Excision and Episomal Replication of Cauliflower Mosaic Virus Integrated Into a Plant Genome 1

Transgenic Arabidopsis ( Arabidopsis thaliana ) plants containing a monomeric copy of the cauliflower mosaic virus (CaMV) genome exhibited the generation of infectious, episomally replicating virus. The circular viral genome had been split within the non-essential gene II for integration into the Arabidopsis genome by Agrobacterium -mediated transformation. Transgenic plants were assessed for episomal infections at flowering, seed set, and/or senescence. The infections were confirmed by western blot for the CaMV P6 and P4 proteins, electron microscopy for the presence of icosahedral virions, and through PCR across the recombination junction. By the end of the test period, a majority of the transgenic Arabidopsis plants had developed episomal infections. The episomal form of the virus was infectious to nontransgenic plants, indicating that no essential functions were lost after release from the Arabidopsis chromosome. An analysis of the viral genomes recovered from either transgenic Arabidopsis or nontransgenic turnip revealed that the viruses contained deletions within gene II and in some cases, the deletions extended to the beginning of gene III. In addition, many of the progeny viruses contained small regions of non-viral sequence derived from the flanking transformation vector. The nature of the nucleotide sequences at the recombination junctions in the circular progeny virus indicated that most were generated by nonhomologous recombination during the excision event. The release of the CaMV viral genomes from an integrated copy was not dependent upon the application of environmental stresses, but occurred with greater frequency with either age or the late stages of plant maturation.


Members of the virus family
in Dahlia (Dahlia variabilis) (reviewed in Harper et al., 2002;Staginnus and Richert-Pöggeler, 2006;Pahalawatta et al., 2008). These integrated viruses have been called plant endogenous pararetroviruses (EPRVs) to distinguish them from pararetroviruses that never have been found integrated into host genomes in nature (Staginnus and Richert-Pöggeler, 2006). None of the episomal forms of these viruses are transmitted through seed, so the processes that lead to episomal infections occur with every generation of plants.
The nucleotide sequences of the integrated viral forms have revealed that the mechanisms that lead to activation of the EPRVs and subsequent episomal infections are complex. Activation of the EPRVs is thought to be associated with epigenetic changes that could (i) allow the production of a greater than full-length transcript, or (ii) lead to homologous recombination and circularization of the ERPV genome (Staginnus and Richert-Pöggeler, 2006). For example, the PVCV genome is integrated in a complete, continuous form in tandem arrays; activation could involve the production of a greater than full-length transcript that could be reverse transcribed into DNA (reviewed in Harper et al., 2002;Staginnus and Richert-Pöggeler, 2006).
On the other hand, the activation of BSV and TVCV would have to involve recombination between at least two integrated viral DNA segments integrated into different loci (Staginnus and Richert-Pöggeler, 2006). Furthermore, activation of the viral DNA and subsequent episomal replication appears to be triggered by changes in day-length or abiotic stresses such as drought or heat stress (Staginnus and Richert-Pöggeler, 2006). These host/virus combinations are valuable because they serve as exquisitely sensitive biosensors for recombination events that occur during the life of the plant, since a single virus genome excised from its chromosomal location could be capable of replication and amplification within the plant.
We have sought to develop a system based on cauliflower mosaic virus (CaMV) that could be used to examine the release of integrated caulimovirus 6 sequences from host chromosomes and subsequent episomal infections. Such a system could be used to identify the genetic and environmental parameters for development of the caulimovirus episomal infections in plants. Although CaMV has never been shown to exist as a natural integrant in its hosts, its complete genome has been introduced into plant chromosomes through Agrobacterium tumefaciensmediated transformation (Young et al., 1987;Gal et al., 1991) and progeny plants were examined for recombinants. In one study, a greater than full-length copy of the CaMV genome integrated into a host chromosome was able to be excised and to replicate episomally (Gal et al., 1991). The duplicated region of CaMV DNA was 989 bp, which provided a suitable substrate for release of an episomal virus. In this instance, the mechanism for release was thought to involve production of the fulllength 35S RNA and its subsequent reverse transcription into DNA. Similarly, infectious CaMV also was generated when Escherichia coli plasmid DNA containing greater then full-length CaMV genomes were directly introduced into plant cells (Grinsley et al., 1986;Stratford and Covey, 1989;Vaden and Melcher, 1990). In another study (Young et al., 1987), only a single copy of the CaMV genome, interrupted within the essential gene V (Fig. 1), was integrated into the host.
Although the gene VI protein product was detected, due to the activity of the CaMV 19S RNA promoter, no episomal infections arose from the insertion of this single, CaMV genomic copy. Similarly, E. coli plasmids containing only a single copy of the CaMV genome were not infectious after inoculation to plants unless the flanking plasmid vector sequences were removed first, regardless of whether the viral genome was split within the essential gene V or the nonessential gene II (Walden and Howell, 1983 (Fig. 1). Since gene II is not required for replication or movement within the plant, recombination events confined within gene II would not abolish infectivity of the virus. These transgenic plants then were used to assess whether recombination events could indeed lead to episomal replication of CaMV, whether the environment influenced the development of episomal infections, and what mechanism for recombination was used. We found that episomal infections readily developed in the transgenic plants, even when they were not subjected to any discernable stress.
The results of our study show that the threshold for release of caulimovirus sequences integrated into plant genomes is surprisingly low.

Detection of Excision and Episomal Replication of CaMV
Transgenic, T 2 -generation Arabidopsis plants transformed with a full-length copy of the circular CaMV genome split at the XhoI site within gene II (Fig. 1B) were used in these experiments. These plants contained a single copy of the CaMV genome, as determined from a combination of the segregation frequency of the Basta resistant plants (data not shown) and product analysis of PCR used to verify that the integration event at one locus was due to a single copy of the CaMV DNA and not to tandem inserts (Supplemental Fig. S1). Since the plants in the T 2 generation still 8 ( Fig. 2) for viral proteins P6 (transactivator protein) and P4 (coat protein), by PCR for viral DNA, and by immuno-capture electron microscopy of CaMV virions (Fig. 3).
A western blot analysis for the presence of the P6 protein revealed that it was detected in the majority of the transgenic plants ( Fig. 2A and data not presented). The CaMV P6 product is expressed from the 19S RNA promoter (Fig. 1); consequently a baseline of expression would be expected to be present in all transgenic plants.
However, we found that there was considerable variation in its concentration. In at least one transgenic plant (CMV-infected #3) the level of P6 protein was comparable to its expression in transgenic plants driven by a 35S RNA promoter (D4-2 samples 1 and 2). Multiple bands were detected for P6, which is in agreement with previous studies indicating the presence of the full length P6 product and several breakdown products (Daubert and Routh, 1990;Schoelz et al., 1991). In contrast, the P4 protein is translated from the polycistronic 35S RNA. Given the structure of the CaMV insert in the T-DNA (Fig. 1), the P4 product could be expressed only after excision and episomal replication of the CaMV viral genome. The P4 protein was detected in fewer transgenic plants ( Fig. 2B and data not shown), but the plants that expressed high levels of P4 protein also expressed high levels of the P6 protein (Figs. 2A and 2B). Taken together, the western blots for the CaMV P6 and P4 proteins indicated that episomal replication of CaMV had occurred in at least some of the transgenic plants.
Episomal replication of the viral DNA was confirmed by PCR (Fig. 2C), using primers that flanked the insertion of CaMV sequences into the T-DNA of the Agrobacterium vector (Fig. 1B). Consequently, a PCR band could be generated only after a recombination event that resulted in recircularization of the viral genome.
Interestingly, PCR products were amplified from every plant except one (Drought #2), in contrast to the western blots for the P4 protein. Furthermore, several of the PCRderived bands were larger in size than the bands derived from CaMV strains H7 and W260, an indication that the viral DNA had either acquired foreign DNA sequences or sustained a rearrangement.
To further confirm the presence of episomally replicating viruses, leaf dips from individual plants were examined by immuno-capture electron microscopy, revealing the presence of icosahedral CaMV particles (Fig. 3). Typical CaMV particles were detected in leaf dips of non-stressed and each type of applied-stressed plants ( Fig. 3 and data not shown).

9
To verify the reproducibility of the above results and to demonstrate that they were not limited to one transgenic line [Line 316], a second transgenic line [Line 318] was evaluated under two environmental conditions for episomally replicating CaMV at 25 days post-planting and again at flowering. Both the western blots for the P4 protein and PCR across the gene II/III junction confirmed the detection of episomal replication of CaMV (data not shown).

Cumulative Development of Episomal Infections in Stressed and Non-stressed Plants
It is conceivable that the removal of a leaf for analysis of transgene status might be a sufficient biotic stress to trigger the excision and episomal replication of CaMV DNA. Hence, in subsequent experiments the plants were not prescreened for the presence of the transgene and they were assessed at different stages of development for episomal replication of CaMV. Consequently, some of the plants would not be expected to carry the CaMV transgene due to segregation of the trait. In one such experiment, episomal replication again was detectable in the Line 316 control plant samples (no applied stress) taken at different times after germination (Table I). In general, the number of plants showing evidence of episomal replication increased as the age of the plants increased (Table I). This may have been due to effects of plant development, aging, or possibly the occurrence increasing with time and number of plant genome replication cycles.
To examine whether environmental stresses were required to induce episomal replication, plants of Line 316 also were propagated under various environmental stress conditions, including high light intensity, high temperature (heat), drought, and infection by CMV. Episomal replication was detected in plants subjected to each of these environmental stresses, but also was detected in a high percentage of the control plants that were no subject to any applied stress (Table I). In fact, the percentage of control plants in which recombinants could be detected by PCR either equaled or exceeded the percentages in plants that had been subjected to stress. A statistical analysis of the data in Table I

Deletions and Insertions in the Sequence of the Progeny Viral Genomes
PCR products obtained from selected plants of Line 316 or Line 318 were cloned and two clones were sequenced from each PCR product. In most, but not all cases, the sister clones had identical sequences. Analysis of the sequences revealed several trends ( Fig. 4 and Supplemental Fig. S2). The sequences of the PCR products indicated that all of the viruses contained a deletion within gene II, and in several viruses the deletions extended into gene III (see clones E318-4, E318-5, and E318-7 in Fig. 4). Although gene II can be deleted without affecting infectivity (Howarth et al., 1981), gene III has an essential role in cell-to-cell movement (Stavolone et al., 2005). Consequently viruses that sustained a deletion within gene III were likely dysfunctional. Some of the progeny viruses obtained from different plants had identical deletions (see E318-1F and E318-6Fa, as well as E318-5F and E318-6Fb), an indication that specific sequences might facilitate recombination events.
Interestingly, in some cases the viruses present in samples extracted before flowering had larger deletions than those taken from the same plants after flowering (see E318-4 vs. E318-4Ft/m/b, and E318-5 vs. E318-5F); in other cases, this could not be ascertained, since the plants either were negative by PCR for the presence of episomal virus before flowering (E318-1, E318-3, E318-6) or died after flowering (E318-7).
These examples are likely indicative of new recombination-excision events in a single plant, but we cannot rule out that they may have been generated by recombination of a defective virus (lacking large parts of genes II and III) with the homologous, integrated, intact CaMV genome sequences. It is important to note that such recombination events would not result in a wild type virus.
Many of the PCR products also contained insertions of DNA sequences that varied in length from 41 to 135 bp and were derived from the downstream, flanking transformation vector that had been incorporated into the plant genome during the transformation of Arabidopsis ( Fig. 4 and Supplemental Fig. S2). The presence of the vector DNA sequences was not correlated with environmental stress, as these sequences were recovered from plants that had not been subjected to stress ( Fig. 4; e.g., E316-10, E316-11, E318-F, E318-4) as well as from stressed plants (e. g., E316-6, E316-7 and E318-1). Some PCR products also contained the presence of small regions (6, 12, or 22 bp) derived from the upstream, flanking vector ( Fig. 4; yellow regions in samples E316-6, E316-7, and E318-4ma). Finally, some viruses contained additional vector DNA that was not present in viruses amplified from the same plant at an earlier timepoint ( Fig. 4; see E318-4Ft and the two clones of E318-4Fm vs. E318-4, respectively). This is further evidence that multiple recombination-excision events occurred in a single plant. Although a number of discreet deletion variants of CaMV were obtained, since CaMV is not seed transmissible (Tompkins, 1937;Squires et al., 2007), these variants would have to be created anew in every generation of plants.
Recombination junctions were identified in the 28 clones recovered from transgenic Arabidopsis plants; 12 were characterized by a region of microhomology of four to six nucleotides (Supplemental Fig. S2), whereas no evidence for microhomology existed in the other 16 clones. No consensus sequence emerged within the 12 junctions that exhibited microhomology. In seven of these clones, the recombination ends were imperfect, but some degree of homology could be seen. For example, in two of the clones with imperfect recombination ends (E318-4F-ba and E318-4F-bb), the sequence on one end of the recombination junction was TTACT, whereas the sequence on the other end was TTAACT. In the five other clones with imperfect recombination ends (E318-6Fb, E318-5a, E318-5b, E318-5Fa, and E318-5Fb), the sequence AGTA was found on one end, whereas on the other end AGTA was located two nucleotides from the recombination junction. In five clones the ends of the recombination junctions were identical: GGGT in clones E316-7a and E316-7b, TCCT in clones E318-3Fa and E318-3Fb, and CCGGG in clone E318-4Fma.

Recovery and Analysis of CaMV After Passage to Nontransgenic Plants
The deletions detected in the excised CaMV variants might affect their infectivity, since CaMV has very strict requirements for translation of its 35S RNA.
To investigate whether the excised CaMV variants were able to replicate and spread The nucleotide sequences of PCR clones isolated from the turnip plants had many similarities to the PCR clones isolated directly from the transgenic Arabidopsis: All clones sustained deletions within gene II ( Fig. 6  on the downstream junction, 15-1 and 3-2 contained a larger segment of the Agrobacterium vector than E316-7 (Supplemental Fig. S3). Interestingly, the changes in all three of these clones retained the gene II reading frame.
The population of viruses recovered directly from Arabidopsis did differ in some ways from the population of viruses recovered from turnips. For example, none of the deletions in viruses recovered from turnips extended into gene III. Since gene III is essential for infection (Stavolone et al., 2005), this class of deletions might be expected to be eliminated upon passage to nontransformed plants. However, we did find one deletion (clone 5-2, Fig. 6 , 1981). There was no evidence of any microhomology around the recombination junctions in the remaining seven clones (Fig. 3). Two of these clones (15-1 and 3-2) contained a single filler nucleotide, a thymidine, which could not be accounted for in either the vector DNA or CaMV DNA sequence, whereas no filler sequences were found in any of the other junctions.

DISCUSSION
Recombination associated with CaMV has been investigated in great detail.
Several studies showed that overlapping fragments of the viral genome, which between them contained the entire viral genome, delivered into the same cells could rapidly recombine to yield the infectious, wild type genome (Howell et al., 1981; www.plantphysiol.org on August 30, 2017 -Published by Downloaded from Copyright © 2011 American Society of Plant Biologists. All rights reserved. Lebeurier et al., 1982;Walden and Howell, 1983;Grimsley et al., 1986;Vaden et al., 1990;). In contrast, plasmids containing a single copy of the viral genome did not yield infectious viral progeny in inoculated plants (Walden and Howell, 1983).
Furthermore, CaMV isolates containing neutral genetic markers were observed to undergo abundant recombination within a single plant; recombinants accounted for up to 50 % of the viral population (Froissart et al., 2005). Later studies showed that recombination was possible between CaMV and transgenes derived from CaMV under strong and weak selection pressures (Schoelz and Wintermantel, 1993;Wintermantel and Schoelz,1996). Furthermore, the release of infectious viral genomes from longer than full-length CaMV genomes integrated into the genome of Brassica napus (Gal et al., 1991) indicated that infectious virus could arise from plant genomes. In most of these studies, the mechanism of recombination could be explained by template switches from one RNA species to another during the reverse transcription process, rather than by a homologous recombination event. However, it is difficult to discern the precise mechanism involved in many of the recombinants, because the recombination events could be explained by either mechanism. For example, Vaden and Melcher (1990) examined the recombination junctions of eight recombinant viruses formed between different CaMV isolates and concluded that their evidence supported a model involving reverse transcription as well as recombination between double-stranded DNA. Our study provides the clearest evidence for a model involving nonhomologous recombination, with strand breakage of DNA, either flanking or within the CaMV sequences contained in the chromosomal DNA, followed by ligation of the CaMV genome into a circular form. Furthermore, our study illustrates how an integrated, monomer copy of a pararetrovirus could be activated from host chromosomes and converted into an infectious form.
Two mechanisms might explain the release of an infectious CaMV virus from a monomer length transgene source. One mechanism involves strand breakage of DNA either flanking or within the CaMV sequences contained in the chromosomal DNA, followed by ligation of the CaMV genome into a circular form (Fig. 7). A second mechanism would require the synthesis of the CaMV reverse transcriptase (RT) and subsequent reverse transcription of viral RNAs into a DNA copy (Fig. 8)  It is important to illustrate the steps that would be necessary for RT-mediated release to evaluate its viability as a model for release of CaMV in our study (Fig. 8).
For example, there are two prerequisites for the RT-mediated mechanism of recombination to occur; one would be production of a polycistronic mRNA through the action of a fortuitously placed Arabidopsis promoter and the second is the production of the CaMV RT (P5). Of the two, the greatest hurdle involves the synthesis of the RT protein, because the RT cistron is in the third position on the CaMV portion of a putative polycistronic transcript (Fig. 8) In addition, it might be impossible to completely eliminate pathogen stress as a contributor to recombination in our study. The CaMV P6 product is expressed as a monocistronic mRNA from the 19S RNA promoter, and several studies have shown that expression of P6 in transgenic Arabidopsis plants elicits virus-like symptoms (Zijlstra and Hohn, 1992;Cecchini et al., 1997;Yu et al., 2003). It is possible that the expression of P6 might trigger the plant genome instability and the excision of the infectious CaMV DNA. We also noted that the incidence of episomal replication of CaMV increased with time or at later stages of maturation. It seems less likely that this was due simply to increased probability associated with increased number of cellular divisions, since there was much less cellular division that would have taken place between flowering and senescence than in the growth stages up to flowering.
Thus, it seems more likely that some event or process associated with age or maturation triggered the events leading to DNA strand breakage, subsequent religation and episomal replication. Ultimately, the factor(s) that contributed to excision and episomal replication, whether it is environmental-, pathogen-, or agerelated, remains to be examined.  (Kohli et al., 1999;Ho et al., 1999Ho et al., , 2000. The evidence for this hotspot was based on a limited number of intrachromosomal recombination events (Kohli et al., 1999), as recombination involving the 35S RNA promoter was detected in four of the 12 transgenic rice lines. Others have used this study to suggest that recombination associated with the 35S RNA promoter could lead to untenable risks to human health and to the environment. These risks include the accidental activation of plant genes or endogenous viruses that could somehow promote cancer in humans, or that the 35S RNA promoter might recombine with mammalian viruses such as HIV, with unexpected consequences (Ho et al., 1999(Ho et al., , 2000. The uncertainty associated with recombination involving the 35S RNA promoter has created an environment of fear within the general public that has damaged the credibility of genetically modified plant products. However, since the initial publication describing the CaMV 35S RNA promoter as a hotspot, new information has emerged regarding the potential for recombination in plants ( Gorbunova et al., 2000;Kovalchuk et al., 2003), indicating that the concept of the recombination hotspot as defined in Kohli et al. (1999) should be re-examined.
In our study we found that recombination centered around gene II nucleotide sequences could be detected by PCR in 29 of 48 transgenic plants (Table I) S2 and S3). In addition, recombination occurred at multiple sites within gene II; no single recombination junction predominated in the clones sequenced. Consequently, it would be incorrect to state that the gene II locus contains a recombination hotspot.
Instead, it appears that strand breakage and re-ligation involving CaMV sequences at a specific locus might occur at least once within the life of a plant and may occur more than once in some plants. Indeed, Kovalchuk et al. (2003) found that (2003) exceed that observed for recombination involving the 35S RNA promoter (Kohli et al., 1999), and collectively, these studies suggest that the 35S RNA promoter should not be considered a recombination hotspot.

Virus Source, Maintenance and Propagation
A full length, infectious clone of CaMV strain H7 (Daubert et al., 1984), a chimera generated between CaMV strains CM1841 and D4, was the source for transformation of A. thaliana. Strain H7 elicits extremely mild symptoms in turnip and is not aphid transmissible (Daubert et al., 1984). The CaMV clones pCaMV10 and H31 have been described previously (Gardner et al., 1981;Schoelz and Shepherd,1988). The CMV used for the environmental stress experiments was Fny-CMV, a CMV strain originally isolated from cucurbits that induces severe symptoms on A. thaliana (Lewsey et al., 2007). CMV was inoculated to A. thaliana using a buffered extract (50 mM phosphate, pH 7.0) taken from a CMV-infected Nicotiana benthamiana plant.

Transformation and Propagation of Arabidopsis
The D4-2 plants are homozygous for expression of P6 from the D4 strain of CaMV (Yu et al., 2003). The P6 gene is under the control of the 35S RNA promoter and rbcS terminator sequences. To insert the full-length CaMV genome into Arabidopsis, the cloned CaMV strain H7 was digested with SalI to release the vector pBR322 and the gel-purified viral DNA was ligated into a circular form. The fulllength, circular viral DNA was subsequently digested with XhoI and ligated into the XhoI site of the Agrobacterium plasmid pGreen 0229 (Hellens et al., 2000). The pGreen plasmid containing the full-length CaMV DNA was transferred into A.
A. thaliana ecotype C24 plants were transformed using the flower dip method (Clough and Bent, 1998). The T 0 seeds were collected from the infiltrated plants and

Extraction of Plant Nucleic Acids from Arabidopsis and PCR Analysis
For examination of episomal replication in Arabidopsis, DNA was extracted from total protein preparations used in parallel for western analysis prepared as described below. DNA was precipitated with addition of potassium acetate, centrifugation and addition of isopropanol to the supernatant and then used as a template for PCR. The forward primer for PCR was 5'-CAAAGACCCTTCGGAGT-3' and the reverse primer was 5'-CCAGAGGATCCTAATTCTTC-3', with an expected product size of 600 bp. PCR products were analyzed on 1.6 % agarose gels stained with ethidium bromide. DNAs present in bands in the gel were extracted using the QIAquick gel extraction kit (Qiagen, U.K.), and were cloned into pGem-T-

Extraction of Proteins and Western Blot Analysis
For protein analysis, samples were extracted from leaf material by grinding in Tris-HCl/SDS buffer and were subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Western blot analysis was carried out following transfer to nitrocellulose membranes as described by (Sambrook et al., 1989). The membranes were immunoprobed with antisera against P6 and the viral coat protein of CaMV.
The antisera to P6 were raised against a synthetic peptide that comprised the 10 amino acids on the C-terminus of P6 (Schoelz et al., 1991), whereas the antisera to the viral coat protein were generated against CaMV virions (Anderson et al., 1991).

Electron Microscopy
Immune-specific electron microscopy was performed on tissues of specific transgenic plant lines testing positive for episomal replication, using immuno-capture onto antibody-labelled grids, to confirm the presence of spherical particles of size appropriate for CaMV (50 nm). These particles were observed for samples E316-6, E316-7, E316-10 and E317-11. Anti-CaMV particle antibody was used to coat the electron microscope grids. Virus extracts were made using phosphate buffer pH 6.5, and a standard antibody-coated grid protocol was used. The grids were stained with ammonium molybdate pH 6.5, or 2 % uranyl acetate and viewed under the electron microscope (Phillips CM10).

Recovery of Infectious CaMV Virions and Viral DNA from Turnip Plants
Plants of transgenic lines 316 and 318 were grown for 45 days and the A. thaliana plants were bulked for isolation of CaMV virions. CaMV virions were partially purified as described previously (Schoelz et al., 1986) and were used for inoculation of turnip plants. Turnip plants were evaluated for CaMV infection by ELISA as described in (Schoelz et al., 1986) using polyclonal antisera directed against CaMV virions (Anderson et al., 1991) (Gardner et al., 1981). CaMV DNA isolated from individual turnip plants was examined on 1 % agarose gels stained with ethidium bromide. PCR was used to determine nucleotide sequence rearrangements that occurred within gene II of the infectious viral DNA. The forward PCR primer was 5'-CGCGGAATTCCAGTGCTTCATCC-TCTAATAC-3' and the reverse primer was 5'TAGGATTTTGGGATCCTAGCA-3'. The PCR-amplified DNA was subsequently cloned into pGem-T-Easy (Promega, Madison WI) and the nucleotide sequence was determined at the DNA Sequencing Core at the University of Missouri, Columbia MO.

Statistical Treatment of Data
The statistical analysis for episomal replication was carried out using a Generalized Linear Model, with a logit link and a dispersal estimation. Data analysis was done using the program GenStat Release 12.2 (copyright 2010, VSN International Ltd.).  showing PCR products generated from episomally replicating CaMV DNAs in plants subjected to various environmental stresses. PCR primers spanned the recombination junction between genes II and III (see Fig. 1  . Electron micrographs of CaMV particles. Anti-CaMV particle antibody was used to coat EM grids. Virus extract were made using phosphate buffer pH 6.5, and a standard antibody-coated grid protocol was used. The grids were stained with ammonium molybdate pH 6.5. Images are virions recovered from (A) Arabidopsis containing CaMV E316-10 (smaller PCR product from Fig. 2C), or (B) a control CaMV-H7 infected turnip. Bars = 50 nm.     29 polycistronic mRNA (mRNA A) would be initiated from the CaMV 35S RNA promoter, although it is unclear where this transcript would terminate, whether in the vector sequence or the Arabidopsis sequence flanking the CaMV insert. A third transcript (mRNA B) would have to be generated from a putative Arabidopsis promoter. This transcript would initiate in the Arabidopsis DNA, then continue through the vector sequence to CaMV gene II, ultimately ending at the 35S RNA termination sequence.
Step 2. The CaMV RT could only be produced through the action of the CaMV translational TAV, which would be synthesized in the transgenic plants through the action of the CaMV 19S RNA promoter. In principle, the TAV might redirect ribosomes to act on the polycistronic mRNA for synthesis of the RT.
Step 3. Nonetheless, if RT was produced, it might initiate synthesis of the first strand of CaMV DNA within mRNA A. This mRNA contains a nucleotide sequence complementary to a methionine tRNA, which serves as the primer binding site for first strand CaMV DNA synthesis by the CaMV RT in natural infections.
When the CaMV RT reached the 5' end of mRNA A, it would then need to switch templates to the 3' end of mRNA B. This step also corresponds to the replication process of CaMV in natural infections.
Step 5. As the RT reaches the end of CaMV sequences within gene II on mRNA B, it would need to initiate a second template switch back to the 3' end of mRNA A. This additional template switch would be necessary for production of an infectious virus by reverse transcription.