VIROLOGY
190,
403-412
(1992)
Regulation of Caulimovirus Gene Expression and the Involvement of c&-Acting Elements on Both Viral Transcripts HERMAN B. SCHOLTHOF,’ Department
of Plant
FANG C. WU, SIDDARAME
GOWDA,
Pathology,
Lexington,
Received
University April
of Kentucky,
15, 1992;
accepted
June
AND
Kentucky
ROBERT I. SHEPHERD* 40546-009
1
5, 1992
In a further analysis of gene regulation of figwort mosaic virus (FMV), a caulimovirus, we studied transient gene expression with modified viral genomes in Nicotiana edwardsonii cell suspension protoplasts. The results demonstrated that the presence of the promoter for the full-length RNA interferes with expression from the separate downstream promoter for gene VI. In addition, expression of gene VI was inhibited by cis-acting sequences within gene VI itself. Both inhibitory effects could be partially relieved by coelectroporation with a plasmid that produces gene VI protein, demonstrating that expression of gene VI is transactivated by its own product. Subsequent expression studies with partially redundant FMV plasmids containing a reporter gene in frame with gene IV showed that efficient transactivation of CAT expression relies on a cis-acting element inside the downstream gene VI. Insertions of a transcriptional terminator upstream of the &-acting element for premature termination of transcription showed that the c/s-acting region is not a DNA element but is active only as a feature of the RNA transcript. We conclude that the k-acting element, together with the transacting gene VI product, enhances expression of all major genes, including gene VI, o WIZ Academic PWSS, IIW. from the polycistronic mRNA and the separate mRNA for gene VI.
al., 1989), and a recent investigation with FMV showed that genes further downstream on the full-length transcript are equally well transactivated (Scholthof et al., 1992). The underlying mechanism of this process remains unclear but it might involve interactions that allow reinitiation by ribosomes for a coupled translation of genes on the polycistronic mRNAs (Futterer and Hohn, 1991; Scholthof et a/., 1992). A &-acting element inside gene VII and an actively expressed gene VI are the only requirements for transactivation of prematurely terminated transcripts containing the leader of FMV and one or two downstream genes (Gowda et a/., 1989, 1991). Since gene VI appears to be crucial for gene expression we have further analyzed its regulation and the molecular interaction of its product during transactivation of viral genes further downstream on the polycistronic transcript. Our results show that expression of gene VI from its own promoter can be down-regulated by the upstream full-length promoter. Moreover, we have found that the gene VI product is able to transactivate its own expression. Our results suggest that gene VI, like all other ORFs, might be translated efficiently from the full-length RNA as well as from its own mRNA. The response to transactivation of the chloramphenicol acetyltransferase (CAT) gene, inside gene IV of a partially redundant plasmid of FMV, depends on the presence of a second &-acting element, in this case located near the 3’ end in the coding
INTRODUCTION Figwort mosaic virus (FMV), a caulimovirus, has a circular double-stranded (ds) DNA genome of 7743 bp (Richins et a/., 1987). Its genome (Fig. 1) is organized much like that of other caulimoviruses such as cauliflower mosaic virus (CaMV) (Gardner et al., 1981) the type member of the group. Caulimoviruses have features in common with animal hepadnaviruses and.accordingly both viruses have been grouped as pararetroviruses (Temin, 1985). These viruses produce a genomic length transcript with direct repeats at each end of the molecule. Their DNA is replicated by reverse transcription of this full-length transcript which, in the case of caulimoviruses, may also function as a polycistronic mRNA for translation of the six closely spaced major genes (Futterer and Hohn, 1991; Scholthof et al., 1992). A separate promoter is responsible for generating a small transcript that spans gene VI (Scholthof ef a/., 1992), which makes it the only FMV gene that appears as a monocistronic mRNA. The product of gene VI was initially shown to activate expression of the 5’ major open reading frames (ORFs) on CaMV (Bonneville et a/., 1989) and FMV (Gowda et
’ Present address: California, Berkeley, ’ To whom reprint
Department of Plant Pathology, CA 94720. requests should be addressed.
University
of
403
0042-6822/92
$5.00
Copyright 0 1992 by Academic Press. Inc. All rights of reproduction in any form reserved.
SCHOLTHOF
FIG. 1. A schematic map of the FMV genome. The interwoven lines represent the circular viral genome of 7743 base pairs of double-stranded DNA. Selected restriction sites are indicated. Of the four interruptions one (a) occurs in the coding strand and three (p) occur in the noncoding strand. The peripheral arrows and Roman numerals indicate the open reading regions. The two intergenic regions, the smaller one between genes V and VI and the larger one between genes VI and VII, contain promoters for the smaller gene VI mRNA and the larger full-length transcript, respectively. These RNAs are represented by the thin lines of the inner circle.
region of gene VI. This k-acting element is active on the full-length RNA as well as on the smaller gene VI transcript.
MATERIAL Standard
AND METHODS
protocols
Standard molecular biology techniques were employed throughout the reported studies (for protocols, see Maniatis et al,, 1982). The procedure described by Gowda et a/. (1989) was used for electroporation and CAT assay of protoplasts from Nicotiana edwardsonii cell suspensions. Quantitative data analyses were performed according to Scholthof et al. (1991).
Plasmid constructs The numbers in this section denote the position of the respective sites on the genomic map of FMV (Richins eta/., 1987) (Fig. 1). However, many plasmids are derived from a naturally occurring deletion mutant devoid of 1237 nucleotides (nt) which encompass most of gene IV and the 5’ end of gene V, resulting in a gene
ET AL.
IV/V fusion (Scholthof et a/., 1991). Construction of pSX103, pH25, pH32, and pH65 (Figs. 2 and 3) has been described previously (Scholthof et al., 1991, 1992). pH55 is similar to pH65 (Fig. 3), except that the promoter for the full-length transcript is deleted (Scholthof, 1990). For construction of pH95, pH32 was cleaved with Sphl (nt 3630) and SnaBl (nt 4795) and the fragment was replaced with the Sphl (nt 6600) to SnaBl (nt 4795) fragment of pDFiPR, a partially redundant clone of the FMV deletion mutant (Scholthof eta/., 1991). pH95 is similar to pH65 except that in the former the polyadenylation signal of the nopaline synthase gene (NT) is positioned immediately downstream of the CAT gene (Fig. 2). For construction of pH32VI (Fig. 3) the WI-BarnHI fragment (nt 3632 and 7083, respectively) of pSXlO3 was inserted in the compatible window of p3’NT (Scholthof et a/., 1992). The resulting plasmid is similar to pH32 (Scholthof et al., 1992) (Fig. 3) except that in pH32VI gene VI sequences are present between the CAT gene and NT, whereas these are absent in pH32. Note that these gene VI sequences are not capable of expressing gene VI because its 5’ terminus is deleted and the downstream portion is not positioned in frame with the CAT gene. Plasmids pH63b and pH63bAS (Scholthof et al., 1992) are FMV derivatives with the CAT gene in frame with the ORF IV/V fusion, followed by either an intact or an inactivated gene VI, respectively (Fig. 4). For construction of pH20, the Nsil fragment (nt 6603-3428) of pH14 (Scholthof et a/., 1991), was inserted in the Pstl site of p3’NT (Scholthof et a/., 1992). The resulting plasmid contains the FMV promoter for the full-length transcript with the downstream large intergenic region and genes VII, I, II, III, and IV/CAT-fusion followed by NT (Fig. 4). pH63bAN (Fig. 5) was generated by introducing a deletion from the Nsil sites (nt 3428 to 6603) in pH63b, which covers gene IV/V and VI. This plasmid is similar to pH20 but it contains the FMV polyadenylation signal instead of NT. Plasmid pH63bAV (Fig. 5) contains the Nsil fragment (nt 6603-3428) of pKScCAT which was inserted into the compatible Pstl site of pJSX6 (Scholthof et a/., 1991). This plasmid contains an inactivated gene VI because the promoter and the 5’ end of this ORF are deleted. In pH93b+NT (Fig. 5) the NT polyadenylation signal was inserted as a Pvull fragment from p3’NT (Scholthof et al., 1992) into the Stul site (nt 5379) of pH63bAV. The orientation was such that transcription would stop immediately downstream of the CAT gene, upstream from gene VI. However, in pH93b-NT (Fig. 5) the NT polyadenylation signal was positioned in the opposite orientation to that in pH93b+NT so that transcription would continue up to the native FMV polyadenylation signal downstream of gene VI.
&-vans
REGULATION
The plasmid used for lransactivation (pGS1 RVI) (Gowda et al., 1989) has gene VI placed between the 35 S promoter of CaMV and the polyadenylation signal of the small subunit of rubisco. This plasmid was added to the electroporation mixture to test for geneVI-induced Iransactivation. A similar plasmid (pGSl), except devoid of gene VI, was used as a control in these experiments. RESULTS Interaction
of full-length
and gene VI promoters
Previously we reported that FMV, like CaMV, contains a promoter upstream of gene VI which is used for transcription of a monocistronic mRNA spanning gene VI (Scholthof et al., 1992). In that report we mentioned the striking difference in CAT gene expression levels from plasmids like pH32 in comparison to that of plasmid pH65, although both have the CAT gene at the same position downstream of the gene VI promoter, as illustrated in Fig. 2. However, pH65 has considerably more FMV genomic sequences upstream and downstream of the CAT gene than pH32. To examine the effect of the upstream sequences, including the promoter for the full-length RNA, we introduced 5’ truncated deletions (Fig. 2) in pH65. The initial results indicated that a deletion of the full-length promoter, as in pH55 and pSXlO3, stimulated expression from the separate gene VI promoter when compared to the CAT expression by pH65. This observation was borne out in three additional comparisons. Similar observations were made with plasmids pH32 and pH95 in which all FMV gene VI sequences downstream of the CAT gene were deleted and termination of transcription was under control of the nopaline synthase poly(A) signal (NT) (Fig. 2). Again, a fivefold decrease in CAT activity, driven by the gene VI promoter, was observed when the full-length promoter was allowed to express from the same plasmid in pH95 versus pH32, as shown in Fig. 2. The results suggest that activity of the fulllength promoter, which results in production of the fulllength RNA, interferes with the expression from the separate gene VI promoter. Identification of gene VI &-acting sequences involved in transactivation by its cognate protein The results of Fig. 2, and similar comparisons (data not shown), indicated that sequences downstream of CAT, in gene VI, had a negative effect on expression of the reporter gene; CAT activity induced with pH55 and pSXlO3 was consistently lower than that observed for pH32 and derivatives. To rule out that the effect was
OF
FMV
GENE
EXPRESSION
405
due to the different poly(A) signals employed by pH55 and pSX103 versus pH32, pH32VI was constructed. Its CAT expression is controlled by the same transcription elements as in pH32 but gene VI sequences that are absent in pH32 are present in cis in pH32VI (Fig. 3). Comparison of expression levels showed that the presence of these cis-acting sequences reduced the level of CAT expression ca. 6- to IO-fold (Fig. 3, pH32 + pGS1 versus pH32VI + pGS1). The aforementioned experiments demonstrated that certain genomic elements inhibit CAT expression from pH65 (Figs. 2 and 3). However, pH25, which is a partially redundant plasmid of FMV similar to pH65, consistently showed five to six times greater levels of CAT expression (Fig. 3) even though the reporter gene was inserted at the identical position within gene VI (Scholthof eta/., 1991). The difference is that pH25 can express a functional gene VI product from an extra copy of this gene upstream from the full-length promoter, whereas pH65 cannot (Fig. 3). The downstream gene VI was inactivated in both plasmids because of the CAT gene insertion. The results suggested that the transactivating role of the gene VI protein in pH25 probably functions for the expression of its own coding region. We therefore examined if the negative effects of promoter interference and the inhibitory gene VI sequences described in previous paragraphs could be relieved by coelectroporation with a plasmid (pGS1 RVI) which would provide the gene VI protein in trans (Gowda et al., 1989). As shown in Fig. 3, all plasmids that had either one (pH55, pH32VI) or both (pH65) negative regulatory elements showed CAT gene expression levels similar to that of pH25, once gene VI was provided in trans by pGS1 RVI. In pH65 and pH25 the transactivation could arise from an enhanced expression of CAT from the full-length transcript. But the examples with related plasmids clearly showed that the expression from the separate gene VI transcript is also enhanced. In addition, the transactivation depends on cis-acting sequences within gene VI because deletion of these, as in pH32, abolished the response to the transactivator (Fig. 3). A &-acting element in gene VI is also required Pansactivation of an upstream gene
for
We next examined whether gene VI in vans (Gowda et al., 1989; Scholthof et al., 1992) and gene VII in cis (Gowda et al., 1991) were the only requirements for efficient expression of the downstream major genes on the FMV genome, such as gene IV, or if the cis-acting element inside gene VI also would be involved. For this purpose we compared the expression of a partially redundant plasmid pH63b to that of its 3’truncated deriv-
406
SCHOLTHOF
ET AL
A 6603
7743/O
2173/3410
5379
7743/o
pH65 6883
Acetylation A
(%) B
pH32:
98 78
pH95:
18
17
FIG. 2. Regulatory effect of the full-length promoter of FMV on expression from the downstream gene VI promoter. (A) Diagram of plasmids used in the transient expression assay in protoplasts. Numbers indicate the nucleotide (nt) position on the FMV genome (Fig. 1) which is presented on a linear scale; position 0 is identical to position 7743 on the circular genome. Note that nt 2173 is fused to nt 3410 to give the IV/V fusion gene of the deletion mutant of FMV (Scholthof eta/., 1991). Viral sequences are represented by the lightly shaded box, vector sequences are omitted. The black circles give the position of the promoter for the full-length RNA followed by the large intergenic region (L-IR). The open circles denote the promoter for gene VI. The position of the CAT gene is indicated. The small darkly shaded box represents the FMV polyadenylation signal, whereas the white box denotes the nopaline synthase gene polyadenylation signal (NT). Omitted from the diagrams are the poly(A) signal immediately downstream of the full-length promoter on the left-hand side and the full-length promoter immediately upstream from the poly(A) signal on the right-hand side. (B) Results of transient expression assays with N. edwardsoniisuspension cell protoplasts electroporated with DNA of the plasmids diagrammed in A. No quantitative CAT assay data are available for the three lanes on the left (50 pg of each plasmid). Numbers in the table on the right give the percentage [‘4C]chloramphenicol that is acetylated by CAT in each case. The values under A are from an experiment in which 50 fig of plasmids pH32 and pH95 were used for electroporation. To allow precise comparisons, equimolar amounts of each plasmid were added for the second experiment with 10 rg pH32 (6047 bp) and 16 pg pH95 (9579 bp), under the assumption that plasmid size does not affect the efficiency of electroporation. The CAT assay results of this experiment are given in the two lanes on the right and the quantitative values for the percentage acetylation are given under B in the table.
atives (Fig. 4). All these plasmids have the CAT gene in frame with the first 108 nucleotides of gene IV. When pH63b was electroporated into protoplasts, a high level of CAT expression was obtained presumably because expression of the intact downstream gene VI could serve to transactivate this expression (Fig. 4). When the promoter and the 5’ end of gene VI were deleted, as in pH63bAS, the CAT expression was practically abolished. However, coelectroporation of this plasmid with pGS1 RVI restored the CAT expression to a higher level (Fig. 4). A relatively low level of transactivation was observed for pH63bAN (Fig. 4) a derivative of pH63b with ORFs IV/V and VI deleted.
Coelectroporation of pH63bAN with pGS1 RVI activated the expression to a level only 25% of that of pH63bAS (Fig. 4). The lower activation efficiency was also observed for pH20 which has identical viral sequences as pH63bAN upstream of CAT but the downstream native FMV polyadenylation signal is replaced by the terminator of the nopaline synthase gene (NT) (Fig. 4). Plasmids similar to pH20, but either with a separate transcription unit for gene VI in cis, or with the CAT gene at different positions, all expressed CATvery poorly even when the gene VI product was provided in trans (data not shown). The results suggested that apart from gene VII in cis and the product of gene VI in
cis-vans
REGULATION
OF
FMV
GENE
407
EXPRESSION
pH55
%Ac:
23
4
2
21
3
27
(A) (9)
94 96
59 60
8 16
21 46
FIG. 3. Self regulation of FMV gene VI expression via transactivation on a c&-acting element. (A) Diagram of plasmrds used in the transient expression assay in protoplasts. Symbols and numbering are as for Fig. 2. (B) CAT assay of N. edwardsonii suspension cell protoplasts electroporated with DNA of the plasmids that are given above each lane. For comparison, equimolar amounts of plasmid DNA were used during electroporation: (left panel), 80 bg pH25 (18,600 bp); 50 pg pH65 (1 1,640 bp); 47 pg pH55 (10,900 bp); (right panel), 10 pg pH32 (6047 bp); 13 wg pH32VI (7751 bp). Coelectroporation was performed with 50 pg of either pGS1 or pGS1 RVI. The quantitative data of this experiment (A) are given under the lanes, whereas another set of data (B) are from an experiment in which 25 pg of pH32 and pH32VI was used.
vans, there was an additional element present in either the ORF IV/V or the ORF VI regions which was required for optimal expression of upstream genes. To further identify this tentative cis-acting element, a plasmid pH63bAV was constructed. This plasmid was similar to pH63bAN except that it retained gene VI sequences (Fig. 5) although the promoter and 5’ end of this gene were deleted. When pH63bAV was electroporated into protoplasts it expressed CAT very poorly but the expression could be transactivated by coelectroporation with pGS1 RVI (Fig. 5). Although the data suggest that the expression of pH63bAV + pGS1 RVI is lower than for pH63bAS + pGS1 RVI (Fig. 5) this difference was not consistent in repeated trials. Nonetheless, the results indicated that an essential element for optimal expression was present inside ORF VI, between positions 5379 and 6603 of the FMV genome, because pH63bAV responded effectively to transactivation, in contrast to pH63bAN (Figs. 4 and 5).
In order to distinguish whether this c&-acting element was active as RNA or DNA, plasmids pH93b+NT and pH93b-NT (Fig. 5) were constructed. In pH93b+NT the polyadenylation signal of the nopaline synthase gene (NT) was inserted upstream of gene VI in pH63bAV in an orientation which would result in premature termination. Consequently, no gene VI sequences would occur at the 3’ end of the viral transcript. In the control plasmid, pH93b-NT, the NT signal was inserted in the opposite orientation to allow transcriptional readthrough. In this case the equivalent of the full-length viral transcript would terminate at the FMV poly(A) signal, downstream of gene VI. Thus, gene VI sequences would be present at the 3’ end of the transcript. Both plasmids were coelectroporated with either pGS1 or pGS1 RVI into N. edwardsonii protoplasts. Transient expression analyses showed that pH93b-NT could be transactivated 96-l 00% (compared to pH63bAS), whereas transactivation of
408
SCHOLTHOF
ET AL.
A 1910
pH63b
m
2173
3410
4735
5379
v,!~::.,..;.:.a .,,.,,, :.d~~@~---&:,,!,,b-,R
6603
‘.:// fl J?.i,qJb,R V,,J
@$@!g - -p. .:, .. ..- _.,..:. :: ,,,: pH63bAS 1~. .,:::E:~..:.:..xs.. :; ;.p.
190,
403-412
(1992)
Regulation of Caulimovirus Gene Expression and the Involvement of c&-Acting Elements on Both Viral Transcripts HERMAN B. SCHOLTHOF,’ Department
of Plant
FANG C. WU, SIDDARAME
GOWDA,
Pathology,
Lexington,
Received
University April
of Kentucky,
15, 1992;
accepted
June
AND
Kentucky
ROBERT I. SHEPHERD* 40546-009
1
5, 1992
In a further analysis of gene regulation of figwort mosaic virus (FMV), a caulimovirus, we studied transient gene expression with modified viral genomes in Nicotiana edwardsonii cell suspension protoplasts. The results demonstrated that the presence of the promoter for the full-length RNA interferes with expression from the separate downstream promoter for gene VI. In addition, expression of gene VI was inhibited by cis-acting sequences within gene VI itself. Both inhibitory effects could be partially relieved by coelectroporation with a plasmid that produces gene VI protein, demonstrating that expression of gene VI is transactivated by its own product. Subsequent expression studies with partially redundant FMV plasmids containing a reporter gene in frame with gene IV showed that efficient transactivation of CAT expression relies on a cis-acting element inside the downstream gene VI. Insertions of a transcriptional terminator upstream of the &-acting element for premature termination of transcription showed that the c/s-acting region is not a DNA element but is active only as a feature of the RNA transcript. We conclude that the k-acting element, together with the transacting gene VI product, enhances expression of all major genes, including gene VI, o WIZ Academic PWSS, IIW. from the polycistronic mRNA and the separate mRNA for gene VI.
al., 1989), and a recent investigation with FMV showed that genes further downstream on the full-length transcript are equally well transactivated (Scholthof et al., 1992). The underlying mechanism of this process remains unclear but it might involve interactions that allow reinitiation by ribosomes for a coupled translation of genes on the polycistronic mRNAs (Futterer and Hohn, 1991; Scholthof et a/., 1992). A &-acting element inside gene VII and an actively expressed gene VI are the only requirements for transactivation of prematurely terminated transcripts containing the leader of FMV and one or two downstream genes (Gowda et a/., 1989, 1991). Since gene VI appears to be crucial for gene expression we have further analyzed its regulation and the molecular interaction of its product during transactivation of viral genes further downstream on the polycistronic transcript. Our results show that expression of gene VI from its own promoter can be down-regulated by the upstream full-length promoter. Moreover, we have found that the gene VI product is able to transactivate its own expression. Our results suggest that gene VI, like all other ORFs, might be translated efficiently from the full-length RNA as well as from its own mRNA. The response to transactivation of the chloramphenicol acetyltransferase (CAT) gene, inside gene IV of a partially redundant plasmid of FMV, depends on the presence of a second &-acting element, in this case located near the 3’ end in the coding
INTRODUCTION Figwort mosaic virus (FMV), a caulimovirus, has a circular double-stranded (ds) DNA genome of 7743 bp (Richins et a/., 1987). Its genome (Fig. 1) is organized much like that of other caulimoviruses such as cauliflower mosaic virus (CaMV) (Gardner et al., 1981) the type member of the group. Caulimoviruses have features in common with animal hepadnaviruses and.accordingly both viruses have been grouped as pararetroviruses (Temin, 1985). These viruses produce a genomic length transcript with direct repeats at each end of the molecule. Their DNA is replicated by reverse transcription of this full-length transcript which, in the case of caulimoviruses, may also function as a polycistronic mRNA for translation of the six closely spaced major genes (Futterer and Hohn, 1991; Scholthof et al., 1992). A separate promoter is responsible for generating a small transcript that spans gene VI (Scholthof ef a/., 1992), which makes it the only FMV gene that appears as a monocistronic mRNA. The product of gene VI was initially shown to activate expression of the 5’ major open reading frames (ORFs) on CaMV (Bonneville et a/., 1989) and FMV (Gowda et
’ Present address: California, Berkeley, ’ To whom reprint
Department of Plant Pathology, CA 94720. requests should be addressed.
University
of
403
0042-6822/92
$5.00
Copyright 0 1992 by Academic Press. Inc. All rights of reproduction in any form reserved.
SCHOLTHOF
FIG. 1. A schematic map of the FMV genome. The interwoven lines represent the circular viral genome of 7743 base pairs of double-stranded DNA. Selected restriction sites are indicated. Of the four interruptions one (a) occurs in the coding strand and three (p) occur in the noncoding strand. The peripheral arrows and Roman numerals indicate the open reading regions. The two intergenic regions, the smaller one between genes V and VI and the larger one between genes VI and VII, contain promoters for the smaller gene VI mRNA and the larger full-length transcript, respectively. These RNAs are represented by the thin lines of the inner circle.
region of gene VI. This k-acting element is active on the full-length RNA as well as on the smaller gene VI transcript.
MATERIAL Standard
AND METHODS
protocols
Standard molecular biology techniques were employed throughout the reported studies (for protocols, see Maniatis et al,, 1982). The procedure described by Gowda et a/. (1989) was used for electroporation and CAT assay of protoplasts from Nicotiana edwardsonii cell suspensions. Quantitative data analyses were performed according to Scholthof et al. (1991).
Plasmid constructs The numbers in this section denote the position of the respective sites on the genomic map of FMV (Richins eta/., 1987) (Fig. 1). However, many plasmids are derived from a naturally occurring deletion mutant devoid of 1237 nucleotides (nt) which encompass most of gene IV and the 5’ end of gene V, resulting in a gene
ET AL.
IV/V fusion (Scholthof et a/., 1991). Construction of pSX103, pH25, pH32, and pH65 (Figs. 2 and 3) has been described previously (Scholthof et al., 1991, 1992). pH55 is similar to pH65 (Fig. 3), except that the promoter for the full-length transcript is deleted (Scholthof, 1990). For construction of pH95, pH32 was cleaved with Sphl (nt 3630) and SnaBl (nt 4795) and the fragment was replaced with the Sphl (nt 6600) to SnaBl (nt 4795) fragment of pDFiPR, a partially redundant clone of the FMV deletion mutant (Scholthof eta/., 1991). pH95 is similar to pH65 except that in the former the polyadenylation signal of the nopaline synthase gene (NT) is positioned immediately downstream of the CAT gene (Fig. 2). For construction of pH32VI (Fig. 3) the WI-BarnHI fragment (nt 3632 and 7083, respectively) of pSXlO3 was inserted in the compatible window of p3’NT (Scholthof et a/., 1992). The resulting plasmid is similar to pH32 (Scholthof et al., 1992) (Fig. 3) except that in pH32VI gene VI sequences are present between the CAT gene and NT, whereas these are absent in pH32. Note that these gene VI sequences are not capable of expressing gene VI because its 5’ terminus is deleted and the downstream portion is not positioned in frame with the CAT gene. Plasmids pH63b and pH63bAS (Scholthof et al., 1992) are FMV derivatives with the CAT gene in frame with the ORF IV/V fusion, followed by either an intact or an inactivated gene VI, respectively (Fig. 4). For construction of pH20, the Nsil fragment (nt 6603-3428) of pH14 (Scholthof et a/., 1991), was inserted in the Pstl site of p3’NT (Scholthof et a/., 1992). The resulting plasmid contains the FMV promoter for the full-length transcript with the downstream large intergenic region and genes VII, I, II, III, and IV/CAT-fusion followed by NT (Fig. 4). pH63bAN (Fig. 5) was generated by introducing a deletion from the Nsil sites (nt 3428 to 6603) in pH63b, which covers gene IV/V and VI. This plasmid is similar to pH20 but it contains the FMV polyadenylation signal instead of NT. Plasmid pH63bAV (Fig. 5) contains the Nsil fragment (nt 6603-3428) of pKScCAT which was inserted into the compatible Pstl site of pJSX6 (Scholthof et a/., 1991). This plasmid contains an inactivated gene VI because the promoter and the 5’ end of this ORF are deleted. In pH93b+NT (Fig. 5) the NT polyadenylation signal was inserted as a Pvull fragment from p3’NT (Scholthof et al., 1992) into the Stul site (nt 5379) of pH63bAV. The orientation was such that transcription would stop immediately downstream of the CAT gene, upstream from gene VI. However, in pH93b-NT (Fig. 5) the NT polyadenylation signal was positioned in the opposite orientation to that in pH93b+NT so that transcription would continue up to the native FMV polyadenylation signal downstream of gene VI.
&-vans
REGULATION
The plasmid used for lransactivation (pGS1 RVI) (Gowda et al., 1989) has gene VI placed between the 35 S promoter of CaMV and the polyadenylation signal of the small subunit of rubisco. This plasmid was added to the electroporation mixture to test for geneVI-induced Iransactivation. A similar plasmid (pGSl), except devoid of gene VI, was used as a control in these experiments. RESULTS Interaction
of full-length
and gene VI promoters
Previously we reported that FMV, like CaMV, contains a promoter upstream of gene VI which is used for transcription of a monocistronic mRNA spanning gene VI (Scholthof et al., 1992). In that report we mentioned the striking difference in CAT gene expression levels from plasmids like pH32 in comparison to that of plasmid pH65, although both have the CAT gene at the same position downstream of the gene VI promoter, as illustrated in Fig. 2. However, pH65 has considerably more FMV genomic sequences upstream and downstream of the CAT gene than pH32. To examine the effect of the upstream sequences, including the promoter for the full-length RNA, we introduced 5’ truncated deletions (Fig. 2) in pH65. The initial results indicated that a deletion of the full-length promoter, as in pH55 and pSXlO3, stimulated expression from the separate gene VI promoter when compared to the CAT expression by pH65. This observation was borne out in three additional comparisons. Similar observations were made with plasmids pH32 and pH95 in which all FMV gene VI sequences downstream of the CAT gene were deleted and termination of transcription was under control of the nopaline synthase poly(A) signal (NT) (Fig. 2). Again, a fivefold decrease in CAT activity, driven by the gene VI promoter, was observed when the full-length promoter was allowed to express from the same plasmid in pH95 versus pH32, as shown in Fig. 2. The results suggest that activity of the fulllength promoter, which results in production of the fulllength RNA, interferes with the expression from the separate gene VI promoter. Identification of gene VI &-acting sequences involved in transactivation by its cognate protein The results of Fig. 2, and similar comparisons (data not shown), indicated that sequences downstream of CAT, in gene VI, had a negative effect on expression of the reporter gene; CAT activity induced with pH55 and pSXlO3 was consistently lower than that observed for pH32 and derivatives. To rule out that the effect was
OF
FMV
GENE
EXPRESSION
405
due to the different poly(A) signals employed by pH55 and pSX103 versus pH32, pH32VI was constructed. Its CAT expression is controlled by the same transcription elements as in pH32 but gene VI sequences that are absent in pH32 are present in cis in pH32VI (Fig. 3). Comparison of expression levels showed that the presence of these cis-acting sequences reduced the level of CAT expression ca. 6- to IO-fold (Fig. 3, pH32 + pGS1 versus pH32VI + pGS1). The aforementioned experiments demonstrated that certain genomic elements inhibit CAT expression from pH65 (Figs. 2 and 3). However, pH25, which is a partially redundant plasmid of FMV similar to pH65, consistently showed five to six times greater levels of CAT expression (Fig. 3) even though the reporter gene was inserted at the identical position within gene VI (Scholthof eta/., 1991). The difference is that pH25 can express a functional gene VI product from an extra copy of this gene upstream from the full-length promoter, whereas pH65 cannot (Fig. 3). The downstream gene VI was inactivated in both plasmids because of the CAT gene insertion. The results suggested that the transactivating role of the gene VI protein in pH25 probably functions for the expression of its own coding region. We therefore examined if the negative effects of promoter interference and the inhibitory gene VI sequences described in previous paragraphs could be relieved by coelectroporation with a plasmid (pGS1 RVI) which would provide the gene VI protein in trans (Gowda et al., 1989). As shown in Fig. 3, all plasmids that had either one (pH55, pH32VI) or both (pH65) negative regulatory elements showed CAT gene expression levels similar to that of pH25, once gene VI was provided in trans by pGS1 RVI. In pH65 and pH25 the transactivation could arise from an enhanced expression of CAT from the full-length transcript. But the examples with related plasmids clearly showed that the expression from the separate gene VI transcript is also enhanced. In addition, the transactivation depends on cis-acting sequences within gene VI because deletion of these, as in pH32, abolished the response to the transactivator (Fig. 3). A &-acting element in gene VI is also required Pansactivation of an upstream gene
for
We next examined whether gene VI in vans (Gowda et al., 1989; Scholthof et al., 1992) and gene VII in cis (Gowda et al., 1991) were the only requirements for efficient expression of the downstream major genes on the FMV genome, such as gene IV, or if the cis-acting element inside gene VI also would be involved. For this purpose we compared the expression of a partially redundant plasmid pH63b to that of its 3’truncated deriv-
406
SCHOLTHOF
ET AL
A 6603
7743/O
2173/3410
5379
7743/o
pH65 6883
Acetylation A
(%) B
pH32:
98 78
pH95:
18
17
FIG. 2. Regulatory effect of the full-length promoter of FMV on expression from the downstream gene VI promoter. (A) Diagram of plasmids used in the transient expression assay in protoplasts. Numbers indicate the nucleotide (nt) position on the FMV genome (Fig. 1) which is presented on a linear scale; position 0 is identical to position 7743 on the circular genome. Note that nt 2173 is fused to nt 3410 to give the IV/V fusion gene of the deletion mutant of FMV (Scholthof eta/., 1991). Viral sequences are represented by the lightly shaded box, vector sequences are omitted. The black circles give the position of the promoter for the full-length RNA followed by the large intergenic region (L-IR). The open circles denote the promoter for gene VI. The position of the CAT gene is indicated. The small darkly shaded box represents the FMV polyadenylation signal, whereas the white box denotes the nopaline synthase gene polyadenylation signal (NT). Omitted from the diagrams are the poly(A) signal immediately downstream of the full-length promoter on the left-hand side and the full-length promoter immediately upstream from the poly(A) signal on the right-hand side. (B) Results of transient expression assays with N. edwardsoniisuspension cell protoplasts electroporated with DNA of the plasmids diagrammed in A. No quantitative CAT assay data are available for the three lanes on the left (50 pg of each plasmid). Numbers in the table on the right give the percentage [‘4C]chloramphenicol that is acetylated by CAT in each case. The values under A are from an experiment in which 50 fig of plasmids pH32 and pH95 were used for electroporation. To allow precise comparisons, equimolar amounts of each plasmid were added for the second experiment with 10 rg pH32 (6047 bp) and 16 pg pH95 (9579 bp), under the assumption that plasmid size does not affect the efficiency of electroporation. The CAT assay results of this experiment are given in the two lanes on the right and the quantitative values for the percentage acetylation are given under B in the table.
atives (Fig. 4). All these plasmids have the CAT gene in frame with the first 108 nucleotides of gene IV. When pH63b was electroporated into protoplasts, a high level of CAT expression was obtained presumably because expression of the intact downstream gene VI could serve to transactivate this expression (Fig. 4). When the promoter and the 5’ end of gene VI were deleted, as in pH63bAS, the CAT expression was practically abolished. However, coelectroporation of this plasmid with pGS1 RVI restored the CAT expression to a higher level (Fig. 4). A relatively low level of transactivation was observed for pH63bAN (Fig. 4) a derivative of pH63b with ORFs IV/V and VI deleted.
Coelectroporation of pH63bAN with pGS1 RVI activated the expression to a level only 25% of that of pH63bAS (Fig. 4). The lower activation efficiency was also observed for pH20 which has identical viral sequences as pH63bAN upstream of CAT but the downstream native FMV polyadenylation signal is replaced by the terminator of the nopaline synthase gene (NT) (Fig. 4). Plasmids similar to pH20, but either with a separate transcription unit for gene VI in cis, or with the CAT gene at different positions, all expressed CATvery poorly even when the gene VI product was provided in trans (data not shown). The results suggested that apart from gene VII in cis and the product of gene VI in
cis-vans
REGULATION
OF
FMV
GENE
407
EXPRESSION
pH55
%Ac:
23
4
2
21
3
27
(A) (9)
94 96
59 60
8 16
21 46
FIG. 3. Self regulation of FMV gene VI expression via transactivation on a c&-acting element. (A) Diagram of plasmrds used in the transient expression assay in protoplasts. Symbols and numbering are as for Fig. 2. (B) CAT assay of N. edwardsonii suspension cell protoplasts electroporated with DNA of the plasmids that are given above each lane. For comparison, equimolar amounts of plasmid DNA were used during electroporation: (left panel), 80 bg pH25 (18,600 bp); 50 pg pH65 (1 1,640 bp); 47 pg pH55 (10,900 bp); (right panel), 10 pg pH32 (6047 bp); 13 wg pH32VI (7751 bp). Coelectroporation was performed with 50 pg of either pGS1 or pGS1 RVI. The quantitative data of this experiment (A) are given under the lanes, whereas another set of data (B) are from an experiment in which 25 pg of pH32 and pH32VI was used.
vans, there was an additional element present in either the ORF IV/V or the ORF VI regions which was required for optimal expression of upstream genes. To further identify this tentative cis-acting element, a plasmid pH63bAV was constructed. This plasmid was similar to pH63bAN except that it retained gene VI sequences (Fig. 5) although the promoter and 5’ end of this gene were deleted. When pH63bAV was electroporated into protoplasts it expressed CAT very poorly but the expression could be transactivated by coelectroporation with pGS1 RVI (Fig. 5). Although the data suggest that the expression of pH63bAV + pGS1 RVI is lower than for pH63bAS + pGS1 RVI (Fig. 5) this difference was not consistent in repeated trials. Nonetheless, the results indicated that an essential element for optimal expression was present inside ORF VI, between positions 5379 and 6603 of the FMV genome, because pH63bAV responded effectively to transactivation, in contrast to pH63bAN (Figs. 4 and 5).
In order to distinguish whether this c&-acting element was active as RNA or DNA, plasmids pH93b+NT and pH93b-NT (Fig. 5) were constructed. In pH93b+NT the polyadenylation signal of the nopaline synthase gene (NT) was inserted upstream of gene VI in pH63bAV in an orientation which would result in premature termination. Consequently, no gene VI sequences would occur at the 3’ end of the viral transcript. In the control plasmid, pH93b-NT, the NT signal was inserted in the opposite orientation to allow transcriptional readthrough. In this case the equivalent of the full-length viral transcript would terminate at the FMV poly(A) signal, downstream of gene VI. Thus, gene VI sequences would be present at the 3’ end of the transcript. Both plasmids were coelectroporated with either pGS1 or pGS1 RVI into N. edwardsonii protoplasts. Transient expression analyses showed that pH93b-NT could be transactivated 96-l 00% (compared to pH63bAS), whereas transactivation of
408
SCHOLTHOF
ET AL.
A 1910
pH63b
m
2173
3410
4735
5379
v,!~::.,..;.:.a .,,.,,, :.d~~@~---&:,,!,,b-,R
6603
‘.:// fl J?.i,qJb,R V,,J
@$@!g - -p. .:, .. ..- _.,..:. :: ,,,: pH63bAS 1~. .,:::E:~..:.:..xs.. :; ;.p.
