Promoter-associated RNA is required for RNA-directed transcriptional gene silencing in human cells

Abstract

siRNAs targeted to gene promoters can direct epigenetic modifications that result in transcriptional gene silencing in human cells. It is not clear whether the antisense strand of the siRNAs bind directly to DNA or to a sense-stranded RNA transcript corresponding to the known promoter region. We present evidence that an RNA polymerase II expressed mRNA containing an extended 5' untranslated region that overlaps the gene promoter is required for RNA-directed epigenetic modifications and transcriptional silencing of the RNA-targeted promoter. These promoter-associated RNAs were detected by their hybridization to the antisense strand of the complementary promoter-directed siRNA. Antisense phosphorothioate oligodeoxynucleotides were used to degrade the promoter-associated RNA transcripts, the loss of which abrogated the effect of siRNA-mediated transcriptional gene silencing, as well as the complexing of the siRNA with the silent state histone methyl mark and the promoter-associated RNA. These data demonstrate that low-copy promoter-associated RNAs transcribed through RNAPII promoters are recognized by the antisense strand of the siRNA and function as a recognition motif to direct epigenetic silencing complexes to the corresponding targeted promoters to mediate transcriptional silencing in human cells.

Fig. 1.

Detection and characterization of a EF1a promoter-associated RNA variant. (A) 5′ biotin-linked antisense RNAs targeted to the EF1a promoter elute with a promoter-associated RNA. Cultures were treated with either EF52 siRNA containing a 5′ biotin-linked sense annealed to the antisense (Control) or 5′ biotin-linked antisense RNA alone (EF52) (100 nM Lipofectamine 2000). Samples from the preelution step (Input) are shown along with the post-biotin/avidin elution (Elution). The elutes were collected 24 h after transfection, DNase treated, subject to RT with either a random hexamer primer (dTT) (BioRad; iScript), or EF1a-specific primer (EF1a) that spans the EF1a promoter/exon 1 junction (SI Table 2, no. 804) and PCR amplified with EF1a-specific primers (SI Table 2, 803/804). (B) The 5′ biotin tagged antisense RNA eluted EF1a promoter-associated RNA is sensitive to RNase A and resistant to RNase H treatment. Sample lanes represent elutes from cultures treated with (1) EF52 sense RNA + 5′ biotin linker, (2) EF52 antisense RNA + 5′ biotin linker, and (3) EF52 sense RNA plus 5′ biotin linker annealed to the EF52 antisense plus 5′ biotin linker. The resultant elutes underwent RT (BioRad; iScript) followed by PCR (SI Table 2, EF1a-specific primers 803/804) after the various RNase treatments. (C) Characterization of the EF1a promoter-associated RNA as determined by bidirectional RT-PCR. RT was performed on DNase treated 293T cell RNA with various primers either sense (1) or antisense (2) (SI Table 2, 803 or 805, respectively) followed by PCR with both primers. (D) RT with primer 2 followed by PCR with primers 1 and 2 generates a ≈156-bp band corresponding to the EF1a promoter whereas RT with primer 1 followed by PCR with primers 1 and 2 does not. The results from three independent cultures are shown. (E) 5′ and 3′ RACE analysis of the EF1a promoter-associated RNA was performed on 293T cell RNA with EF1a primers specific for the promoter-associated RNA. The 5′ RACE analysis was performed with primer 5 (SI Table 2, EFpRNARev), whereas the 3′ RACE analysis was performed with the 3′RACE F primer (SI Table 2). The resulting RACE products were cloned, sequenced, and shown to be spliced and processed accordingly and to contain a 5′ UTR which overlaps ≈230 bp of the EF1a promoter.

Fig. 2.

Promoter-associated RNA variant and mRNA expression in siRNA treated cultures. (A) The EF1a promoter-associated RNA variant is reduced (P = 0.352) along with EF1a mRNA expression (P = 0.038) when the promoter is targeted by the EF52 siRNA (EF1a) relative to the CCR5-specific siRNA (Control). Measurements of mRNA or promoter-associated RNA levels from three independent transfections standardized to GAPDH expression with the standard error of the means (SEM) are shown with P values reported for a single-sided F test (mRNA) and double-sided t test (promoter-associated RNA). (B) Treatment with an EF1a-specific (EF1a) antisense phosphorothioate ODN suppresses EF1a promoter-associated RNA variant expression whereas the CCR5-specific ODN (Control) does not. 293T cells were transfected with either an ODN targeted to the EF1a promoter (EF1a) or to the CCR5 promoter (CCR5) (100 nM Lipofectamine 2000). Twenty-four hours later, the cultures were collected, cell RNA was isolated, Dnase was treated, and EF1a promoter-associated RNAs were assessed by qRT-PCR relative to GAPDH expression. Triplicate treated cultures were assessed and the standard error of the means is shown. (C) Treatment of cells with the EF1a promoter-targeted ODN (EF1a) suppresses siRNA (EF52)-mediated transcriptional silencing of the EF1a gene, whereas the control CCR5 ODN (Control) does not. EF1a mRNA expression was measured by qRT-PCR at 18 h after siRNA transfection in untreated (Mock) and cells treated with either EF52 (EF52) or control (CCR5) siRNAs, which was 42 h after ODN transfection and standardized to GAPDH expression. Results are from triplicate transfections, and the standard deviations are shown.

Fig. 3.

The EF1a promoter-associated RNA is required for RNA-directed epigenetic modifications at the EF1a promoter. (A) The EF1a promoter-targeted ODN (Control) suppresses EF52 siRNA-directed H3K9me2 at the siRNA-targeted EF1a promoter, whereas treatment with the CCR5 promoter-specific ODN followed by treatment with EF52 siRNA (EF1a) does not. Results from two independent experiments with the respective ranges are shown. (B) Treatment of cultures with the EF1a promoter-targeted ODNs (EF1a) suppresses the dual pull-down (ChIP) of H3K9me2 followed by 5′ biotin-linked antisense RNA, whereas the treatment with a CCR5 promoter-specific ODN (CCR5) does not. Undiluted inputs (Input), no antibody controls (No Ab), a positive PCR control using an EF1a containing plasmid (+), and molecular weight standards are also (MW) shown. Results from a single experiment are shown and demonstrate that the H3K9me2, 5′ biotin-linked antisense EF1a-specific RNA, and EF1a promoter (DNA) coimmunoprecipitate. (C) The EF1a promoter-associated RNA variant coimmunoprecipitates with the H3K27me3 epigenetic mark and the 5′ biotin-linked antisense EF1a-specific RNA. Dual pull-down assays were performed on 293T cultures treated with either the EF1a-specific (EF1a) or CCR5-specific (CCR5) ODNs (100 nM) transfected 4 h later with the 5′ biotin-linked EF1a-specific antisense RNA (EF52as, 100 nM). Twenty-four hours later, cultures were collected and an H3K27me3 immunoprecipitation (ChIP) was performed followed by a biotin/avidin pull-down on the H3K27me3 elutes. The resultant elutes were then subject DNase treatment, and a reaction was carried out with or without reverse transcriptase (+RT or −RT, respectively). The samples were then subject to nested PCR specific for the promoter-associated EF1a RNA variant. The result from one of two such experiments is shown.

Fig. 4.

Model for RNA-directed TGS in human cells. (A) The promoter-associated RNA model of RNA-mediated TGS proposes that a variant species of mRNA, a promoter-associated mRNA, essentially containing an extended 5′ UTR, is recognized by the antisense strand of siRNAs or possibly endogenous antisense RNAs during RNAPII-mediated transcription of the RNA-targeted promoter. (B) The antisense strand of the siRNA might then guide a putative transcriptional silencing complex (possibly composed of DNMT3A, Ago-1, HDAC-1, and/or EZH2) to the targeted promoter where histone modifications result and the initial gene-silencing event. (C) The initial silencing event or prolonged suppression of the siRNA-targeted promoter may result in heterchromatization of the local siRNA-targeted genomic region and is not, based on these data, thought to be the result of slicing of the low-copy promoter-associated RNA but rather due to a recruitment of chromatin remodeling factors or complexes to the targeted promoter that result in the gene silencing event.