key: cord-0734199-so5jooj0
authors: Bertholet, Christine; Van Meir, Erwin; ten Heggeler-Bordier, Béatrice; Wittek, Riccardo
title: Vaccinia virus produces late mRNAs by discontinuous synthesis
date: 1987-07-17
journal: Cell
DOI: 10.1016/0092-8674(87)90211-x
sha: e54c0a07b76e9b5ae145e81a0b44cde27b923f76
doc_id: 734199
cord_uid: so5jooj0

Abstract We describe the unusual structure of a vaccinia virus late mRNA. In these molecules, the protein-coding sequences of a major late structural polypeptide are preceded by long leader RNAs, which in some cases are thousands of nucleotides long. These sequences map to different regions of the viral genome and in one instance are separated from the late gene by more than 100 kb of DNA. Moreover, the leader sequences map either upstream or downstream of the late gene, are transcribed from either DNA strand, and are fused to the late gene coding sequence via a poly(A) stretch. This demonstrates that vaccinia virus produces late mRNAs by tagging the protein-coding sequences onto the 3′ end of other RNAs.

The 5' flanking sequences of cellular genes transcribed by RNA polymerase II contain conserved sequence elements that are responsible for correct initiation of transcription and regulation of gene expression (Breathnach and Chambon, 1981) . In the case of DNA viruses that utilize the host-cell transcription machinery, the 5' flanking regions of the genes closely resemble those of the host cell.

Vaccinia virus, a member of the poxvirus family, belongs to the small group of DNA viruses that replicate in the cytoplasm of the host cell. Gene expression in vaccinia virus is temporally well regulated and occurs in two distinct phases. Early genes are transcribed before the complete uncoating of the viral DNA, and some encode enzymes that are subsequently used for DNA replication. Late genes encode most structural polypeptides and are expressed after DNA replication. Vaccinia virus thus offers a unique opportunity to study eukaryotic transcription and mechanisms of gene regulation that presumably have evolved independently from those of the host cell. All enzymes required for the production of mRNAs are carried in the virion (reviewed in Moss, 1985) and, although the mRNAs are made in the cytoplasm, they show such typical eukaryotic features as methylated caps (Wei and Moss, 1975) and poly(A) tails (Kates and Beeson, 1970) . The mRNAs produced early in infection have an average size of about 1500 nucleotides (Oda and Joklik, 1967; Cooper et al., 1981; Mahr and Roberts, 1984) . Furthermore, these molecules are not spliced (Wittek et al., 1980; Cooper et al., 1981) and processing at the 5' end has also been ruled out (Venkateaan and Moss, 1981) .

Late transcription is characterized by several unusual features. First, a large fraction of late RNA can selfhybridize to form double-stranded structures, indicating extensive symmetrical transcription late in infection (Colby and Duesberg, 1989; Duesberg and Colby, 1969; Colby et al., 1971; Boone et al., 1979) . Second, late mRNAs are on average about twice as long as early mRNAs and very heterogeneous in size (Oda and Joklik, 1967; Cooper et al., 1981; Mahr and Roberts, 1984) . These properties have been explained by a failure of the virus to terminate transcription specifically late in infection (Moss, 1985) .

Several studies have demonstrated that the sequences involved in gene regulation in vaccinia virus reside in the 5'flanking region of early and late genes (Weir and Moss, 1984; Bertholet et al., 1985; Cochran et al., 1985) . Recent analyses (Bertholet et al., 1986; Hanggi et al., 1966) have revealed unexpected features of the putative promoter region of a strongly expressed late gene that encodes a major structural polypeptide of molecular weight 11,000 (11K gene). Surprisingly, very short stretches, of about 20 nucleotides, preceding the translation initiation codon are sufficient to regulate late gene expression. All mutations around the putative mRNA start site, however, have abolished transcription. In several late genes the putative transcription initiation site has been located within, or very close to, the highly conserved sequence TAAATG, which also includes the translation initiation codon (Weir and Moss, 1984; Bertholet et al., 1985; Hirt et al., 1986; Weinrich and Hruby, 1986) . Thus late mRNAs appear to have extremely short nontranslated leader sequences. In all these studies, however, the 5' ends of the mRNAs were mapped by the nuclease Sl procedure. In this communication we show that the 5' end of the 11K mRNA, as defined by this technique, does not represent the true 5'end. Instead, the protein-coding sequences of the late gene are preceded by long, polyadenylated RNAs.

Primer Extension of the 11K RNA To define the 5'end of the 11K late mRNA by an alternative procedure, we performed primer extension experiments. As a control, we included the thymidine kinase (TK) early mRNA, which showed identical map positions for the 5 end by both Sl analysis and primer extension (Bajszar et al., 1983: Weir and Moss, 1983) . Appropriate 5' endlabeled DNA fragments ( Figure 1A ) were hybridized to early or late RNA from infected cells and then either extended with reverse transcriptase or treated with nuclease Sl ( Figure 1B ). For the TK mRNA, primer extension and Sl nuclease analysis yielded DNA fragments of identical length and of the expected size. A protected fragment of 300 nucleotides was obtained by the nuclease Sl procedure with the 11K mRNA. This places the putative 5' end B. 4500 Figure 1 . Primer Extension and Nuclease Sl Analysis of the TK Early and 11K Late mRNAs (A) shows the coding sequences of the genes (thick lines); the drrection of transcription is indicated by the arrows. The map positions of the 5' end-labeled fragments used as primers and Sl probes are also indicated. For analysis of the TK and 11K mRNAs, the primers were hybridized to early and late RNA, respectively (8, lanes P2). or to tRNA as a control (6, lanes Pl), and extended with reverse transcriptase.

Appropriate fragments were also hybridized to early or late RNA and then treated with Sl nuclease (B, C, lanes S,). Resulting DNA fragments were analyzed on a 6% polyacrylamide sequencing gel. The sizes (in nucleotides) of DNA fragments are indicated at right. To sequence the RNA, primer extensions of the 11K late mRNA were also performed in the presence of dideoxynucleotides (C. lanes G. A, T, C). The DNA sequence complementary to the RNA around the translabon initiation codon. as read from the gel, IS shown at the right in (C of the late transcript very close to the A residue of the ATG translation initiation codon, a map position similar to the one reported previously (Bertholet et al., 1985) . Primer extension, however, yielded a different and unexpected result. A series of closely spaced bands was observed and these fragments were on average about 20 nucleotides longer than the corresponding nuclease Sl-protected fragment ( Figure 1B ). In addition, intense bands were seen around the position of 1500 nucleotides. This material was resolved into multiple bands upon agarose gel electrophoresis (not shown), indicating considerable length heterogeneity.

To exclude the possibility that the primer extension result was due to nonspecific hybridization of the DNA fragment, the primer was hybridized to RNA under the same conditions but was extended in the presence of dideoxynucleoside triphosphates ( Figure 1C ). An easily readable sequence was obtained up to the position of the Slprotected fragment, and this sequence corresponded to that reported previously for the 11K gene (Bertholet et al., 1985) . Beyond this point, bands were observed in all four lanes but were most intense in the '7" track, indicating that this region is rich in A residues (mRNA-like strand). In fact, at least 3-4 A residues immediately upstream of the 11K ATG translation initiation codon could be read. This experiment thus clearly demonstrated that the primer hybridized to the expected RNA but that the RNA contained additional sequences upstream of the 5' end defined by nuclease Sl analysis, and that this stretch was heterogeneous in length and possibly also in sequence.

To characterize these extra sequences, a cDNA library was made from total poly(A)-containing late RNA. We wish to emphasize that cDNA clones were selected on the basis of two criteria only. First, clones were isolated by colony hybridization using as probe an EcoRI-Hindlll fragment of 135 bp from the very beginning of the 11K coding sequences (note that the EcoRl site starts at the G residue of the 11K ATG). Second, only those cDNA clones in which that EcoRl site was present were chosen for further analysis. All cDNA clones that fulfilled these criteria are shown schematically in Figure 2 , where they are compared with the corresponding genomic DNA around the 1lK gene. In addition to the EcoRl site, Hindlll and Clal sites characteristic of the 11K gene coding sequence and 3'flanking region were also present in all six cDNA clones. Furthermore, the four cDNA clones with long 3' flanking regions also had the expected BamHl site. Thus, whereas the sequences downstream of the EcoRl site are characteristic of the 11K gene and its 3' flanking region, this is not true for the sequences upstream of it. In this region the restriction maps of the cDNA clones differ from the corresponding region of genomic DNA. Furthermore, these upstream sequences vary considerably in length between different cDNA clones, ranging in size from less than 100 nucleotides to more than 2000 nucleotides (the latter for cDNA clone 9). The presence of short as well as long sequences upstream of the 11K ATG thus reflects the primer extension result. Moreover, the different restriction maps of the three long cDNAs (clones 3,8, and 9) suggest that these sequences differ.

of cDNA-Derived Probes to RNA The extra sequences added onto the 11K coding sequences are either of viral or cellular origin. We reasoned that hybridization of cDNA-derived probes exclusrvely to RNA from infected cells would argue that the RNAs are of viral origin, although such a result would not exclude the possibility that they represent cellular RNAs that are induced upon virus infection. RNA was isolated from noninfected or infected cells and then hybridized to 32P-labeled probes isolated from the 5' end region, upstream of the 11K coding sequences of cDNA clones 3, 8, and 9 (see Figure 2 ). As a control, a fragment containing the coding region of the 11K gene and Yflanking sequences was also hybridized to various nitrocellulose-bound RNAs. Weak hybridization to RNA from uninfected cells was only observed when the filters were washed at low stringency ( Figure 3 ). In contrast, the probes from cDNA clones 3 and 8 hybridized strongly to both early and late RNA from infected cells. Whereas these probes hybridized somewhat more strongly to early than to late RNA, the opposite is true for the probe derived from cDNA 9. Finally, the probe specific for the 11K gene hybridized strongly to late RNA, as expected, but also to early RNA. This may be due to the presence of long early RNAs of unknown significance; these RNAs have been found to be transcribed from various parts of the vaccinia virus genome (Wittek et al., 1980; Cooper et al., 1981) , and they may traverse early or late genes on the same or opposite DNA strand.

Probes to Viral DNA The hybridization experiment suggested that the sequences upstream of the 11K coding sequences are transcribed from the vaccinia virus genome. To confirm this, and to identify the regions from which they are transcribed, the cDNA-derived probes used in the previous experiment were hybridized to cloned DNA restriction fragments representing the entire vaccinia virus genome ( Figure 4 ). Each probe hybridized to only one fragment; in the case of probes p3 and p8, the fragments map more than 100 kb apart on the genome. Moreover, with respect to the position of the 11K gene, which is located at the junction of the Hindlll F and Hindlll E fragments (Wittek et al., 1984) and is transcribed from left to right, probe p3 hybridized to sequences located, upstream (see below), whereas probes p8 and p9 clearly hybridized to sequences downstream of it. %P-labeled DNA probes (~3, ~9, p9, pllK) derived from the 5' region of the corresponding cDNA clones or from genomic DNA (see Figure  2 ) were hybridized lo RNA isolated from uninfected (HeLa) or infected (Early, Late) cells, or to tRNA immobilized on a nitrocellulose membrane. After hybridization, parallel strips were washed at 38% in 50% formamide. 0.1% SDS, and the indicated SSC concentration (see Experimental Procedures). N.D., not determined.

Size Heterogeneity of the 11K RNA Population The previous experiments demonstrated that various RNAs transcribed from different regions of the viral genome serve as leader RNAs. The 11K mRNA population should thus exhibit considerable length heterogeneity. This was tested by Northern blot analysis ( Figure 5 ). When a DNA fragment from the 11K coding region was used as a hybridization probe (pllK), two long RNA species were detected with early RNA. These presumably represent RNAs transcribed from the DNA strand oppo- A synthetic oligonucleotide complementary to nucleotides 44 to 65 downstream of the G of the 1lK ATG of the noncoding strand was used as a primer for sequencing.

Plasmid DNA was used to sequence clones 1 and 3; sequences for clones 8 and 9 were obtained from fragments cloned into single-stranded phage DNA (see Figure 2 for clone designations).

The genomic sequence (mRNA-like strand) around the 11K ATG is compared with the corresponding region of the cDNA (bottom). Boldface, underlined letters indicate common sequences, and the bar represents the DNA region to which the 5'end of the 11K mRNA was mapped by nuclease Sl analysis.

site to the 11K gene previously detected by nuclease Sl analysis (Wittek et al.,1984) . With late RNA, a smear was observed over a wide size range. This size heterogeneity is certainly in part due to the well-established length heterogeneity at the 3' end of late RNAs (reviewed by Moss, 1985) . On the basis of the previous experiments, however, it is clear that the S'end also contributes to the overall size heterogeneity. Total early and late RNAs were also hybridized to probes p3, p8, and p9 derived from the leader RNAs. Particularly with probes p8 and p9, smears were seen with late RNA, demonstrating that late RNAs encoded in these parts of the genome also vary in size. Contrary to what one might have expected, probe pllK did not detect smears that comigrated with those detected with the leader-RNA-derived probes. This is not surprising since, as shown by cDNA cloning, the length heterogeneity contributed by the 5' end of the 11K RNA population results from fusion of different leader RNAs to the 11K coding region. In addition, a given leader RNA as represented by probe p3, p8, or p9 may be fused to various other RNA% of which the 11K RNA population presumably only represents a small subset. Furthermore, some RNAs detected by the leader RNA probes may themselves be acceptor RNAs to which other leader RNAs are fused.

Sequence at the Junction of Leader RNA and 11K Coding Region To characterize the junction at the 11K and foreign sequences, the relevant portions of the cDNAs were se-quenced. The cDNA clones 3, 8, and 9 were chosen for this analysis, as was clone 1, which contained only a short stretch of additional DNA upstream of the EcoRl site (Figure 8 ). An identical sequence, corresponding to the 11K gene (Bertholet et al., 1985) , was obtained for all four cDNA clones up to the position of the ATG codon. Beyond this point, clone one had a long poly(A) stretch (on the mRNA-like strand) of about 40 residues, which led to the typical problems in the sequencing reactions described for long homopolymer stretches (Smith, 1980) and which presumably also caused the problems in the sequencing of the RNA ( Figure 1C ). This region was followed by a poly(G) tract, as expected from the cDNA cloning protocol used. The cDNA clones 8 and 9 also contained long poly(A) sequences of at least 80 residues. This is less obvious for cDNA 3, in which only the first three bases can clearly be identified as A residues.

In the three long cDNAs, the homopolymer stretches appeared to be followed by other sequences, although the actual sequences could not be read in this part of the gel. Appropriate fragments from the 5'end of the cDNA clones were therefore subcloned, and in each case stretches of 300 to 400 nucleotides were sequenced in the opposite direction (not shown). This confirmed that the leader RNAs contain additional sequences upstream of the poly(A) tract.

Nucleaee Sl Protection of a cDNA-Derived Probe The primer extension and cDNA cloning experiments described above depended on the use of reverse transcriptase. To rule out the possibility that the unusual structure at the Vend of the 11K mRNA as revealed by these experiments is due to an artifact generated by reverse transcriptase, we searched for means independent of this enzyme to confirm the observed structures. A nuclease Sl protection experiment was performed using as hybridization probe a DNA fragment derived from cDNA clone 1 ( Figure  7A ). This fragment was 5' end-labeled at the Hindlll site within the coding sequences of the 11K gene, and it contained a stretch of 30 T residues (complementary strand) immediately upstream of the AAATG genomic sequence (see also Figure 8 ) comprising the 11K translation initiation codon. This sequence is preceded by 15 C residues resulting from the cDNA cloning procedure, and the fragment was isolated after cleavage at the Pstl site just upstream of this sequence. A second probe consisted of a Xbal-Hindlll fragment derived from genomic DNA and was also labeled at the Hindlll site. The genomic ( Figure  78 , lane 5) and cDNA-derived probes (lane 8) were analyzed either directly or after hybridization to early RNA and Sl treatment (lanes 1 and 2). As expected, no protected fragments were observed with early RNA and either probe. Note that in contrast to the probe used for hybridization to RNA (Figure 3) , the Sl probe was labeled on one strand only and therefore did not detect the early RNA transcribed from the DNA strand opposite to the 11K gene (Wittek et al., 1984) . Total late RNA from infected cells protected fragments of 127 and 128 nucleotides of the genomic probe, consistent with the previous experiments, which demonstrated that the sequence complementarity between the genomic DNA and 11K mRNA ends just upstream of the 11K ATG codon. Significantly, protected fragments of about 158 nucleotides were observed with the cDNA-derived probe, as well as some reannealing of the probe (Figure 78, lane 4) . These bands can best be explained by the presence of a poly(A) sequence upstream of the 11K ATG codon. The ladder pattern in the lower part of the gel probably results from nuclease Sl cleavage at internal positions within the poly(A) sequence. However, very little, if any, material was observed at the position of the protected fragments obtained with the genomic probe, suggesting that the majority of the 11K mRNA molecules contain such a poly(A) stretch upstream of the translation initiation codon.

To confirm that the 11K mRNA molecules also contain sequences upstream of the poly(A) tract, we performed electron microscopy of DNA-RNA hybrids. A Clal DNA fragment consisting of about 700 bp of DNA downstream of the 11K ATG codon and about 600 bp upstream of the translation initiation codon was isolated. This DNA was hybridized to 11K RNA purified by hybrid selection to DNA of the 11K coding region, and the resulting molecules were analyzed by electron microscopy. Many molecules with a "Y" structure characteristic of the two hybrids shown in Figure 8 were examined. Four different regions can be distinguished, two of which were constant in size and two of which varied in size. We interpret the different regions as follows: Region 1, which is constant in size, represents the single-stranded DNA of the 11K 5' flanking sequence. Region 2 varies in size from 890 to 6070 nucleotides and represents the leader RNA. The double-stranded region 3 results from hybridization of the 11K coding sequence and 3' flanking sequence to the corresponding region of the RNA. Finally, region 4 is the 3'portion of the RNA. The fact that this region again varies considerably in length is not unexpected from the 3' length heterogeneity of late vaccinia virus transcripts (see Moss, 1985) . Thus the results obtained by electron microscopy are consistent with the unusual structure of the 11K mRNA proposed on the basis of the previous experiments.

The following experiment was designed to map precisely the genes encoding the leader RNAs on the vacciniavirus DNA and to determine their direction of transcription. The restriction sites present in cDNA clones 3, 8, and 9 were first mapped within the large Hindlll fragments to which the corresponding cDNA-derived probes had hybridized (see also Figure 4 ). Figure 9A shows the map positions of these sequences within the Hindlll fragments F, 8, and I. These are shown in the correct left-to-right orientation with respect to the conventional orientation of the genome (Figure 98) . Thus, the sequences in cDNA clone 3 map to the left-hand half of the Hindlll F fragment, whereas the 11K gene is located at the extreme right-hand end (Wittek et al., 1984) . The sequences in cDNA clone 8 map at the lefthand end of the 30 kb Hindlll B fragment, and those present in clone 9 map toward the right of the Hindlll I fragment. Furthermore, from the 5'-to-3 polarity of the leader RNAs established by restriction analysis of the cDNA clones (see also Figure 2 ) and as indicated by the poly(A) se-A. EPsBl C  BB  E  c CCEH  I  III,  ,I,  , I I I Figure 9A ), it is obvious that the leader RNAs in cDNA clones 3 and 9 are transcribed from the leftwardreading strand of the vaccinia virus genome ( Figure 9B ). On the other hand, the 11K mRNA is transcribed in the opposite direction (Wittek et al., 1984) , as is the leader RNA of cDNA clone 8.

We have shown that late in infection vaccinia virus produces chimeric mRNAs consisting of the coding sequences for a major structural polypeptide downstream of poly(A) sequences. We have also found cDNA clones that contain only the poly(A) stretch, and thus lack the sequences preceding it. It is possible that these molecules result from premature termination of reverse transcription during cDNA cloning. It is known, at least for the Klenow fragment of DNA polymerase, that long homopolymer stretches are not easily copied (Smith, 1980) . This interpretation is also consistent with the results obtained by electron microscopy, where we have found no evidence for RNA molecules that had only very short sequences upstream of the 11K coding sequences. On the contrary, most molecules observed possessed even longer leader RNAs than those found by cDNA cloning. Both procedures also confirmed the 3' length heterogeneity of late RNA. We have sequenced the S'ends of five cDNA clones and have found stretches of between 30 and 100 A residues (not shown). These ends, therefore, most likely represent the true 3'ends and did not result from priming at internal positions in the RNA molecules. On the other hand, it is clear from the data presented here that the 5' end also contributes significantly to the overall length heterogeneity of late RNA.

An important question is whether this unusual mechanism of RNA synthesis is unique to the particular late gene examined. A preliminary analysis of cDNA clones for another late gene that we have recently mapped and sequenced (Hirt et al., 1988) has shown that they also contain a poly(A) stretch and additional sequences upstream of the translation initiation codon that are not transcribed from DNA in the immediate vicinity of the gene.

Perhaps the most intriguing aspect of the bizarre structure of the 11K mRNA concerns its translation. In most eukaryotic mRNAs the AUG closest to the 5' end is used for translation initiation (Kozak, 1978) , although exceptions to this rule, particularly for mRNAs of viral origin (reviewed in Kozak, 1988) have been described. If the 11K polypeptide is indeed translated from the mRNAs described in this paper, the translation initiation codon would be located in several cases thousands of nucleotides downstream of the 5' end. Furthermore, the leader sequences certainly contain several AUGs. One might therefore expect a poor translation efficiency of the 11K sequences. However, the opposite appears to be true. From the abundance of the polypeptide, the 11K gene is judged to be one of the most strongly expressed vaccinia virus genes, since its product contributes about 10% to the total protein mass present in purified virions (Sarov and Joklik, 1972; Moss, 1974) . Furthermore, in vitro translation of hybrid-selected RNA, which was used to map the 11K gene (Wittek et al., 1984) RNA that was first selected by hybridization to the cDNA of one of the leader RNAs and then on the 11K coding sequences, suggests that such chimeric RNAs may indeed be functional 11K mRNAs. Another interesting question is whether the leader RNA sequences are also translated. At present this question cannot be answered, since in the previous mapping of the 11K gene (Wittek et al., 1984) we used immunoprecipitation of the in vitro translation products of hybrid-selected RNA to identify the 11K polypeptide.

Any additional polypeptides translated from 11K mRNA would therefore have been lost. It should be borne in mind, however, that these unusual mRNAs are translated in infected cells, and results obtained in a reticulocyte lysate might not reflect the in vivo situation. mRNAs, this could explain the shutoff of early gene expression late in infection, which has been particularly well studied in the case of the viral TK gene (McAuslan, 1963a (McAuslan, , 1963b Jungwirth and Joklik, 1965; Zaslavsky and Yakobson, 1975; Hruby and Ball, 1981 tides of cellular mRNAs, including the cap structure, to prime the synthesis of its mRNAs (Plotch et al., 1981; Herz et al., 1981; reviewed in Krug, 1985) , which thus consist of cellular and viral sequences. A similar mechanism has also been reported for bunyavirus (Patterson et al., 1984) .

Coronaviruses use a virus-encoded leader RNA of about 70 nucleotides to prime the synthesis of the individual mRNAs on the template molecule (Lai et al., 1984; Baric et al., 1985; reviewed in Krug, 1985) . A nonviral example is the mRNAs of trypanosomes, which contain an identical 35 nucleotide leader sequence (reviewed in Borst, 1986; Van der Ploeg. 1986 ). It is not clear how the leader is added to the body of the mRNA, but frans splicing appears to be the most likely mechanism (Murphy et al., 1986; Sutton and Boothroyd, 1986) . We can only speculate on the mechanism by which the long polyadenylated RNAs are fused to the late coding sequences in vaccinia virus. Splicing of a single, very long precursor molecule can be excluded, since the sequences added to the body of the late mRNA are transcribed from either DNA strand and are located either upstream or downstream of the 11K gene. A priming mechanism can be envisaged in which the poly(A) sequence of the donor RNA interacts with sequences on the DNA coding strand. An alternative hypothesis is that the two RNAs are joined together by ligation.

One would then have to explain how transcription of the late coding sequences is initiated. Although it cannot be excluded that the upstream sequences act as promoters, we prefer an alternative model in which the highly conserved 5'-TAA-ATG-3' sequences represent processing sites at the RNA level. It is well established that very long transcripts are made particularly late in infection, and from both DNA strands (reviewed in Moss, 1985) . These transcripts might represent precursor molecules from which the late coding sequences are excised by a site-specific endoribonuclease. Interestingly, an enzyme that appears to possess the required properties has been isolated from vaccinia virions and partially characterized (Paoletti and Lipinskas, 1978) . Clearly, further work is needed to understand the mechanism by which donor and acceptor RNAs become joined. However, in comparison with trypanosomes, vaccinia virus represents a relatively simple system that should greatly facilitate the unraveling of this unusual mechanism of RNA production.

Virus and Cells

The WR strain of vaccinia virus was obtained from Bernard Moss, National Institutes of Health, Bethesda, Md. HeLa cells were grown as monolayer cultures in Dulbecco's modified Eagle's minimal essential medium supplemented with 5% fetal calf serum.

Total RNA from infected HeLa cells was isolated after lysis of the cells with 8 M guanidinium hydrochloride exactly as previously described (Wittek et al., 1964) . Early RNA was isolated at 4 hr after infection from cells maintained in medium containing either 100 pglml of cycloheximide or 40 pglml of cytosine arabinoside.

Late RNA was extracted at 7 hr after infection from cells that were not treated with either inhibitor. 

Sl Analysis The 5' ends of RNA were mapped by Sl analysis (Berk and Sharp, 1977) using 5' end-labeled DNA fragments as hybridization probes (Weaver and Weissman, 1979) . The experimental details were as previously described (Wittek et al., 1984) .

cDNA Cloning and Identification of cDNA Clones Poly(A)containing cytoplasmic RNA from infected cells isolated late in infection was annealed to oligo(dT)-tailed plasmid p&V-1 vector DNA (Okayama and Berg, 1983 ) purchased from Pharmacia. All subsequent steps were performed exactly as described in the detailed protocol supplied by Pharmacia, except that JMlO9 host bacteria were used for the initial plasmid amplification.

Desired cDNA clones were identified by colony hybridization using an EcoRI-Hindlll DNA fragment located at the beginning of the 11K coding sequences (Bertholet et al., 1985) ; the fragment was 32P-labeled by repair synthesis (Maniatis et al., 1982) . DNA from positive colonies was isolated, and used to transform competent DH5.1 host cells (Vector Cloning Systems, San Diego, Cal.). Slot Blot Hybridization DNA or RNA was immobilized on nitrocellulose membranes using a minifold apparatus (Schleicher and Schuell). The filters were dried at 60% for 2 hr and prehybridized in a solution consisting of 150 uglml of denatured herring sperm DNA, 5x SSC (SSC is 0.15 M NaCI, 0.015 M sodium citrate), 25 mM sodium pyrophosphate, Ix Denhardt's solution (Denhardt, 1966) and 50% formamide, at 40% for 4-12 hr. Hybridizations were carried out at 40°C for 16-90 hr with =P-labeled DNA probes (Rigby et al., 1977) that were denatured and diluted into prehybridization solution. For RNA-DNA hybridizations(Figure3) parallel filters were washed at 38OC in 50% formamide, 0.1% SDS, and either 2x SSC, 0.5x SSC, or'0.2x SSC. For DNA+DNA hybridizations ( Figure 4 ) the.filters were washed at 38°C in 60% formamide, 0.1% SDS, 0.2x SSC.

Blot Analysis I .

For Northern blot analysis, 10 ug of total early or late RNA was denatured with glyoxal as previously described (McMaster and Carmichael, 1977) and then size-fractionated by electrophoresis in a 1% agarose gel. The RNA was then transferred to nitrocellulose membranes (Thomas, 1980) and hybridized to DNA fragments that had been 32P-labeled by the random-primer procedure (Feinberg and Vogelstein, 1983) . Prehybridization, hybridization, and washing of the membranes in 0.5x SSC were performed as described for slot blot hybridization.

DNA Sequencing Sequencing of DNA was performed by the chain-termination method (Sanger et al., 1977) either on linearized, denatured plasmid DNA or after appropriate DNA fragments were subcloned on single-stranded recombinant phage WB238 DNA ( Barnes and Bevan, 1983; Sahli et al., 1985) .

Microscopy For electron microscopy, llK-specific RNA was selected by hybridization to a single-stranded Hindlll-Clal fragment containing the 1 IK coding sequences.

The DNA was covalently bound to Sephacryl S-1000 (Pharmacia) by the procedure of Seed (1982) with modifications (BOnemann et al., 1982; BOnemann, 1982) . AClal DNAfragment spanning the 11K gene (Wittek et al., 1984) was purified from an agarose gel. DNA-RNA hybrids were formed as described by Brack et al. (1976) . RNA and DNA were dissolved at 10 &ml, and 1 ul of each was diluted into 14 PI of 76% formamide, 100 mM PIPES-NaOH (pH 7.8) 20 mM Tris-HCI (pH 7.8) 4 mM EDTA, 500 mM NaCI. Samples were heated to 8oOC for IO min and then incubated at 49% for 1 hr. The mixture was then diluted with 10 volumes of 70% formamide, 100 mM Tris-HCI (pH 8.5) 10 mM EDTA, 50 &ml of cytochrome C, and spread on distilled water as hypophase. The samples were processed for electron microscopy as previously described (Brack et al., 1978) . Plasmid pBR322 and phage (pX174 DNA were included as double-and singlestranded length standards, respectively.

Immobilization of denatured DNA to macroporous supports: I. Efficiency of different coupling procedures

In vitro mutagenesis of the promoter region for a vaccinia virus gene: evidence for tandem early and late regulatory signals

Double-stranded RNA in vaccinia virus infected cells

Mechanism of synthesis of vaccinia virus double-stranded ribonucleic acid in vivo and in vitro

Extension of the transcriptional and translational map of the left end of the vaccinia virus genome to 21 kilobase pairs

A member filter technique for the detection of complementary DNA

On the biosynthesis and structure of double-stranded RNA in vaccinia virus-infected cells

A technique for radiolabeling DNA restriction endonuclease fragments to high specific activity

Conserved TAAAT motif in vaccinia virus late promoters: overlapping TATA box and site of transcription initiation

Influenza virus, an RNA virus, synthesizes its messenger RNA in the nucleus of infected cells

Localization and fine structure of a vaccinia virus gene encoding an envelope antigen

Control of expression of the vaccinia virus thymidine kinase gene

Studies on "early" enzymes in HeLa cells infected with vaccinia virus

Ribonucleic acid synthesis in vaccinia virus. II. Synthesis of polyriboadenylic acid

How do eucaryotic ribosomes select initiation regions in messenger RNA?

Bifunctional messenger RNAs in eukaryotes

The role of RNA priming rn viral and trypanosomal mRNA synthesis

Characterization of leader RNA sequences on the virion and mRNAs of mouse hepatitis virus, a cytoplasmic RNA virus

Arrangement of late mRNAs transcribed from a 7.1-kilobase EcoRl vaccinia virus DNA fragment

Molecular Cloning: A Laboratory Manual

Control of induced thymidine kinase activity rn the poxvirus infected cell

The induction and repression of thymidine kinase in the poxvirus-infected HeLa cell

Analysis of single-and double-stranded nucleic acids on polyacrylamide and agarose gels by using glyoxal and acridine orange

Reproduction of poxviruses

Replication of poxviruses

Identification of a novel Y branch structure as an intermediate in trypanosome mRNA processing: evidence for Pans splicing

Hybridization and sedimentation studies on "early" and "late" vaccinia messenger RNA

A cDNA clomng vector that permrts expression of cDNA inserts in mammalian cells

Soluble endoribonuclease activity from vaccinia virus: specific cleavage of virion-assocrated highmolecular weight RNA

La crosse virions contain a primer-stimulated RNA polymerase and a methylated cap-dependent endonuclease

A unique cap (m7GpppXm)-dependent influenza vrnon endonuclease cleaves capped RNAs to generate the primers that rmtiate vrral RNA transcnption

Labeling deoxyribonucleic acid to high specific activity in vitro by nick translation with DNA polymerase I

DNAsequence comparison between two tissue-specific variants of the autonomous parvovirus, minute virus of mice

DNA sequencing with chain-terminating inhibitors

Studies on the nature and location of the capsid polypeptides of vaccinia virions

Diazotizable arylamine cellulose papers for the coupling and hybridization of nucleic acids

DNA sequence analysis by primed synthesis

Evidence for trans splicing in trypanosomes

Hybridization of denatured RNA and small DNA fragments transferred to nitrocellulose

Discontinuous transcription and splicing in trypanosomes

In vitro transcription of the inverted terminal repetition of the vaccinia virus genome: correspondence of initiation and cap sites

Mapping of RNA by a modification of the Berk-Sharp procedure: the 5' termini of 6-globin mRNA have identical map coordinates

Methylated nucleotides block S-terminus of vaccinia virus mRNA

A tandemly-oriented late gene cluster within the vaccinia virus genome

Nucleotide sequence of the vaccinia virus thymidine kinase gene and the nature of spontaneous frameshift mutations

Regulation of expression and nucleotide sequence of a late vaccinia virus gene

Expression of the vaccinia virus genome: analysis and mapping of mRNAs encoded within the inverted terminal repetition

Mapping of agene coding for a mafor late structural polypeptide on the vaccinia virus genome

Control of thymidine kinase synthesis in IHD vaccinia virus-infected thymidine kinase-deficient LM cells

We are grateful to Anne SeilerTuyns and Philippe Walker for the

Bertholet, C., Drillien, R., and Wittek, R. (1985) . One hundred base pairs of 5'flanking sequence of a vaccinia virus late gene are sufficient to temporally regulate late transcription.Proc. Natl. Acad. Sci. USA 82, 2096 -2100 . Bertholet, C., Stocco, t?, Van Meir, E., and Wittek, R. (1986 . Functional analysis of the 5' flanking sequence of a vaccinia virus late gene. EMB~ J. 5, 1951 EMB~ J. 5, -1957 Boone, Ft. F, Parr, R. P, and Moss, B. (1979) . Intermolecular duplexes formed from polyadenylated vaccinia virus RNA. J. Virol. 30, 365-374. Borst, l? (1986) . Discontinuous transcription and antigenic variation in trypanosomes.Ann. Rev. Biochem. 55, 701-732. Brack, C., Hirama, M., Lenhard-Schuller, R., and Tonegawa, S. (1978