1. Gene silencing and reactivation in transgenic rice 
S.P. Kumpatla, W. Teng, W.G. Bucholz and T.C. Hall

Institute of Developmental and Molecular Biology and Department of Biology, Texas A & M University, College Station, TX 77843-3155, USA

Although Agrobacterium-mediated transformation of cereals is now feasible, direct DNA delivery procedures remain in widespead use for the transformation of monocots. In particular, Biolistics, a transformation procedure that introduces naked DNA into plant cells, became the method of choice for cereal transformation in the recent years (Christou 1996). Although a large number of transgenic plants can be produced in a short span of time, plants generated using this method frequently contain multiple, often rearranged,

Rice Genetics Newsletter Vol. 14

copies of the introduced genes. We have investigated several hundreds of transgenic rice plants in our laboratory and found that such multicopy transformants frequently display aberrant segregation of the foreign genes or their expression. In the present study we have investigated a rice line that contained a complex transgene locus and displayed aberrant segregation of transgene expression.

Transgenic rice plants containing Btt cryIIIA (a synthetic gene that codes for an insecticidal protein) and bar (that confers resistance to the herbicide, bialaphos) genes were generated using biolistics-mediated transformation procedure as detailed before (Buchholz et al. 1997). Three independent co-bombardments of rice (Oryza sativa L., ssp. japonica cv. Taipei 309) embryos with plasmids JKA (that consists of a 1794 bp synthetic Btt cryIIIA gene fused to a CaMV 35S promoter bearing an alfalfa mosaic virus translational enhancer and a nopaline synthase, nos, terminator) and pUbil-bar, a selectable marker plasmid that contains a bar gene under the control of a maize ubiquitin (Ubil) promoter and a nos terminator, resulted in ten bialaphos-resistant and PCR-positive (with Btt cryIIIA-specific primers) transgenic plants. Genomic DNA blots of these ten primary transformants revealed multiple, full length and rearranged, fragments and three distinct hybridization patterns when the Btt cryIIIA coding region was used as a probe (data not shown). On the basis of these patterns, the plants were designated group I (plants JKA 50, 51, 52, 53 and 54), group II (JKA 56 and 58) and group III (JKA 59, 60 and 61), consistent with the three bombardments used to generate them. Although all of the ten plants expressed the bar gene (as indicated by their resistance to bialaphos), only one of them, JKA 52, expressed the Btt cryIIIA gene, based on sensitive RNase protection assays (RPAs). However, when the seedlings obtained from the selfed seed of this line (JKA 52) were subjected to insect bioassays, heterogeneous results were obtained. These results, combined with RPAs on the progeny plants, revealed the non-Mendelian segregation of Btt cryIIIA expression in the progeny, since the data were deviant from the expected 3:1 ratio (analyses of progeny revealed that JKA 52 had the transgene at a single locus). In order to conveniently evaluate a large number of progeny plants for transgene expression and to understand the basis of the observed lack of expression in several progeny plants, the bar gene was chosen for further detailed analysis as its expression can be easily assayed by testing the leaves for herbicide resistance. Of 108 progeny (R1) seedlings obtained from selfing of JKA 52, only 53 were resistant and 55 were sensitive (a 1: 1 ratio), indicating that the bar gene also segregated aberrantly in this transgenic line.

Fig. 1 A shows the pUbil-bar construct, probes used for genomic DNA blot analyses, and the fragments expected to be present following digestion with the indicated restriction endonucleases. Genomic blot analysis of ten primary transformants using a bar-coding region-specific probe (Fig. 1A, probe 2) revealed 3 distinct hybridization patterns (Fig. 1B shows the data for representative lines JKA 52, 58 and 60), suggesting that they were derived from three independent transformation events. As mentioned before, the plants fell into the same three groups when the Btt cryIIIA coding region was used as a probe on the genomic blots. Two things are evident from Fig. 1B. First, all three lines displayed multiple, rearranged, fragments. Second, only JKA 52 contained the expected (full length)

Research Notes 157

Fig. 2. Genomic DNA blot analysis of primary transformants. (A) Proportional map of pUbil-bar showing functional regions. Restriction sites used for genomic analysis and expected fragments are shown at the top and locations of HpaII/MspI sites and expected fragments arc shown below. Solid bars denote radioactive fragments used as hybridization probes. Genomic DNA was digested with HindIII and KpnI (B) or BamHI and EcoRI (B/E) or undigested (U) (C) and hybridized with probe 2. The blot shown in (B) was sequentially stripped and re-hybridized with probes 3 (promoter) and 4 (bar coding region). P and C indicate fragments hybridized to probes 3 and 4, respectively (P/C, strong signal; p/c, weak signal). P and pcpositions of fragments that hybridized only to probe 3; wt, untrasformed T309: 2x, 2 copy reconstruction of pUbil-bar DNA: arrows indicate locations of expected fragments.

 2552 bp fragment. In order to further confirm the authenticity of the 2552 bp fragment in JKA 52 and to understand the promoter and coding region composition of the observed bands in Fig. 1B, the blot was sequentially tripped and hybridized with a probe corresponding to the 5' portion of Ubi1 promoter (Fig. 1A, probe 3) followed by a probe that spans the 3' portion of the bar coding region (Fig. 1A, probe 4). These analyses indicate the promoter (P/p) or coding region (C/c) composition of the HindIII/KpnI fragments and that rearrangements are involved in generating the chimeric fragments. However, in all lines, the bar gene was integrated into the genome, since strong hybridization was observed to high molecular weight DNA (lanes marked U in Fig. 1C). Although the presence of expected 852 bp BamHI/EcoRI fragments in all lanes (Fig. 1C) indicate the exisi- 158 Rice Genetics Newsletter Vol. 14 Fig. 2. Genomic blot analysis of JKA 52 progeny. (A) Genomic DNA from JKA 52 and 12 R, plants was digested with HindIII and KpnI and hybridized with probe 2. The expected 2552 hp fragment containing the UbiI promoter and the bar coding regon is indicated by an arrow. (B) Methylation status of promoter in silenced and reactivated plants. Genomic DNA was digested with either HpaII or MspI and hybridized with probe 1. Note the absence of 756 bp and 263 bp fragments in the DNA of bialaphos-sensitive plants digested with HpaII and the presence of these fragments when DNA was digested with MspI. These fragments were partially (52-10-8) or fully (52-10-16) restored in R2 seedlings grown in the presence of AzaC, wt, DNA from untrasformed T309: 2x, two copy reconstruction of pUbil-bar DNA: *denotes that bar is physically present but silenced: arrows indicate locations of expected fragments.
Table 1. Reactivation of bar gene expression in R2 seedlings by 5-azacytidine (AzaC) treatment
Control (No AzaC)
50 mg/L AzaC
75 mg/L AzaC
R1 plant number Resistant  Sensitive Total No. tested Resistant Sensitive Total No. tested Resistant Sensitive Total No. tested
52-6* 35 (76%) 11 (24%) 46 ND ND ND ND ND ND
52-9* 0 41 41 8 (24%) 26 (76%) 34 15 (58%) 11 (42%) 26
52-10* 0 42 42 12 (29%) 30 (71%) 42 19 (70%) 8 (30%) 27
52-15* 0 41 41 9 (25%) 27 (75%) 36 ND ND ND

Plant 52-6 contained a functional bar gene; *, plants containing an intact but silenced bar gene; ND. not determined.

Research Notes 159

ence of intact bar coding region and terminator, reference to Fig. 1B indicates that only JKA 52 has the full length Ubi1-bar-nos fragment.

A random sample of 20 JKA 52 progeny plants were subjected to genomic DNA blot analysis (Fig. 2A shows the data for 12 plants) to evaluate whether the bar gene might have been inactivated due to DNA rearrangements). As expected, those that lacked the gene were sensitive to bialaphos and most that had the gene were resistant. However, three of the plants that had the gene (plants 52-9, 52-10 and 52-15) were sensitive to bialaphos. After confirming that rearrangement of the terminator was not responsible for bar gene inactivation (data not shown), methylation analysis was performed on these three lines using a resistant R1 line (52-6) and parent (JKA 52) as controls (Fig. 2B shows promoter methylation analysis). A comparison of HpaIII and MspI hybridization patterns indicates (i) the absence or diminution of the expected 756 bp and 263 bp promoter fragments in the bialaphos sensitive lines but their presence in the resistant lines (when HpaIII was used for digestion) and (ii) a shift in the mobility of several fragments due to methylation. Analysis of coding region methylation also indicated the involvement of methylation but at least a single copy of the expected fragments were present in the sensitive lines (data not shown). These results established that silencing of the bar gene is methylation-based and that promoter inactivation most probably led to its silencing. Nuclear run-on assays confirmed that transcriptional inactivation was responsible for bar gene silencing (data not shown).

To further confirm our view that transcriptional silencing of the bar gene resulted from cytosine methylation, R2 progeny of the silenced plants were grown in the presence of 5-azacytidine (AzaC) (Table 1). Without AzaC, all the seedlings from the silenced lines were also sensitive to bialaphos indicating the meiotic stability of the silenced state. However, in the presence of AzaC, 24 to 70% of the seedlings were resistant to bialaphos, indicating reactivation of the bar gene, which correlated with the demethylation of the promoter (see lane 52-10-16 in Fig. 2B).

These findings strongly support the concept that gene silencing can result from the introduction of multiple homologous sequences, a feature common to direct DNA delivery-mediated transformation procedures, and emphasize the need for the careful molecular analysis and multi-generation testing of the transgenic lines before they are subjected to large scale programs or plant breeding purposes.

References

Buchholz, W.G., W. Teng, D. Wallace, J. Ambler and T.C. Hall, 1997. Production of transgenic rice (Oryza sativa L. ssp. japonica cv. Taipei 309). In Methods in Molecular Biology: From Virus Isolation to Transgenic Resistance. (Foster, G. and Taylor, S., eds.) Christou, P. 1996. Transformation technology. Trends Plant Sci. 1: 423-431.