Transcriptional silencing induced by Arabidopsis T-DNA mutants is associated with 35S promoter siRNAs and requires genes involved in siRNA-mediated chromatin silencing

Abstract

The utility of many T-DNA insertion mutant lines of Arabidopsis is compromised by their propensity to trigger transcriptional silencing of transgenes expressed from the CaMV 35S promoter. To try to circumvent this problem, we characterized the genetic requirements for maintenance of 35S promoter homology-dependent transcriptional gene silencing induced by the dcl3-1 (SALK_005512) T-DNA insertion mutant line. Surprisingly, even though DCL3 and RDR2 are known components of the siRNA-dependent transcriptional gene silencing pathway, transcriptional gene silencing of a 35S promoter-driven GUS hairpin transgene did occur in plants homozygous for the dcl3-1 T-DNA insertion and was unaffected by loss of function of RDR2. However, the transcriptional gene silencing was alleviated in dcl2 dcl3 dcl4 triple mutant plants and also by mutations in AGO4, NRPD2, HEN1 and MOM1. Thus, some T-DNA insertion mutant lines induce 35S promoter homology-dependent transcriptional silencing that requires neither DCL3 nor RDR2, but involves other genes known to function in siRNA-dependent transcriptional silencing. Consistent with these results, we detected 35S promoter siRNAs in dcl3-1 SALK line plants, suggesting that the 35S promoter homology-dependent silencing induced by some T-DNA insertion mutant lines is siRNA-mediated.

Figure 1

Transcriptional silencing of 35S promoter-driven transgenes is triggered by the dcl3-1 T-DNA insertion. Accumulation of GUS mRNA and of antisense strand GUS siRNAs in wild type (wt) and homozygous dcl3-1 plants carrying the L1 (a) or 306 (b) GUS transgene was determined using RNA gel blot analysis. Aerial tissue from three to four plants was pooled for RNA isolation, and each RNA sample is from separately pooled plants. The positions of 21- to 24-nucleotide RNA size markers are indicated. Ethidium bromide (EtBr) stained rRNA and the major RNA species in low molecular weight RNA are shown as loading controls. (a) The diagram shows the coding region of the silenced GUS sense transgene in line L1, plus the location of the sense and anti-sense probes used. The position and length in base pairs (bp) of the stem of the 306 GUS hairpin construct are also shown. Dcl3-1 RNA samples are from plants hemizygous (lanes 1–2) or homozygous (lanes 3–4) for the L1 transgene; the wt RNA samples (lanes 5–6) are from the homozygous L1 line. (b) The diagram shows the coding region of the 306 hairpin (hp) transgene drawn to illustrate the double-stranded configuration of the transcript. The locations of the hybridization probes are indicated. Probe 1 is the same as in Figure 1a. The wt RNA samples are from the homozygous 306 line. (c) Relative levels of transcription of the GUS transgene in wt (lane 1) and dcl3-1 mutant (lane 2) homozygous L1 GUS plants were determined in isolated nuclei. Nuclear transcripts were labeled with 32P by run-off transcription and then hybridized to slot blots loaded with plasmid DNA containing GUS, actin, and pUC19 empty vector sequence.

Figure 2

Silencing pathway mutations impair maintenance of 35S promoter homology-dependent transcriptional silencing triggered by the dcl3-1 T-DNA insertion. Accumulation of GUS mRNA (panels a; c–h) and of antisense strand 35S promoter siRNAs (panel b) in wild type (wt) and mutant plants carrying the 306 GUS locus was determined using RNA gel blot analysis. Plants were homozygous for the mutations shown except in the case of samples from dcl3-1 hemizygous plants (panel c, lane 1 and panel b, lanes 6–7). Ethidium bromide (EtBr) stained rRNA (panels a; c-h) and the major RNA species in low molecular weight RNA (panel b) are shown as loading controls. All samples in panels a and c-h are from plants that contain the 306 GUS transgene. The wt controls (panels a; c-h, last lane) are from the homozygous line 306 and were isolated from tissue pooled from three to four plants. Except as noted, other samples in panels a and c-h are from individual F2 progeny of crosses between our dcl3-1/306 line and the following lines: (a) dcl2-1 and dcl4-2; (c) rdr2-2, the rdr2/306 controls (lanes 6–7) are from a direct cross between rdr2-2 and 306; (d) nrpd2a-2, the nrpd2a-2/306 control (lane 6) is from pooled plants; (e) hen1-1, the hen1-1/306 controls (lanes 9–10) are from a direct cross between hen1-1 and 306; (f) ago4-1, the ago4-1/306 control (lanes 6) is from pooled plants; (g) mom1-2, the mom1-2/306 control (lane 6) is from pooled plants; and (h) ler. (b) The Columbia (Col) control (lane 1) is from pooled, non-transgenic plants and the 306 controls (lanes 2–3) are from pooled homozygous line 306 plants. The remaining samples are from individual plants and are from the homozygous dcl3-1 SALK line (lanes 4–5) and from F2 progeny of a cross between dcl3-1 and 306 (lanes 6–9). The positions of 21- to 24-nucleotide RNA size markers are indicated.

Figure 3

Model for 35S promoter homology-dependent TGS induced by the dcl3-1 T-DNA insertion. The results of the present study are summarized in the context of the siRNA-dependent transcriptional silencing pathway. RDR2 was not required for the dcl3-1 SALK line-induced TGS, suggesting that a complex insertion of the T-DNA in this line promotes direct production of 35S promoter (35S-P) dsRNA, either due to overlapping transcription or to transcription of an inverted repeat. Detection of 24-nt 35S promoter siRNAs in dcl3-1 hemizygous plants and 21- to 22-nt ones in dcl3-1 homozygous plants, plus the partial loss of the TGS in the dcl2 dcl3 dcl4 triple mutant, suggest that DCL3 is likely the enzyme responsible for siRNA production in initiation of the TGS, but that DCL2, DCL4, and possibly DCL1 maintain it in the absence of DCL3. Partial loss of the TGS in the hen1 mutant suggests maintenance involves HEN1 stabilization of the 35S promoter siRNAs. Impaired maintenance in the ago4 mutant suggests that siRNA-directed AGO4 methylation of 35S promoter DNA is also involved in maintenance. Because NRPD2 is a subunit of both POLIV and POLV, however, impaired maintenance in the nrpd2 mutant cannot distinguish whether the POLV role in methylation or the POLIV role in amplification of siRNA production or both are involved. Lastly, impaired maintenance in the mom1 mutant implicates MOM1-directed histone (H3K9) methylation in maintenance or in reinforcement of the TGS.