key: cord-0745802-xq1dxmb9
authors: KRISHNAN, RAJESH; CHANG, RUEY-YI; BRIAN, DAVID A.
title: Tandem Placement of a Coronavirus Promoter Results in Enhanced mRNA Synthesis from the Downstream-Most Initiation Site
date: 1996-04-15
journal: Virology
DOI: 10.1006/viro.1996.0210
sha: 76787d76b67af412075bcb38ec51f6057f61cd19
doc_id: 745802
cord_uid: xq1dxmb9

Abstract Insertion of the 17-nucleotide promoter region for the bovine coronavirus N gene as part of a 27-nucleotide cassette into the open reading frame of a cloned synthetic defective-interfering (DI) RNA resulted in synthesis of subDI RNA transcripts from the replicating DI RNA genome. Duplicating and triplicating the promoter sequence in tandem caused a progressive increase in the efficiency of subgenomic mRNA synthesis despite a concurrent decrease in the rate of DI RNA accumulation that was not specific to the promoter sequences being added. Although initiation of transcription (leader fusion) occurred at each of the three promoter sites in the tandem construct, almost all of the transcripts were found as a product of the most downstream (3′-most on the genome) promoter. These results show that enhancement of subgenomic mRNA synthesis is a property that can reside within sequence situated near the promoter. A possible role for the plus strand in the downstream promoter choice is suggested.

During bovine coronavirus replication, a 3-coterminal coworkers (11) have shown with cloned replicating defective-interfering (DI) RNAs of MHV that the rate of mRNA nested set of subgenomic mRNAs that coordinately peak synthesis from an engineered intergenic promoter does not in abundance at around 6 hr postinfection is made. At necessarily correlate with the degree of sequence complethis time, the mRNA species are not equal in number, but mentarity. Sequences distantly upstream from the promoter rather the shorter 3-proximal species are progressively in MHV, furthermore, may enhance transcription (12). more abundant. The molar differences range from 1 for A second model used to explain the origin of differing genome (mRNA 1) to 1000 for mRNA 7 (N gene mRNA), mRNA levels (13) was based on the observations that the smallest mRNA species (1) . Although exceptions ocsubgenomic mRNA-length minus strands function as cur, this general pattern of mRNA regulation holds true components of replicative intermediates (13, 14) . This for all coronaviruses (reviewed in 2). model predicted that subgenomic minus-strand RNAs Explanations for the differing mRNA levels have reflected are generated by an interruption of minus-strand syntheviews on how mRNAs are generated. In the leader-primed sis at attenuating intergenic sites. A greater abundance transcription model, mRNAs are generated from a genomeof 5-proximal minus-strand templates would be exlength minus strand (antigenome) by priming from a free pected that would, in turn, generate progressively more leader at intergenic promoter sites (3, 4) . The original model abundant 3-proximal mRNAs. Transcription initiation predicted that the degree of base complementarity between would result from the use of 5-terminal AGAUUUG prothe promoter region and the priming free leader would moter motifs or perhaps from a spliced antileader on the determine promoter strength. Although this prediction was subgenomic minus strands (13) . supported by the degree of base pairing and mRNA accu-A third model was also based on the presence of mulation rates in murine hepatitis virus (MHV) (5, 6) , the subgenomic minus-strand RNAs (1, 14, 15) but predicted same correlation was not found in infectious bronchitis that mRNAs, once generated (by whatever mechanism), virus (7) , calling this mechanism of regulation into question.

would undergo amplification by replication through the In addition, Makino and coworkers (8-10) and Spaan and use of terminal promoter sequences (1, 14) . Conceivably, for either model 1 or 2, subgenomic molecules possessing intergenic promoters could function as tem- the initial cloning) encoding the 1-to 23-aa neutralizing epitope of herpes simplex virus (HSV) gD (18) into the mung bean nuclease blunt-ended Bsu36.1 site of pDrep1 (17) . A 3-nt overdigestion at the Bsu36.1 site caused an in-frame insertion of 81 rather than 84 nt.

For transfection and transcription analysis, cells infected with BCV were transfected as previously described (17) , except that the helper virus was a stock that contained no wild-type DI RNA (as determined by Northern analysis), and 1 mg RNA per 35-mm plate was used for transfection. At 24 hr posttransfection supernatant was collected and used to infect fresh HRT cells from which cytoplasmic RNA was isolated at 9 hr postinfection. RNA was analyzed by quantitative Northern blot hybridization using the AMBIS photoanalytic imaging system (AMBIS, Inc., San Diego, CA) following hybridization with end-labeled oligonucleotide 8 which hybridizes to the plus strand of the TGEV reporter sequence (17) .

For primer extension analysis, approximately 1 pmol cDNA synthesis from cytoplasmic RNA and sequencing of asymmetrically amplified PCR products were carried out as previously described (20) except that oligonucleotide 8b, 5CATGGCACCATCCTTGGCA3, a 3-truncated In this study we have directly examined the effect of a multimerized internal promoter (termed intergenic se-version of oligonucleotide 8, was used for the reverse transcriptase reaction. For sequencing the 5 end of DI quence, IS) on mRNA abundance. For this, a cDNA clone of a naturally occurring bovine coronavirus (BCV) DI RNA RNA subgenomic mRNAs, oligonucleotide L(0), 5GCG-GGATGCACGCACGCAAATCGCTC3, which binds to engineered to contain a reporter, an in-frame 30-nt sequence from the N gene of porcine transmissible gastro-bases 7 to 33 of BCV leader plus strand, and oligonucleotide 8 were used. For sequencing genomic DI RNA, oligo-enteritis virus (TGEV), and called pDrep1 (17) was used for intergenic sequence insertion ( Fig. 1) . We have shown nucleotide N(0), 5AGAGCGTCCTTTGGAAATCGTTCT-GG3, which binds to bases 34 to 59 in the N ORF, and that T7 RNA polymerase-generated transcripts of pDrep1 linearized at the Mlu1 site undergo replication and subse-oligonucleotide 8 were used. Cloning of PCR-amplified products was carried out as recommended by the manu-quent packaging after transfection into helper virus-infected cells (17) . To make pDrepIS1, two 27mer oligonu-facturer for TA cloning (Invitrogen). From studies of genomic sequences for the mammalian cleotides, A1 (5TCAGGAATATCTAAACTTTAAGATGAA-3) and A2 (5TGATTCATCTTAAAGTTTAGATATTCC3), coronaviruses BCV (Table 1) , MHV, and TGEV, the heptameric intergenic sequence UCUAAAC has emerged as a were annealed (creating a Bsu36.1-compatible site at its 5 end, virus sense) and inserted at the unique Bsu36. 1 consensus element postulated to be the core promoter for directing viral subgenomic transcription. This was directly site 52 nt upstream of the reporter sequence in pDrep1 (Fig. 1) . Upon insertion the Bsu36.1 site was regenerated tested by Makino and Joo (9) in a study that demonstrated the UCUAAAC heptad to be sufficient for subgenomic tran-at the 5 end. To construct pDrepIS2 and pDrepIS3, oligonucleotides A1 and A2 were phosphorylated, annealed, scription from a site within the ORF of a replicating DI RNA of MHV. Promoter activity, however, became optimal when and ligated with Bsu36.1-digested, dephosphorylated pDrepIS1. pDrep-gpD81 was generated by cloning an an intergenic sequence of 18 nt from the N gene, a region showing full complementarity to the 3 end of MHV geno-87-nt blunt-ended fragment (5ccAAATATGCCTTG-GTGGATGCCTCTCTCAAGATGGCCGACCCCAATCGC-mic leader, the postulated primer of transcription in the leader priming model (3, 4, 23) , was used. TTTCGCGGCAAAGACCTTCCGgtcctggacgggaatt 3; lowercase letters refer to primer sequences used in BCV is a close relative of MHV (24, 25), yet differs mRNA and DI genome accumulation for these were compared to those for pDrep1 and pDrepIS1 ( Fig. 2A This demonstrated an apparent progressive increase in 12.7 kDa 5uuuaggUAgACcuuaua3 S 5auaaUCUAAACauguug3 the efficiency of transcription (Fig. 3) . pDrepIS3 but without subgenomic transcription was c Underlined region in the 17-nucleotide intergenic sequence found found for pDrep-gpD81 RNA (Fig. 2B) , suggesting that immediately preceding the N gene. The engineered start codon for subgenomic mRNA is boxed.

the inhibition of replication in pDrepIS constructs was not necessarily a result of the transcription but was due to an added sequence at this site in the DI RNA genome. By probing a blot identical to that in Fig. 2A , lanes 1-4, from MHV in two respects with regard to the relationship between genomic leader and intergenic sequence. (i) with an oligonucleotide recognizing the N gene sequence, and hence all viral mRNA species, it was deter-The region of contiguous sequence identity between the extended genomic leader and the intergenic sequence mined that virus from passage 1 replicated nearly equally well in all infected plates ( Fig. 2A, lanes 5-8) . preceding the N gene is 12 not 18 nt (Table 1 ). (ii) Repetitive UCUAA sequences, thought to yield alternative fu-

The pattern of Northern blot analysis ( Fig. 2A) suggested that the subgenomic transcripts of pDrepIS1, sion sites in MHV (6, 26-28), do not exist in the BCV leader or its flanking sequence and hence may be a pDrepIS2, and pDrepIS3 were the same size since incremental differences of 27 nt would have been revealed in sequence phenomenon related to the invariant mRNA fusion junction patterns found in BCV (20). To determine, an agarose gel of this design. These differences, for example, could be discerned among the DI genomic therefore, whether the analogous region in BCV could direct subgenomic transcription, the analogous 17-nt in-RNAs. In the first approach used to characterize the leader-mRNA fusion sites for subgenomic mRNAs, RT-tergenic sequence from the BCV genome (a base was deleted from the 3 end to conform to the homologous PCR analysis (Fig. 2C ) was carried out on RNA samples prepared from the first virus passage, the same RNA region as defined in Ref. 7) ( Table 1 ) was inserted as part of a 27-nt cassette into the unique Bsu36.1 site of used for Northern blot hybridization analysis ( Fig. 2A) . As suggested by Northern hybridization data, there was pDrep1 to form pDrepIS1 (Fig. 1 ), and transcripts were tested for replication and subgenomic transcription. The revealed a preferential usage of the 3-most promoter for leader fusion. The identity of the abundant DI mRNA cassette was designed to maintain the DI RNA ORF, a cis-acting requirement for BCV DI RNA replication (17), band 1 in Fig. 2C, lanes 2, 3, and 4 , was further ascertained by direct PCR sequencing, and that of DI mRNAs and to insert an AUG start codon just downstream of the leader on the subgenomic transcript. Both replication of 1 and 2 by the sequencing cDNA-cloned cDNA products (data not shown). The DI mRNA 3 band could not be the DI RNA and generation of subgenomic mRNA from pDrepIS1 were observed (results were similar to those cloned as cDNA. These results showed that the downstream-most promoter site was predominantly used by in Fig. 2A, lane 2) .

To determine what effects multimerization of the pro-pDrepIS2 and pDrepIS3 and that the junction sequence in each subgenomic transcript demonstrated no hetero-moter sequence would have on subgenomic mRNA synthesis, the cassette was placed in two (pDrepIS2) and geneity and was therefore the same as for N mRNA (20; Fig. 1 , and data not shown). The DI mRNA 3 fusion site three (pDrepIS3) adjacent copy repeats within the pDrep1 ORF sequence, and the rates of subgenomic could not be confirmed by sequencing but the position virus harvested from the respective transfection experiment were analyzed with 32 P-labeled oligonucleotide probe. Lanes 1 through 4 were probed with reporter plus-strand-detecting probe (oligonucleotide 8) to identify DI RNA genome and subgenomic transcripts, and lanes 5 through 8 were probed with N plus-strand-detecting probe (oligonucleotide N/) to identify helper virus RNA as well as DI RNA species (17) . (B) Northern blot analysis showing inhibition of DI RNA replication as a function of the 81-nt HSV gD (nonpromoter) insert. Probing was done as described in A, lanes 1 through 4. (C) RT-PCR analysis showing the relative amounts of genomic DI RNA and subgenomic mRNAs. Cytoplasmic RNA described in A was used in an RT-PCR with primers 1 and 2, and the products were analyzed on an agarose gel of 3% and stained with ethidium bromide. (D) Primer extension analysis. Lanes 1 through 6, radiolabeled primer (oligonucleotide 8) which binds to the reporter sequence 52 nt downstream from the IS1 promoter sequence was extended on cytoplasmic RNA obtained from cells as indicated. The extended products were then electrophoresed on a denaturing sequencing gel of 6% polyacrylamide. Lanes 7 through 10, products of dideoxynucleotidyl DNA sequencing using oligonucleotide 8 as the radiolabeled primer and pDrepIS3 DNA as the template. The predicted mRNA positions were determined from the DNA sequence. of the PCR band (Fig. 2C ) indicated its fusion site would genomic RNA accumulated with each additional promoter sequence; (ii) there was an apparent progressive be at or near the upstream-most position.

In the second approach to characterize leader-mRNA increase in the efficiency of transcription as judged by the DI mRNA to genome ratios of 0.56, 1.63, and 3.24, fusion sites, a primer extension experiment was done on the RNA using radiolabeled oligonucleotide 8 which for pDrepIS1, pDrepIS2, and pDrepIS3, respectively; and (iii) the vast majority of transcripts came from the down-binds to the DI RNA reporter sequence. Since it can be expected that the primer would bind alike to DI mRNAs stream-most (DI mRNA 1) position in pDrep1S2 and pDrepIS3. For pDrepIS2 no DI mRNA 2 was found, and and DI genomic RNA, extended products should reflect the relative amounts of template species. Except for the for pDrepIS3 little DI mRNA 2 and no DI mRNA 3 transcripts were found. Although these results support the pDrep1 RNA, RNA samples in this experiment were the same as those used in the Northern blot in Fig. 2A . The trend identified by Northern analysis (Fig. 2A) , they cannot be given a strict quantitative interpretation since extended products (Fig. 2D ) showed the following: (i) There was a progressive decrease in the amount of DI shorter products are favored in enzymatic extension re-however, demonstrate that a local sequence surrounding a promoter region in the DI RNA genome can direct an enhanced synthesis of subgenomic mRNA when synthesis of that RNA is quantitated as a ratio of DI template to subDI mRNA. How might the downstream initiation site be determined in a tandem construct? An explanation we favor is a scheme that is most consistent with the model in which minus-strand RNA synthesis is attenuated at intergenic sites to yield subgenomic minus-strand templates for transcription (13; model 2 above). The nascent mRNA might then continue to amplify (1, 14; model 3 RNAs, genomic and subgenomic, are found as components of membrane-bound replication complexes in which the plus and minus strands exist in nearly equimo-actions. To ensure that the replicating molecules had not acquired mutations in the unused promoter regions that lar amounts (31) as apparent components of transcriptionally active double-stranded molecules (13, 32) . In ad-might have harmed promoter function, asymmetric RT-PCR sequencing was carried out on cytoplasmic DI ge-dition, at least one viral protein, N, and a 55-kDa cellular protein have been shown to bind to plus-strand leader nomic RNA bands in Fig. 2A with oligonucleotides 8 and N(0). No mutations were found (not shown).

in the region of the UCUAAAC consensus sequence (33) (34) (35) , suggesting that they may take part in transcription One possible determinant for the strong initiation of transcription from the IS1 promoter (fusion) site might initiation. The potential of N to multimerize (24, 26) might also contribute to protein-protein interactions of the kind have been a linkage between the transcription initiation and the closely associated translation start site. This postulated for assembly of coronavirus transcription complexes (11, 12). Many features of this view find prece-possibility was tested by converting the AUG at base 23 in the IS1 promoter to ACG in the pDrepIS1 construct dent in poliovirus for which both viral and cellular proteins bind to the 5 terminus of the plus-strand genome (Fig. 1 ) by site-directed mutagenesis. Subgenomic transcription rates were identical with the AUG and ACG and contribute mechanistically to initiation of plus-strand synthesis (37, 38) . codons at this site (data not shown). Thus, the mutation did not prevent subgenomic transcription, making the Thus, transcription factors could bind to plus strands at intergenic sites and interact with polymerase and other translation start codon an unlikely cause for the preference of downstream transcription initiation in pDrepIS2 cofactors to initiate transcription on nascent subgenomic minus strands as they are generated. In this scheme, the and pDrepIS3.

Our studies with a BCV DI RNA show results consis-most downstream site in the promoter series would be chosen because it would have associated with it the first tent with those of Makino and coworkers (8) (9) (10) 29) and Spaan and coworkers (11, 30) with MHV DI RNA in which assembled complex encountered by the approaching polymerase during minus-strand synthesis. On the basis it was shown first that synthesis of a subgenomic mRNA can be induced by insertion of an intergenic promoter of recent studies with high-frequency leader recombination in BCV (39), we further envision that the polymerase sequence into the DI RNA and second that enhancement of downstream subgenomic mRNA synthesis resulted might switch templates near the region of the intergenic promoter to copy a leader (possibly one on genomic or when two or more promoters were placed in close proximity. This pattern of mRNA synthesis from engineered subgenomic mRNAs) and thereby generate an antileader. The nascent antileader-containing minus strands DI RNAs of two separate coronaviruses, therefore, reflects the general pattern observed from the coronavirus would then serve as templates for synthesis of leadercontaining subgenomic plus strands and could conceiv-genome and would seem to validate the use of DI RNA molecules for deciphering the mechanisms of mRNA ably be repeatedly used as template were mRNA to replicate. A recent demonstration of in vivo transcription from abundance regulation. The results of our experiments did not allow us to discern among the postulated tem-the termini of subgenomic minus-strands would seem consistent with this model (40). Our experimental system, plates for transcription, which include the DI antigenome, nonreplicating subgenomic minus strands, and replicat-furthermore, may reflect a mechanism used by BCV in a natural case of strong downstream selection between ing subgenomic minus strands (corresponding to the three models of transcription described above). They do, two adjacent promoters (20). During synthesis of the

In mRNA for the BCV 12.7-kDa putative nonstructural pro-''Positive-Strand RNA Viruses

quence mapping just 15 nucleotides upstream

a locally high concentration of transcription factors is

Molecular Cloning

Proc. Natl. Acad. Sci. Wimmer, E