The transcriptional termination of unstable non-coding RNAs (ncRNAs) is poorly understood compared to coding transcripts. We recently identified ZC3H4-WDR82 ("restrictor") as restricting human ncRNA transcription, but how it does this is unknown. Here, we show that ZC3H4 additionally associates with ARS2 and the nuclear exosome targeting complex. The domains of ZC3H4 that contact ARS2 and WDR82 are required for ncRNA restriction, suggesting their presence in a functional complex. Consistently, ZC3H4, WDR82, and ARS2 co-transcriptionally control an overlapping population of ncRNAs. ZC3H4 is proximal to the negative elongation factor, PNUTS, which we show enables restrictor function and is required to terminate the transcription of all major RNA polymerase II transcript classes. In contrast to short ncRNAs, longer protein-coding transcription is supported by U1 snRNA, which shields transcripts from restrictor and PNUTS at hundreds of genes. These data provide important insights into the mechanism and control of transcription by restrictor and PNUTS.

SUMMARYThe human genome encodes thousands of non-coding RNAs. Many of these terminate early and are then rapidly degraded, but how their transcription is restricted is poorly understood. In a screen for protein-coding gene termination factors, we identified ZC3H4. However, its depletion causes upregulation and extension of hundreds of unstable transcripts, particularly antisense RNAs and those transcribed from so-called super-enhancers. These loci are occupied by ZC3H4, suggesting that it directly functions in their transcription. Consistently, engineered tethering of ZC3H4 to reporter RNA promotes its degradation by the exosome. ZC3H4 is metazoan-specific - interesting when considering its impact on enhancer RNAs that are less prominent in single-celled organisms. Finally, ZC3H4 loss causes a substantial reduction in cell proliferation, highlighting its overall importance. In summary, we identify ZC3H4 as a factor that plays an important role in restricting non-coding transcription in multi-cellular organisms.

The human genome encodes thousands of non-coding RNAs. Many of these terminate early and are then rapidly degraded, but how their transcription is restricted is poorly understood. In a screen for protein-coding gene transcriptional termination factors, we identified ZC3H4. Its depletion causes upregulation and extension of hundreds of unstable transcripts, particularly antisense RNAs and those transcribed from so-called super-enhancers. These loci are occupied by ZC3H4, suggesting that it directly functions in their transcription. Consistently, engineered tethering of ZC3H4 to reporter RNA promotes its degradation by the exosome. ZC3H4 is predominantly metazoan -interesting when considering its impact on enhancer RNAs that are less prominent in single-celled organisms. Finally, ZC3H4 loss causes a substantial reduction in cell proliferation, highlighting its overall importance. In summary, we identify ZC3H4 as playing an important role in restricting non-coding transcription in multicellular organisms.

The allosteric and torpedo models have been used for 30 yr to explain how transcription terminates on protein-coding genes. The former invokes termination via conformational changes in the transcription complex and the latter proposes that degradation of the downstream product of poly(A) signal (PAS) processing is important. Here, we describe a single mechanism incorporating features of both models. We show that termination is completely abolished by rapid elimination of CPSF73, which causes very extensive transcriptional readthrough genome-wide. This is because CPSF73 functions upstream of modifications to the elongation complex and provides an entry site for the XRN2 torpedo. Rapid depletion of XRN2 enriches these events that we show are underpinned by protein phosphatase 1 (PP1) activity, the inhibition of which extends readthrough in the absence of XRN2. Our results suggest a combined allosteric/torpedo mechanism, in which PP1-dependent slowing down of polymerases over termination regions facilitates their pursuit/capture by XRN2 following PAS processing.

Many RNA polymerases terminate transcription using allosteric/intrinsic mechanisms, whereby protein alterations or nucleotide sequences promote their release from DNA. RNA polymerase II (Pol II) is somewhat different based on its behavior at protein-coding genes where termination additionally requires endoribonucleolytic cleavage and subsequent 5'→3' exoribonuclease activity. The Pol-II-transcribed small nuclear RNAs (snRNAs) also undergo endoribonucleolytic cleavage by the Integrator complex, which promotes their transcriptional termination. Here, we confirm the involvement of Integrator but show that Integrator-independent processes can terminate snRNA transcription both in its absence and naturally. This is often associated with exosome degradation of snRNA precursors that long-read sequencing analysis reveals as frequently terminating at T-runs located downstream of some snRNAs. This finding suggests a unifying vulnerability of RNA polymerases to such sequences given their well-known roles in terminating Pol III and bacterial RNA polymerase.

Cell-based studies of human ribonucleases traditionally rely on methods that deplete proteins slowly. We engineered cells in which the 3′→5′ exoribonucleases of the exosome complex, DIS3 and EXOSC10, can be rapidly eliminated to assess their immediate roles in nuclear RNA biology. The loss of DIS3 has the greatest impact, causing the substantial accumulation of thousands of transcripts within 60 min. These transcripts include enhancer RNAs, promoter upstream transcripts (PROMPTs), and products of premature cleavage and polyadenylation (PCPA). These transcripts are unaffected by the rapid loss of EXOSC10, suggesting that they are rarely targeted to it. More direct detection of EXOSC10-bound transcripts revealed its substrates to prominently include short 3′ extended ribosomal and small nucleolar RNAs. Finally, the 5′→3′ exoribonuclease, XRN2, has little activity on exosome substrates, but its elimination uncovers different mechanisms for the early termination of transcription from protein-coding gene promoters.

Termination is a ubiquitous phase in every transcription cycle but is incompletely understood and a subject of debate. We used gene editing as a new approach to address its mechanism through engineered conditional depletion of the 5' → 3' exonuclease Xrn2 or the polyadenylation signal (PAS) endonuclease CPSF73 (cleavage and polyadenylation specificity factor 73). The ability to rapidly control Xrn2 reveals a clear and general role for it in cotranscriptional degradation of 3' flanking region RNA and transcriptional termination. This defect is characterized genome-wide at high resolution using mammalian native elongating transcript sequencing (mNET-seq). An Xrn2 effect on termination requires prior RNA cleavage, and we provide evidence for this by showing that catalytically inactive CPSF73 cannot restore termination to cells lacking functional CPSF73. Notably, Xrn2 plays no significant role in either Histone or small nuclear RNA (snRNA) gene termination even though both RNA classes undergo 3' end cleavage. In sum, efficient termination on most protein-coding genes involves CPSF73-mediated RNA cleavage and cotranscriptional degradation of polymerase-associated RNA by Xrn2. However, as CPSF73 loss caused more extensive readthrough transcription than Xrn2 elimination, it likely plays a more underpinning role in termination.