Invasion of the Arabidopsis genome by the tobacco retrotransposon Tnt1 is controlled by reversible transcriptional gene silencing

Abstract

Long terminal repeat (LTR) retrotransposons are generally silent in plant genomes. However, they often constitute a large proportion of repeated sequences in plants. This suggests that their silencing is set up after a certain copy number is reached and/or that it can be released in some circumstances. We introduced the tobacco (Nicotiana tabacum) LTR retrotransposon Tnt1 into Arabidopsis (Arabidopsis thaliana), thus mimicking the horizontal transfer of a retrotransposon into a new host species and allowing us to study the regulatory mechanisms controlling its amplification. Tnt1 is transcriptionally silenced in Arabidopsis in a copy number-dependent manner. This silencing is associated with 24-nucleotide short-interfering RNAs targeting the promoter localized in the LTR region and with the non-CG site methylation of these sequences. Consequently, the silencing of Tnt1 is not released in methyltransferase1 mutants, in contrast to decrease in DNA methylation1 or polymerase IVa mutants. Stable reversion of Tnt1 silencing is obtained when the number of Tnt1 elements is reduced to two by genetic segregation. Our results support a model in which Tnt1 silencing in Arabidopsis occurs via an RNA-directed DNA methylation process. We further show that silencing can be partially overcome by some stresses.

Figure 1.

Tnt1 is silenced in plants containing numerous elements. A, Schematic representation of Tnt1 and Tnt1-GUS T-DNA. The NdeI restriction site present in the Tnt1 coding sequence is unique. B, DNA and RNA gel-blot analyses of 2 and 10 μg of DNA and total RNA, respectively, extracted from mature rosette leaves of S15, S17, T2, and S14 primary transformants self-pollinated for two generations. DNAs were digested with NdeI. The Southern blot was probed with DNA complementary to the Gag region of Tnt1, and the northern blot was successively probed with the same DNA and DNA corresponding to a β-tubulin gene for a loading control. Each fragment of the Southern blot corresponds to at least one Tnt1 insertion locus (Tnt1 insertion #) in the hemizygous or homozygous state. C, RNA gel-blot analysis of 10 μg of total RNA extracted from mature rosette leaves of a T2-derived transformant (genotype T2.2.4.28.20; see Supplemental Fig. S2 for a genealogy of the plants used in this study), a S14-derived transformant (genotype S14.6.10), and a F1 hybrid resulting from a cross between the two. The blots were successively probed with DNA complementary to the Gag region of Tnt1 and a β-tubulin gene for a loading control. D, Histochemical staining of seedlings, leaves, and flowers of plants expressing a Tnt1-GUS fusion (genotype LTR-GUS; homozygous for the Tnt1-GUS insertion) or resulting from a cross between a T2-derived plant and LTR-GUS (genotype T2.2.52 × LTR-GUS). Similar staining patterns were observed for the following F1 lines: T2.2.53, T2.2.60, and T2.2.61 × LTR-GUS (data not shown).

Figure 2.

Tnt1 is inactivated by transcriptional gene silencing. A, Run-off experiments (see “Materials and Methods”) performed on 10⁶ to 10⁷ isolated nuclei of roots from Tnt1-expressing plants (genotype S14.6.10), F1 hybrids resulting from crosses between the C24 ecotype and Tnt1-GUS-expressing plants (genotype C24 × LTR-GUS), and a F1 hybrid resulting from crosses between a T2-derived plant and Tnt1-GUS-expressing plants (genotype T2.2.61 × LTR-GUS). In vitro labeled RNAs were hybridized with dot-blotted DNAs corresponding to Tnt1, GUS, 25S, and β-tubulin genes. The pKS vector DNA used to clone the different DNA fragments was used as a control. B, RNA gel-blot analysis of 10 μg of total RNAs prepared from mature rosette leaves of plants expressing Tnt1-GUS (LTR-GUS line), S14-derived plants, and T2-derived plants containing zero, four, or more than 20 Tnt1 copies, respectively. The blots were probed with an antisense radiolabeled RNA probe transcribed in vitro complementary to the LTR region of Tnt1. U6 hybridization was used as a loading control. nt, Nucleotides.

Figure 3.

Tnt1 silencing is reversed after segregation of the element. RNA gel-blot analyses of 10 μg of total RNA extracted from mature rosette leaves of plants derived from selfing T2-derived plants backcrossed several times to the C24 ecotype or from S14-derived plants crossed to LTR-GUS [S14-derived, containing only one homozygous insertion of Tnt1; genotype (S14.6.10 × LTR-GUS)3.17.1]. F3, F4, and F5 indicate the number of generations after segregation from high to low copy numbers of the element, as determined by Southern-blot analyses of DNAs digested by NdeI and probed with DNA complementary to the Gag gene of Tnt1 (data not shown). c, h, and e designate insertion loci as mentioned in Table I. The RNA blots were probed with DNAs complementary to the Gag gene of Tnt1. Ethidium bromide (EtBr) staining of the tRNA and 5S rRNA species is shown as a loading control. In A, genotypes of the plants are, from left to right, T2C24#6.1.n, T2C24#5.n, T2C24#5.1.n, and T2C24#5.1.1.n (three different plants); in B, genotypes are, from left to right, (S14.6.10 × LTR-GUS)3.17.1, T2C24#7.5.4.3, and T2C24#5.1.n. The origins of these plants are described in Supplemental Figure S2.

Figure 4.

Methylation states of Tnt1 cytosine sites. A, Localization of Tnt1 restriction sites and probes used. HincII (restriction site, GTPyPuAC) does not cleave when the 3′ C is methylated; the restriction site in Tnt1 is GTTGAC. HindIII (restriction site, AAGCTT) does not cleave when the C is methylated. HpaII (restriction site, CCGG) does not cleave when the second C is methylated. NdeI (restriction site, CATATG) is insensitive to methylation. HincII and HindIII both reveal non-CG methylation, while HpaII reveals both CG and non-CG methylation. B and C, DNA gel-blot analyses of 2 μg of DNA extracted from mature rosette leaves of S14-derived (genotype S14.6.10) and T2-derived (genotype T2.2.4.28.20) plants, a plant containing two reactivated copies of Tnt1 derived from backcrossing T2-derived plants to the C24 ecotype followed by selfing (reactivated; genotype T2C24#5.1.1), and a hybrid between T2- and S14-derived plants (genotype T2.2.4.28.20 × S14.6.10). DNAs were digested with the restriction enzymes indicated (HincII, HindIII, HpaII, and NdeI) to reveal the methylation state of non-CG sites in the Tnt1 promoter (B) or the methylation states of both CG and non-CG sites in the coding sequence of Tnt1 (C). Blots were probed with DNAs complementary to the regions indicated in A (Gag probe, A probe, B probe). The expression levels of Tnt1 (noted as + or −) were determined by northern blot (Supplemental Fig. S1).

Figure 5.

Methylation state of Tnt1-GUS cytosine sites. A, Localization of Tnt1-GUS restriction sites and probes used. The enzymes are described in the legend to Figure 4A. B, DNA gel-blot analyses of 2 μg of DNA extracted from rosette leaves of plants resulting from crosses between T2-derived plants and LTR-GUS plants (containing a homozygous Tnt1-GUS T-DNA). Genotypes of the plants are, from left to right, (T12C24#3 × LTR-GUS)4, (T12C24#2 × LTR-GUS)7, -9, and -10, (T2C24#5.1.1 × LTR-GUS)1, and (T2.2.4.28.20 × LTR-GUS)1. DNAs were digested with HincII, and blots were hybridized with the C probe (A) to reveal the methylation states of non-CG sites present in the LTR region. Expression of Tnt1-GUS (noted as + or −) was determined by histochemical staining (data not shown). C, DNA gel-blot analyses of 2 μg of DNA extracted from mature and young rosette leaves of a plant resulting from a cross between a T2-derived plant and LTR-GUS [genotype (T2.2.4.28.20 × LTR-GUS)13]. DNAs were digested with HincII or HpaII, and blots were hybridized with the C and D probes (A) to reveal the methylation state of non-CG sites in both LTR and coding sequences of Tnt1-GUS (HincII digestion; C+D probes) or CG and non-CG sites in the coding sequence of Tnt1-GUS (HpaII digestion; D probe). Expression of Tnt1-GUS (noted as + or −) was determined by histochemical staining (data not shown).

Figure 6.

MET1 is not involved in Tnt1 silencing. DNA and RNA gel-blot analyses of 2 μg of DNA and 10 μg of total RNA extracted from S14-derived plants (S14-derived; genotype S14.6.10), a T2-derived plant (T2-derived; genotype T2.2.4.28), and a T2-derived plant transformed with a MET1 antisense construct (T2 AS-MET1; genotype T2.2 AS-MET1). A and B, DNAs were restricted with HpaII and probed with the 180-bp centromeric probe (Finnegan et al., 1996; A) or digested with HpaII or HincII and NdeI and then hybridized with the probes indicated (B). C, The northern blot was probed with DNAs complementary to the Gag region of Tnt1. Ethidium bromide (EtBr) staining of the tRNA and 5S rRNA species is shown as a loading control.

Figure 7.

Silencing of Tnt1 is released in ddm1 mutants. RNA and DNA gel-blot analyses of 10 μg of total RNA and 2 μg of DNA extracted from mature rosette leaves of F2 plants (T2-derived ddm1/ddm1 or DDM1/DDM1) obtained by crossing a T2-derived plant (genotype T2.2.52) and a ddm1-2/DDM1 plant. The number of Tnt1 insertion loci (Tnt1 insertion #) was determined by Southern-blot analysis (data not shown). S14-derived plants (S14-derived; genotype S14.6.10.6.10) and T2-derived plants (T2-derived; genotype T2.2.4.28.23.10 in A and T2.2.4.28.20.3.1 in B) were used as controls. A, Northern blots were probed with DNAs complementary to the Gag gene of Tnt1. Ethidium bromide (EtBr) staining of the tRNA and 5S rRNA species is shown as a loading control. B, DNAs were digested by HindIII and NdeI to reveal the methylation states of non-CG sites or by HpaII to reveal the methylation states of both CG and non-CG sites. Blots were probed with DNAs complementary to the regions indicated in Figure 4A (Gag and A probes).

Figure 8.

Silencing of Tnt1 is released in nrpd1a mutants. Expression analysis of Tnt1 mRNA by RT-PCR in the nrpd1a-2 mutant containing six copies of Tnt1. Total RNAs were isolated from 5-week-old plants and used as templates for RT. A 336-bp DNA fragment corresponding to a region of the Tnt1 mRNA was amplified with specific primers in c/c h/h l/l nrpd1a-2 plants but not in c/c h/h l/l or wild-type (WT; ecotype Columbia) plants. Control primers were used to amplify a region of the Arabidopsis mRNA (after RT) encoding elongation factor 1α (GenBank accession no. AY039583). −RT designates negative control experiments in which reverse transcriptase was omitted before the final PCR amplification step. cDNA synthesis and RT-PCR were repeated 10 times on three different batches of plants grown independently, and identical results were obtained.

Figure 9.

Leaf wounding partially releases Tnt1 silencing. Histochemical staining performed on leaf discs from young and mature leaves of plants containing a homozygous Tnt1-GUS construct (LTR-GUS) or resulting from a cross between a T2-derived plant and LTR-GUS (T2-derived × LTR-GUS; genotype T2.2.52 × LTR-GUS). Leaves were stained immediately after wounding (0 h) or 48 h later (48 h).