CD-61 was conserved. The amounts of CAT produced by CD-53CATXcmI were similar to those of CD-61CATPmaCI (Table 1, experiment 2). Thus, the results showed that D-RNAs lacking most of domain II (replicase 1b gene) were rescued very inefficiently and that domain II could not be successfully replaced by structural gene sequences.

Sequences required for RNA replication.

Our data show that the only sequences required for the replication of IBV D-RNAs are within the 5′ 544 nt and 3′ 388 nt, i.e., essentially within the 5′ and 3′ UTRs of the IBV genome. The first 544 nt of the genome include the first 15 nt of ORF 1a; it is possible that these ORF 1a nucleotides are part of the sequence required for replication. Our results are in contrast to those for MHV D-RNA DIssF, which requires a 58-nt region, approximately 3 kb from the 5′ end of ORF 1a, that folds into a stem-loop secondary structure (15, 17, 26). Our findings are similar to those with the D-RNAs of MHV A-59 (19) and BCoV (7), which do not require internal regions from the replicase 1a gene for replication.

A small (2.1 kb) D-RNA of TGEV was efficiently replicated (13). The first 5′ continuous part of this D-RNA was 1.3 kb in length, the first 315 nt being the 5′ UTR (9). It is possible that only a part of the first 1.3 kb is essential; smaller fragments were not investigated. No sequences corresponding to replicase 1b sequence were required for the replication of the IBV D-RNAs, in keeping with the findings for small D-RNAs of TGEV (13), MHV-JHM (21), and BCoV (7).

The BCoV pDrep1 D-RNA required almost the entire N gene for replication (6). Our D-RNAs CD-12, CD-12CAT, CD-31, CD-51, and CD-38CATstem+ did not contain any N gene sequence nor did the 9.7-kb TGEV DI-C or the small (3.3-kb) derivative DI-M33 which was replicated by helper TGEV (13). Defective RNAs of MHV contain part of the N gene, much of which can be removed without impairment of replication. Thus, some 378 to 463 nt of the 3′-terminal end of the genome are required for replication, depending on the strain of MHV and the particular D-RNA, most being derived from the 3′ UTR (14, 17, 32).

Predicted stem-loop in the 3′ UTR.

Comparison of the 3′ UTRs of 19 isolates of IBV and 3 of TCoV has confirmed that the UTR comprises a highly variable region I, adjacent to the N gene, and a much more conserved region II, proximal to the poly(A) tract. Structure analysis predicted that a stem-loop structure exists at the 5′ end of region II (Fig. 3), and deletion analysis indicated that this structure is essential for replication of the IBV D-RNAs (Fig. 2). It is also possible that the fusion of the specific sequences in CD-38CATstem−, which lacked the predicted stem-loop, affected replication, as has been seen elsewhere for MHV A-59 D-RNAs (19). However, the phylogenetic data presented in Fig. 3 strongly suggest that there is a stem-loop in this region, and we propose that this potential stem-loop is essential for IBV replication. This potential stem-loop is analogous in position to the stem-loops found in MHV and BCoV D-RNAs (12).

Situated 9 nt downstream of the predicted IBV stem-loop structure is a potential pseudoknot identified by Williams et al. (35), analogous to a pseudoknot structure demonstrated to be present in BCoV and essential for RNA replication. CD-38CATstem−, which lacked the predicted stem-loop structure and retained the sequence corresponding to the predicted IBV pseudoknot, was not replicated. It is possible that the predicted pseudoknot might have been impaired in CD-38CATstem− as a consequence of removal of the sequence ordinarily adjacent to it. The rescue of CD-38CATstem+, which lacked region I of the 3′ UTR, showed that this region was not essential for replication or packaging of this IBV D-RNA. No other strong candidate secondary structures were identified at the 3′ end of the genome.

We have previously identified three potential stem-loop structures within the 5′ UTR (29), one of which (nt 7 to 30) would appear to be analogous to the stem-loop identified by Chang et al. (7) in BCoV and concluded to be essential for the cis-acting replication signal associated with the leader sequence.

Sequences required for packaging of the D-RNAs.

The results of our experiments investigating packaging led us to the conclusion that only the sequences in the 5′ UTR and/or region II of the 3′ UTR were specifically required for packaging. This is similar to the findings for BCoV (6) and TGEV (13) D-RNAs but in contrast to other findings with MHV (10, 22, 31), in which a small part of gene 1b was found to act as an RNA packaging signal. However, the D-RNA DIssE of MHV-JHM lacks this 1b packaging signal but is incorporated into defective-interfering particles, at low efficiency, and is detectable by Northern blot analysis after multiple passages (21).

IBV D-RNAs CD-12 and CD-12CAT, which contain only 5′- and 3′-terminal sequences, totaling 1.2 kb of IBV sequence, were replicated, but overall rescue was poor. This might have been attributed in part to the small size of these D-RNAs, which at 1.2 and 1.9 kb (inclusive of the CAT gene), respectively, are smaller than any coronavirus D-RNA so far successfully rescued. However, our results showed that increasing the size of CD-12 with IBV sequence corresponding to the structural protein genes did not improve rescue. A D-RNA of TGEV, M21 (2.1 kb), was rescued but with a lower efficiency than that of those D-RNAs that contained one or the other of two fragments (F1 and F2) of the ORF 1b sequence (13). Izeta and colleagues (13) concluded that either both F1 and F2 fragments contain a packaging signal or their presence is required for the proper folding of a packaging signal in one or the other terminal region. Our results were similar to those of Izeta et al. (13) in that parts of the replicase 1b sequence—but not any specific part—were required for efficient rescue. Replacement of region II of CD-61 by sequence downstream of gene 1 (CD-84CATXcmI-struct) (Fig. 6) did not result in efficient rescue (Table 1). Thus, a large size alone was not sufficient for efficient packaging. The results suggest that either some D-RNAs folded in a way that was not conducive to packaging or the D-RNAs were highly unstable.