Gene silencing from plant DNA carried by a Geminivirus

The Plant Journal (1998) 14(1), 91–100

Gene silencing from plant DNA carried by a Geminivirus

Susanne Kjemtrup1,†, Kim S. Sampson2,

Charles G. Peele1, Long V. Nguyen2,‡, Mark A. Conkling2,

William F. Thompson1,2 and Dominique Robertson1,*

Departments of 1Botany and 2Genetics, North Carolina

State University, Raleigh, NC 27695, USA

Summary

The geminivirus tomato golden mosaic virus (TGMV)

replicates in nuclei and expresses genes from high copy

number DNA episomes. The authors used TGMV as a

vector to determine whether episomal DNA can cause

silencing of homologous, chromosomal genes. Two

markers were used to assess silencing: (1) the sulfur allele

(su) of magnesium chelatase, an enzyme required for

chlorophyll formation; and (2) the firefly luciferase gene

(luc). Various portions of both marker genes were inserted

into TGMV in place of the coat protein open-reading frame

and the constructs were introduced into intact plants

using particle bombardment. When TGMV vectors carrying

fragments of su (TGMV::su) were introduced into leaves

of wild-type Nicotiana benthamiana, circular, yellow

spots with an area of several hundred cells formed after

3–5 days. Systemic movement of TGMV::su subsequently

produced variegated leaf and stem tissue. Fragments that

caused silencing included a 786 bp 59 fragment of the

1392 bp su cDNA in sense and anti-sense orientation, and

a 403 bp 39 fragment. TGMV::su-induced silencing was

propagated through tissue culture, along with the viral

episome, but was not retained through meiosis. Systemic

downregulation of a constitutively expressed luciferase

transgene in plants was achieved following infection with

TGMV vectors carrying a 623 bp portion of luc in sense

or anti-sense orientation. These results establish that

homologous DNA sequences localized in nuclear episomes

can modulate the expression of active chromosomal

genes.

Introduction

Genes introduced at ectopic positions in plant genomes

can be unpredictably silenced (Flavell, 1994; Jorgensen,

Received 2 September 1997; revised 5 January 1998; accepted 16 January

1998.

*For correspondence: Dominique Robertson, Department of Botany, Box

7612, North Carolina State University, Raleigh, NC 27695–7612, USA (fax

11 919 515 3436; e-mail [email protected]).†Current address: Department of Biology, 108 Coker Hall CB#3280, UNC

Chapel Hill, NC 27599, USA.‡Current address: Department of Molecular Biology, Massachusetts General

Hospital, Boston, MA 02114, USA.

© 1998 Blackwell Science Ltd 91

1995; Matzke and Matzke, 1995; Thompson et al., 1996).

If the ectopic sequences are homologous to endogen-

ous plant genes, silencing of the endogenous gene also

frequently occurs. Both cytoplasmic and nuclear events

have been implicated in gene silencing. Post-transcriptional

gene silencing may require accumulation of a threshold

level of mRNA, after which degradation of all homologous

gene transcripts occurs and/or may be potentiated by

aberrant mRNAs (English et al., 1996; Goodwin et al., 1996;

Lindbo and Dougherty, 1993; Metzlaff et al., 1997; Mueller

et al., 1995). Some examples of pathogen-derived host

resistance to RNA viruses have been attributed to a post-

transcriptional gene silencing mechanism (Covey et al.,

1997; Mueller et al., 1995; Ratcliff et al., 1997; Tanzer et al.,

1997). Transcriptional gene silencing has been hypothe-

sized to involve DNA/DNA pairing, DNA methylation or

heterochromatinization (Kumpatla et al., 1997; Neuhuber,

1995; Park et al., 1996). Repeated DNA has a tendency to

undergo transcriptional silencing, which may be associated

with changes in chromatin structure (Meyer, 1996; Ye and

Signer, 1996), as well as to induce certain types of post-

transcriptional silencing (Stam et al., 1997).

Geminiviruses are single-stranded DNA viruses that

replicate through double-stranded DNA intermediates

using the plant DNA replication machinery (Hanley-

Bowdoin et al., 1997; Laufs et al., 1995). Geminiviruses

are useful as plant episomal vectors because they

replicate to high copy numbers in the nucleus and, in

some geminivirus/host combinations, approximately 0.8–

1 kb of foreign DNA can be stably integrated into

the viral genomes without significantly affecting their

replication or movement. Several geminiviruses have

been used to carry foreign genes in plants (reviewed in

Timmermans et al., 1994). Geminivirus genes resemble

repeated DNA in that they are present in multiple copies

in a single nucleus, and it is not clear how geminivirus

genes escape silencing. Although geminivirus DNA itself

is not methylated (Brough et al., 1992; Ermak et al., 1993),

little is known about the silencing potential of non-viral

DNA inserted into the geminivirus genome.

To begin to understand the relationship between

episomal DNA and gene silencing, we sought to determine

if silencing of nuclear genes could be triggered by homo-

logous sequences carried by a geminivirus episome. To

do this, we targeted both a nuclear transgene encoding

luciferase (luc) and an endogenous gene coding for an

enzyme required in chlorophyll biosynthesis. Magnesium

chelatase is a multi-subunit protein that catalyzes the

insertion of magnesium into protoporphyrin IX (Jensen

et al., 1996). In tobacco, a mutated allele (Su) of one subunit

causes the phenotype known as ‘sulfur’ (Nguyen, 1995).

92 Susanne Kjemtrup et al.

Nicotiana tabacum plants homozygous for this allele are

yellow (Su/Su), and heterozygous plants are yellow-green

(Su/su). The gene responsible for this phenotype was

identified in tobacco by transposon tagging (Nguyen, 1995)

and encodes the nucleotide binding subunit of magnesium

chelatase. We report here the results of introduction of

gene fragments from luc and from the wild-type allele, su,

in TGMV-derived vectors bombarded into N. benthamiana.

Our results demonstrate that DNA carried on episomes

can silence active, chromosomal gene expression.

Results

Plants infected with TGMV::su show a variegated

phenotype

Tomato golden mosaic virus (TGMV) is a bipartite gemini-

virus with a genome of two circular molecules, A and

B (Figure 1a, TGMV-A and TGMV-B). TGMV-A replicates

autonomously (Rogers et al., 1986; Sunter et al., 1987), and

TGMV-B is required for movement (Sunter et al., 1987).

The TGMV coat protein (AR1) is dispensable for replication

and movement in N. benthamiana (Brough et al., 1988;

Gardiner et al., 1988; Pooma et al., 1996) and can be

replaced with up to 800 bp of foreign DNA, which is stably

maintained in the viral genome (Elmer and Rogers, 1990).

To determine if TGMV could be used as a vector to

silence plant gene expression, we tested fragments of su

inserted into TGMV-A in place of the AR1 coding sequence,

transcribed from the AR1 promoter (Figure 1a). We used

sequences from su because down-regulation of the endo-

genous gene was likely to produce a visible phenotype:

lack of chlorophyll. Although the cDNA we used was from

N. tabacum (Nguyen, 1995), the sulfur alleles of N. tabacum

and N. benthamiana showed 96% nucleotide sequence

conservation over the 59 and 39 fragments used in this

study (C. Peele, unpublished observations).

N. benthamiana plants were infected using particle gun

bombardment to deliver cloned viral DNA. Wt TGMV pro-

duced chlorotic, irregular spots on inoculated leaves

approximately 5 days after bombardment. Leaf curling and

chlorosis were apparent at 5–7 days. In contrast, leaves

inoculated with TGMV::su5S (Figure 1b) showed discrete,

round, yellow spots after 3–5 days (Figure 2a). Not all

yellow spots produced by TGMV::su5S were circular, but

most had a distinct border between green and yellow

tissue (Figure 2b) that was absent in plants infected with

wt TGMV and in TGMV vectors carrying luc fragments.

‘Spots’ produced by inoculation of TGMV::luc and wt TGMV

occurred later, were infrequent, and had mixtures of green

and yellow cells (Figure 2b,c). Individual spots produced

by TGMV::su5S varied in size from 1 to 3 mm and had an

area of cells lacking chlorophyll that corresponded to at

least 500–800 epidermal cells (Figure 2a,b).

© Blackwell Science Ltd, The Plant Journal, (1998), 14, 91–100

Figure 1. Wild-type TGMV genome and DNA fragments used for silencing.

(a) The TGMV A and B components each contain a common region (CR)

that includes the origin of replication. AL1, AL2 and AL3 are viral genes

needed for replication and gene expression. The AR1 gene encodes the

coat protein and was replaced with the DNA fragments shown in (b) and

(c) to produce the TGMV-derived episomes TGMV::su or TGMV::luc. The B

component encodes two movement proteins, BL1 and BR1.

(b) The full length su cDNA includes 23 bp of upstream, non-coding

sequence and 1392 bp of coding sequence. A 786 bp 59 fragment in sense

orientation (su5S) was used to make TGMV::su5S. TGMV::su5F has a

frameshift in the su coding sequence. TGMV::su5 A contains a 786 bp 59

fragment in inverse orientation. A 403 bp 39 fragment in sense orientation

was used to make TGMV::su3S.

(c) A full length gene for luc is shown along with two fragments derived

from this gene used to make TGMV::luc constructs. A 623 bp fragment

extending from the 59 end of the gene to the EcoR1 site was cloned into

TGMV in sense (luc5S) and anti-sense (luc5 A) orientations.

Figure 2 (c,d,e,g) show variegation produced by systemic

infection with the TGMVA::su constructs. As new leaves

emerged subsequent to inoculation, large yellow-white

or light green sectors developed, often emanating from

vascular tissue (Figure 2c,g). The patterns of variegation

differed from plant to plant and from leaf to leaf (e.g.

Figure 2e). In many plants, yellow or white tissue was

confined to areas adjacent to the veins. In other plants,

large sectors of mesophyll tissue were affected. White

tissue likely resulted from bleaching of carotenoids. Micro-

scopic examination of fixed tissue showed no evidence of

Episomal silencing of plant genes 93

necrosis or cell death in areas lacking chlorophyll

(Figure 2h,i). Replacement of the AR1 gene by foreign

DNA, such as su, is known to attenuate symptoms (Gardiner

et al., 1988). Wild-type TGMV also produced chlorotic

lesions (Figure 2c), but these lesions usually contained

collapsed cells. A striking feature of the variegated plants

was the stem tissue, which was either striped yellow and

green, uniformly yellow, or white (Figure 2e,g). Neither

chlorotic lesions or any other visible symptoms (other than

stunting) were ever seen on stems of wild-type TGMV-

infected plants, although stem tissue of such plants occa-

sionally contains collapsed cells (Nagar et al., 1995).

Systemic variegation required TGMV-B, which has two

genes required for viral movement. When TGMV-A::su5S

was bombarded in the absence of the TGMV-B component,

yellow spots were produced but systemic variegation was

not detected (Figure 2f). Wild-type TGMV-A inoculated

without TGMV-B did not show visible spots, probably

because the TGMV-B component is required for symptom

formation (von Arnim and Stanley, 1992).

Variegation is not caused by mutant proteins

The mutant Su allele cloned from N. tabacum has a single

missense mutation causing an altered amino acid (Nguyen,

1995). This polypeptide has been hypothesized to stably

interfere with the function of the heteromeric magnesium

chelatase complex, causing a light green phenotype in

heterozygous plants. Translation of RNA from the wild

type 786 bp su fragment included in TGMV:su5S may

produce a truncated polypeptide that would similarly dis-

rupt assembly of the heteromeric complex. To test this

hypothesis, we introduced a frameshift mutation into the

su fragment 104 bp downstream of the start codon to

produce TGMV::su5F. The wild-type su polypeptide is 424

amino acids, including the putative transit sequence. The

frameshift mutation was predicted to yield a truncated

polypeptide with only 35 amino acids homologous to the

su protein. If translation were to re-initiate at the next AUG

(position 99), it would produce a polypeptide lacking a

transit sequence that would be unlikely to be imported

into the chloroplast. There were no repeatable differences

in the variegation produced in plants inoculated with

TGMV::su5S or TGMV::su5F (data not shown) suggesting

that defective protein interactions in the chloroplast were

not responsible for the yellow phenotype.

Plants bombarded with TGMV::su5 A or TGMV::su3S also

produced variegation similar to that seen with TGMV::su5S.

The 59 anti-sense and the 39 sense fragments lack

sequences for functional transit peptides. Because both

constructs were as effective as the 59 sense fragment of

su in producing yellow spots and systemic variegation, we

conclude that down-regulation of su gene function is more

likely to result from interactions at either the RNA or DNA


level, rather than from defective protein–protein inter-

actions.

We tested the original su cDNA in a plasmid vector

incapable of replicating in plant cells (pLVN44) to determine

if TGMV was required for homology dependent silencing

of su. No systemic variegation occurred and we never

observed yellow spots on the bombarded leaves.

To determine if tissue recovered from TGMV::su-

infected tissue retained the variegated phenotype, plants

were regenerated from sectors of systemically infected

N. benthamiana. Explants were taken from yellow/white

(no chlorophyll), variegated, or green tissue of plants

infected with TGMV::su5S. Variegated plants were

recovered from yellow/white or variegated tissue

(Figure 2j). In addition, one plant regenerated from

yellow/white tissue was fully green, with no indication of

variegation. Unlike the variegated regenerants, the green

plant lacked viral DNA, as determined by DNA gel blot

hybridization analysis (data not shown).

To determine if the TGMV::su constructs had integrated

into chromosomal DNA, regenerated plants that retained

variegation were allowed to set seed. Seed harvested from

flowers with white stems and sepals (which are normally

green) produced green, wild-type seedlings, as did all seed

harvested from variegated plants. This result suggested

that the TGMV::su sequences did not integrate into

chromosomal DNA at significant frequencies and/or that

any chromosomal form of TGMV::su was not sufficient to

cause silencing of the endogenous su alleles.

Endogenous su mRNA levels are reduced in yellow

sectors of TGMV/su-infected plants

Some phloem-limited geminiviruses cause yellow mosaic

leaf tissue. We considered the possibility that TGMV::su-

induced variegation may reflect altered symptomology of

the engineered virus rather than a reduction of su mRNA.

Total RNA was isolated from systemically infected leaf and

stem tissues from control plants and plants infected with wt

TGMV, TGMV::su5S, TGMV::su5F, and TGMV::luc. Whereas

TGMV::su5S-infected stem tissue was uniformly yellow,

leaf tissue from TGMV::su5S- and TGMV::su5F-infected

tissue contained some green areas, as it was difficult to

dissect yellow sectors from variegated tissue. Figure 3

shows that a transcript of the endogenous su gene at the

expected size was present in control and wt TGMV-infected

stem tissue (lanes 1, 4, 5, and 10) but lacking in TGMV::su5S-

infected leaf and stem tissue (lanes 2, 6, and 9). RNA from

tissue infected with TGMV::su5S also contained a transcript

corresponding to the viral 786 bp su fragment. Accumula-

tion of this transcript was attenuated in leaf tissue at a

later stage of viral infection (compare lanes 2 and 6, from

tissue 2 weeks post-infection with lane 9, at 4 weeks post-

infection). Lane 8 shows that tissue infected with TGMV


carrying the 786 bp fragment with a frameshift mutation

also lacked the full length su transcript from the endogen-

ous gene.

The level of endogenous su transcription was also tested


in plants infected with TGMV::luc5S. TGMV::luc has no

homology to plant DNA; it shows stable movement but

attenuated symptoms. The level of su transcript in

TGMV::luc-infected leaf material (Figure 3, lanes 3 and 7)


Figure 3. TGMV::su-infected plants show reduced accumulation of

endogenous su mRNA.

RNA was isolated from particle bombarded tissue (control) and from plants

infected with TGMV::su5S, TGMV::su5F, TGMV::luc, and wt TGMV. Leaf

material from TGMV::su-infected plants contained some green tissue but

the majority was yellow or white, while stem tissue was uniformly green

(control, wild-type) or white (TGMV::su5S and TGMV::su5F). 15 µg of total

RNA isolated from leaf or stem tissue was loaded in each lane. The blot

was probed with a 32P-labeled RNA transcript made from pLVN44 containing

the su cDNA clone. The 18S rRNA band served as a loading control. The

episomal transcript is labeled ‘TGMV::su5S9 and corresponds to the su

transcript from the AR1 promoter of TGMV::su. ‘2 weeks, leaf’ and ‘2 weeks,

stem’ show RNA harvested from plants 14 d.p.i. ‘4 week’ leaf tissue consists

of leaves at a similar developmental stage as 2 weeks (young, expanded)

but was taken from older plants, 28 d.p.i. Lanes 1, 5, and 10 show RNA

from uninfected control tissue; lanes 2, 6, and 9, TGMV::su5S-infected

tissue; lanes 3 and 7, TGMV::luc5S-infected tissue; lane 4, wtTGMV-infected

tissue; and lane 8, TGMV::su5F-infected tissue.

was similar to that of the control (Figure 3, lanes 1, 5,

and 10).

Downregulation of transgene expression

We used transgenic luciferase plants to determine if episo-

mal sequences could silence expression of a foreign gene.

N. benthamiana plants stably transformed with luc driven

by the 35S CaMV promoter were bombarded with

TGMV::luc5S and TGMV::luc5 A, containing a 623 bp 59

transcribed luc sequence in sense or anti-sense orientation.

Areas with both high and low luciferase activity were found

in leaves from uninfected control plants and in plants

systemically infected with wt TGMV 2 weeks after inocula-

tion (Figure 4). Only low levels of luciferase activity were

found in leaves infected with TGMV::luc5S and

TGMV::luc5 A (Figure 4). The complete absence of high

level luciferase activity was surprising as viral infections

are non-uniform.

We analyzed RNA from uninfected control and infected

Figure 2. N. benthamiana plants inoculated with TGMV::su show variegation.

(a) N. benthamiana leaf 3 d.p.i. with TGMV::su5S. Round and oval yellow spots (arrow) are apparent.

(b) Yellow spots 3 weeks after inoculation. Left, TGMV::luc5S-inoculated leaf with a mixture of green and yellow cells and an irregular border. Right,

TGMV::su5S with a circular spot and discrete borders. Bar is 2 mm.

(c) N. benthamiana 28 d.p.i. with wt TGMV (left) or TGMV::su5S (right). The plant on the left shows symptoms of TGMV infection including leaf curling

(upper leaves) and chlorosis. The short arrow shows chlorosis on inoculated leaf. The plant on the right shows variegation on upper leaves associated with

the presence of the TGMV::su5S. Inoculated leaves have circular, yellow spots (long arrow) and do not show variegation.

(d) Leaves taken from plants 24 d.p.i. Top: Control (left) and wt TGMV-infected leaf (right). Bottom: TGMV::su5S-infected leaf (left), TGMV::su5 A-infected leaf

(right). Yellow/white sectors are seen only in TGMV::su-infected leaves.

(e) Variegation patterns in leaves from a TGMV::su5S-infected plant.

(f) Plant bombarded with TGMV-A::su5S in the absence of TGMV-B shows only the initial, round, yellow spots (arrows) on inoculated leaves 28 d.p.i.

(g) Stem tissue of TGMV::su5S-infected plant has large areas of white tissue. Arrow shows a non-clonal green stripe.

(h) Cross-section of leaves from TGMV::su5S-infected (left) and control (right) plants. Left: TGMV::su5S-infected leaf with a uniform lack of chlorophyll. Right:

Mock-inoculated control leaf.

(i) Same as (h), viewed with fluorescence to detect DAPI-stained DNA. Both leaf sections have intact nuclei.

(j) Plants regenerated in tissue culture from variegated tissue of TGMV::su5S-infected tissue retain variegation.


plants to determine if the reduced luciferase activity was

caused by a reduction in mRNA accumulation. Figure 5

shows total RNA isolated from the same plants shown in

Figure 4, 3 weeks after infection. RNA was isolated from

the top 30 of each of the plants and included systemically

infected leaves, stems and meristematic areas, but not

inoculated leaves. The amount of luc transcript was greatly

reduced in plants bombarded with TGMV::luc compared

to both control and wt TGMV-infected plants. Accumulation

of the luc transcript in two of the wt TGMV-infected plants

was higher than that of uninfected controls. Increased

luciferase activity, detected by photon imaging, has

also been noted in TGMV-infected plants 3–4 weeks after

inoculation (K.S. Sampson, unpublished observations)

suggesting an enhancement of 35S CaMV promoter activity

by TGMV.

To determine whether the reduction in luc transcript

reflected a general reduction in cytoplasmic transcript

accumulation, we examined the levels of mRNAs encoding

elongation initiation factor 4A (eIF4A) and histone H1. The

eIF4A gene is constitutively transcribed in meristematic,

leaf, stem and root tissues of tobacco (Mandel et al., 1995).

An RNA gel blot was hybridized with probes for eIF4A

from tobacco and histone H1 from pea (Gantt and Key,

1987) (Figure 5). The level of both the eIF4A and histone

H1 transcripts were similar for wt TGMV-infected and

TGMV::luc-infected plant tissue. These results demonstrate

that the TGMV::luc constructs cause a specific reduction in

luc mRNA accumulation.

Tissue with incomplete silencing contains higher viral

DNA levels

To determine if there was a direct relationship between

the presence of viral DNA and silencing, we performed gel

blot hybridization analysis of DNA isolated from leaves

systemically infected with TGMV::su5S, TGMV::su5A or

TGMV::su5F (Figure 6). Variegated leaf tissue was separated

into three categories: green tissue (G), yellow tissue (Y)

and mixed (M). Green tissue had no apparent silencing


Figure 4. Luciferase transgene expression is downregulated by TGMV::luc.

Leaves were imaged from N. benthamiana containing a 35S CaMV promoter-luciferase transgene 14 d.p.i. with (a) particles only (control), (b) wild-type

TGMV, (c) TGMV::luc5S, or (d) TGMV::luc5 A. Five leaves were imaged from each plant, in the same pattern as shown for the control. Photon density from

low to high is shown as blue, green, yellow, red, and white.

phenotype or viral symptoms. The yellow category

included yellow and white tissue that lacked chlorophyll,

while mixed tissue contained light green regions and

borders between yellow and green tissue. Green, healthy

tissue showed a low amount of viral DNA (Figure 6)

indicating that some cells in the green tissue were infected.

Viral DNA levels were higher in mixed tissues than they

were in yellow tissue. The stability of the viral su insert

was shown by a lack of smaller fragments corresponding

to deleted forms of TGMV::su (Figure 6a). Each of the

TGMV::su constructs tested showed the same pattern, with

the highest viral DNA levels in mixed tissues (Figure 6).


This experiment was repeated several times with variable

results. However, none of the plants ever had higher viral

DNA levels in yellow tissue than in mixed tissue.

The ratio of ssDNA to dsDNA in plants infected with the

TGMV::su constructs, which lack the AR1 coat protein gene,

is lower than it is in wild-type TGMV infections (Figure 6).

The AR1 protein encapsidates ssDNA, forming inclusion

bodies in plant nuclei (Rushing et al., 1987), and the

absence of the AR1 protein is known to reduce ssDNA

accumulation (Pooma et al., 1996; Sunter and Bisaro, 1992).

In principle, viral DNA in the yellow tissue might have

been degraded as a result of a host response to viral


Figure 5. The luciferase transcript is downregulated by TGMV::luc.

RNA was harvested from plants 21 d.p.i. and 5 µg of total RNA loaded in

each lane. Blotted RNA was probed with a 32P-labeled luc RNA probe (top

panel) or probes for the constitutively expressed genes for eIF4 A (bottom

panel, top) and histone H1 (bottom). RNA is shown from two particle

bombarded plants (control), and three plants each infected with wtTGMV

(3–5), TGMV::luc5S (6–8), or TGMV::luc5 A (9–11). Degradation of the luc

message in some of the samples is seen in the smear below the full

length transcript.

Figure 6. TGMV::su DNA accumulation is lower in yellow tissue than in

yellow-green borders.

DNA was isolated from plants showing systemic symptoms or variegation.

Uncut DNA was electrophoresed, blotted to membranes and probed with

an 850 bp TGMV-A-specific DNA fragment. Lane 1 shows wt TGMV DNA

with three DNA forms: open circular (OC), closed circular (CC), and ssDNA.

Lanes 2–4 show DNA isolated from plants infected with TGMV::su5S. Y is

yellow-white tissue and M is mixed (borders of green and yellow areas).

G is green tissue which lacked visible traces of variegation or viral

symptoms. Lanes 5–7 show viral DNA from N. benthamiana inoculated

with TGMV::su5F and lanes 8–10, TGMV::su5 A. Bottom panel: Same blot

as above but probed with a 786 bp fragment from su.

infection. However, microscopic examination of DAPI-

stained vibratome sections from yellow-white sectors

showed intact, healthy cells (Figure 2h,i), and viral DNA

isolated from such tissue did not appear more degraded

than DNA isolated from ‘mixed’ or green tissue (Figure 6a).

Furthermore, ethidium bromide-stained gels of uncut total

DNA from yellow tissue showed no signs of chromosomal

DNA degradation (data not shown). To exclude wild-type

contamination, the same blots were stripped and reprobed

with the 786 bp su fragment. This probe hybridized to both

ssDNA and dsDNA from each of the TGMV::su constructs

(Figure 6b). These experiments suggest that su gene

silencing is not simply proportional to viral DNA accumu-

lation.


Discussion

There are many examples of gene silencing mediated by

stably integrated homologous DNA, or by homologous

nuclear gene sequences carried by cytoplasmic RNA

viruses (e.g. English et al., 1996; Flavell, 1994; Mueller

et al., 1995). Here, we show that a nuclear-localized DNA

virus carrying sequences complementary to chromosomal

genes can silence all copies of the chromosomal gene.

We used the TGMV vector to test sense, anti-sense

and frameshifted versions of a 786 bp 59 fragment of

the magnesium chelatase gene (su), as well as a non-

overlapping 403 bp 39 sense fragment. In each case similar

variegated phenotypes were produced after infection. The

lack of chloroplast transit sequences in two of the su

fragments used for silencing (the anti-sense fragment and

the 39 fragment), as well as the anti-sense result, argue

that defective protein–protein interactions in the multi-

subunit magnesium chelatase complex are unlikely to be

responsible for the lack of chlorophyll formation. In addi-

tion, silencing of the luciferase transgene by TGMV::luc is

also likely to be at the RNA or DNA level, as both sense

and anti-sense versions of the luc fragment downregulate

transgene expression.

We used luciferase to test downregulation by episomal

sequences because we could visualize its activity non-

destructively. Whereas the stability of the magnesium

chelatase enzyme complex is not well characterized, the

activity of luciferase is known to be directly proportional

to the accumulation of a single gene product (Millar et al.,

1992). Both luciferase activity and mRNA accumulation

were markedly decreased in TGMV::luc-infected plants. We

were surprised at the extent of downregulation of luciferase

activity in TGMV::luc-infected plants (Figure 4).

Recently, silencing caused by diffusible factors has been

reported. Evidence for a phloem-mobile component

involved in silencing was provided by graft transmissiblity

experiments (Palauqui et al., 1997). In a different set of

experiments, introduction of sequences for the green

fluorescent protein (GFP) into a single leaf of a transgenic

GFP plant using Agrobacterium infiltration caused silencing

of the GFP gene in upper leaves remote from the site of

infiltration (Voinnet and Baulcombe, 1997).

The pattern of su silencing by TGMV::su resembled the

pattern of silencing of the GFP gene mediated by infiltration

of Agrobacterium carrying a GFP gene sequence into an

intact leaf (Voinnet and Baulcombe, 1997). In both cases,

initially silenced areas were flanked by non-silenced areas

at the distal portion of the inoculated leaf blade while

systemic silencing appeared in upper leaves and often

emanated from vascular tissue. Our experiments differ in

that we used a mobile vector whereas the Agrobacterium

vector is not known to move in plants. In contrast to the

work reported by Voinnet and Baulcombe, we could not


separate long distance silencing effects from the presence

of vector DNA. When TGMV::su5S was inoculated in the

absence of the B component, yellow spots formed only on

inoculated leaves and there was no systemic variegation

(Figure 2f). Because microprojectile bombardment targets

only a few cells, but the silenced area comprised several

hundred cells, we suspect that a diffusible factor may be

involved. Cell-to-cell movement of TGMV is known to

require the B component (Jeffrey et al., 1996). Although

we cannot rule out the formal possibility that single-

stranded viral DNA may be moving cell-to-cell, we think it

unlikely because there was no evidence of further viral

movement from the spots, even after 4 weeks.

The fact that yellow tissue contained lower amounts of

TGMV::su DNA than mixed green and yellow tissue

(Figure 6) was surprising. This suggests that a threshold

level of viral DNA may be reached beyond which both viral

DNA replication and su gene expression are down-

regulated. Time course studies are in progress to distin-

guish this possibility from a transient high level of viral

DNA followed by exit from the silenced tissue.

Previous reports have demonstrated gene expression

from geminivirus-derived episomes (reviewed in

Timmermans et al., 1994). A direct correlation between

episome copy number and gene expression was shown in

cultured cells for TGMV carrying the neo gene (Kanevski

et al., 1992), suggesting that episomal gene expression

is not affected by factors that often cause silencing of

chromosomal transgenes. Our experiments differ in that

there was no selection for gene expression, there was

homology between the episomal and chromosomal

sequences, and only partial copies of the su or luc genes

were carried in episomes. We demonstrated that episomal

DNA can silence chromosomal gene expression, but have

not resolved whether episomal genes have the potential

to be silenced themselves. There is accumulation of a viral

transcript corresponding to the inserted su fragment in

TGMV::su-infected plants examined 2 weeks after inocula-

tion (Figure 3), suggesting that episomal transcription

was active. Experiments to further characterize episomal

silencing are now in progress.

Experimental procedures

The plasmid used for construction of the TGMV-A-derived vectors

was pMON1655, a pUC-based plasmid with 1.5 tandem copies of

TGMV-A containing the AR1 coding sequence replaced by a short

polylinker. In this plasmid, the AR1 promoter and terminator

sequences are retained. pLVN44 is a full length 1392 bp cDNA of

the nucleotide-binding subunit of magnesium chelatase (Nguyen,

1995) isolated from Nicotiana tabacum cv. SR1. A 786 bp Acc65I/

EcoRV fragment from the 59 end of the su cDNA (including the

start codon and 17 bp of 59 non-coding sequence) was cloned into

pMON1655 in both orientations, resulting in two TGMV constructs:

the 59 su fragment driven by the TGMV AR1 promoter in the

sense (TGMV-A::su5S) or anti-sense orientation (TGMV-A::su5A)


(Figure 1b). The sense reading frame of the su gene was disrupted

by cleaving and filling an EcoNI restriction enzyme site to create

a stop codon 104 bp from the initiator ATG to produce TGMV-

A::su5F. TGMV-A::su5S contains a 403 bp EcoRV fragment in sense

orientation that ends 108 bp upstream of the putative stop codon.

The plasmids TGMV-A::luc5S and TGMV-A::luc5 A contain a 623 bp

EcoR1/BglII fragment of the 59 end of the 1650 bp firefly luciferase

gene (Ow et al., 1996) cloned into pMON1655 in the sense and

anti-sense orientations, respectively (Figure 1c). This fragment

contains 26 bp of 59 non-coding sequence and retains the luciferase

start codon. A plasmid containing 1.4 tandem copies of the TGMV

B component (pTG1.4B, Fontes et al., 1994) was used with the

TGMV-A constructs for inoculation unless otherwise noted.

Three-week-old plants were used for bombardment. The

BIOLISTIC® Particle Delivery System (Bio-Rad, Hercules, CA,

USA) was used to infect N. benthamiana plants as described

previously (Nagar et al., 1995). Total DNA from infected and

healthy plants was isolated (Dellaporta et al., 1983), blotted and

probed with labeled DNA corresponding to the TGMV-A AL1, AL2,

AL3 reading frames, or 786 bp of su. Total RNA was isolated

from infected or mock-infected leaf and stem tissue as described

(Kaufman et al., 1985). Glyoxylated RNA was fractionated, blotted

and hybridized with RNA probes as described (Frances et al., 1992).

Transgenic plants were imaged using a photon-counting

imaging system (Hamamatsu Photonic Systems, Bridgewater, NJ,

USA) as described (Millar et al., 1992). Systemically infected or

control leaves were excised and immediately submerged in a

0.01 mg ml–1D-luciferin solution for precisely 5 min before being

placed in the imaging chamber.

Acknowledgments

We thank Monsanto for providing pMON1655, William Folk for

the transgenic luciferase plants and Cris Kuhlemeier for the eIF4A

probe. We are grateful to Steve Nagar for help with microscopy,

and Linda Hanley-Bowdoin and Tim Petty for reading the manu-

script. This work was supported by a North Carolina Biotechnology

Center ARIG to D.R. and by US Department of Agriculture National

Research Initiative Grant no. 97–35303–4538 to D.R. and W.F.T.

Kim Sampson was supported by the McKnight Foundation.

References

von Arnim, A. and Stanley, J. (1992) Determinants of tomato

golden mosaic virus symptom development located on DNA B.

Virology, 186, 286–293.

Brough, C.L., Gardiner, W.E., Inamdar, N.M., Zhang, X.Y., Ehrlich,

M. and Bisaro, D.M. (1992) DNA methylation inhibits

propagation of tomato golden mosaic-virus DNA in transfected

protoplasts. Plant Molec. Biol. 18, 703–712.

Brough, C.L., Hayes, R.J., Morgan, A.J., Coutts, R.H.A. and K.W.B.

(1988) Effects of mutagenesis in vitro on the ability of cloned

tomato golden mosaic virus DNA to infect Nicotiana

benthamiana plants. J. Gen. Virol. 69, 503–514.

Covey, S.N., Al-Kaff, N.S., Langara, A. and Turner, D.S. (1997)

Nature, 385, 781.

Dellaporta, S.L., Wood, J. and Hicks, J.B. (1983) A plant DNA

minipreparation:version II. Plant Molec. Biol. Rep. 1, 19–21.

Elmer, S. and Rogers, S.G. (1990) Selection for wild type size

derivatives of tomato golden mosaic virus during systemic

infection. Nucl Acids Res. 18, 2001–2006.

English, J.J., Mueller, E. and Baulcombe, D.C. (1996) Suppression


of virus accumulation in transgenic plants exhibiting silencing

of nuclear genes. Plant Cell, 8, 179–188.

Ermak, G., Paszkowski, U., Wohlmuth, M., Scheid, O.M. and

Paszkowski, J. (1993) Cytosine methylation inhibits replication

of African cassava mosaic virus by 2 distinct mechanisms. Nucl

Acids Res. 21, 3445–3450.

Flavell, R.B. (1994) Inactivation of Gene Expression in Plants as a

Consequence of Specific Sequence Duplication. Proc. Natl Acad.

Sci. USA, 91, 3490–3496.

Fontes, E.P.B., Gladfelter, H.J., Schaffer, R.L., Petty, I.T.D. and

Hanley-Bowdoin, L. (1994) Geminivirus replication origins have

a modular organization. Plant Cell, 6, 405–416.

Frances, S., White, M., Edgerton, M.D., Jones, A.M., Elliott, R.C.

and Thompson, W.F. (1992) Initial cahracterization of a pea

mutant with light-independent photomorphogenesis. Plant Cell,

4, 1519–1530.

Gantt, J.S. and Key, J.L. (1987) Molecular cloning of a pea H1

histone cDNA. Eur. J. Biochem. 166, 119–125.

Gardiner, W.E., Sunter, G., Brand, L., Elmer, J.S., Rogers, S.G. and

Bisaro, D.M. (1988) Genetic analysis of tomato golden mosaic

virus: The coat protein is not required for systemic spread or

symptom development. EMBO J. 7, 899–904.

Goodwin, J., Chapman, K., Swaney, S., Parks, T.D., Wernsman,

E.A. and Dougherty, W.G. (1996) Genetic and biochemical

dissection of transgenic RNA-mediated virus ressistance. Plant

Cell, 8, 85–105.

Hanley-Bowdoin, L., Eagle, P.A., Orozco, B.M., Robertson, D. and

Settlage, S.B. (1997) Geminivirus rReplication. In Biology of

Plant Microbe Interactions (Stacey, B., Mullin, B. and Gresshoff,

P.M., eds). St Paul: International Society of Molecular Plant

Microbe Interactions, pp. 287–292.

Jeffrey, J.L., Pooma, W. and Petty, I.T.D. (1996) Genetic

requirements for local and systemic movement of tomato

golden mosaic virus in infected plants. Virology, 223, 208–218.

Jensen, P.E., Willows, R.D., Petersen, B.L., Vothknecht, U.C.,

Stummann, B.M., Kannangara, C.G., von Wettstein, D. and

Henningsen, K.W. (1996) Structural genes for Mg-chelatase

subunits in barley: xantha-f,-g and -h. Molec. Gen. Genet. 250,

383–394.

Jorgensen, R. (1995) Cosuppression, flower color patterns, and

metastable gene expression states. Science, 268, 686–691.

Kanevski, I.F., Thakur, S., Cosowsky, L., Sunter, G., Brough, C.,

Bisaro, D. and Maliga, P. (1992) Tobacco lines with high copy

numbers of replicating recombinanat geminivirus vectors after

biolistic DNA delivery. Plant J. 2, 457–463.

Kaufman, L.S., Briggs, W.R. and Thompson, W.F. (1985)

Phytochrome control. of specific mRNA levels in developing

pea buds. Plant Physiol. 78, 388–393.

Kumpatla, S.P., Teng, W., Buchholz, W.G. and Hall, T.C. (1997)

Epigenetic transcriptional silencing and 5-azacytidine-mediated

reactivation of a complex transgene in rice. Plant Physiol. 115,

361–373.

Laufs, J., Jupin, I., David, C., Schumacher, S., Heyraudnitschke, F.

and Gronenborn, B. (1995) Geminivirus replication: Genetic and

biochemical characterization of Rep protein function, a review.

Biochimie, 77, 765–773.

Lindbo, J.A. and Dougherty, W.G. (1993) Untranslatable transcripts

of the tobacco etch virus coat protein gene sequence can

interfere with tobacco etch virus replication in transgenic plants

and protoplasts. Virology, 189, 725–733.

Mandel, T., Fleming, M.T., Krahenbuhl, R. and Kuhlemeier, C.

(1995) Definition of constitutive gene expression in plants: the

translation initiation factor 4A gene as a model. Plant Mol. Biol.

29, 995–1004.


Matzke, M.A. and Matzke, A.J.M. (1995) How and why do plants

inactivate homologous (trans) genes? Plant Physiol. 107, 679–

685.

Metzlaff, M., O’Dell, M., Cluster, P.D. and Flavell, R.B. (1997) RNA-

mediated RNA degradation and chalcone synthase A silencing

in petunia. Cell, 88, 845–854.

Meyer, P. (1996) Repeat-induced gene silencing: Common

mechanisms in plants and fungi. Biol. Chem Hoppe Seyler, 377,

87–95.

Millar, A., Short, S.R., Chua, N.-H. and Kay, S.A. (1992) A novel

circadian phenotype based on firefly luciferase expression in

transgenic plants. Plant Cell, 4, 1075–1087.

Mueller, E., Gilbert, J.E., Davenport, G., Brigneti, G. and

Baulcombe, D.C. (1995) Homology-dependent resistance:

transgenic virus resistance in plants related to homology-

dependent gene silencing. Plant J. 7, 1001–1013.

Nagar, S., Pedersen, T.J., Carrick, K.M., Hanley-Bowdoin, L. and

Robertson, D. (1995) A geminivirus induces expression of a

host DNA synthesis protein in terminally differentiated plant

cells. Plant Cell, 7, 705–719.

Neuhuber, F. (1995) Susceptibility of transgene loci to homology-

dependent gene silencing. Mol Gen Genet. 247, 264.

Nguyen, L.V. (1995) Transposon tagging and isolation of the sulfur

gene in tobacco (Nicotiana tabacum). PhD thesis, North Carolina

State University.

Ow, D.W., Wood, K.V., Deluca, M., De Wet, J.R., Helinski, D.R. and

Howell, S.H. (1996) Transient and stable expression of the firefly

luciferase gene in plant cells and transgenic plants. Science,

234, 856–859.

Palauqui, J.C., Elmayan, T., Pollien, J.M. and Vaucheret, H. (1997)

Systemic acquired silencing: Transgene-specific post-

transcriptional silencing is transmitted by grafting from silenced

stocks to non-silenced scions. EMBO J. 16, 4738–4745.

Park, Y.D., Papp, I., Moscone, E.A., Iglesias, V., Vaucheret, H.,

Matzke, A.J. and Matzke, M.A. (1996) Gene silencing mediated

by promoter homology occurs at the level of transcription and

results in meiotically heritable alterations in methylation and

gene activity. Plant J. 9, 183–194.

Pooma, W., Gillette, W.K., Jeffrey, J.L. and Petty, I.T.D. (1996) Host

and viral factors determine the dispensability of coat protein

for bipartite geminivirus systemic movement. Virology, 218,

264–268.

Ratcliff, F., Harrison, B.D. and Baulcombe, D.C. (1997) A similarity

between viral defense and gene silencing in plants. Science,

276, 1558–1560.

Rogers, S.G., Bisaro, D.M., Horsch, R.B., Fraley, R.T., Hoffman,

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.

Rushing, A.E., Sunter, G., Gardiner, W.E., Dute, R.R. and Bisaro,

D.M. (1987) Ultrastructural aspects of tomato golden mosaic

virus infection in tobacco. Phytopathology, 77, 1231–1236.

Stam, M., deBruin, R., Kenter, S., vanderHoorn, R.A.L.,

vanBlokland, R., Mol, J.N.M. and Kooter, J.M. (1997) Post-

transcriptional silencing of chalcone synthase in Petunia by

inverted transgene repeats. Plant J. 12, 63–82.

Sunter, G. and Bisaro, D.M. (1992) Transactivation of geminivirus-

AR1 and geminivirus-BR1 gene expression by the viral-AL2

gene product occurs at the level of transcription. Plant Cell, 4,

1321–1331.

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. Plant Mol. Biol.

8, 477–484.


Tanzer, M.M., Thompson, W.F., Law, M.D., Wernsman, E.A. and

Uknes, S. (1997) Characterization of post-transcriptionally

suppressed transgene expression that confers resistance to

tobacco etch virus infection in tobacco. Plant Cell, 9, 1411–1423.

Thompson, W.F., Allen, G.C. Jr, G.H. and Spiker, S. (1996) Matrix

attachment regions and transgene expression. In Genomes

(Proceedings of the 21st Stadler Genetics Symposium)

(Gustafson, P. and Flavell, R.B., eds). New York: Plenum

Publishing Corp., pp. 243–269.


Timmermans, M., Das, O. and Messing, J. (1994) Geminiviruses

and their uses as extrachromosomal replicons. Annu. Rev. Plant

Physiol. 45, 79–112.

Voinnet, O. and Baulcombe, D.C. (1997) Systemic signalling in

gene silencing. Nature, 389, 553–553.

Ye, F. and Signer, E.R. (1996) RIGS (repeat-induced gene silencing)

in Arabidopsis is transcriptional and alters chromatin

configuration. Proc. Natl Acad. Sci. USA, 93, 10881–10886.

Gene silencing from plant DNA carried by a Geminivirus

Documents

Transcript of Gene silencing from plant DNA carried by a Geminivirus