Chromodomains direct integration of retrotransposons to heterochromatin

Abstract

The enrichment of mobile genetic elements in heterochromatin may be due, in part, to targeted integration. The chromoviruses are Ty3/gypsy retrotransposons with chromodomains at their integrase C termini. Chromodomains are logical determinants for targeting to heterochromatin, because the chromodomain of heterochromatin protein 1 (HP1) typically recognizes histone H3 K9 methylation, an epigenetic mark characteristic of heterochromatin. We describe three groups of chromoviruses based on amino acid sequence relationships of their integrase C termini. Genome sequence analysis indicates that representative chromoviruses from each group are enriched in gene-poor regions of the genome relative to other retrotransposons, and when fused to fluorescent marker proteins, the chromodomains target proteins to specific subnuclear foci coincident with heterochromatin. The chromodomain of the fungal element, MAGGY, interacts with histone H3 dimethyl- and trimethyl-K9, and when the MAGGY chromodomain is fused to integrase of the Schizosaccharomyces pombe Tf1 retrotransposon, new Tf1 insertions are directed to sites of H3 K9 methylation. Repetitive sequences such as transposable elements trigger the RNAi pathway resulting in their epigenetic modification. Our results suggest a dynamic interplay between retrotransposons and heterochromatin, wherein mobile elements recognize heterochromatin at the time of integration and then perpetuate the heterochromatic mark by triggering epigenetic modification.

Figure 1.

Retrotransposons with CHDs and CR motifs are preferentially located in gene-poor, transposon-rich regions of their host genomes. (A) The genomic distribution of M. grisea MAGGY (group I) insertions. For each of the 75 MAGGY insertions in the completed genome sequence (Dean et al. 2005), 20-kb windows at the site of insertion (10 kb upstream and downstream) were surveyed for genes and transposable elements. Seventy-five randomly selected sites were similarly surveyed as a control. (B) The distribution of group II, CR, and Tos17 insertions on chromosome 1 of O. sativa. Insertions of group II chromoviruses and Tos17 are distributed along the chromosome arms; however, as indicated in C below, these two element families occupy different genomic contexts. Only a few CR elements are found in the assembled rice genome sequence and these are near the pericentromeric regions (gap in chromosome). Many other CR insertions have been revealed by FISH analysis or are present in the recent sequence of the centromere from chromosome 8 (Cheng et al. 2002; Nagaki et al. 2004). (C) Os (class II) insertions are in transposon-rich, gene-poor regions scattered across the 12 rice chromosomes. For each Os insertion site, a 40-kb window (20 kb upstream and 20 kb downstream) was surveyed for genes and transposable elements. Windows containing Os insertions have more transposable elements and fewer genes than random windows or windows centered on Tos17 insertions, a retrotransposon that integrates preferentially into gene-rich regions of the rice genome (Miyao et al. 2003).

Figure 2.

Subnuclear localization of retrotransposon CHDs and the CR motif. (A) YFP fusion proteins expressing retrotransposon CHDs or the CR motif localize to sites of heterochromatin in A. thaliana cells. Constructs expressing fusion proteins between YFP and either group I or group II CHDs or the CR motif were transformed into A. thaliana suspension cell protoplasts and visualized by confocal microscopy. The fusion proteins formed punctate foci in the nucleus and were enriched in the nucleolus. Localization was coincident with a fusion between CFP and TFL2—the A. thaliana HP1 homolog—but not the negative control, namely the Tnt1 IN C terminus. (B) Mutations in a conserved aromatic amino acid in group I and group II CHDs that is predicted to interact with the methyl group on H3 K9 abrogate subnuclear localization of YFP-CHD fusion proteins. Residues modified in the representative group I and group II CHDs are highlighted in red in Supplemental Fig. S1. In each case, residues were mutated to valine.

Figure 3.

The MAGGY CHD recognizes histone H3 methyl-K9. (A) MAGGY CHD interacts with histone H3. GST-MAGGY CHD and GST-Tma CHD fusion proteins were mixed with a calf thymus histone extract and then pulled down using glutathione agarose beads. Both the input and pull-down reactions were separated by SDS-PAGE and visualized by SYPRO Ruby protein gel stain. H3, as indicated by the asterisk, was enriched in the pull-down with the MAGGY CHD. No interaction was observed with the MAGGY CHD RW mutation or with wild-type or mutant variants of the Tma CHD. (B) The MAGGY CHD specifically interacts with H3 dimethyl-K9. Biotin-labeled histone peptides were incubated with His₆-tagged MAGGY CHD, and then pulled-down with streptavidin agarose beads. Pull-down reactions were separated by SDS-PAGE and transferred to nylon membranes. The presence of the CHD was detected using an anti-His₆ antibody. The MAGGY CHD specifically interacts with histone H3 dimethyl-K9, and not with H3 dimethyl-K4. Mutations in conserved residues of the CHD abrogate the interaction. (C) MAGGY CHD localizes to sites of H3 methyl-K9 in S. pombe. The MAGGY CHD localizes to punctate sites within the nucleus as does Chp2, a CHD protein found at sites of centromeric heterochromatin in S. pombe (Sadaie et al. 2004) (top). Colocalization of the MAGGY CHD and Chp2 is observed in a swi6-115 strain, which lacks the S. pombe HP1 homolog (Nakagawa et al. 2002). The swi6-115 strain should allow expression of genes in centromeric heterochromatin and was used in transposition assays that require expression of a marker gene carried by Tf1 (see Fig. 4).

Figure 4.

Target site choice of the Tf1 retrotransposon in S. pombe is altered by adding the MAGGY CHD to integrase. (A) Transposition and cDNA recombination frequencies of Tf1 and Tf1-Mac. The MAGGY CHD was fused to the C terminus of Tf1 integrase, creating Tf1-Mac. Tf1-Mac transposes, but at a lower frequency than wild-type Tf1. Integrase frameshift mutants (Tf1-fs and Tf1-Mac-fs) abolish Tf1 or Tf1-Mac integration, and therefore serve to infer frequencies of cDNA recombination. (B) A screen for insertions of Tf1 and Tf1-Mac in heterochromatin. A diagram of S. pombe centromeric repeats and mating type region depicting the location of the PCR primers used to identify heterochromatic Tf1 insertions. (C) Targeting of Tf1-Mac to heterochromatin. Centromeric heterochromatin in S. pombe, which is enriched in histone H3 methyl-K9, becomes a target site for Tf1-Mac in the swi6-115 strain. In the swi6-115 strain, there are eight integration events into heterochromatin out of 2200 Tf1-Mac transposition events. This number of transposition events includes elements incorporated into the genome by either integration or recombination. The number that arose specifically by integration was inferred by normalizing the data using the value for the ratio of transposition to recombination (4:1). The normalized heterochromatin targeting frequency is eight out of 1650 integration events. Tf1-Mac does not target into heterochromatin in the clr4Δ strain, which lacks H3 methyl-K9. (D) Native Tf1 target specificity is retained by the Tf1-Mac construct. Fusion of the MAGGY CHD to Tf1 does not abolish Tf1’s native targeting specificity in the swi6-115 strain. Sites of insertion for nine randomly selected Tf1 integration events were determined by inverse PCR. Eight were found within promoters and one near a transcription terminator, consistent with previously documented Tf1 targeting patterns (Behrens et al. 2000; Singleton and Levin 2002).

Figure 5.

Self-perpetuating model for heterochromatin expansion. Repetitive elements such as retrotransposons produce dsRNAs that trigger the RNAi pathway. This results in the targeting of DNA and histone methyltransferases to retrotransposons resident on host chromosomes. Histone modification (e.g., H3 methyl-K9) establishes heterochromatin and, in turn, creates recognition sites for retrotransposon integrases. The targeting of retrotransposons to the domain of heterochromatin serves to reinforce the epigenetic mark.