VIROLOGY

187, 555-564

Inhibition

(1992)

of African Cassava Mosaic Virus Systemic Infection by a Movement from the Related Geminivirus Tomato Golden Mosaic Virus

Protein

ALBRECHT VON ARNIM AND JOHN STANLEY’ John lnnes Institute, John lnnes Centre for Plant Science Research, Received September

24, 199 1; accepted

Colney Lane, Norwich November

NR4 WH, United Kingdom

25, 199 1

Plant viruses encode proteins that mediate their movement through the host plant leading to the establishment of a systemic infection. We have analyzed the effect of tomato golden mosaic virus (TGMV) genes BLi and BRl, which are thought to be involved in the process of virus movement, on the infectivity of African cassava mosaic virus (ACMV) in Nicotiana benthamiana. Recombinant genomes were constructed by replacing the ACMV coat protein coding sequence with those of either BLl or BRl. Replication of recombinants containing BLI and BRl coding sequences in the sense orientation with respect to the coat protein promoter was detected in the inoculated leaves only when the constructs were co-inoculated, suggesting that both genes are being expressed and act in a cooperative manner. Co-inoculated recombinants induced localized symptoms on inoculated leaves but did not spread systemically, either because of a defect in BLl and/or BRl expression or due to the inability of the TGMV gene products to functionally complement their ACMV counterparts. Systemic spread of ACMV was inhibited when the recombinant containing the BLl coding sequence in the sense, but not in the antisense, orientation was co-inoculated with ACMV DNA B. Disruption of the BLl coding sequence by a frameshift mutation restored the ability of the recombinant to spread systemically, suggesting that the gene product is responsible for the inhibitory effect. The inhibitory phenotype was mimicked by a chimera containing amino-terminal sequences of TGMV BLl and carboxy-terminal sequences of its ACMV homologue, BCl. The chimera has characteristics of a dominant negative mutant. We suggest that dominant negative 0 1992 Academic press, I~C. I’IIUtSIItS Of VirUS ITIOVeITEnt genes may provide a IJOWl source for VirUS ESiStanCe genes.

INTRODUCTION

ever, because of the need to co-inoculate whole viruses, assignment of the complementing phenotype to a particular gene of the helper virus was usually not possible. In addition, the degree to which the movement function of the rescued virus was impeded often remained uncertain. All of the reports to date describe positive complementation. A negative effect of one virus movement gene on another virus has not been described, probably due to the fact that the target virus can readily escape any inhibitory effect during co-inoculation experiments involving whole plants. Negative effects of one virus on the ability of another to coinfect or superinfect have been described and are referred to as cross-protection, but have not been traced back to negative interaction between movement functions. We have chosen to investigate the compatibility of movement genes using members of the geminivirus group of plant viruses. There has been only a single report describing complementation of a presumptive geminivirus movement function; ultrastructural observations on the cytopathological effects of co-infection revealed that the tissue tropism of the geminivirus bean golden mosaic virus (BGMV) was modified by the tobamovirus tobacco mosaic virus (TMV) (Carr and Kim, 1983).

Plant viruses encode proteins responsible for the spread of infection between cells and possibly for long distance spread through the vascular system of the host (reviewed by Hull, 1989). Genes dispensable for replication but otherwise necessary for infection may be functionally classified as movement genes. Such genes play a major role in viral pathogenesis due to their influence on host range (Taliansky et al. 1982a) and symptom development (Ziegler-Graff et al., 1991; von Arnim and Stanley, 1992). Some plant resistance genes target virus movement in a manner reminiscent of gene-for-gene interactions (Meshi et al., 1989). Numerous reports of classical complementation experiments suggest that virus movement genes are, to a certain degree, interchangeable. Avirulence of a virus in a particular host was sometimes overcome by co-infection with an often unrelated virulent helper virus (Taliansky eta/., 1982a), and viruses with a defective, mutationally impaired movement function could be rescued by co-infection with certain other movement-competent viruses (Taliansky eta/., 1982b). How’ To whom reprint requests dressed.

and correspondence

should be ad-

555

0042-6822192

$3.00

Copyright 0 1992 by Academic Press. Inc. All rights of reproduction in any form reserved.

556

VON ARNIM SP

277911

Bg 200

N 2000

Bg

FIG. 1. Genome map of African cassava mosaicvirus. The genome of TGMV is organized on a similar basis (Hamilton et al., 1984). ACMV coding regions are referred to as complementary-sense (C) and virion-sense (V) orientation. The equivalent coding regions in TGMV are referred to as leftward (L) and rightward (R) orientation. The shaded areas symbolize the common region, a stretch of high sequence similarity between the genomic components. Numbers indicate nucleotide positions and the sizes of the genome components B, BarnHI; Bg, Bglll; N, Ncol; Sp, Sphl; Xb, Xbal.

Geminiviruses are small DNA viruses with circular single-stranded genomes which replicate in the nucleus (reviewed in Davies and Stanley, 1989). In addition to the problem of movement between neighboring cells, geminiviruses therefore also face the challenge of movement into and out of the nucleus. The genomes of the geminiviruses African cassava mosaic virus (ACMV; Fig. 1) and tomato golden mosaic virus (TGMV) each consist of two DNA circles of ca. 2.8 and 2.6 kb, respectively (Stanley, 1983; Stanley and Gay, 1983; Hamilton et a/., 1984). Replication and mutagenesis studies have established that ACMV and TGMV contain two bona fide movement genes on DNA B (ACMV BCl and BVl, TGMV BLl and BRl) and one on DNA A (ACMV AC2 and TGMV AL2) (Rogers et al., 1986; Townsend et a/., 1986; Brough et al., 1988; Elmer et al., 1988; Etessami et a/., 1988, 1991). Both genes located on DNA B are essential for wild-type infection of Nicotiana benthamiana but have no direct role in viral DNA replication (Sunter eta/., 1990) or virus assembly (Sunter et al., 1987; Klinkenberg and Stanley, 1990). We have previously demonstrated that gene BLl determines symptom characteristics of phenotypically distinct strains of TGMV in N. benthamiana (von Arnim and Stanley, 1992), reflecting its role in virus movement. Because the DNA B gene products of ACMV and TGMV show some degree of homology (51% conserved homology between BVl and BRl , 70% between BCl and BLl ; Hamilton eta/., 1984), the present study was initiated to determine whether the gene products of one virus can functionally complement those of the other and, if so, what effect they might have on symptom expression. Viable pseudorecombinants have been produced by the reassortment of cloned genomic components of closely related

AND STANLEY

strains of ACMV (Stanley et a/,, 1985) and TGMV (von Arnim and Stanley, 1992). Unfortunately, pseudorecombinants produced by mixing ACMV and TGMV components are not viable, precluding their use in this investigation. Unlike the majority of plant viruses, coat protein mutants of ACMV are unimpaired in their ability to systemically infect plants (Stanley and Townsend, 1986; Etessami et a/., 1989) and coat protein replacement vectors have been used to express foreign genes in plants. ACMV DNA A-based replicons allow gene expression in the sense, but not in the antisense, orientation with respect to the coat protein promoter (Ward et a/., 1988), and the functional expression of the two ACMV DNA B proteins from such replicons has been demonstrated (Etessami et al., 1988). In this report we have used similar constructs to analyze the effect of TGMV DNA B proteins on ACMV infection in their common host N. benthamiana. MATERIALS Maintenance

AND METHODS

and inoculation

of plants

N. benthamiana plants were maintained and mechanically inoculated with cloned DNA as described (Stanley, 1983; Stanley et al., 1990). To avoid having to excise clone inserts prior to inoculation, all viral DNAs were introduced into plants as partial or tandem repeats. Plasmids harboring partial repeats of genomic components (described below) were purified over CsCl gradients and 1 pg of each plasmid per plant was inoculated onto celite-dusted expanding leaves. Viruses were held and manipulated under MAFF License Numbers PHF 49A/85 and PHF 49A/41 under the Plant Pests (Great Britain) Order 1980. Clone construction techniques

and recombinant

DNA

All recombinant DNA techniques were essentially as described (Sambrook et al,, 1989). Enzymes were supplied by GIBCO-BRL, Pharmacia (Klenow fragment of DNA polymerase) and Northumbria Biologicals Ltd. (Taq polymerase). For the construction of a partial repeat of the coat protein replacement vector pJS172 (Etessami et al., 1988), the 1-kbp Bg/ll (206 1)-BamHI (29 1) fragment of ACMV DNA A from pJS092 (Stanley, 1983) was cloned into BarnHI-cut plC19H (Marsh et a/., 1984). The unique BamHl site of the resulting clone pCLV0.4A was filled in with T4 DNA polymerase, and the coat protein replacement vector from pJSl72 was inserted as a 1.8-kbp Sphl fragment into the unique Sphl site.

INHIBITION 1 AC1

I

OF ACMV

g(p= AC2 m

BLl+ BLI BLI/BCl+ BLI/BClBCl/BLl+ BCI/BLlBLla

AP

BRI+ BRlBVl + BCI +

II I

FIG. 2. Maps of ACMV DNA A recombinants based on a partial repeat of the coat protein replacement vector plC172. Inserted coding regions, their orientation, and the names of the resulting clones are indicated. TGMV coding regions are indicated by heavy (BLl) and intermediate (BRl) shading, ACMV coding regions by light (BVl) and no (BCl) shading. Note that the coordinates of DNA A repeat sequences of clone BCl+ differ from those of plC172. The shaded boxes symbolize the common region. B, BarnHI; X, Xhol. The map locations of two oligonucleotides used in this study are indicated by small arrows. Solid and open triangles indicate the positions of virion-sense consensus CAAT and TATA boxes, respectively.

The Xhol site within the ACMV sequence was made unique by in-fill of an Ssrl site in the polylinker of plC19H, which destroyed a neighboring Xhol site, resulting in clone plC172 (Fig. 2). The BRl coding region of the common strain of TGMV was isolated from csTBBg (von Arnim and Stanley, 1992) by the polymerase chain reaction (PCR) with Xhol-containing oligonucleotides 5’-CATATTTACTCGAGCAGGAT-3’ and 5’%AATTGGCTCGAGTATGACA-3’ (map positions 428-447 and 1276-1257, respectively, using the nucleotide numbering of Hamilton et al., 1984). The product was digested with Xhol and the resulting 828-bp fragment was inserted into plC172 in the sense and antisense orientations to give clones

INFECTION

557

BRl+ and BRl-, respectively. Similarly, the BLl coding region was isolated from csTB (von Arnim and Stanley, 1992) with the Xhol-containing oligonucleotides 5’-TGTCATACTCGAGCCAATTT-3’ and 5’-ACGCTGCCTCTCGAGGCCAG-3’ (map positions 12571276 and 2225-2206, respectively) and then digested with Xhol and the resulting 946-bp fragment was inserted into plC172 to give clones BLl+ and BLl-. Two clones each of BLl+ and BLl- were analyzed further. Mutant BLl +AP was constructed by partial digestion of BLl + with Pstl, deletion of 4 bp with T4 DNA polymerase, and religation. Deletion of the Pstl site disrupts the BLl coding region at codon position 33. The deletion was confirmed by sequence analysis after subcloning the relevant fragment into Ml 3. To construct chimeric ACMV/TGMV coding regions, we exploited a conserved lvcol site within the coding regions of ACMV BCl and TGMV BLl, encoding the dipeptide proline-tryptophan at codon positions 1 16 (ACMV) and 1 14 (TGMV). The BLl coding region was subcloned as a Xhol fragment into Bluescript (Stratagene) after PCR amplification from csTB to give clone BLIBS, and the amino-terminal fragment was excised as an SstllNcoI fragment (using the Sstl site in the Bluescript polylinker) and used to replace the analogous fragment of the BCl coding region in clone pCLV039 (Etessami et a/., 1988). The resulting BLl/ BCl chimeric coding region was inserted in both orientations into plC172 as a Xhol fragment to give clones BLl/BCl+ and BLl/BCl-. The reciprocal chimera BCl/BLl was created by replacement of the aminoterminal SstllNcol fragment of the BLl coding region in clone BLIBS with that of the BCl coding region from pCLV039, and the chimeric coding region was inserted in both orientations into plC172 as a Xhol fragment to give clones BCl/BLl+ and BCl/BLl-. The construction of recombinants pCLVO43 and pCLVO52, containing the coding regions of ACMV BCl and BVl , respectively, inserted in the sense orientation into derivatives of the coat protein replacement vector pJSl72, has been described (Etessami et al., 1988). A partial repeat of the insert of pCLVO52 was constructed by cloning the full-length fllcol insert into pCLVO.4A to give clone BV1-t. Clone BCl+, a partial repeat of the insert of pCLVO43, was assembled in Bluescript from a 0.7-kbp C/al (2403)lHindlII fragment (the /+fldlll site is located within the BCl coding region) and the full-length C/al insert of pCLVO43. The construction of pCLVl.3A and pCLV2B, containing a partial repeat of ACMV DNA A in Bluescript and a tandem repeat of DNA B in pATl53, respectively, has been described (Klinkenberg eta/., 1989; Etessami et al., 1991).

VON ARNIM

558

Isolation

and characterization

of nucleic acids

Total nucleic acids were isolated from leaf tissue as described by Stanley (1983) and quantified by uv spectrophotometry. Southern blot analysis was performed after electrophoresis of nucleic acid samples in 1% agarose gels with TBE (45 mM Tris-borate, 1 mM EDTA, pH 8.3) or TNE (40 mM Tris-acetate, 20 mM sodium acetate, 2 mM EDTA, pH 7.5) buffer. DNA blotted onto GeneScreen Plus (NEN) was detected by hybridization with radiolabeled probes prepared by random priming (Feinberg and Vogelstein, 1983). Probes were specific to ACMV DNAA (A&l (734)~Sphl (2581)) and DNA B (Pstl (245)-EcoRV (2550)), the coding regions of ACMV BCl (HindIll (1494)-HindIll (2281)) and BVl (X&l (608)-Sphl (1303)) and the coding regions of TGMV BLl and BCl (gel-purified PCR products, described above). A 1.8-kbp fragment of ACMV DNA A was amplified from total nucleic acid extracts using the oligonucleotides 5’-TTTGGGTCTAGAACACTTTG-3’ and 5’AGACAAGTCTAGAAGCGAGC-3’, which correspond to nucleotide positions 2685-2704 and 171 l-l 692, respectively, on ACMV DNAA(Fig. 2). The oligonucleotides contain Xbal sites to facilitate subcloning of PCR products. Sequences were determined by Sequenase dideoxy-sequencing (U.S. Biochemical Corp.) on singlestranded templates, using universal primer or ACMV DNA A-specific primers, after subcloning appropriate restriction fragments into M 13. RESULTS Phenotypes of ACMV chimeras DNA B coding regions

containing

TGMV

Introduction of the coding region of TGMV BRl, the homologue of ACMV BVl , into ACMV DNA A in either orientation (clones BRl+ and BRl-) had no discernable effect on the infectivity of ACMV when co-inoculated with DNA B (Table 1). Also, the clone combinations of BCl+ with either DNA B or BVl+ are highly infectious (Etessami et a/., 1988; Table 1). However, the clone combinations BCl+ with BRl+ and BLl+ with BVl+ failed to infect as judged by the lack of symptoms, suggesting that BRl and BLl do not complement their ACMV counterparts. Surprisingly, the ability of BLl + to systemically infect plants when co-inoculated with DNA B was greatly reduced in comparison with that of BLl-. The statistical significance of this difference was determined in a t test (99.9% confidence level). Two duplicate clones of BLl+ and BLltested exhibited the same phenotypes (data pooled in Table l), suggesting that the inhibitory phenotype was not an artifact introduced during PCR amplification of the BLl coding region. Because the phenotype is orien-

AND STANLEY

tation dependent, the results suggest that BLl has an inhibitory effect on ACMV infection when expressed from the coat protein promoter. BLl+ infection movement

is blocked at the level of virus

Interference of ACMV infection by TGMV sequences could occur by disrupting either viral DNA replication or the ability of the virus to move through the plant. Inspection of the inoculated leaves showed that chlorotic lesions, typical of ACMV infection, regularly developed between 4 and 9 days after co-inoculation of DNA B with any of the DNA A recombinants described in this study, even when systemic infection did not subsequently develop. Single- and double-stranded viral DNA forms of both genomic components were detected in these localized infections, indicating that the DNA A recombinants are replication-competent and replicate DNA B in trans. The presence of BLl+ and BLl- progeny in the inoculated leaves was confirmed by restriction analysis. For both constructs, progeny are linearized at the unique EcoRV site located within the BLl coding sequence (Fig. 3). Progeny also contain three Bglll sites, one of which is located asymmetrically within the BLl coding sequence. Two of the three Bglll fragments hybridize to the DNA A-specific probe in Fig. 3. This assay system, however, is not suitable for a quantitative analysis of replicative ability because symptoms and viral DNAs could only be observed when DNA A recombinants were co-inoculated with DNA B which amplifies the viral DNAs by providing functions for virus movement. This is also the case for wild-type DNA A of ACMV and TGMV which could not be detected in inoculated leaves unless co-inoculated with its cognate DNA B (see, for example, Fig. 6). The level of single-stranded DNA was highly variable in all infections analyzed, presumably because of the absence of coat protein which, in a wild-type infection, sequesters and protects this DNA species. However, throughout our experiments we did not observe any consistent influence of orientation of the TGMV DNA inserts on the relative levels of ACMV genomic components or their DNA forms. We conclude that the TGMV BLl sequence affects systemic spread of the recombinant virus, although the latter retains the ability to replicate and induce pathogenic effects in the inoculated leaf. BLi+ systemic infection is correlated with recombinational rescue Systemic infection following co-inoculation of BLl+ with DNA B is not completely abolished but greatly reduced (Table 1). Systemic symptoms induced in N. benthamiana were tentatively classified as being either

INHIBITION

OF ACMV

559

INFECTION

TABLE 1 INFECTIVITYOFACMV COATPROTEIN REPLACEMENTCONSTRUCTS Analysis of progeny Systemic infectivityb

Constructsa BRl+ BRlBCl+ BCl+ BCl+ BLl+ BLl+

DNA DNA DNA BVlf BRl+ BVl+ DNA

B B B

BLl-

DNA B

44/45 [6]

BLl +AP BLl/BCl+

DNA B DNA B

19/20 [2] 2130 [3]

BLlIBCl-

DNA B

12/25 [3]

BCl/BLl t BCl/BLlDNA A

DNA 8 DNA B DNA B

20/20 [3] 1O/l 0 [2] 25/25 [3]

B

16/17 14/l 5 22/25 49/54 O/20 o/20 1 l/70

[3] [3] [3] [9] [2] [3] [lo]

Symptom type

Number of plants

Number Analyzed

ori’

recc

1 2 1 0

1 2 1

0 0 0

6 3 6

0 1 6

10

7

6 2 0 3

Severe Severe Severe Severe Severe Mild Severe Mild Severe Severe Mild Severe Mild Severe Severe Severe

6 5 44 0 0 2 7 5

a DNA A and DNA B were supplied by clones pCLV1.3A and pCLV2B, respectively. b Number of plants infected/inoculated [number of experiments]. c Progeny virus was analyzed as described in the text and classified as either indistinguishable nant produced in viva (ret).

mild or severe. Severe symptoms include stunting of the plant, widespread chlorosis, and severe leaf curling typical of ACMV infection. In contrast, chlorosis in mild infections was confined to one or a few spots on systemically infected leaves with no stunting and hardly any leaf deformation. While inoculation with BLl+ often led to mild systemic infections, infections with BLl- were exclusively of the severe type. We analyzed viral nucleic acids isolated from systemically infected leaves by restriction digestion in order to determine whether the DNA A-based replicons had retained their original structure, as found in the inoculated leaves. The analysis was initially done by restriction digestion and Southern blot analysis of total nucleic acids isolated from systemically infected leaves, using probes specific for ACMV DNA A or the TGMV gene inserts. Subsequently, viral DNA was synthesized by PCR amplification with a pair of primers designed to amplify part of the ACMV DNA A replicons encompassing the foreign DNA inserts. As AC1 is the only virus-encoded gene product essential for replication, we chose primers within this coding sequence for our analysis (Fig. 2). Nucleic acid preparations from plants inoculated with DNA A alone did not give rise to PCR products (data not shown). Amplified fragments were mapped by restriction digestion and gel electro-

2 1 0 2 1 3

from the original construct

0 1 2 1 3

2 0 0 0 0

(ori) or a recombi-

phoresis (Fig. 4). Several samples analyzed by Southern blot analysis and PCR amplification gave similar results, summarized in Table 1. No major modifications were detected in the progeny of BLl- isolated from systemically infected tissues. In contrast, we were unable to detect viral DNA with the original structure in any of the plants systemically infected with BLl + and showing severe symptoms. In each case, recombination events had modified sequences within the aminoterminus of TGMV BLl (Fig. 4A). Recombination also occurred in mild infections with BLl+ although, on one occasion, progeny with a restriction pattern indistinguishable from the original genotype was detected. In this instance it is possible that a minor modification might have escaped detection. The results indicate that disruption of the TGMV sequence in BLl + is necessary for induction of severe symptoms and is probably prerequisite for the efficient spread of the virus. The nature of the recombination event was investigated by sequence analysis of PCR-amplified viral DNA. Sequences encompassing the 5’Xhol site and extending into the BLl coding region were found to be replaced by Escherichia co/i /acZ sequences. The 3’ endpoint of the replacement was located near the middle of the BLl coding region at codon position 126. In several other cases, restriction sites indicative of a simi-

560

VON ARNIM

A

B

BLi+ II u Ev Bg

A

BLIu Ev

BLl+

Bg

-

1633 1326

-

651

BLl-

AND STANLEY

fectivity of ACMV. However, the effect may be a consequence of either BLl expression or the presence of disruptive k-acting TGMV sequences, although the orientation-specific nature of the inhibitory effect suggests the former. To resolve this point, the BLl coding sequences were disrupted by the introduction of a 4bp deletion in clone BLl +AP. BLl +AP was highly infectious when co-inoculated with DNA B and routinely induced severe symptoms (Table 1). The deletion mutant frequently retained its structure during systemic movement, as judged by restriction analysis. To confirm this, the entire TGMV DNA insert was sequenced after subcloning PCR-amplified BLl +AP progeny. Apart from the 4-bp deletion introduced in vitro, the sequence was identical to the original sequence determined for the parental TGMV DNA B clone csTB (von Arnim and Stanley, 1992). The results indicate that expression of BLl is responsible for the suppression of ACMV movement. Construction

of a chimeric

dominant

inhibitor

gene

In an attempt to define the TGMV BLl domain responsible for the suppression of systemic movement, FIG. 3. Southern blot analysis of nucleic acids isolated from inoculated leaves 5 days after co-inoculation of ACMV DNA A (lane A), BLl f, or BLI - with ACMV DNA B. Samples were either undigested (u) or digested with EcoRV (Ev) or Bglll (Bg); all samples in B were undigested. Blots were hybridized with probes specific for DNA A (A) and DNA B (B). The positions of linear (lin), supercoiled (SC), singlestranded (ss), and open-circular (oc) viral DNAs and size markers (bps) are indicated. Note that different samples were used in A and B and the relative positions of the viral DNA forms are different in TNE (A) and TBE (B) electrophoresis buffers.

lar replacement involving /acZ sequences were found. The data are consistent with excision of replicons from the inoculum by nonhomologous recombination in parallel with the more efficient recombination events involving homologous sequences (Fig. 5). Products of nonhomologous recombination can retain the common region and complementary-sense genes required for replication, together with a part of the /acZ gene located upstream of the carboxy-terminal coding sequences of BLl Residual /acZ and BLl sequences incorporated into the ACMV DNA serve to maintain the size of the replicon, a prerequisite for systemic spread (Etessami et a/., 1989). Such products, containing a disrupted BLl coding region, are selected in plants due to their ability to move systemically. TGMV BLI acts as an inhibitor movement

of ACMV systemic

Our results indicate that the inserted TGMV BLl coding sequences impose a negative effect on the in-

Xbal

Xhol

EcoRV I

L+‘+ ,I

CM

868

AP 390

Xhol

Xbal

1

I

997 951

524

FIG. 4. Restriction analysis of the progeny of ACMV DNA A chimeras following PCR fragment amplification from total nucleic acids isolated from systemically infected leaves. (A) Progeny of BLl+ (+) and BLl- (-) digested with Xhol. Markers (M) were produced by PCR amplification of inoculum BLl+. (B and C) Progeny of BLl +AP (AP) digested with Xhol (B) and double-digested with Xbal and fcoRV (C). Marker(M) was produced by PCR amplification of inoculum BLl.+AP. The structures of the nonrecombined marker fragments are indicated on the right, with the common region shown as a shaded box. The sizes of restriction fragments (in basepairs) are indicated.

INHIBITION

homologous recombination

A t

BLl + BX V -I

OF ACMV

XB

$g,:

non-homologous recombination

B 1

561

INFECTION

the chimera BLl/BCl. Assuming that BLl can functionally replace BCl, removal of this potential target from the ACMV genome might restore to the virus the ability to systemically infect plants. However, co-inoculation of BCl+ with BRl+, BLl+ with BVl+ (Table l), and BLl/BCl+ with BVl+ (five plants) showed that this condition is not met. No accumulation of viral DNA in either the inoculated or the upper leaves was detected by Southern blot analysis (for example, see Figs. 6B, 6C) for the combinations of BCl+ with BRl+ (four samples), BLl+ and BVl+ (nine samples over three experiments), and BLl +/BCl+ and BVl+ (three samples). We believe that the lack of complementation is not entirely due to some deficiency in the expression of the TGMV genes from ACMV DNA A because co-inoculation of BLl+ and BRl+ consistently produced both single- and double-stranded viral DNA forms in

A

A N

B u

B

A B N

Cl c1 - Ll c1 - Rl Vl Vl VI I I - u u u u Xb

FIG. 5. Models for recombinational release of virus replicons from BLl + inoculum DNA. (A) Homologous recombination to produce an intact copy of the chimeric geminivirus DNA containing TGMV sequences inhibitory to virus spread. Sequences within the common region (shaded boxes) might play an important role in homologous recombination (Etessami et al., 1989). (B) Nonhomologous recombination involving sequences in TGMV BLl and f. co/i /acZ genes, producing a movement-competent replicon. B, BarnHI; X, Xhol.

- lin - SC - ss I

we constructed a pair of chimeras from ACMV BCl and TGMV BLI coding regions (Fig. 2). Co-inoculation of BCl/BLl+ or BCl/BLlwith DNA B produced severe symptoms, and both chimeras were judged to be stable by restriction analysis (Table 1). BCl/BLl+ still depended on the presence of ACMV BCl for systemic spread. However, the phenotype of BLl+ was mimicked by the construct BLl/BCl+ containing 116 amino-terminal codons from TGMV and 184 carboxyterminal codons from ACMV. Because inoculations were performed in the presence of a wild-type ACMV BCl gene on DNA B, BLl/BCl resembles a dominant negative mutant of the natural movement protein. BLl/ BCl+ progeny retained their structure in inoculated leaves (data not shown) but induced chlorotic lesions less efficiently than BLl+. No systemic spread of BLl/ BCl + progeny was observed without prior recombination. BLi+ and BRl+ produce inoculated leaves

a localized

infection

in

We speculated that ACMV BCl is a possible target for inhibition of systemic movement by TGMV BLl and

I A

1 B

C Rl

-(I

I A

Cl

probe

Ll Rl

L-z-T&

Cl Ll Rl -

Vl

Ll Rl

Ll AP Vl RI

UlulJ

RI

probe

u

Ll

FIG. 6. Southern blot analysis of viral DNA forms in extracts of inoculated leaves 10 days after inoculation. Inoculated DNAs and hybridization probes are indicated above and below the blots, respectively. Inocula: A, ACMV DNA A; B, ACMV DNA B; Cl, BCl f; Ll, BLl f; Rl, BRl+; Vl , BVl+; AP, BLl + AP. Extracts were either undigested (u) or digested with EcoRV (Ev), /Vcol (N) (both of which linearize the respective viral DNAs), Xhol (X), Xbal (Xb), and Bglll (Bg). The positions of linear (lin), supercoiled (SC), and single-stranded (ss) viral DNAs and size markers (in basepairs) are indicated.

562

VON ARNIM AND STANLEY

localized infections on inoculated leaves (in all eight samples analyzed over four experiments; for example, see Fig. 6C). The presence of BLl+ and BRl+ progeny was confirmed by restriction analysis. In some instances, these localized infections were associated with the production of chlorotic lesions on the inoculated leaves, although systemic spread from these primary points of infection was not detected. Unfortunately, symptoms induced by ACMV and TGMV on inoculated leaves are indistinguishable and so whether TGMV DNA B genes can impose TGMV-type symptoms on what is essentially an ACMV infection remains undetermined. The data support the view that BLl and BRl are being expressed from the recombinant virus and act in a cooperative manner to form an infection in inoculated leaves, although they cannot fully functionally replace their ACMV homologues. DISCUSSION DNA B coding regions from the common strain of TGMV (von Arnim and Stanley, 1992) have been inserted into an ACMV coat protein replacement vector in both orientations with respect to the coat protein promoter. In the sense orientation (clones BRl+ and BLl+), neither coding region could complement its ACMV homologue (clones BVl+ and BCl+) in co-infection experiments. At present, we cannot rule out the possibility of impaired expression of the TGMV proteins from the ACMVvector. However, we feel that this is unlikely because similar transcription fusion constructs have been successfully employed to express ACMV DNA B proteins in this manner (Etessami el a/., 1988) and the inserted fragments do not drastically alter the size of the genomic component (BRl+ and BLl+ replicons differ from ACMV DNA A by -62 and +56 nucleotides, respectively), which might otherwise affect the stability of the constructs and their ability to spread (Stanley and Townsend, 1986; Etessami et al., 1989; Elmer and Rogers, 1990). Furthermore, co-inoculation of BRl+ and BLl+ resulted in viral DNA ampiification and the induction of localized symptoms on the inoculated leaves. This behavior has not been observed when these constructs are inoculated individually, suggesting that the TGMV genes are being expressed and are acting in a cooperative manner. However, the gene products cannot support ACMV systemic infection, which might be a consequence of their inefficient expression from the ACMV vector or their inability to recognize a feature of the ACMV genome or an ACMV-encoded or induced gene product. In view of the wealth of successful complementation experiments between different RNA viruses (reviewed by Atabekov and Taliansky, 1990) it is somewhat sur-

prising that geminivirus DNA B movement proteins are not interchangeable in Iv, benthamiana, a common host for ACMV and TGMV. However, complementation was classically demonstrated by co-inoculation of whole viruses, and the dependent virus might still have possessed a partially functional movement protein complemented by the helper virus. In our experiments, entire homologous coding regions have been exchanged, excluding this possibility. Experiments involving co-inoculation of viruses will usually only reveal positive complementation. However, expression of foreign proteins from an ACMV DNA A replicon ensures their presence in every infected cell, which has allowed us to observe a negative effect on virus infection. We have demonstrated that ACMV infection is largely suppressed when BLl+ is co-inoculated with ACMV DNA B. Given the orientation-dependent nature of the phenotype and the fact that infectivity is restored after disruption of the BLl coding region, we conclude that suppression is mediated in Vans by the gene product BLl rather than by a cis-acting feature of the inserted TGMV sequence. The effect is almost certainly operating at the level of long-distance movement rather than cell-to-cell spread because viral DNA replication seems to be largely unaffected and localized infections develop on inoculated leaves. This is not simply the host response to a large number of independent primary infection sites because other ACMV DNA A-based replicons do not elicit this type of response when inoculated individually. The inhibitory effect is highly significant and restoration of efficient systemic spread is correlated with disruption of the BLl coding sequence by recombination. Inhibition by BLl was mimicked by a chimera comprising the amino-terminus of BLl and the carboxy-terminus of BCl (clone BLl/BCl+), but not bythe reciprocal construct (clone BCl/BLl +). Inhibition of systemic spread by a gene product that is essentially a modified version of ACMV BCl is strongly reminiscent of a dominant negative mutation. Dominant negative mutants can inhibit transcriptional regulation, cytoskeleton function, and signal transduction (reviewed by Herskowitz, 1987). More recently, they have been used to disrupt signal transduction at the cell surface (Mercola et al., 1990) and HIV infectivity (Malim et al., 1989). The crucial criterion for the classification of a gene as a dominant negative mutant is interference with the function of the wild-type gene, and overexpression of the target gene may sometimes overcome the effect of the dominant inhibitor (Amaya et al., 1991). In the absence of a specific biochemical function assigned to ACMV BCl this cannot be rigorously tested in our case, although the fact that BLl and the chimera BLl/BCl specifically interfere with systemic

INHIBITION

OF ACMV

movement of the virus, a function attributed to a DNA B gene, suggests that this is indeed the case. Geminiviruses are recognized as serious pathogens of a variety of economically important crops (reviewed by Harrison, 1985). We recently demonstrated the potential of defective interfering DNA as a means to control the pathogenic effects of these viruses by interfering with their ability to replicate in transgenic plants (Stanley eta/., 1990; Frischmuth and Stanley, 1991). It is possible that dominant negative mutants that interfere with virus movement, when expressed directly from the host genome, or from constitutive or virus-inducible geminivirus replicons (Hayes et al., 1988; Stanley et al,, 1990), will contribute to the development of plants resistant to geminiviruses and other plant viruses.

ACKNOWLEDGMENTS A.v.A. was supported by a studentship from the John lnnes Foundation. We thank Professor Jeff Davies for critical reading of the manuscript.

REFERENCES AMAYA, E., MUSCI. T. J., and KIRSCHNER,M. W. (1991). Expression of a dominant negative mutant of the FGF receptor disrupts mesoderm formation in Xenopus embryos. Cell 66, 257-270. ATABEKOV,J. G., and TALIANSKY, M. E. (1990). Expression of a plant virus-coded transport function by different viral genomes. Adv. !hus Res. 38, 201-248. BROUGH, C. L., HAYES, R. G., MORGAN, A. J.. Courts, R. H. A., and BUCK, K. W. (1988). Effects of mutagenesis in vitro on the ability of cloned tomato golden mosaic virus DNA to infect Nicotiana benthamiana plants. J. Gen. Viral. 69, 503-514. CARR, R. J., and KIM, K. S. (1983). Evidence that bean golden mosaic virus invades non-phloem tissue in double infections with tobacco mosaic virus. J. Gen. Viral. 64, 2489-2492. DAVIES,J. W., and STANLEY.J. (1989). Geminivirus genes and vectors. Trends Genet. 5, 77-81. ELMER, J. S., BRAND, L., SUNTER, G., GARDINER,W. E., BISARO, D. M., and ROGERS,S. G. (1988). Genetic analysis of the tomato golden mosaic virus. II. The product of the AL1 coding sequence is required for replication. Nucleic Acids Res. 16, 7043-7060. ELMER, S., and ROGERS, S. G. (1990). Selection for wild type size derivatives of tomato golden mosaic virus during systemic infection. Nucleic Acids Res. 18, 2001-2006. ETESSAMI.P., CALLIS, R., ELLWOOD, S., and STANLEY,J. (1988). Delimitation of essential genes of cassava latent virus DNA 2. Nucleic Acids Res. 16, 481 l-4829. ETESSAMI, P., SAUNDERS,K., WATTS, J., and STANLEY,J. (1991). Mutational analysis of complementary-sense genes of African cassava mosaic virus DNA A. /. Gen. Viral. 72, 1005-l 012. ETESSAMI, P., WATS, J., and STANLEY, J. (1989). Size reversion of African cassava mosaic virus coat protein gene deletion mutants during infection of Nicotiana benthamiana. 1. Gen. Viral. 70, 277289. FEINBERG.A. P., and VOGELSTEIN,B. (1983). A technique for radiolabeling DNA restriction endonuclease fragments to high specific activity. Anal. Biochem. 132, 6-13. FRISCHMUTH,T., and STANLEY,J. (1991). African cassava mosaicvirus

INFECTION

563

DI DNA interferes with the replication of both genomic components. virology 183, 539-544. HAMILTON, W. 0. O., STEIN. V. E., COUTTS, R. H. A., and BUCK, K. W. (1984). Complete nucleotide sequence of the infectious cloned DNA components of tomato golden mosaic virus: Potential coding regions and regulatory sequences. EMBO J. 3, 2197-2205. HARRISON, B. D. (1985). Advances in geminivirus research. Annu. Rev. Phytopathol. 23, 55-82. HAYES, R. J., PE~~v, I. T. D., Cou-rrs, R. H. A., and BUCK, K. W. (1988). Gene amplification and expression in plants by a replicating geminivirus vector. Nature 334, 179-l 82. HERSKOWIZ, I. (1987). Functional inactivation of genes by dominant negative mutations. Nature 329, 219-222. HULL, R. (1989). The movement of viruses in plants. Annu. Rev. Phytopathol. 27, 213-240. KLINKENBERG,F. A., ELLWOOD, S., and STANLEY, J. (1989). Fate of African cassava mosaic virus coat protein deletion mutants after agroinoculation. /. Gen. Viral. 70, 1837-l 844. KLINKENBERG, F. A., and STANLEY, J. (1990). Encapsidation and spread of African cassava mosaic virus DNA A in the absence of DNA B when agroinoculated to Nicotiana benthamiana. 1. Gen. Viroi. 71, 1409-1412. MALIM. M. H., BGHNLEIN, S., HAUBER, J., and CULLEN, B. R. (1989). Functional dissection of the HIV-1 Rev trans.activator-Derivation of a trans.dominant repressor of Rev function. Cell 58, 205-214. MARSH, J. L., ERFLE, M., and WYKES, E. J. (1984). The plC plasmid and phage vectors with versatile cloning sites for recombinant selection by insertional inactivation. Gene 32, 481-485. MERCOLA, M., DEININGER,P. L., SHAMAH, S. M., and PORTER,J. (1990). Dominant-negative mutants of a platelet derived growth factor gene. Genes and Dev. 4,2333-2341. MESHI, T., MOTOYOSHI, F., MAEDA, T., YOSHIWOKA,S., WATANABE, H., and OKADA, Y. (1989). Mutations in the tobacco mosaic virus 30 kDa protein gene overcome Tm-2 resistance in tomato. P/ant Cell 1, 5 15-522. ROGERS, S. G., BISARO, D. M., HORSCH, R. B., FRALEY, R. T., HOFFMANN, N. L., BRAND, L., ELMER, J. S., and LLOYD, A. M. (1986). Tomato golden mosaic virus A component DNA replicates autonomously in transgenic plants. Cell 45, 593-600. SAMBROOK, J., FRITSCH, E. F., and MANIATIS, T. (1989). “Molecular Cloning: A Laboratory Manual.” Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. STANLEY, J. (1983). Infectivity of the cloned geminivirus genome requires sequences from both DNAs. Nature 305, 643-645. STANLEY,J., and GAY, M. R. (1983). Nucleotide sequence of cassava latent virus DNA. Nature 301, 260-263. STANLEY,J., FRISCHMUTH,T., and ELLWOOD, S. (1990). Defective viral DNA ameliorates symptoms of geminivirus infection in transgenic plants. Proc. Nat/. Acad. Sci. USA 87, 6291-6295. STANLEY,J., and TOWNSEND, R. (1986). Infectious mutants of cassava latent virus generated in vivo from intact recombinant DNA clones containing single copies of the genome. Nucleic Acids Res. 14, 5981-5998. STANLEY,J., TOWNSEND, R., and CURSON, S. J. (1985). Pseudorecombinants between cloned DNAs of two isolates of cassava latent virus. 1. Gen. Viral. 66, 1055-l 061. SUNTER, G., GARDINER, W. E., RUSHING, A. E., ROGERS, S. G., and BISARO, D. M. (1987). Independent encapsidation of tomato golden mosaic virus A component DNA in transgenic plants. P/ant fvlol. Biol. 8, 477-484. SUNTER, G., HARTITZ, M. D., HORMUZDI, S. G.. BROUGH, C. L., and BISARO, D. M. (1990). Genetic analysis of tomato golden mosaic virus: ORF AL2 is required for coat protein accumulation while

564

VON ARNIM

ORF AL3 is necessary for efficient DNA replication. Virology 179, 69-77. TALIANSKY, M. E., MALYSHENKO,S. I., PSHENNIKOVA,E. S., and ATABEKOV, J. G. (1982a). Plant virus-specific transport function. II. A factor controlling virus host range. Virology 122, 327-331. TALIANSKY, M. E., MALYSHENKO, S. I., PSHENNIKOVA, E. S., KAPLAN, I. B., ULANOVA, E. F., and ATABEKOV,.I. G. (1982b). Plant virus-specific transport function. I. Virus genetic control required for systemic spread. Virology 122, 318-326. TOWNSEND, R., WATTS, J., and STANLEY, J. (1986). Synthesis of viral DNA forms in Nicotiana plumbaginifoha protoplasts inoculated with cassava latent virus (CLV): Evidence for the independent repli-

AND STANLEY cation of one component of the CLV genome. Nucleic Acids Res. 14, 1253-l 265. VON ARNIM, A., and STANLEY, .I. (1992). Determinants of tomato golden mosaic virus symptom development located on DNA 6. Virology 186, 286-293. WARD, A., ETESSAMI,P., and STANLEY,J. (1988). Expression of a bacterial gere in plants mediated by infectious geminivirus DNA. EMBO J. 7, 1583-l 587. ZIEGLER-GRAFF,V., GUILFORD, P. J., and BAULCOMBE, D. C. (1991). Tobacco rattle virus RNA-1 29K gene product potentiates viral movement and also affects symptom induction in tobacco. Virology 182,145-155.