Differential concurrent transgene silencing associated with distinct types of nonspreading cytosine methylation

Transgene silencing is a common phenomenon in transgenic plants. Transgene expression may be blocked in the primary transformant, or the silencing may occur de novo in subsequent generations. Silencing is unpredictable in that it tends to affect some plants but not others, even if all plants carry the same construct. Silencing may occur at the transcriptional or post-transcriptional levels and may involve homology between multiple transgene copies or between the transgene and an endogenous gene. Indeed, in the latter case, both the transgene and endogenous gene can be silenced, a phenomenon termed cosuppression. Homology dependent silencing may reflect in cis or in trans DNA pair- ing, or interaction between DNA and RNA molecules (Gallie 1998). However, not all silencing events are homology dependent. There are increasing reports of silencing affecting single copy transgenes, which must involve distinct homology-independent mechanisms (Elyman and Vaucheret 1996). It has been suggested that silencing may result from position-dependent spreading of methylation or condensed chromatin structure from surrounding genomic DNA, or may represent a genomic scanning mechanism that identifies and methylates �invading� DNA. Indeed, silencing in plants is often correlated with increased methylation at the silenced locus. Different methylation patterns may be associated with different forms of silencing. For example, methylation of symmetrical (CpG and CpNpG) cytosines in the promoter region is often associated with transcriptional silencing, while symmetrical methylation in the coding region is often associated with post-transcriptional silencing. There have been sporadic reports of non-symmetrical cytosine methylation associated with silencing (Meyer et a!. 1994)

To investigate the silencing of single copy transgenes, and the relationship between methylation patterns and single copy silencing, we characterized over 100 plants from one plant line carrying a single copy of a three-gene transgenic locus. The transgenic line carried three heterologous transgenes: hpt, gusA and bar, each driven by a copy of the CaMV 35S promoter. We screened the primary transformant, its Rl progeny, and selected R2 and R3 progeny for GUS, HPT and PAT activities, for the presence of gusA, hpt and bar rnRNAs and (by nuclear run-on assays) for gusA, hpt and bar transcription. We also sequenced bisuiphite-treated DNA from each of the plants to determine the methylalion status of the transgenic locus.

Early on in the series of experiments, we found a consistent lack of PAT activity in the primary transformant and all progeny plants, corresponding to the absence of bar mRNA and bar transcription. On further investigation, we found that there was a destructive rearrangement at the 5� border of the transgenic locus, which inactivated the CaMV 35S promoter driving bar. However, the other two genes were intact and we found many plants with gusA and hpt expression. The primary transformant showed both GUS and HPT activities, but our screening procedure revealed a number of R 1 plants showing various forms of silencing. One R1 plant showed stable, epigenetic silencing of hpt, which was associated with symmetrical cytosine methylation in the promoter. The silencing was transmitted faithfully to all R2 and R3 plants, while the gusA gene remained active. Another R1 plant expressed both hpt and gusA, but of 11 R2 plants derived therefrom, one showed gusA silencing. This de novo silencing was stably transmitted to all R3 progeny, and the promoter region showed the same type of methylation seen in the hptsilenced Ri plant described above.

56. Differential concurrent transgene silencing associated with distinct types of nonspreading cytosine methylation X. Fu, A. Kohli, R.M. Twyman and P. Christou Molecular Biotechnology Unit, John Innes Centre, Colney Lane, Norwich NR4 7UH, UK Transgene silencing is a common phenomenon in transgenic plants. Transgene expression may be blocked in the primary transformant, or the silencing may occur de novo in subsequent generations. Silencing is unpredictable in that it tends to affect some plants but not others, even if all plants carry the same construct. Silencing may occur at the transcriptional or post-transcriptional levels and may involve homology between multiple transgene copies or between the transgene and an endogenous gene. Indeed, in the latter case, both the transgene and endogenous gene can be silenced, a phenomenon termed cosuppression. Homology dependent silencing may reflect in cis or in trans DNA pair- ing, or interaction between DNA and RNA molecules (Gallie 1998). However, not all silencing events are homology dependent. There are increasing reports of silencing affecting single copy transgenes, which must involve distinct homology-independent mechanisms (Elyman and Vaucheret 1996). It has been suggested that silencing may result from position-dependent spreading of methylation or condensed chromatin structure from surrounding genomic DNA, or may represent a genomic scanning mechanism that identifies and methylates �invading� DNA. Indeed, silencing in plants is often correlated with increased methylation at the silenced locus. Different methylation patterns may be associated with different forms of silencing. For example, methylation of symmetrical (CpG and CpNpG) cytosines in the promoter region is often associated with transcriptional silencing, while symmetrical methylation in the coding region is often associated with post-transcriptional silencing. There have been sporadic reports of non-symmetrical cytosine methylation associated with silencing (Meyer et a!. 1994) To investigate the silencing of single copy transgenes, and the relationship between methylation patterns and single copy silencing, we characterized over 100 plants from one plant line carrying a single copy of a three-gene transgenic locus. The transgenic line carried three heterologous transgenes: hpt, gusA and bar, each driven by a copy of the CaMV 35S promoter. We screened the primary transformant, its Rl progeny, and selected R2 and R3 progeny for GUS, HPT and PAT activities, for the presence of gusA, hpt and bar rnRNAs and (by nuclear run-on assays) for gusA, hpt and bar transcription. We also sequenced bisuiphite-treated DNA from each of the plants to determine the methylalion status of the transgenic locus. Early on in the series of experiments, we found a consistent lack of PAT activity in the primary transformant and all progeny plants, corresponding to the absence of bar mRNA and bar transcription. On further investigation, we found that there was a destructive rearrangement at the 5� border of the transgenic locus, which inactivated the CaMV 35S promoter driving bar. However, the other two genes were intact and we found many plants with gusA and hpt expression. The primary transformant showed both GUS and HPT activities, but our screening procedure revealed a number of R 1 plants showing various forms of silencing. One R1 plant showed stable, epigenetic silencing of hpt, which was associated with symmetrical cytosine methylation in the promoter. The silencing was transmitted faithfully to all R2 and R3 plants, while the gusA gene remained active. Another R1 plant expressed both hpt and gusA, but of 11 R2 plants derived therefrom, one showed gusA silencing. This de novo silencing was stably transmitted to all R3 progeny, and the promoter region showed the same type of methylation seen in the hptsilenced Ri plant described above. The most remarkable R 1 plant, however, showed two novel forms of silencing. The hpt transgene was silenced at the transcriptional level (nuclear run-on negative; Fig. 1) but the only evidence of methylation was at asymmetrical sites in the coding region (Fig. 2). There was no methylation in the promoter (data not shown). Moreover, all methylated residues were present on the same strand (the sense strand) and would therefore be transmitted to only half the products of meiosis. Indeed, of 20 R2 progeny plants examined, 10 showed hpt silencing and 10 showed hpt activity. Furthermore, this unusual transmission ratio was perpetuated. All R3 progeny of hpt-expressing R2 plants showed hpt expression, but characterization of R3 progeny of the hpt-silenced R2 plants once again showed that one half had inherited the silenced phenotype, while the other half expressed hpt. This is the first report of transcriptional silencing associated with a) coding region methylation, b) methylation at asymmetrical sites and c) hemimethylation, resulting in incomplete epigenetic inheritance. In the same line, the gusA gene underwent a distinct form of silencing. In the Ri plant, the gusA gene was silenced at three weeks post-germination but reactivated three weeks later. The reversible gusA silencing was post-transcriptional and corresponded to symmetrical methylation in the gusA coding region. After six weeks, the region was completely demethylated and expression reactivated (Fig. 3). This unstable form of silencing was not transmitted to progeny � all R2 plants showed gusA expression, although three of 40 R3 progeny underwent constitutive de novo gusA silencing. These results provided a great deal of information about the manner in which silencing mechanisms occur and interact in transgenic plants. Firstly, since Southern blot experiments were carried out on all plants, we knew that the transgenic locus was stable. Therfore, the variable silencing effects we observed occurred in the context of a stable single copy locus with constant position effects in all plants. Secondly, we characterized at least four distinct modes of silencing, including two that have not been reported before. Thirdly, we showed that silencing could occur de novo in a line of plants showing stable transgene expression. Finally, and perhaps most remarkably, we showed that two distinct forms of silencing could occur concurrently in the same plant in adjacent heterologous transgenes separated by less than 100 bp of DNA. There was no evidence of silencing states or methylation patterns spreading between the adjacent hpt and gusA genes. Further studies on this line of transgenic plants may reveal important information that will allow us to elucidate the mechanisms of transgene silencing, and perhaps derive constructs that will allow more efficient and stable transgene expression. References Elmayan, T. and H. Vaucheret, 1996. Expression of single copies of a strongly expressed 35S transgene can be silenced post-transcriptionally. Plant J. 9: 787-797. Gallie, DR., 1998. Controlling gene expression in transgenics. Curr. Opin. Plant. Biol. 1: 166-172. Meyer, p., I. Niedenhif and M. ten Lohuis, 1 994. Evidence for cytosine methylation of non-symmetrical sequences in transgenic Petunia hybrida. EMBO J. 13: 2084-2088.

The most remarkable R 1 plant, however, showed two novel forms of silencing. The hpt transgene was silenced at the transcriptional level (nuclear run-on negative; Fig. 1) but the only evidence of methylation was at asymmetrical sites in the coding region (Fig. 2). There was no methylation in the promoter (data not shown). Moreover, all methylated residues were present on the same strand (the sense strand) and would therefore be transmitted to only half the products of meiosis. Indeed, of 20 R2 progeny plants examined, 10 showed hpt silencing and 10 showed hpt activity. Furthermore, this unusual transmission ratio was perpetuated. All R3 progeny of hpt-expressing R2 plants showed hpt expression, but characterization of R3 progeny of the hpt-silenced R2 plants once again showed that one half had inherited the silenced phenotype, while the other half expressed hpt. This is the first report of transcriptional silencing associated with a) coding region methylation, b) methylation at asymmetrical sites and c) hemimethylation, resulting in incomplete epigenetic inheritance. In the same line, the gusA gene underwent a distinct form of silencing. In the Ri plant, the gusA gene was silenced at three weeks post-germination but reactivated three weeks later. The reversible gusA silencing was post-transcriptional and corresponded to symmetrical methylation in the gusA coding region. After six weeks, the region was completely demethylated and expression reactivated (Fig. 3). This unstable form of silencing was not transmitted to progeny � all R2 plants showed gusA expression, although three of 40 R3 progeny underwent constitutive de novo gusA silencing.

These results provided a great deal of information about the manner in which silencing mechanisms occur and interact in transgenic plants. Firstly, since Southern blot experiments were carried out on all plants, we knew that the transgenic locus was stable. Therfore, the variable silencing effects we observed occurred in the context of a stable single copy locus with constant position effects in all plants. Secondly, we characterized at least four distinct modes of silencing, including two that have not been reported before. Thirdly, we showed that silencing could occur de novo in a line of plants showing stable transgene expression. Finally, and perhaps most remarkably, we showed that two distinct forms of silencing could occur concurrently in the same plant in adjacent heterologous transgenes separated by less than 100 bp of DNA. There was no evidence of silencing states or methylation patterns spreading between the adjacent hpt and gusA genes. Further studies on this line of transgenic plants may reveal important information that will allow us to elucidate the mechanisms of transgene silencing, and perhaps derive constructs that will allow more efficient and stable transgene expression.

References

Elmayan, T. and H. Vaucheret, 1996. Expression of single copies of a strongly expressed 35S transgene can be silenced post-transcriptionally. Plant J. 9: 787-797.

Gallie, DR., 1998. Controlling gene expression in transgenics. Curr. Opin. Plant. Biol. 1: 166-172.

Meyer, p., I. Niedenhif and M. ten Lohuis, 1 994. Evidence for cytosine methylation of non-symmetrical sequences in transgenic Petunia hybrida. EMBO J. 13: 2084-2088.