A CPF-like phosphatase module links transcription termination to chromatin silencing

Abstract

RNA-mediated chromatin silencing plays important roles in development and environmental response in many organisms. However, it is still unclear how the RNA, and/or the transcriptional processes producing them, trigger chromatin silencing. Through the study of developmental timing in plants we have found co-transcriptional RNA processing modulates a chromatin regulatory mechanism controlling quantitative expression of Arabidopsis FLOWERING LOCUS C (FLC). Here, we show that the phosphatase module of the mRNA 3′ processing CPSF complex is central to this mechanism. The phosphatase module components APRF1 (yeast Swd2/human WDR82), TOPP4 (Glc7/PP1) and LD (which we show is functionally equivalent to Ref2/PNUTS) all function genetically downstream of the CPSF-mediated cleavage and polyadenylation and are necessary for chromatin modification that results in FLC transcriptional silencing. LD has been previously shown to stoichiometrically co-associate in vivo with the histone demethylase FLD. We show both proteins result in removal of H3K4 monomethylation in the central section of the FLC locus. This links transcription termination to delivery of chromatin modifications that then influence subsequent transcription. Thus, physical association of the CPSF phosphatase module with chromatin modifiers generates a transcription-coupled silencing mechanism. The conservation of the factors involved suggest this mechanism may be generally relevant to transcription-mediated chromatin silencing.

Proximal termination generates a transcriptional state that determines the rate of establishment of Polycomb silencing

“…The model allows us to explain this counter-intuitive behaviour of FLD. Since FLD function is coupled to transcription and termination in the model, FLD targeting must be co-transcriptional, consistent with the molecular and genetic analysis in the accompanying paper (Mateo-Bonmatí, 2023). FLD levels at the locus must therefore be low in a Polycomb silenced state, where transcription events are infrequent.…”

“…FLD also physically associates in vivo with LD, a TFII-S domain containing protein and homolog of the CPF phosphatase subunit, PNUTS in human and Ref2 in yeast. This strong association of chromatin modifiers such as FLD with RNA Pol II regulators shown in the accompanying paper, identifies a mechanism where FCAmediated proximal cleavage and polyadenylation triggers FLD-mediated removal of H3K4me1 (Mateo-Bonmatí, 2023).The Arabidopsis H3K36 methyltransferase SDG8 is known to recognise H3K4me1, so H3K4me1 removal might be expected to affect H3K36 methylation. FLC upregulation in an fca mutant is rescued in an sdg8 fca double mutant (Fang et al, 2020), indicating that H3K36me3 may be essential to set up an active transcriptional state at FLC.…”

“…The separate roles of the co-transcriptional repression and Polycomb silencing as predicted by the model are further validated by examining a mutant of a Polycomb methyltransferase. The model clearly elaborates how FCA/FLD-mediated transcription-coupled repression, induced by cotranscriptional processing steps as described in the accompanying paper (Mateo-Bonmatí, 2023), can influence the timing of the digital switch to Polycomb silencing. The combination of molecular genetics and the theoretical modelling described here provide a quantitative description of how cotranscriptional processing can quantitatively set transcriptional states that determine rates of Polycomb silencing, a mechanism that is likely to be generally relevant in many organisms.…”

“…Genetic analysis also shows that the plant Polycomb methyltransferase CURLY LEAF (CLF) is essential for FCA-mediated silencing of FLC (Fang et al, 2020). However, despite the detailed molecular data on 3' processing-mediated silencing described above, and in the accompanying paper (Mateo-Bonmatí, 2023), how the various elements linked to Polycomb silencing was unknown, due in part to the complex feedbacks involved. To bridge this gap, we therefore constructed a computational model based on the above features (Fig.…”

“…The experimental validation of the model assumption of FLD targeting coupled to transcription termination, is presented in the accompanying paper (Mateo-Bonmatí, 2023), where we elucidate the links between the FLD complex and CPF components. By ChIP-qPCR analysis of FLD enrichment at FLC, comparing high transcription states induced by two very different genetic backgrounds (fca-9 and FRI) with a Polycomb silenced state (Col-0), we show that high FLD enrichment at FLC accompanies high transcription, irrespective of FCA function.…”