Dual agonistic and antagonistic roles of ZC3H18 provide for co-activation of distinct nuclear RNA decay pathways

Abstract

The RNA exosome is a versatile ribonuclease. In the nucleoplasm of mammalian cells, it is assisted by its adaptors the nuclear exosome targeting (NEXT) complex and the poly(A) exosome targeting (PAXT) connection. Via its association with the ARS2 and ZC3H18 proteins, NEXT/exosome is recruited to capped and short unadenylated transcripts. Conversely, PAXT/exosome is considered to target longer and adenylated substrates via their poly(A) tails. Here, mutational analysis of the core PAXT component ZFC3H1 uncovers a separate branch of the PAXT pathway, which targets short adenylated RNAs and relies on a direct ARS2-ZFC3H1 interaction. We further demonstrate that similar acidic-rich short linear motifs of ZFC3H1 and ZC3H18 compete for a common ARS2 epitope. Consequently, while promoting NEXT function, ZC3H18 antagonizes PAXT activity. We suggest that this organization of RNA decay complexes provides co-activation of NEXT and PAXT at loci with abundant production of short exosome substrates.

Keywords: ARS2; CP: Molecular biology; NEXT; PAXT; ZC3H18; activation; inhibition; nuclear RNA decay.

Graphical abstract

Figure 1

The ZFC3H1 N terminus harbors important information for its function in RNA decay (A) Left: schematic representation of the PAXT connection, highlighting its ZFC3H1-MTR4 core. See text for more detail. Right: experimental design of the ZFC3H1 mutational analysis where the endogenous protein was replaced by one of the MYC-tagged variants from (B). (B) Schematic map of full-length ZFC3H1 (Z1^FL), depicting its predicted domains: SER, serine-rich; CC, coiled coil; PRO, proline-rich; ZnF, zinc finger; TPR, tetratricopeptide repeat and the regions of ZFC3H1 covered by the generated Z1 variants. (C) qRT-PCR analysis of selected PAXT substrates (proASH1L, proSAMD4B, FOXD2-AS1) and a negative control NEXT substrate (proSTAT3) from total RNA isolated from HeLa cells treated with control (siLUC) or ZFC3H1-targeting siRNA, while stably expressing one of the siRNA-immune ZFC3H1 variants from (B). The RT step was performed using a mixture of random and oligo d(T)₂₀VN primers. qPCR amplicons were positioned in TU 5′-end regions. Results were normalized to 18S rRNA levels and plotted as fold change relative to siLUC control samples. Columns represent the mean values of three technical replicates, which are depicted as individual data points, with error bars showing the standard deviation (SD). (D) Top: schematic representation of the mouse Z1^1-1363 variant and its N-terminal truncations used for further characterization of ZFC3H1. Bottom: western blotting (WB) analysis of expression levels of mouse ZFC3H1 variants in mZfc3h1^−/− mESCs, using antibodies against ZFC3H1 and vinculin (VCL) as a loading control. (E) qRT-PCR analysis of selected mouse PAXT substrates and OCT4 mRNA (negative control) from total RNA isolated from cells used in (D). Results were normalized to GAPDH mRNA levels and plotted as fold changes relative to WT control samples. Data representation as in (C).

Figure 2

A conserved short linear motif connects ZFC3H1 to the CBC via ARS2 (A) Top left: schematic representation showing the conserved acidic short linear motif (SLiM) in Z1^FL(WT) and the point mutations introduced in the Z1^FL(MUT) construct. Top right: WB analysis of Z1^FL(WT) and Z1^FL(MUT) expression in siZFC3H1-treated cells, using antibodies against ZFC3H1 and tubulin alpha (TUBA) as a loading control. siLUC- and siZFC3H1-treated cells served as controls. Irrelevant samples between the “Z1^FL(WT)” and “Z1^FL(MUT)” lanes were deleted. Bottom: qRT-PCR analysis as in Figure 1C but on RNA from the above cell lines. Results were normalized to GAPDH mRNA levels. (B) Volcano plot of mass spectrometry (MS) analysis of three biological replicates of MYC IPs from lysates of cells expressing MYC-tagged Z1^1-209(WT) relative to Z1^1-209(MUT). Protein signals from both samples were normalized to bait protein levels. (C) WB analysis of Z1^1-209(WT) and Z1^1-209(MUT) MYC-IP samples from (B). The input and IP samples were probed with antibodies against MYC and ARS2. Irrelevant samples between control (−) and “Z1^1-209(WT)” lanes were deleted. Note that signals from input and IP samples were captured at different exposure times. (D) Coomassie-stained SDS-PAGE gel showing input and amylose resin pull-down samples from assays of His-ARS2^147-871 incubated with His-MBP-Z1^11-35(WT) or His-MBP-Z1^11-35(MUT). “+” on top indicates which proteins were added. Protein markers are indicated to the left.

Figure 3

The ZFC3H1-ARS2 connection facilitates the decay of short, adenylated transcripts (A) Violin boxplots depicting transcription unit (TU) size distribution of PAXT-sensitive and -insensitive RNAs further categorized into ARS2-dependent and -independent groups as indicated. The following criteria were applied to define ARS2-dependent TUs: log₂ fold change (siARS2/CTRL) > 0 and adjusted p value > 0.1 in total RNA-seq dataset generated from HeLa cells depleted of ARS2. TUs exhibiting NEXT sensitivity were excluded from both transcript categories. p values, calculated using two-sided Student’s t test and indicating the difference between ARS2-dependent and -independent groups, are displayed on top of the plots. (B) qRT-PCR analysis of selected PAXT substrates derived from TUs of different sizes as estimated from RNA-seq data from siZFC3H1 cells. RNA samples and data representation as in Figure 2A. qPCR amplicons for proDIP2A and ZNF250 were positioned in TU 3′-end regions to avoid detection of early terminated RNA isoforms. (C) Violin boxplots depicting the log₂ fold change distributions of ARS2-dependent and -independent PAXT-sensitive and -insensitive RNAs from total RNA-seq datasets generated from siZFC3H1-treated HeLa cells, expressing Z1^FL(MUT) relative to Z1^FL(WT). TUs exhibiting NEXT sensitivity were included in all categories as NEXT-mediated RNA decay was not disrupted in these samples. p values were calculated as in (A). (D) Genome browser views of representative ARS2-dependent and -independent PAXT-sensitive transcripts from RNA-seq dataset from (C). Tracks show the average signal of three biological replicates in the relevant genomic coordinates. Signals from “+” and “−” strands are displayed as positive and negative values, respectively. The irrelevant strand signal has reduced opacity. An in-house custom HeLa-specific transcriptome annotation is displayed for both strands below the tracks. (E) Model of PAXT-mediated RNA decay. The ARS2-dependent branch utilizes the ZFC3H1-ARS2 connection to enhance targeting of primarily short, unspliced substrates (top). The ARS2-independent branch targets mainly longer and spliced RNAs (bottom). Whether ARS2 still binds the ARS2-independent substrates is unclear (?).

Figure 4

ZC3H18 antagonizes the ARS2-dependent PAXT pathway (A) Schematic representation of the CBCA-NEXT and CBCA-PAXT connections. ZC3H18 bridges the CBCA and NEXT complexes, while its role for CBCA-PAXT was evaluated here. (B) WB analysis showing depletion of ZC3H18-3F-mAID in HeLa cells with (+) or without (−) expression of TIR1-HA and treatment with IAA as indicated. Membranes were probed with antibodies against FLAG, HA, and ACTIN as a loading control. (C) qRT-PCR analysis of selected NEXT-sensitive (left) and ARS2-dependent (middle) or -independent (right) PAXT-sensitive RNAs from total RNA isolated from cells used in (B). Results were normalized to RPO mRNA levels and plotted as fold change relative to -TIR1, -IAA control samples. Data representation as in Figure 1C. (D) Bar plots showing TU counts of NEXT-sensitive (orange) and PAXT-sensitive (blue) RNAs upregulated (>0) or downregulated (<0) in total RNA-seq datasets generated from HeLa cells depleted of ARS2 or ZC3H18 relative to control cells.

Figure 5

ZFC3H1 and ZC3H18 compete for ARS2 binding (A) WB analysis of MYC IPs from lysates of HeLa cells stably expressing MYC-ARS2 and treated with either control siRNA (siLUC) or siZC3H18. Input and IP samples were probed with antibodies against MYC, ZC3H18, ZCCHC8, and ZFC3H1 as indicated. (B) WB analysis of MYC IPs from lysates of HeLa cells expressing MYC-tagged ZC3H18^WT, MYC-tagged ZC3H18^MUT, and untagged control cells. Input and IP samples were probed with antibodies against MYC, ARS2, NCBP2, ZCCHC8, MTR4, and tubulin alpha (TUBA) as a control. The membrane probed for ZCCHC8 was incompletely stripped and subsequently probed for MTR4 resulting in detection of residual ZCCHC8 signal as indicated. (C) Schematic representation of the suggested inhibition of a common ZFC3H1/ZC3H18 binding site on ARS2 by overexpression of Z1^1-209(WT). The zoom-ins depict the conserved ARS2-binding SLiM in the indicated proteins and the corresponding binding site on ARS2. (D) qRT-PCR analysis as in Figure 4E but of total RNA isolated from WT HeLa control cells or from cells following 2 days of overexpression of stably integrated Z1^1-209(WT) or Z1^1-209(MUT). (E) Violin boxplots depicting log₂ fold change distribution of PAXT- (left) and NEXT-sensitive (right) RNAs in a total RNA-seq dataset generated from cells used in (D). Both RNA groups were further categorized into ARS2 dependent and independent as in Figure 3A. p values, calculated as in Figure 3A, indicate the difference between ARS2-dependent and -independent groups. (F) Heatmaps displaying log₂ fold changes of 3′ end 4-thiouridine RNA-seq data from siZFC3H1 (left) and siZCCHC8 (right) samples relative to their control. Data were plotted in 2-kb regions centered around annotated TESs of the top 50% of ARS2-dependent PAXT- and NEXT-sensitive TUs. The average signals from three replicates are shown. RNA samples were in vitro polyadenylated to capture both pA⁺ and pA⁻ transcripts. (G) Genome browser views of representative TUs from (F). Tracks contain the average signal of three biological replicate samples showing the relevant strand and genomic coordinates. The in-house annotation from Figure 3D is displayed for the relevant strand below the tracks.

Figure 6

Dual agonistic and antagonistic roles of ZC3H18 provides for co-activation of NEXT and PAXT pathways (A) At TUs with basal NEXT activity ZC3H18 is in excess, resulting in promotion of CBCA-NEXT formation, while inhibiting the CBCA-PAXT connection. (B) Higher substrate load demands increased NEXT activity, which will titrate ZC3H18 from its CBCA-PAXT inhibition and lead to co-activation of the ARS2-dependent PAXT- and NEXT-mediated pathways.