RSRC2 is a novel RNA-binding protein that safeguards mitotic fidelity by interacting with the lncRNA C1QTNF1-AS1
Parnia Babaei, Alice O Coomer, Kaliya Georgieva, Giulia Guiducci, Elisa Vitiello, Martin Dodel, Eleni Maniati, Hanya Elsayed Eid, Anisha Thind, Sneha Krishnamurthy, Anna Nawrocka, Sam Wallis, Jun Wang, Alena Shkumatava, Faraz K Mardakheh, Lovorka Stojic
TL;DR
RSRC2, an RNA-binding protein, works with the long non-coding RNA C1QTNF1-AS1 to ensure accurate cell division by maintaining centrosome and spindle function.
Contribution
RSRC2 is identified as a novel RNA-binding protein that interacts with C1QTNF1-AS1 to safeguard mitotic fidelity.
Findings
RSRC2 and C1QTNF1-AS1 are essential for proper chromosome alignment and spindle formation during mitosis.
RSRC2 regulates splicing and centriole integrity by interacting with centrosomal scaffold proteins PCNT and CDK5RAP2.
C1QTNF1-AS1 directs RSRC2 to the centrosome, where it promotes PCNT mRNA recruitment.
Abstract
Mitotic fidelity requires proper chromosome alignment at the spindle equator, a process known as chromosome congression, mediated by well-established protein networks. Although RNA-binding proteins (RBPs) and non-coding RNAs (ncRNAs) have been implicated in cell division, their functional interplay remains unclear. Here, we show that RSRC2, a poorly characterized RBP, is essential for proper cell division through its interaction with the long ncRNA C1QTNF1–AS1. Loss of either RSRC2 or C1QTNF1–AS1 causes mitotic defects. RSRC2 associates with distinct protein sets involved in splicing and centrosome biogenesis, regulating mitotic gene splicing and maintaining centriole integrity. RSRC2 depletion impairs recruitment of the centrosomal scaffold proteins PCNT and CDK5RAP2, which are essential for organizing microtubules to form the mitotic spindle. While C1QTNF1–AS1 loss does not alter…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8- —Barts Charity10.13039/100015652
- —Cancer Research UK10.13039/501100000289
- —Royal Society Research
- —Academy of Medical Science Springboard
- —AIRC10.13039/100018202
- —Fondation pour la Recherche Médicale10.13039/501100002915
- —LABEX10.13039/501100004100
- —MRC10.13039/100018645
- —BBSRC10.13039/501100000268
- —Barts Cancer Institute10.13039/501100019154
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsRNA Research and Splicing · Microtubule and mitosis dynamics · Nuclear Structure and Function
Introduction
Chromosome congression, the process of aligning chromosomes at the spindle equator, is a prerequisite for faithful chromosome segregation during cell division [1]. In human cells, this process lasts 15–20 min and facilitates the formation of the metaphase plate, which is spatially and temporally coordinated with the assembly of the mitotic spindle, a microtubule-based structure that mediates kinetochore–microtubule interactions required for chromosome movement during cell division. Over 100 proteins, including kinases, phosphatases, motor proteins, kinetochore, and centrosomal factors, contribute to chromosome congression [2]. Any perturbations in these pathways can alter microtubule dynamics or kinetochore function, leading to chromosomal instability (CIN), a hallmark of cancer [3].
Beyond well-characterized protein networks, emerging evidence indicates that RNA-binding proteins (RBPs) [4, 5] and various RNAs, including protein-coding mRNAs [6, 7] and non-coding RNAs (ncRNAs) [8–13], contribute to mitotic spindle structure and function [4]. Many RBPs regulate cell division post-transcriptionally [14], either by localizing to the mitotic spindle [15, 16] or by interacting with centrosomes [15–19], spindle microtubules [20], and/or the Ndc80 complex, which is essential for kinetochore–microtubule attachment [21]. Additionally, RBPs indirectly regulate cell division by controlling the splicing of pre-mRNAs required for cell division [22–24]. For example, the RNA splicing factor (SF) SON [22, 23] is not present on the mitotic spindle yet its depletion causes mitotic defects through mis-splicing of mitotic pre-mRNAs. Similarly, the PRP19 splicing complex has both direct and indirect roles in cell division [25, 26]. Genome-wide RNAi screens and proteogenomic analyses further demonstrate that SFs and spliceosome components contribute to centrosome function, revealing that many splicing proteins have “moonlighting” roles during mitosis independent of splicing [27–31]. However, the mechanisms that coordinate these dual functions in time and space remain largely unknown.
Long non-coding RNAs (lncRNAs) also regulate cell division [32] often through interactions with RBPs. RBPs have a key role in driving the functions of lncRNAs, with their binding specificity typically determined by RNA motifs or structural elements present within those lncRNAs [33]. Despite some mechanistic studies where lncRNAs were implicated in mitotic progression [9, 10], very little is known about whether (lnc)RNA-mediated regulation of RBPs contributes to the fidelity of cell division.
Here, we show that the lncRNA C1QTNF1–AS1 regulates cell division by interacting with RSRC2, an RBP with a largely unknown function. Depletion of either C1QTNF1–AS1 or RSRC2 causes mitotic delay, chromosome congression defects, and spindle assembly checkpoint (SAC) activation. RSRC2, but not C1QTNF1–AS1, also regulates the splicing of transcripts enriched for cell cycle functions, and its interactome includes both splicing and centrosomal proteins. Co-depletion of RSRC2 and C1QTNF1–AS1 does not produce additive effects in mitotic delay, indicating that they operate in the same pathway during cell division. Mechanistically, RSRC2 regulates proper localization of centrosomal mRNAs and its recruitment to centrosomes is guided by C1QTNF1–AS1, which localizes near centrosomes in a microtubule-dependent manner. These findings reveal a dual role for RSRC2 as a “moonlighting” protein [31] coordinating splicing in the nucleus and RNA localization at the centrosome, and highlight a novel mechanism by which a lncRNA modulates RBP function to ensure error-free mitosis.
Materials and methods
Cell culture
Human normal retinal pigment epithelial hTERT-RPE1 (ATCC) and hTERT-RPE1 H2B-GFP (provided by Prof. David Pellman, USA) cells were maintained in Dulbecco's modified Eagle's medium (DMEM) F12 medium (Sigma, D8437), supplemented with 10% foetal bovine serum (FBS, A5256801, Gibco). Human colorectal carcinoma HCT116 TP53+/+ (provided by Prof. Bert Vogelstein, USA) and HCT116 H2B-GFP cells (provided by Prof. Zuzana Storchova, Germany) were cultured in McCoy’s 5A medium (Gibco, 36600-021), supplemented with 10% FBS. The HepG2 human liver cancer cell line (provided by Vinothini Rajeeve, QMUL), human breast cancer cell line MDA-MB-231 (provided by Prof. Sarah McClleland, QMUL), and human embryonic kidney HEK293T cells (kindly provided by Prof. Masashi Narita, UK) were maintained in DMEM (Gibco, 41966-029) and supplemented with 10% FBS. The OE19 human oesophageal adenocarcinoma cell line (provided by Prof. Francesca Ciccarelli, QMUL) was maintained in RPMI-1640 Medium (Gibco, 21875034) and supplemented with 10% FBS. All cell lines were verified by short tandem repeat (STR) profiling and were regularly tested for mycoplasma contamination. All the cell lines used in this work were cultured in a sterile, humidified incubator at 37°C with 5% CO_2_.
Single-molecule RNA FISH using Stellaris probes
Cells were grown on 12-mm coverslips (ECN 631-1577, VWR, 150 μm thickness) in 12- or 6-well plates, washed briefly with 1× nuclease-free PBS, and fixed with methanol/acetic acid (75%/25%) at room temperature (RT) for 10 min. Following fixation, the cells were washed with phosphate-buffered saline (PBS) and permeabilized in 70% ethanol for at least 1 h at 4°C. For knockdown experiments, cells were transfected with small interfering RNAs (siRNAs) or Locked Nucleic Acid (LNA) gapmers on the day after plating and were fixed 48 h after transfection. Post-fixation, coverslips were washed with Stellaris^®^ RNA FISH wash buffer A (SMF-WA1-60, LGC Biosearch Technologies) and supplemented with 10% deionized formamide (AM9342, Ambion) for 5 min at RT. C1QTNF1–AS1 FISH exonic probes (labeled with Quasar^®^ 570 dye, Supplementary Information) were added to Stellaris^®^ RNA FISH hybridization buffer (SMF-HB1-10, LGC Biosearch Technologies), supplemented with 10% deionized formamide at a final concentration of 125 nM. Hybridization occurred overnight for up to 16 h at 37°C in a dark, humid chamber. After hybridization, coverslips were washed with wash buffer A for 30 min at 37°C in the dark. This was followed by a 30-min wash with 4,6-diamidino-2-phenylindole (DAPI) in wash buffer A to counterstain the nuclei (5 ng/ml; D9542 Sigma) at 37°C in the dark. The last wash was done with Stellaris^®^ wash buffer B (SMF-WB1-20, LGC Biosearch Technologies) at RT for 5 min. For signal enhancement, coverslips were equilibrated with Base Glucose Buffer composed of 2× saline-sodium citrate (SSC) buffer (1051520320× UltraPure^™^ Invitrogen^™^, Fisher Scientific), 0.4% glucose solution (∼20% in H_2_O BioUltra, 49163, Sigma–Aldrich), and 20 mM RNase-free Tris pH 8.0 (1 M, 10259194, Fisher Scientific) in RNase-free H_2_O. Then, coverslips were incubated for 5 min in Base Glucose Buffer supplemented with a 1:100 dilution of glucose oxidase (3.7 mg/ml; G7141, Sigma–Aldrich) and catalase (4 mg/ml; C1345, Sigma–Aldrich). Finally, the coverslips were mounted with ProLong^™^ Gold antifade (P36934, Invitrogen) on a glass slide and left to cure overnight before image acquisition. Z-stacks with 200 nm z-step capturing the entire cell volume were acquired with a GE wide-field DeltaVision Elite microscope (Cytiva) with an Olympus UPlanSApo 100×/1.40-numerical aperture oil immersion objective lens and a PCO Edge sCMOS camera using appropriate filters. The built-in SoftWoRx Imaging software (Applied Precision) was used to deconvolve the three-dimensional stacks. Images were max projected using FIJI ImageJ. RNA FISH signal was counted manually using the Multi point tool in FIJI.
ViewRNA® FISH
Cells were grown on 12 mm coverslips (ECN 631-1577, VWR, 150 μm thickness) in 24-well plates and treated either with siRNA (as described) or with dimethyl sulfoxide (DMSO, 0.8%), Nocodazole (M1404, Sigma) (3 µg/ml for 2 h), or Ciliobrevin D (250401, Sigma–Aldrich) (50 µM for 1.5 h). Slides were fixed and stained using the ViewRNA^®^ Cell Plus Assay (88-19000, Invitrogen) following the manufacturer’s protocol for a 24-well plate. C1QTNF1–AS1 (Supplementary Information) and PCNT (VA1-3000760, Invitrogen) Type 1 (Alexa Fluor 546) FISH probes (VX-01, Invitrogen) were used. Slides were stained with DAPI according to the instructions and mounted with ProLong^™^ Gold antifade (P36934, Invitrogen) on a glass slide. Slides were cured at 37°C for 2 h before image acquisition. Z-stacks with 200 nm z-step capturing the entire cell volume were acquired using an Eclipse Ti‐E inverted microscope (Nikon) equipped with a CSU‐X1 Zyla 4.2 camera (Ti‐E, Zyla; Andor), including a Yokogawa Spinning Disk, a precision motorized stage, and Nikon Perfect Focus, all controlled by NIS‐Elements Software (Nikon) using a CFI Plan Apochromat Lambda D 100× 1.45‐NA oil objective (Nikon). C1QTNF1-AS1 FISH quantification: C1QTNF1–AS1 distance from γ-tubulin was measured using Imaris software (v9.9.1) to fit the protein signal as surfaces and the messenger RNA (mRNA) signal as spots in a 3D view. The distance of each of the C1QTNF1–AS1 foci within a 10 µm radius of the centrosomes was measured to their nearest γ-tubulin signal in each cell.
PCNT FISH quantification: PCNT-positive centrosomes were quantified in FIJI ImageJ using maximum projected images. A box was drawn around the centrosomes using γ-tubulin as a marker, and the raw integrated density of PCNT mRNA was measured. Background signal was accounted for by taking a further measurement close to the centrosome to subtract from the intensity measurements. A threshold was established for PCNT-positive and PCNT-negative centrosomes based on control cells and applied to PCNT- and RSRC2 siRNA-treated cells. The specificity of PCNT mRNA probe was confirmed using PCNT-depleted cells.
RNA isolation, cDNA synthesis, and qPCR
RNA (1 µg) was extracted from cells using the RNeasy^®^ mini kit (74106, QIAGEN) and treated with DNase I (79254, QIAGEN), according to the manufacturer’s instructions. Complementary DNA (cDNA) was prepared using the QuantiTect^®^ Reverse Transcription Kit (205313, QIAGEN), including an additional wipeout step to eliminate genomic DNA (gDNA) contamination.Quantitative PCR (qPCR) was performed on a QuantStudio^™^ 7 Flex (ThermoFisher Scientific) using PowerUp^™^ SYBR^™^ Green Master Mix (A25742, Applied Biosystems). Thermocycling parameters were set to 50°C for 2 min, 95°C for 2 min, followed by 40 cycles of 95°C for 1 s and 60°C for 30 s. Two reference genes (GAPDH and RPS18) were used to normalize expression levels using the 2^−ΔΔCT^ method. C1QTNF1/CTRP1 primers for total C1QTNF1/CTRP1 expression targeting multiple transcript isoforms (NM_198593, NM_030968, NM_153372, XM_006721667, XM_006721665, XM_006721663, XM_006721666, and XM_006721664) were supplied by QIAGEN (GeneGlobe ID: QT00044443, Cat. No. 249900, QuantiTect Primer Assays). The sequences of qPCR primers were designed using the web interface Primer3Plus [34] and are provided in the Supplementary Information.
siRNA and LNA depletion experiments
Cells were transfected at 60%–70% confluency with Lipofectamine RNAiMAX (13778150, Invitrogen) following the manufacturer’s instructions. siRNAs (a pool of four; Horizon Discovery) and antisense LNA GapmeRs (QIAGEN) were used at final concentrations of 5 and 10 nM, respectively. Lipofectamine RNAiMAX and siRNA or LNA gapmer were diluted separately in Opti-MEM^™^ medium (Gibco, 31985-047). Cells were transfected 24 h after plating and harvested for further experimental procedures 24, 48, or 96 h after transfection, depending on experimental requirements. Two negative controls, silencer^™^ Negative Control No. 1 siRNA (Ambion, 4390843) and non-targeting LNA A (300611-00, QIAGEN), were used. For double knockdown experiments, HCT116 cells were plated and transfected the next day with either negative control siRNA (Ctl, from Ambion) and siRNAs targeting RSRC2, in combination with C1QTNF1-AS1 LNA 1 or Ctl LNA A. siRNAs were combined at 5 nM each to make up a final concentration of 10 nM. For a combination of siRNA with LNA gapmers, siRNAs were used at 5 nM and LNAs at 10 nM, resulting in a final concentration of 15 nM. Cells were always transfected with an equal final concentration of siRNA and LNA gapmers for double knockdown experiments. The list of siRNAs and LNA gapmers is provided in the Supplementary Information.
ASO delivery using Lipofectamine
HCT116 cells (180 000 cells/12-well) were reverse-transfected using Lipofectamine 3000 (Thermo Fisher, L3000015) following the manufacturer’s instructions. Antisense oligonucleotides (ASOs) were diluted in sterile IDTE pH 7.5 buffer (IDT, 1× TE solution). The final concentration of ASO (IDT) was 50 nM after initial titration of several different concentrations. The PCNT ASO was Hs_PCNT_E19 3′SS (-11) (ASO 2′MOE, IDT, Supplementary Information). The cells were collected 48 h post-transfection for phenotyping, RNA or protein extraction.
Cas13-mediated depletion of C1QTNF1–AS1
Guide RNAs (gRNAs) against C1QTNF1–AS1 were designed using gRNA design tool (https://cas13design.nygenome.org) [35]. Single-stranded gRNA oligonucleotides (Sigma) were 5′-phosphorylated with T4 Polynucleotide Kinase (M0201S, NEB) at 37°C for 30 min. Forward and reverse oligonucleotides were mixed in a 1:1 molar ratio and annealed at gradually decreasing temperatures from approximately 95–100°C to RT in a glass beaker with boiling hot water. Five micrograms of pxR003 CasRx gRNA cloning backbone (pUC19; 109053, Addgene) was linearized with 10 Units of BbsI restriction enzyme (R3539, NEB) at 37°C for 4 h. BbsI was then heat-inactivated at 65°C for 20 min. The linearized plasmid was run on 1% agarose gel, and the corresponding band was cut and extracted using QIAquick^®^ Gel Extraction Kit (28 704, QIAGEN). Annealed oligonucleotides were ligated into linearized pxR003 with Quick Ligation Kit (M2200S, NEB) for 20 min at RT. NEB5α (C2987H, NEB) cells were transformed with the ligated pxR003 vector according to the manufacturer’s guidelines for transformation. Four bacterial colonies were chosen per gRNA construct and DNA was purified with QIAprep^®^ Spin Miniprep Kit (27104, QIAGEN). All clones were verified by Sanger sequencing. Finally, the HCT116 cells were plated at 80 000 cells/well in 24-well plates for Cas13-mediated depletion of C1QTNF1–AS1. Twenty-four hours after plating, the cells were transfected with the following combination of vectors: pxR003 gRNA vector (200 ng) + pxR001 EF1a-CasRx-2A-EGFP (active CasRx, 200 ng) (109049, Addgene) using Mirus TransIT-X2^®^ (MIR6004, Mirus Bio). Cells were collected for RNA extraction 48 h after transfection and qPCR was performed to verify the C1QTNF1–AS1 depletion. The list of C1QTNF1–AS1 guide and negative control sequences is provided in the Supplementary Information.
Cell proliferation assays
Cells were seeded at 10 000 cells/well (HCT116) on 24-well plates (Corning). The following day, the cells were transfected with C1QTNF1–AS1 LNAs, including the negative control LNA A. The plates were placed in an IncuCyte^®^ S3 Live-cell Analysis System (Essen BioScience, Ltd; Sartorius Group) 24 h post-transfection to monitor cell proliferation in real time. The IncuCyte^®^ S3 System was set up to take 4 phase-contrast images per well (4× objective) every 3 h for up to 7 days. For analysis, confluence masking was performed in the IncuCyte^®^ S3 Software to estimate cell confluence (%).
Doubling time assessment
HCT116 cells (15 000 cells/well) were seeded into 24-well plates. Live and dead cells were manually quantified every 2 days for 7 days using trypan blue exclusion on a hemocytometer. Doubling time was calculated from cell counts obtained on day 0 post-seeding, with percentage live/dead cells determined concurrently throughout the time course.
Immunofluorescence (IF)
The cells were seeded in 12- or 6-well plates on 12 mm coverslips (ECN 631-1577, VWR, 150 μm thickness). On the following day, the cells were transfected with siRNAs, Cas13, or LNAs as described above. After 48 h, the coverslips were washed once with 1× PBS and fixed immediately depending on the antibody: methanol, PTEM-F buffer or paraformaldehyde (PFA). Methanol fixation: slides were fixed in 99.9% ice‐cold methanol (Acros Organics, 167830025) and incubated at −20°C for 10 min. PFA fixation: slides were fixed in 4% PFA + PBS (28 908, Thermo Scientific) and incubated at RT for 15 min. PTEM-F fixation: slides were fixed in PTEM-F buffer (20 mM PIPES, 0.2% Triton X-100, 10 mM EGTA, 1 mM MgCl_2_, and 4% PFA) at RT for 15 min. Slides fixed in methanol or PFA were permeabilized in PBS/0.5% Tween20/0.5% Triton X-100, while the slides fixed in PTEM-F were permeabilized in PBS/0.2% Triton X-100. All permeabilization took place at RT for 5 min. Slides were blocked in 5% BSA/PBS at 4°C for 1 h, and the primary antibodies (Supplementary Information) were added overnight at 4°C. Cells were washed 3× for 10 min with PBS/0.1% Tween-20 and then incubated with secondary antibodies diluted in blocking-buffer for 1 h at RT. After washing again 3× for 10 min with PBS/0.1% Tween-20, the cells were stained with Hoescht (1 µg/ml, 33258, Sigma–Aldrich, diluted in PBS) at RT for 10 min. Coverslips were washed with 1× PBS and water before being mounted in ProLong^™^ Diamond antifade mountant (P36961, Invitrogen).
Image processing and quantification
Congression defects, nuclear speckle colocalization and nuclear intensity analysis were performed on slides imaged with a GE widefield DeltaVision Elite High-Resolution Microscope (Cytiva) using an Olympus UPlanSApo 100× 1.40-numerical aperture oil immersion objective lens. Analysis of RSRC2, PCNT, CDK5RAP2, and γ-tubulin localization at the centrosomes was imaged with an Eclipse Ti‐E inverted microscope (Nikon) equipped with a CSU‐X1 Zyla 4.2 camera (Ti‐E, Zyla; Andor), including a Yokogawa Spinning Disk, a precision motorized stage, and Nikon Perfect Focus, all controlled by NIS‐Elements Software (Nikon) using a CFI Plan Apochromat Lambda D 100× 1.45‐NA oil objective (Nikon). For CREST staining, optical slices were taken at 200 nm intervals and 100× oil immersion objective was used. For other antibodies, optical slices were set to 500 nm thickness, and either 100× or 60× oil objective was used.
Congression defects
Cells at the metaphase stage of mitosis were imaged using CREST staining to identify misaligned chromosomes. Metaphases were classified as either normal (aligned to the metaphase plate) or congression defects (unaligned chromosomes towards the spindle poles). Congression defects were presented as a percentage of the total metaphase count.
Colocalization of RSRC2 with SC35, a nuclear speckle marker
RSRC2 and SC35 colocalization analysis was quantified on images stained with both antibodies using the JACoP BIOP plugin for FIJI ImageJ [36]. Sum-projected images were segmented using thresholding to separate the target areas. The colocalization was reported as the Manders’ coefficient, by measuring the level of overlap between the thresholded pixels from the RSRC2 and SC35 staining. Costes’ randomization was performed to ensure colocalization was genuine where we performed 200 rounds of shuffling for the pixels in the measured area. Only data points which met significance following Costes’ randomization were included in the final analysis.
Nuclear RSRC2 intensity
Fluorescence intensity of RSRC2 in the nucleus was analysed using CellProfiler [37]. Images were z-projected using ‘sum slices’, and CellProfiler was used to identify the nuclear area and measure the fluorescence intensity of RSRC2 signal within the nucleus. The following modules were used: RescaleIntensity (for both DAPI and RSRC2 channels); IdentifyPrimaryObjects based on the DAPI signal, using minimum cross-entropy thresholding to identify nuclei; FilterObjects, using AreaShape perimeter, eccentricity, and form factor to remove poorly segmented nuclei; MeasureObjectIntensity using the filtered nuclei objects to measure fluorescence intensity (Integrated Density) of the RSRC2 channel.
Localization to centrosomes
RSRC2, centrin (Cen 2/3), PCNT, CDK5RAP2, and γ-tubulin localization to centrosomes was analysed in FIJI ImageJ using sum-projected images. A box was drawn around the centrosomes using centrin as a marker, measuring the raw integrated density of RSRC2/PCNT/CDK5RAP2/γ-tubulin and centrin (Cen 2/3). Background signal was accounted for by taking a further measurement close by the centrosome to subtract from the intensity measurements.
Mitotic index (number of cells in mitosis)
To assess the percentage of Histone-H3-pS10-positive cells, the total cell number was determined by counting the DAPI-stained nuclei. This was repeated for the Histone-H3-pS10 channel, and a percentage was calculated.
Measuring pole-to-pole distance and spindle angle in metaphase cells
We measured the pole-to-pole distance of mitotic HCT116 and RPE1 cells and the spindle angle relative to the coverslip using IF images where centrosomes were labelled with an antibody against Cen2/3. The straight-line distance between the centres of the two centrosomes in both xy and z was measured. The following formulas were then used:
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} \begin{eqnarray*} \textit{pole}\ to\ \textit{pole}\ \textit{distance}\ \left( {\mu m} \right) = \ \sqrt {\left( {x{{y}^2} + {{z}^2}} \right)} \end{eqnarray*}\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} \begin{eqnarray*} \textit{spindle}\ \textit{angle}\ \left( ^\circ \right) = \ \frac{{180}}{\pi }\left( {{{{\tan }}^{ - 1}}\left( {\frac{z}{{xy}}} \right)} \right) \end{eqnarray*}\end{document}Ultrastructure expansion microscopy (U-ExM)
RPE1 and HCT116 cells were prepared for ultrastructure expansion microscopy as previously described [38, 39, 40]. The reagents used in U-ExM experiments included sodium acrylate (sc-236893, ChemCruz), formaldehyde (FA, 36.5%–38%, F8775, SIGMA), acrylamide (AA, 40%, A4058, SIGMA), N,N′-methylenbisacrylamide (BIS, 2%, M1533, SIGMA), sodium acrylate (SA, 97%–99%, 408220, SIGMA), ammonium persulfate (APS, 17874, ThermoFisher), tetramethylethylendiamine (TEMED, T9281, Sigma), nuclease-free water (AM9937, Ambion-ThermoFisher), and poly-D-Lysine (P8920, Sigma). Briefly, cells were grown on 12 mm coverslips, treated with RSRC2 siRNAs for 48 h, and then fixed in ice-cold methanol for 10 min, rehydrated in PBS for 10 min at RT and permeabilized in 0.5% Tween, 0.5% Triton X-100, and 0.05% SDS for 10 min at RT, before blocking with 1% BSA. Coverslips were incubated in 2% AA + 1.4% FA diluted in PBS for 3 h at 37°C. Subsequently, a gelation solution was prepared in the monomer (19% sodium acrylate, 0.1% BIS, 10% acrylamide, 0.5% TEMED, and 0.5% APS) and added to the coverslips. These were then left to polymerize in a humid chamber for 1 h at 37°C. Denaturation was performed for 1.5 h at 95°C, and gels were stained with acetylated tubulin (overnight at 4°C, 1:200) and secondary antibodies at 37°C for 3 h (1:300 dilution). Images were acquired with a ×63 water-based objective on a Zeiss 780 confocal microscope.
Time-lapse microscopy imaging
HCT116 GFP-H2B (30 000 cells/well) and hTERT-RPE1 GFP-H2B cells (15 000 cells/well) were seeded in 4-compartment glass-bottom cell culture dishes (CELLview^™^ cell culture dish, 627870) with a glass thickness of 175 µm ± 15 µm. Cells were transfected the following day, and time-lapse imaging was performed for 24–48 h, depending on the experiment, after transfection, using a GE widefield DeltaVision Elite High-Resolution Microscope (Cytiva). Microscope settings for imaging hTERT-RPE1 GFP-H2B cells were as follows: 10% laser intensity of the FITC (Fluorescein isothiocyanate) channel with a 0.05 s exposure time, 10 optical sections of 2 µm thickness, 4× gain, and 2 × 2 binning. HCT116 GFP-H2B cells were imaged using similar settings, apart from the exposure time being changed to 0.1 s. Cells were imaged every 3 min at 4 or 5 positions per compartment for 10 to 12 h. Cells were maintained in a microscope stage incubator at 37°C and 5% CO_2_ humidified atmosphere throughout the experiment. The mitotic duration was analysed by visual inspection of the images, counting the time from nuclear envelope breakdown to the onset of anaphase. For the nocodazole experiments, HCT116 GFP-H2B cells were treated with 300 nM nocodazole (M1404, Sigma) for 2 h before filming for an additional 12 h.
FACS preparation and analysis
Cultured cells were trypsinized and washed with ice-cold PBS, followed by centrifugation. Cells were fixed in 70% ethanol and incubated overnight at 4°C. Fixed cells were pelleted by centrifugation and washed twice in PBS before being resuspended in PBS. RNase A (GE101-01, Generon) was added to a final concentration of 50 µg/ml and the samples were incubated at 37°C for 30 min. The cell suspension was filtered using the 40 µm sterile cell strainers, (FisherBrand, 22363547) and incubated with propidium iodide (20 µg/ml, P4170, Sigma–Aldrich) on ice in the dark for 30 min. Cells were analysed using a LSR Fortessa (BD Biosciences) and analysed with the FlowJo (v10.8.1).
Cell synchronization
The HCT116 cells were grown to 50% confluency and transfected with either siRNAs or LNAs the following day. After 24 h, thymidine (2 mM, T9250, Sigma) was added, and the cells were incubated for 18 h. They were then washed three times with PBS and released into thymidine-free medium for an additional 9 h. Thymidine (2 mM) was then re-added for another 15 h. The cells were subsequently washed three times with PBS. At this point (G1/S, T0), samples were collected at 2 (T2), 4 (T4), 6 (T6), 8 (T8), and 11 (T11) h. Cells were then harvested and analysed by PI-FACS as described above.
Western blot analysis
HCT116 cells were plated at 300 000 cells/dish on 60 mm² dishes (Corning). On the following day, the cells were transfected with siRNA or LNA gapmers, and after 48 h, they were trypsinized, washed twice with ice-cold 1× PBS, and pelleted at 1000 rpm for 3 min at 4°C. The pellets were lysed in NP-40 lysis buffer composed of 50 mM Tris–HCl pH 8 (15568025, Invitrogen), 125 mM NaCl (71386, Sigma–Aldrich), 1% NP-40 (85124, Thermo Scientific), 2 mM EDTA (15575020, Invitrogen), 1 mM Phenylmethylsulfonyl fluoride (PMSF) (93482, Sigma–Aldrich), cOmplete^™^ Mini EDTA-free protease inhibitor cocktail (11836170001, Roche), and phosphatase inhibitors (2 mM sodium fluoride, 201154, Sigma–Aldrich; and 1 mM sodium orthovanadate, S6508, Sigma–Aldrich). The samples were incubated on ice for 25 min and then centrifuged for 3 min at 12 000 × g at 4°C. The supernatant was collected, and the protein concentration was estimated using the Bio-Rad Protein Assay (5000006, Bio-Rad). Proteins (20 µg) were denatured in 6× SDS buffer [0.375 M Tris pH 6.8, 12% SDS, 60% glycerol, 0.6 M dithiothreitol (DTT), and 0.06% bromophenol blue] at 95°C for 5 min. The proteins were then separated using Bolt^®^ 4%–12% Bis–Tris Plus Gel (Thermo Fisher Scientific, NW04120BOX) in MOPS buffer (Thermo Fisher Scientific, B0001-02). Precision Plus Protein Standards (161-0373, Bio-Rad) were used as a protein standard. The proteins were then transferred to a nitrocellulose membrane (1620115, Bio-Rad) using 1× Tris-Glycine transfer buffer with 20% methanol and blocked with 5% non-fat milk in TBS-T (50 mM Tris, 150 mM NaCl, and 0.1% Tween-20) for 1 h at RT. The membranes were incubated with primary antibodies in 5% milk in TBS-T at 4°C overnight. The membranes were washed three times the next day with TBS-T and incubated with horseradish peroxidase secondary antibodies (Agilent Dako, P0447 and P0448, dilution 1:2000). Immunobands were detected with SuperSignal^™^ West Pico PLUS Chemiluminescent Substrate (34580, Thermo Scientific), and the signal was developed using CL-XPosure films (34089, Thermo Scientific). Quantification of immunoblots normalized against appropriate loading controls was performed using ImageJ. The list of all antibodies is provided in the Supplementary Information, along with uncropped immunoblot scans shown in Supplementary Fig. S17.
Lentiviral overexpression of C1QTNF1–AS1 lncRNA
To generate lentivirus, HEK293T cells were plated and transfected with 15 µg of DNA, composed of 9 µg of the lentiviral vector DNA containing the transgene (e.g. lincXpress), 4 µg of psPAX.2 packaging vector (12260, Addgene), and 2 µg of pMD2.G envelope expressing vector (12259, Addgene) in the final transfection volume of 1.5 ml (including 45 µl of Trans-Lt1 transfection reagent, MIR6004, Mirus) using OptiMEM medium (31985070, Gibco). The transfection reaction was incubated at RT for 30 min before adding it to the cells. Twenty-four hours post-transduction, the medium was refreshed. Viral supernatant was collected 48 and 72 h post-transduction, spun down at 1800 × g for 5 min at 4°C, filtered through a 45 µm filter, and stored in cryovials at −80°C. For overexpression of C1QTNF1–AS1 in HCT116 cells, we used C1QTNF1–AS1 RNA sequence amplified from cDNA (Labomics, based on Gencode vs30) and a negative control vector (scrambled C1QTNF1–AS1 sequence) cloned into the pLenti6.3/TO/V5-DEST vector (also known as lincXpress; kindly provided by John Rinn, University of Colorado) using the Gateway cloning strategy. HCT116 cells were seeded at 70 000 cells/well in 12-well plates, and the following day, viral supernatant and appropriate cell medium without antibiotics were added dropwise in the presence of polybrene (5 μg/ml, TR100-3, Sigma–Aldrich). Cells were harvested 48 h after infection. In the rescue experiment, HCT116 cells were plated at the same density, and the next day, the cells were transfected with control and RSRC2 siRNAs at a final concentration of 10 nM. The following day, cells were transduced with lentiviral particles containing the lincXpress C1QTNF1–AS1 construct (full) or the negative control scrambled C1QTNF1–AS1 (scr). Twenty-four hours post-viral infection, cells were harvested for RNA extraction, live cell imaging, or immunofluorescence. The list of C1QTNF1–AS1 full and scrambled sequences is provided in the Supplementary Information.
In cell protein–RNA interaction (incPRINT)
The incPRINT experiment was performed as described previously [41]. Briefly, the C1QTNF1–AS1 mature RNA sequence was cloned from the Labomics vector into the 10xMS2 vector (kindly provided by Dr Alena Shkumatava, University of Edinburgh). First, we linearized the 10xMS2 vector with BstBI restriction endonuclease (R0519S, NEB) at 65°C for 1 h followed by its dephosphorylation using the Quick CIP (M0525S, NEB) at 37°C for 10 min. The linearized vector was run on a 1% agarose gel, and the products were extracted with the QIAquick Gel Extraction Kit (28704, QIAGEN) according to the manufacturer’s instructions. Secondly, the C1QTNF1–AS1 sequence (Labomics) was PCR-amplified using C1QTNF1–AS1 specific primers designed with NEBuilder online software (listed in the Supplementary Information). PCR amplification was performed with Q5 High-Fidelity DNA Polymerase (M0491S, NEB). Amplified insert and linearized 10xMS2 vector were assembled in a 1:3 molar ratio. C1QTNF1–AS1 insert was cloned upstream of the MS2 stem-loop sequence of the 10xMS2 vector using a Gibson assembly cloning kit (E5510S, NEB) at 50°C for 15 min. The assembled C1QTNF1–AS1–10xMS2 vector was transformed into competent Escherichia coli cells NEB5α, and the DNA was purified using the QIAprep^®^ Spin Miniprep Kit. The confirmed positive C1QTNF1–AS1 10xMS2 construct was validated with Sanger sequencing. In the C1QTNF1–AS1 small scale incPRINT experiment, all FLAG-tagged proteins were tested in four replicates. Interaction intensity values (in relative light units, RLU) between C1QTNF1–AS1–MS2 and the tested proteins were plotted as the average luminescence across four luciferase replicates. EGFP (enhanced green fluorescent protein) was used as a negative control, and PABPC3 (a polyadenylated RBP) was included to control for RNA expression. The Xist (C)-MS2 vector were used alongside C1QTNF1AS1–MS2 as a positive control for RNA–protein interactions.
RNA immunoprecipitation (RIP)-qPCR
HCT116 cells were grown on 15 cm^2^ dishes up to ∼90% confluency. The cells were trypsinized and pelleted at 800 × g for 4 min at 4°C. The cell pellets were lysed in ice-cold RIP buffer containing 25 mM Tris–HCl pH 7.5, 5 mM EDTA, 0.5% NP-40, 150 mM KCL (AM9640G, Invitrogen), 0.5 mM DTT supplemented with 100 U/ml RNaseOUT Recombinant Ribonuclease Inhibitor (10777019, Invitrogen), EDTA-free protease inhibitor cocktail (11836170001, Roche), and 1 mM PMSF (93 482, Sigma–Aldrich) for 30 min. 1.5 mg of protein total cell lysate was first incubated with 10 μg antibody for 2 h at 4°C (IgG, 2729, Cell Signaling and RSRC2, NBP1-83787, Novus). Pre-washed Dynabeads^™^ Protein G beads (10003D, Invitrogen) were then added to the antibody-lysate mix for 2 h at 4°C. After incubation, the beads were washed with RIP buffer 3 × 10 min at 4°C on a rotating wheel. RNA was extracted directly from the beads using TRIzol reagent (15596018, Invitrogen) as per the manufacturer’s instructions. Dry RNA pellets were resuspended in 15 μl of RNase-free water and subjected to qPCR using the Power SYBR^™^ Green RNA-to-CT^™^ 1-Step kit (4389986, Applied Biosystem) as per manufacturer’s guidelines.
UV-RIP
HCT116 cells were plated in 15 cm^2^ dishes and harvested at 90% confluency. Cells were irradiated with 150 mJ/cm^2^ of UV-C (254 nm, Boekel Scientific UV Crosslinker, 234100) in ice-cold 1× PBS on ice, then scraped in ice-cold PBS, spun at 400 × g for 5 min at 4°C, and stored at −80°C. Cells were lysed in ice-cold UV-RIP lysis buffer (50 mM Tris–HCl, 100 mM NaCl, 1% NP-40, 0.1% SDS, and 0.5% sodium deoxycholate) supplemented with EDTA protease inhibitor cocktail (11836170001, Roche) and SUPERaseIn^™^ RNase Inhibitor (AM2696, Invitrogen) on ice for 25 min. Pre-washed Dynabeads^™^ Protein G beads (10003D, Invitrogen) were coupled with 3 µg of the indicated antibodies (IgG, 2729, Cell Signaling and RSRC2, NBP1-83787, Novus) for 1 h at RT. One milligram of protein lysate was incubated with antibody-coupled beads overnight at 4°C. Beads were then washed with UV-RIP buffer (not containing inhibitors) 3 × 1 min at 4°C and eluted in PK elution buffer (50 mM Tris–HCl, 100 mM NaCl, 1% NP-40, 1% SDS, and 0.5% sodium deoxycholate) incubated for 1 h at 65°C. RNA was extracted with TRI Reagent^®^ LS (T3934, Sigma–Aldrich). One-step RT-qPCR was performed using Power SYBR^™^ Green RNA-to-CT™ 1-Step Kit (4 389 986, Applied Biosystems) according to the manufacturer’s guidelines.
Generation of RSRC2 CRISPR–Cas9 clones
Edit-R predesigned synthetic gRNAs (crRNA, 65 117) targeting human RSRC2 exon 4 and exon 8 (NM_023012.6) were ordered from Horizon along with Edit- R trans-activating CRISPR RNA (tracrRNA, U-002005-05-Edit-R CRISPR–Cas 9 Synthetic, 5 nmol, Horizon) and Edit-R EGFP Cas9 Nuclease mRNA (CAS11860, 20 µg, Horizon). The day before transfection, HCT116 cells were plated 200 000 cells/well in a six-well plate, and the following day, the cells were co-transfected with Cas9 mRNA, tracrRNA, and RSRC2 crRNA or non-targeting control crRNA (U-007501-01-05, Horizon) using DharmaFECT Duo (T-2010-02, Horizon) according to the manufacturer’s instructions. Forty-eight hours after transfection, the top 10% GFP-positive cells were single-sorted using FACS Aria II/Infusion cell sorter (BCI, Flow Cytometry Core Facility) into 96-well plates with DMEM supplemented with 20% FBS. Single-cell clones were expanded, and gDNA was extracted using Direct PCR Lysis reagent (Viagen, 201-Y) with Proteinase K (25530-049, Thermo Fisher Scientific), as previously described [9]. gDNA was used for screening by PCR amplification (Phusion^®^ High- Fidelity PCR Master Mix with GC Buffer, NEB, M0531S) of the targeted genomic region. PCR conditions for RSRC2 guides were: 98°C for 30 s, 30× (98°C for 30 s, 62°C for 30 s, and 72°C for 30 s), 72°C for 90 s. Half of the PCR reaction was analysed on a 2% agarose gel, and the remaining half was ligated into pJet 1. 2/Blunt and transformed into bacteria using CloneJET (K 1231, Thermo Fisher Scientific). To ensure representation of both alleles, plasmids were isolated and sequenced from at least 10 bacterial colonies. RSRC2 wild-type control clones were chosen from sorted GFP-positive cells, which did not contain the required RSRC2 mutation. The list of gRNA and PCR primers sequences is provided in the Supplementary Information.
Construction of minigene reporter assay
gDNA was extracted from HCT116 cells using the DNeasy Blood & Tissue Kit (QIAGEN, 69504). Event exons (exon 18–19 within PCNT, and exon 17 within CENPE) and their flanking introns were amplified from the extracted gDNA using the high-fidelity Q5^®^ polymerase (NEB, M0491) and exon-specific primers. Thermocycling parameters were set to: 98°C for 30 s, 35× (98°C for 10 s, Tanneal 30 s, 72°C for 30 s) 72°C for 2 min (where Tanneal is 71°C for PCNT and 60°C for CENPE). Amplified insert DNA was excised and extracted from the agarose gel using the QIAquick Gel Extraction Kit (QIAGEN, Cat. No. 28704). Amplified PCR products were then digested with EcoRI restriction endonuclease (NEB, R3101) according to the manufacturer’s instructions, except for the following modifications: maximum volume of eluted insert DNA was used instead of the recommended 1 μg; the reaction mixture was incubated at 37°C for 1 h instead 5–15 min. In parallel, we purified the pXJ41 vector (kindly provided by Prof. David Elliott, Newcastle University) using the Plasmid Maxi Kit (QIAGEN, 12 162). MfeI-HF (NEB, R3589) was used to digest the pXJ41 vector, which was then purified and excised upon agarose gel electrophoresis. Antarctic phosphatase (NEB, M0289) was used for dephosphorylation to prevent re-circularization during ligation. The reaction was scaled up to 50 μl and incubated for 30 min at 37°C. Quick Ligase (NEB, Cat. No. M2200) was used to ligate each amplified event exon insert into the generated MfeI site within the β-globin intron of pXJ41. The ligations were then re-cleaved with *MfeI-*HF restriction endonuclease to remove empty vectors. Each minigene reporter was verified by Sanger sequencing. Minigenes were sequenced using primers within the β-globin gene of pXJ41. All primers are listed in the Supplementary Information.
In vitro mutagenesis of RSRC2 constructs
Plasmid DNA was extracted from the pcDNA3.1 (+)-N-eGreen fluorescent protein (GFP) vector (Genscript) with an RSRC2 insert (Genscript, Clone ID: OHu12299C, NM_023012.6 ORF Sequence,1317 bp) (pcDNA3.1 + N-eGFP-RSRC2) using the Plasmid Maxi Kit (QIAGEN, 12 162). In vitro mutagenesis kit (Q5^®^ Site-Directed Mutagenesis Kit, NEB, E0552) was used on the vector per the manufacturer’s instructions to produce deletions at the following bases: 1–687 for deletion of the intrinsically disordered region (ΔIDR), 1057–1236 for deletion of the small acidic protein-like domain (ΔSMAP), 691–795 for deletion of coiled-coil region 1 (ΔCC1), and 1174–1236 for the coiled coil region 2 (ΔCC2). These domains were determined using UniProt. Primers for these deletions were designed annealing back-to-back using NEBaseChanger. Thermocycling conditions were: 98°C for 30 s, 35 cycles (98°C for 10 s, Tanneal for 30 s, 72°C for 4 min), 72°C for 10 min, where Tanneal is: 59°C for IDR, 54°C for SMAP, 66°C for CC1 and 57°C for CC2. The amplified products were circularized, and the template was removed using a KLD enzyme mix (Q5^®^ Site-Directed Mutagenesis Kit, E0552). The mixture was incubated at RT for 15 min, then at 37°C for an additional 15 min. The PCR purification was performed using the QIAquick PCR Purification Kit (QIAGEN, 28106). All the constructs have been verified by Sanger sequencing using primers in the Supplementary Information.
Minigene splicing experiments
HCT116 cells were co-transfected with PCNT minigene and pcDNA3.1 (+)-N-eGFP, or the full-length or mutated RSRC2 insert. For subsequent RNA and protein extraction, 9 × 10^4^ and 1.8 × 10^5^ cells were seeded in 12- and 6-well plates, respectively. After 24 h, 800 and 2000 ng of DNA were transfected into each well of 12-well and 6-well plates, respectively, using Lipofectamine 3000. Upon overexpression of RSRC2 alone, 800 ng of DNA was used, and cells were collected for IF or protein analysis after 48 h.
RNA ASO pulldown coupled with mass spectrometry (MS): HCT116 cells were plated in 15 cm^2^ dishes and harvested at 90% confluency. 100 million HCT116 cells were irradiated with 500 mJ/cm^2^ of UV-C (254 nm) in ice-cold 1× PBS on ice. Cells were scraped in ice-cold PBS, spun at 800 × g for 5 min at 4°C, and stored at −80°C. Amino-C12-LNA-containing oligonucleotides against C1QTNF1-AS1 and luciferase (negative control) (QIAGEN, custom design, listed in the Supplementary Information) were coupled to Dynabeads^™^ MyOne^™^ Carboxylic Acid (65011, Invitrogen) according to the manufacturer’s instructions using 5 nmol LNA oligo per 100 μl beads in the presence of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) (E1769, Sigma–Aldrich) dissolved in MES buffer pH 4.8 (69892, Sigma–Aldrich). Before hybridization with cell lysate, the oligo-coated beads were equilibrated in RNA pulldown (RP) buffer containing 50 mM Tris–HCl pH 7.5 (15567-027, Gibco), 5 mM EDTA, 500 mM LiCl (L7026, Sigma–Aldrich), 0.5% DDM (n-dodecyl β-d-maltoside, D4641, Sigma–Aldrich), 0.2% SDS (15553027, Invitrogen), 0.1% Na-deoxycholate (D6750, Sigma–Aldrich), 4 M Urea (U5378, Sigma–Aldrich), 2.5 M TCEP (tris (2-carboxyethyl)phosphine hydrochloride solution, 646547, Sigma–Aldrich), and protease inhibitors (11836170001, Roche). Then, the oligonucleotide-bead complexes were blocked with RP buffer supplemented with salmon sperm DNA (ssDNA) 200 μg/ml (AM9680, Invitrogen), BSA 1 mg/ml (AM2616, Invitrogen), and yeast RNA 200 μg/ml (AM7118, Invitrogen). Cell pellets were lysed in 5 ml of RP buffer with murine RNase inhibitor in a 1:500 dilution (M0314L, NEB). Lysates were sonicated in 15 ml tubes with MSE Soniprep 150 Ultrasonicator (5 cycles of 10 s, with 30 s intermittent breaks, at setting 15% at 4°C) and centrifuged at 16 000 × g for 5 min at 4°C. At this point, 1% input was collected. Supernatants were pre-heated to 65°C with shaking, after which oligo-coated beads were incubated with the lysates for 4 h at 65°C with shaking at 1200 rpm. The beads were washed four times with RP buffer and then three times with 50 mM TEAB buffer (triethylamonium bicarbonate, 90114, Thermo Scientific) at RT to remove detergents. Beads were resuspended in a final volume of 1 ml of 50 mM TEAB buffer from which 50–100 μl were collected for RNA enrichment analysis. The rest of the buffer (900–950 μl) was discarded, and the beads were stored at −80°C. Proteins were then subjected to on-bead digestion where the beads were resuspended in 50 mM ABC buffer (ammonium bicarbonate, A6141, Sigma–Aldrich) with 8 M urea and reduced by adding DTT (10197777001, Roche) at a final concentration of 10 mM. After a 30 min incubation at RT with 1200 rpm shaking, samples were alkylated by adding 55 mM iodoacetamide (A3221, Sigma–Aldrich) for 30 min at RT in the dark. Trypsin digestion was performed overnight at RT using 2 μg trypsin (T4799, Sigma–Aldrich) per sample. The next day, samples were desalted using the Stage Tip procedure and recovered in a buffer containing 0.1% TFA (trifluoroacetic acid), 0.5% acetic acid, and 2% acetonitrile for MS analysis. LC-MS analysis was performed on a Q Exactive-plus Orbitrap mass spectrometer coupled with a nanoflow ultimate 3000 RSL nano HPLC platform (Thermo Scientific). The data were analysed using the MaxQuant (version 1.6.3.3) for all MS searches [42]. Raw data files were searched against a FASTA file of the Homo sapiens proteome, extracted from Uniprot (2016). Downstream data analyses, including data filtering, log transformation, data normalization, one-sample, or two-sample t-test analysis, category annotation, and data visualizations by scatter or volcano plots, were performed in Perseus software (version 1.6.2.1). Fold change values of median peptide intensities were calculated using three independent oligonucleotides (oligo 1, 3, and 5), and missing values were imputed using the minimal intensity value detected. Evaluation of the MS data was performed to reveal significantly enriched proteins in C1QTNF1–AS1 pulldown samples over control luciferase pulldown based on a one-sample t-test with Benjamini–Hochberg FDR calculation. To confirm the C1QTNF1–AS1 enrichment in the oligonucleotide pulldown, RNA from beads from each sample was eluted in 100 μl elution buffer (0.2% SDS and 2 mM EDTA) at 95°C for 5 min at 1200 rpm. Supernatant was collected and mixed with one volume of proteinase K buffer containing 100 mM NaCl, 10 mM Tris–HCl pH 7 (AM9850G, Invitrogen), 1 mM EDTA, and 0.5% SDS with the addition of 1 mg/ml proteinase K (25530049, Invitrogen). The reaction was incubated at 37°C for 30 min. RNA was extracted with TRIzol^™^ Reagent (15596018, Invitrogen) and Direct-zol^™^ RNA Miniprep kit (R2050, Cambridge Bioscience). RNA concentrations were measured using the NanoDrop 2000c UV/IV Spectrophotometer. One-step RT-qPCR was performed to quantify the RNA enrichment on the beads using Power SYBR^™^ Green RNA-to-CT^™^ 1-Step kit (4389986, Applied Biosystems) according to manufacturer’s guidelines where GAPDH was used as a housekeeping control.
RSRC2 immunoprecipitation (IP) followed by LC-MS/MS-based protein analysis
HCT116 cells (1.4 million) were plated in 10 cm dishes, and the cells were collected at 90% confluency. In the case of transfection, the same number of cells was plated, and the following day, the cells were transfected with the corresponding LNA gapmers at 25 nM. Four biological replicates were performed for each IP-MS in cells ± RNase A/T1 Mix or cells treated with Ctl LNA A and C1QTNF1–AS1 LNA 1 gapmer. The next day, the media was refreshed, and on day four, the cells were washed with ice-cold 1× PBS and lysed in 1 ml of lysis buffer composed of 50 mM Tris–HCl, pH 7.4, 100 mM NaCl, 1% NP40 detergent, 0.1% SDS, and 0.5% sodium deoxycholate, supplemented with SUPERase-IN (Invitrogen, AM2694, 40U), phosphatase (Roche PhosStop, 04906837001), and protease inhibitors (Roche Diagnostic GmbH, 14549800), which were directly added to each 10 cm dish. The lysates were transferred into 1.5 ml Eppendorf tubes, cleared with the QIAGEN Shredder (Qiagen, 79656), and spun down at 12 000 × g for 10 min at 4°C. The lysates were then transferred into new tubes, and the concentration was determined with the BCA Assay (Thermo Scientific, 23227). In the meantime, 60 µl of protein G Dynabeads (Dynal, 100.02) were washed three times with lysis buffer before the addition of 5 µg of IgG (2729, Cell Signaling) and RSRC2 (NBP1-83787, Novus). The mixture was incubated at RT for 1 h on a rotating wheel. Two milligrams of total cell extracts were incubated with 60 µl of protein G Dynabeads coupled to antibodies in a total volume of 1 ml of lysis buffer overnight at 4°C. For the IP-MS sample treated with RNase A/T1 Mix (EN0551, ThermoFisher), 5U of RNase A/T1 Mix was added overnight at 4°C. RNA was extracted from total cell extracts, and the quality of the RNA was analysed on the TapeStation (Agilent) to demonstrate the efficiency of RNA degradation in the lysates treated with RNase A/T1 Mix. The next day, the beads were washed three times with 1 ml of ice-cold lysis buffer for 5 min at 4°C on a rotating wheel. To proceed with the IP-MS, the enriched proteins were resuspended with 60 µl of SDS buffer for each sample (2% SDS, 100 mM Tris pH 7.5, and 100 mM DTT). Samples were heated at 95°C for 5 min, spun down quickly, and by using the magnetic rack, the eluate was transferred into a new Eppendorf tube. Ten microlitres was used for the western blot to show the efficiency of the IP. The eluates were then subjected to alkylation, detergent removal, and trypsin digestion using the Filter Aided Sample Preparation protocol, followed by desalting using StageTips. Desalted peptides were subsequently lyophilized by vacuum centrifugation, resuspended in 7 μl of A*buffer (2% acetonitrile, 0.5% acetic acid, and 0.1% trifluoroacetic acid in water) and analysed on a Q-Exactive plus Orbitrap mass spectrometer (MS) coupled with a nanoflow ultimate 3000 RSL nano HPLC platform (ThermoFisher). The instrument was operated using the Thermo Xcalibur v.4.5 SP1 software. Briefly, 6 μl of each peptide sample was resolved at 250 nl min^−1^ flow rate on an Easy-Spray 50 cm × 75 μm RSLC C18 column (Thermo Fisher), using a 123 min gradient of 3%–35% of buffer B (0.1% formic acid in acetonitrile) against buffer A (0.1% formic acid in water). LC-separated samples were infused into the mass spectrometer by electrospray ionization (1.95 kV, 255°C). The mass spectrometer was operated in data-dependent positive mode, using a TOP15 method in which one MS scan is followed by 15 MS2 scans. The scans were acquired at a range of 375–1500 m/z, with a resolution of 70 000 (MS) and 17 500 (MS/MS). A 30-s dynamic exclusion was applied. MaxQuant (version 1.6.3.3) was used for the MS search and protein quantifications [42] against a FASTAfile of the Homo sapiens proteome extracted from Uniprot (2016). The search for differential phosphorylated peptides (STY) was performed using Maxquant, with phospho (STY) added as a variable modification. All downstream MS data analysis was performed using Perseus software (version 1.6.2.1). The same protocol was used for Co-IP, but instead of proceeding with IP-MS, the beads were resuspended in 4× Laemmli protein sample buffer (Bio-Rad, 1610747) with XT reducing agent (Bio-Rad, 1610792).
Splicing analysis
RNA was extracted from control- and RSRC2–siRNA-treated HCT116 cells using the RNeasy Kit (74106, QIAGEN) followed by the DNase I (79254, QIAGEN) treatment. cDNA was prepared from 1 μg of total RNA using the QuantiTect Reverse Transcription Kit (205313, QIAGEN), following the manufacturer’s instructions. To analyse the splicing pattern of the alternative events, primers flanking the variable exon (exon 17 within CENPE, exon 19 within PCNT, and exon 19 within CDK5RAP2) were designed and regions of interest amplified by RT-PCR (primer sequences are provided in the Supplementary Information). PCR reactions were performed using One Taq Hot Start 2× Master Mix with Standard Buffer guideline (M0484S, NEB) as per manufacturer’s instructions. The PCR conditions were: 94°C for 30 s, 35 cycles (94°C for 20 s, Tanneal for 30 s, and 68°C for text), 68°C for 5 min where Tanneal is: 53°C for CDK5RAP2, 58°C for PCNT and CENPE; and text is 1 min for CDK5RAP2, 30 s for PCNT and CENPE. The RT-PCR products were analysed both by normal agarose gel electrophoresis (not shown) and quantified using the QIAxcel DNA High Resolution Kit (1200) (929002, QIAGEN) with the QIAxcel capillary electrophoresis system 100 (9001941, QIAGEN). An alignment marker was injected and run simultaneously with the samples with fragments of 50 bp and 5 kb (929529, QIAGEN). The QIAxcel Size Marker ranging from 50 to 800 bp (929561, QIAGEN) was used for size and concentration estimation. Samples were analysed using Method OM500. Percentage splicing inclusion (PSI) for each event was calculated by dividing the concentration of the inclusion isoform’s amplicon by the sum of the concentrations of the inclusion and exclusion products.
RNA library preparation, sequencing, and analysis
RNA-seq libraries were prepared from HCT116 cells using the CORALL^™^ Total RNA-Seq V2 Library Prep Kit with UDIs set B1 (184.96, Lexogen) with Lexogen RiboCop rRNA Depletion Kit for Human/Mouse/Rat V2 following the workflow for generation of long insert sizes. Four biological replicates of cell populations were generated after depleting C1QTNF1–AS1 or RSRC2, with an equal number of replicates for the respective negative controls. The number of PCR amplification cycles was determined by performing a test qPCR using the Lexogen PCR Add-on Kit V2 (020.96, Lexogen) according to the manufacturer’s instructions. Libraries were quantified with Qubit^™^ dsDNA HS Assay Kit (Q3285, Invitrogen), and their quality was assessed with the Tapestation (Agilent). Samples were pooled in equimolar ratios and sequenced using 150 bp paired-end reads on Illumina NovaSeq 6000 S4 instrument (Novogene). Each library was sequenced to a depth of >40 million read pairs. The quality of the raw sequencing reads was assessed with FastQC v0.11.9 [43]. Extraction of UMIs was performed using the UMI tools v1.1.4 [44] and quality trimming of fastq files using trimgalore v0.6.5. Paired-end reads were aligned to the hg38 build of the human genome using STAR 2.7.0f [45], and the number of read pairs was counted for each library with rsem v1.3.1 using gencode.v38. Only genes that achieved at least 10 counts in at least four samples were included for downstream analysis. Approximately 80% of uniquely aligned reads were mapped to the human reference genome. Differential gene expression analyses were performed using R package DESeq2 [46]. For the analyses of alternatively spliced events, we used rMATS v4.1.2 [47]. The same version of Gencode was used for rMATS and RNA-Seq (Gencode release 38, which is GRCh38p13). Further downstream, Ensembl IDs of differentially expressed genes (DEGs) were annotated to gene names (gene symbols) using BioMart Ensembl Genes version 108 (GRCh38p13).
Data and code availability
RNA-sequencing data generated during this study are available in the GEO database (https://www.ncbi.nlm.nih.gov/geo/) under the accession number GSE284007. The mass spectrometry raw files and their associated MaxQuant output files are available in the PRIDE partner repository (http://www.ebi.ac.uk/pride/archive/) under accession numbers: PXD059546 (C1QTNF1–AS1 oligonucleotide pulldown), PXD059512 (RSRC2 IP-MS in C1QTNF1–AS1 depleted cells), and PXD059507 (RSRC2 IP-MS with or without RNase). This paper did not report any original code.
Statistical analysis
Graphs and statistical significance were determined using Prism 10 (GraphPad Software) or R version 4.4.2. Results are presented as mean ± SEM unless otherwise stated. Statistical analysis was performed on average values for each experiment using: unpaired t-test with Welch’s correction, unpaired two-tailed t-test, one-sample t-test for normalized data, Mann–Whitney test, and one-way ANOVA with a Dunnett’s multiple comparison test. Normal distribution of all data was assessed using Shapiro–Wilk normality test in Prism before selecting appropriate statistical analysis. P-values > 0.05 were considered statistically not significant. Details of statistical analysis and the number of replicates can be found in the figure and dataset legends.
Results
C1QTNF1–AS1 lncRNA interacts with the RNA-binding protein RSRC2
C1QTNF1–AS1, which we previously identified as essential for mitotic progression in HeLa cells [9], is a conserved antisense lncRNA transcribed opposite the locus (Fig. 1A). To assess its role in cell division across distinct cellular backgrounds, we employed chromosomally stable and near-diploid RPE1 cells together with HCT116 colon carcinoma cells, two widely used models with well-characterized mitotic behaviours [48–52]. In both cell lines, smRNA FISH and ViewRNA FISH confirmed that C1QTNF1–AS1 is predominantly nuclear (Fig. 1B and Supplementary Fig. S1A–C). Multiple LNA gapmers [53] targeting different regions of the transcript efficiently depleted C1QTNF1–AS1 and abolished its RNA FISH signal (Fig. 1C, and Supplementary Fig. S1B and C), providing robust loss-of-function tools for mechanistic assays.
*C1QNTF1–AS1 interacts with RSRC2, an RBP of unknown function. (A) Schematic representation of the C1QTNF1–AS1 and C1QTNF1/CTRP1 genomic landscape (C1QTNF1–AS1 annotated as NR_040 018/NR_040 019 in RefSeq; Gencode gene ENSG00000265096; chr17:79019209-79027601, hg38). (B) Maximum intensity projections of representative images of C1QTNF1–AS1 exon smRNA FISH in HCT116 showing its nuclear localization. Nuclei were stained with DAPI (magenta) and outlined with a dashed circle. Right panel: quantification of total transcript in the nucleus (N) and cytoplasm (C), solid line represents the mean. N = 3 (n(HCT116) =398). (C) Expression levels of C1QTNF1–AS1 in HCT116 cells following depletion with three LNA gapmers targeting either exon 3 (LNA 1) or first intron of C1QTNF1–AS1 (LNA 2, 3), as measured by qPCR. Primers spanning mature (ex2-3) C1QTNF1–AS1 were used. Results are presented relative to negative control (Ctl A) LNA; N = 3. (D) Schematic representation of workflow for the ASO pulldown of C1QTNF1–AS1 in HCT116 cells. Five different ASOs targeting different regions of the C1QTNF1–AS1 locus were used, with luciferase (Luc) ASOs as a negative control. Pulldown efficacy was assessed by qPCR (Supplementary Fig. S1D), and proteins were identified using LC-MS analysis (see Materials and methods). (E) The volcano plot highlighting proteins enriched in C1QTNF1–AS1 pulldown using ASO 1, 3, and 5 versus Luc. Significant C1QTNF1–AS1 protein interactors are highlighted in red (FDR 5%). (F) RIP-qPCR from HCT116 extracts. Left panel: Western blot of RSRC2 in the input and IP samples to show RSRC2 IP efficiency compared to IgG. Right panel: RIP-qPCR showing association of RSRC2 with C1QTNF1–AS1 transcript. GAPDH was used as negative control RNA for RSRC2 RIP. RIP enrichments are presented as % of input RNA (normalized to IgG); N = 3. (G) Relative expression of RSRC2 in HCT116 cells following siRNA-mediated depletion of RSRC2, as measured by qPCR. Results are presented relative to control siRNA (Ctl); N = 3. (H) Representative western blot showing RSRC2 protein expression in HCT116 cells following siRNA-mediated depletion of RSRC2. β-Tubulin and Ponceau staining were used as loading controls. An asterisk indicates an unspecific RSRC2 protein band. (I) Densitometric analysis of RSRC2 levels from panel (H) relative to control siRNA (Ctl); N = 3. (J) Interaction intensities between C1QTNF1–AS1 and the indicated proteins show that C1QTNF1–AS1 interacts with RSRC2. In-cell interactions were measured in an incPRINT experiment where MS2-tagged C1QTNF1–AS1 RNA was co-expressed with a set of FLAG-tagged proteins in HEK293T cells harbouring a luciferase detector fused to the MS2 coat protein (MS2CP). Upon the formation of FLAG–protein–RNA–MS2–MS2CP ternary complexes, RNA–protein interactions were measured by luciferase activity [41]. eGFP (enhanced green fluorescent protein) was used as a negative control and PABPC3 (a polyadenylated RBP) was used to control for RNA expression. Xist(C)-MS2 vector was used alongside C1QTNF1AS1–MS2 as a positive control for RNA–protein interactions. RLU are relative light unit; N = 4. (K) Protein expression levels were estimated from horseradish peroxidase ELISA of FLAG-tagged proteins in the same experiment as in panel (J). Error bars in all panels are shown as mean ± S.E.M.; scale bar: 5 μm; N = number of cells analysed. An unpaired t-test with Welch’s correction was applied in panels (C), (G), and (I). Unpaired t-test was used in panel (F). * <0.05, **<0.01, and ***<0.0001.
As the first step towards defining C1QTNF1–AS1’s cellular function and molecular mechanism of action, we identified proteins interacting with the RNA by ASO RNA pulldown followed by mass spectrometry (Fig. 1D). Three independent oligonucleotide probes specifically purified C1QTNF1–AS1 from UV-crosslinked HCT116 lysates (Supplementary Fig. S1D). Two RBPs [54], RSRC2 and SRRM1, were significantly enriched relative to a luciferase (Luc) RNA control (Fig. 1E and Supplementary Table S1). Because SRRM1 depletion did not cause mitotic defects (Supplementary Fig. S1E–G), whereas RSRC2 has been implicated in mitotic progression in the Mitotic Cell Atlas [28], we focused on RSRC2 as the primary functional interactor.
Several orthogonal approaches validated the C1QTNF1–AS1–RSRC2 interaction. RNA immunoprecipitation (RIP) followed by qPCR (RIP-qPCR) with anti-RSRC2 enriched C1QTNF1–AS1 relative to GAPDH (Fig. 1F), and the specificity of the antibody was confirmed by RSRC2 knockdown at both the RNA and protein levels (Fig. 1G–I). We further confirmed the interaction in living cells using the incPRINT assay [41]. C1QTNF1–AS1–MS2 specifically interacted with RSRC2–FLAG but not with unrelated FLAG-tagged proteins, whereas the Xist-MS2–hnRNPU interaction served as a positive control (Fig. 1J) [55]. Parallel ELISA confirmed comparable expression of the tested proteins across all conditions (Fig. 1K). Together, these data demonstrate that C1QTNF1–AS1 directly interacts with the RBP RSRC2.
Loss of either C1QTNF1–AS1 or RSRC2 causes defects in chromosome alignment and mitotic progression in multiple cell lines
To examine the roles of C1QTNF1–AS1 and RSRC2 in cell division, we performed a systematic comparison of normal and cancer cell lines stained for kinetochores (CREST). C1QTNF1–AS1 depletion caused chromosome congression defects in 27%–36% of mitotic HCT116 cells, compared to 14% in controls (Fig. 2 A, B). Because lncRNA loci may function at either the RNA or DNA level [56], we employed CRISPR–Cas13 [35] to target the C1QTNF1–AS1 transcript independently. Cas13-mediated depletion of C1QTNF1–AS1 similarly induced chromosome congression defects (Supplementary Fig. S2A–C), confirming an RNA-dependent phenotype.
*C1QTNF1–AS1 and RSRC2 are critical for chromosome alignment in multiple cell lines. (A) Representative images of metaphase cells stained for kinetochores (CREST, cyan) and DNA (Hoescht, magenta) after LNA-mediated depletion of C1QTNF1–AS1 in HCT116 cells. (B) Characterization of mitotic phenotypes in HCT116 cells after LNA-mediated depletion of C1QTNF1–AS1. Percentage of mitotic cells with normal metaphase plate and congression defect using the same antibodies as in panel (A); N = 3 (n(Ctl A) =627; n(C1QTNF1–AS1LNA1) =578; n(C1QTNF1–AS1LNA2) =560; n(C1QTNF1–AS1LNA3) =586). (C) Representative images of metaphase HCT116 cells stained for kinetochores (CREST, blue) and DNA (Hoescht, cyan) following siRNA-mediated depletion of RSRC2. (D) Characterization of mitotic phenotypes in HCT116 cells after siRNA-mediated depletion of RSRC2. Percentage of mitotic cells with normal metaphase plate and congression defect using the same antibodies as in panel (C); N = 3 (n(Ctl si) =289; n(RSRC2 si) = 271). (E) Representative images of metaphase cells stained for kinetochores (CREST, cyan) and DNA (Hoescht, magenta) after LNA-mediated depletion of C1QTNF1–AS1 in RPE1 cells. (F) Characterization of mitotic phenotypes in RPE1 cells after LNA-mediated depletion of C1QTNF1–AS1. Percentage of mitotic cells with normal metaphase plate and congression defect using the same antibodies as in panel (E); N = 3 (n(Ctl A) =93; n(C1QTNF1–AS1LNA1) = 91; n(C1QTNF1–AS1LNA3) =63). (G) Representative images of metaphase RPE1 cells stained for kinetochores (CREST, blue) and DNA (Hoescht, cyan) following siRNA-mediated depletion of RSRC2. (H) Characterization of mitotic phenotypes in RPE1 cells after siRNA-mediated depletion of RSRC2. Percentage of mitotic cells with regular metaphase plate and congression defect using the same antibodies as in panel (G); N = 4 (n(Ctl si) =116; n(RSRC2 si) =65). (I) Representative images of metaphase cells stained for kinetochores (CREST, cyan) and DNA (Hoescht, magenta) after LNA-mediated depletion of C1QTNF1–AS1 in OE19 cells. (J) Characterization of mitotic phenotypes in OE19 cells after LNA-mediated depletion of C1QTNF1–AS1. Percentage of mitotic cells with normal metaphase plate and congression defect using the same antibodies as in panel (I); N = 3 (n(Ctl A) =192; n(C1QTNF1–AS1LNA1) = 198; n(C1QTNF1–AS1LNA3) = 224). (K) Representative images of metaphase OE19 cells stained for kinetochores (CREST, blue) and DNA (Hoescht, cyan) following siRNA-mediated depletion of RSRC2. (L) Characterization of mitotic phenotypes in OE19 cells after siRNA-mediated depletion of RSRC2. Percentage of mitotic cells with normal metaphase plate and congression defect using the same antibodies as in panel (K); N = 3 (n(Ctl si) =201; n(RSRC2 si) = 194). (M) Representative images of metaphase cells stained for kinetochores (CREST, cyan) and DNA (Hoescht, magenta) after LNA-mediated depletion of C1QTNF1–AS1 in HepG2 cells. (N) Characterization of mitotic phenotypes in HepG2 cells after LNA-mediated depletion of C1QTNF1–AS1. Percentage of mitotic cells with normal metaphase plate and congression defect using the same antibodies as in panel (M); N = 4 (n(Ctl A) = 172; n(C1QTNF1–AS1LNA1) =174; n(C1QTNF1-AS1LNA3) = 184). (O) Representative images of metaphase HepG2 cells stained for kinetochores (CREST, blue) and DNA (Hoescht, cyan) following siRNA-mediated depletion of RSRC2. (P) Characterization of mitotic phenotypes in HepG2 cells after siRNA-mediated depletion of RSRC2. Percentage of mitotic cells with normal metaphase plate and congression defect using the same antibodies as in panel (O); N = 5 (n(Ctl si) =261; n(RSRC2 si) = 307). Error bars in all panels are shown as mean ± S.E.M.; scale bar: 5 μm. An unpaired t-test with Welch’s correction was applied in panels (B), (D), (F), (H), (J), (L), (N), and (P). * <0.05, **<0.01, and **<0.001. C1 = C1QTNF1–AS1.
RSRC2-depleted HCT116 cells displayed a comparable increase in misaligned chromosomes (Fig. 2C and D) and the phenotypes were recapitulated in RPE1 cells (Fig. 2E–H, and Supplementary Fig. S2D and E). We further validated the phenotype in two cancer cell lines in which RSRC2 has been linked to proliferation [57, 58] (OE-19 and MDA-MB-231) and in HepG2, which expresses C1QTNF1–AS1 at high levels [59]. Depletion of either gene again induced chromosome congregation defects (Fig. 2I–P and Supplementary Fig. S2F–L), indicating that C1QTNF1–AS1 and RSRC2 function in mitosis is not cell-type dependent.
We further confirmed that C1QTNF1–AS1 and RSRC2 downregulation causes mitotic delay by performing time-lapse imaging of HCT116 cells stably expressing GFP-tagged histone H2B (H2B–GFP). Depletion of C1QTNF1–AS1 or RSRC2 markedly delayed anaphase onset relative to controls (from ∼37 min to 45–60 min and 57 min, respectively; Supplementary Fig. S3A–D and Supplementary Movies S1–5). In RPE1 H2B–GFP cells, the delay was smaller (2–4 min) but highly reproducible (Supplementary Fig. S3E–H and Supplementary Movies S6–10). Although modest in absolute terms, even brief delays are well documented to compromise chromosome segregation fidelity and to allow checkpoint adaptation in RPE1 cells [60, 61]. The difference in severity between HCT116 and RPE1 likely stems from differences in cell cycle regulation and checkpoint stringency rather than in knockdown efficiency. Overexpression of full-length C1QTNF1–AS1 did not affect mitotic timing or its subcellular localization (Supplementary Fig. S2M–O), indicating that the lncRNA acts through its endogenous transcript rather than dosage effects. To test whether C1QTNF1–AS1 and RSRC2 function in the same pathway, we performed co-depletion experiments. Simultaneous loss of both factors did not exacerbate mitotic duration beyond single depletions (Supplementary Fig. S2P and Q), consistent with C1QTNF1–AS1 and RSRC2 acting within a shared regulatory mechanism rather than in parallel pathways.
C1QTNF1–AS1 and RSRC2 loss elevates the mitotic index and activates SAC
Mitotic duration is sensed by the p53-based mitotic stopwatch pathway [62–64]. To determine whether mitotic defects caused by C1QTNF1–AS1 or RSRC2 loss feed into this checkpoint response and impair long-term cell proliferation, we performed Incucyte analysis, which measures cell confluence, and a cell-doubling assay. Loss of C1QTNF1–AS1 had no major effect on cell proliferation, whereas RSRC2-depleted cells displayed a small but not significant reduction in cell coverage at 48–72 h (Supplementary Fig. S4A). In contrast, cell-doubling assays revealed reduced cell numbers at days 5–7 in both conditions, with a more pronounced effect upon RSRC2 depletion (Supplementary Fig. S4B). We also observed a reduction in cell viability, confirming increased cell death by days 5–7, particularly in RSRC2-depleted cells (Supplementary Fig. S4C and D).
To assess potential cell-cycle defects, we analysed profiles at early time points and found no major alterations upon C1QTNF1–AS1 or RSRC2 depletion (Supplementary Fig. S5A–D). Furthermore, neither C1QTNF1–AS1 nor RSRC2 caused a G2/M arrest at 72 or 96 h after depletion (Supplementary Fig. S5E–H), or in synchronized cells (Supplementary Fig. S5I–L). Thus, modest mitotic delay can impair long-term proliferation and cell viability without substantially altering cell cycle progression.
We next asked whether mitotic defects observed upon loss of C1QTNF1–AS1 or RSRC2 activate the SAC [65], which can delay mitotic progression without triggering a G2/M arrest. IF for MAD2, a core member of the SAC signalling machinery, revealed robust SAC activation: 67%–73% of C1QTNF1–AS1-depleted cells and 65% of RSRC2-depleted cells exhibited MAD2-positive kinetochores on uncongressed chromosomes, compared to 28% in control cells (Fig. 3A–F). This phenotype was also observed in RPE1 cells (Supplementary Fig. S6A–D). Consistent with SAC proficiency, nocodazole treatment caused prolonged mitotic arrest in both control and RSRC2- or C1QTNF1–AS1-depleted HCT116 H2B cells (Fig. 3G). These data, combined, indicate that SAC is fully functional upon depletion of both RSRC2 and C1QTNF1–AS1. Furthermore, we quantified the percentage of mitotic cells (mitotic index), as an increase in this index suggests a delay or arrest in mitosis [9]. The mitotic index increased following depletion of either RSRC2 or C1QTNF1–AS1 (Fig. 3H), consistent with a mitotic delay rather than a cell cycle arrest. Given the modest rise in mitotic index, it is unlikely that these delays exceed the threshold required to induce G2/M arrest [63, 64]. Thus, loss of C1QTNF1–AS1 or RSRC2 activates the SAC and elevates the mitotic index, providing a mechanistic explanation for the observed mitotic delay.
Loss of C1QTNF1–AS1 or RSRC2 results in increased mitotic index and intact SAC signalling. (A) Representative images of metaphase cells stained for kinetochores (CREST, cyan), SAC (Mad2, green), and DNA (Hoescht, magenta) after LNA-mediated depletion of C1QTNF1–AS1 in HCT116 cells. (B) Insets show examples of MAD2 and CREST staining from C1QTNF1–AS1-depleted cells, as in panel (A); scale bar: 0.5 µm. (C) Characterization of SAC activity in HCT116 cells after LNA-mediated depletion of C1QTNF1–AS1. Percentage of mitotic cells with Mad2 signal on misaligned chromosomes using the same antibodies as in panel (A); N = 3 (n(Ctl A) =67; n(C1QTNF1–AS1LNA1) = 69; n(C1QTNF1–AS1LNA2) = 72; n(C1QTNF1–AS1LNA3) =93). (D) Representative images of metaphase HCT116 cells stained for kinetochores (CREST, cyan), SAC (Mad2, green), and DNA (Hoescht, magenta) following siRNA-mediated depletion of RSRC2. (E) Insets show examples of MAD2 and CREST staining from RSRC2-depleted cells, as in panel (D); scale bar: 0.5 µm. (F) Characterization of SAC activity in HCT116 cells after siRNA-mediated depletion of RSRC2. Percentage of mitotic cells with Mad2 signal on misaligned chromosomes using the same antibodies as in panel (C); N = 4 (n(Ctl si) = 108; n(RSRC2 si) = 126). (G) Quantification of mitotic duration from time-lapse imaging microscopy following LNA-mediated depletion of C1QTNF1–AS1 (left panel) and siRNA-mediated depletion of RSRC2 (right panel) in HCT116 H2B–GFP cells treated with nocodazole. N = 3 (n(Ctl A) = 49; n(C1QTNF1–AS1LNA1) = 47; n(C1QTNF1–AS1LNA2) = 50; n(C1QTNF1–AS1LNA3) =51, n(Ctl si) = 87; n(RSRC2 si) = 51). (H) Changes in mitotic index after LNA-mediated depletion of C1QTNF1–AS1 (left panel) and siRNA-mediated depletion of RSRC2 (right panel) in HCT116 cells. Percentage of cells in mitosis shown based on Histone–H3–pS10 staining; N = 3 (n(Ctl A) = 119; n(C1QTNF1–AS1LNA1) = 141; n(C1QTNF1–AS1LNA3) = 136; n(Ctl si) = 129; n(RSRC2 si) = 204). Error bars in all panels are shown as mean ± S.E.M.; scale bar: 5 μm. An unpaired t-test with Welch’s correction was applied in panels (C) and (F). The Mann–Whitney test was used in panel (G). A one-way ANOVA (left panel) and an unpaired t-test (right panel) were used in panel (H). <0.05, **<0.01, ***<0.001 and ***<0.0001.
Finally, we considered whether these defects may relate to CIN, a hallmark of cancer that arises from mitotic defects [66]. Pan-cancer TCGA analysis revealed that RSRC2 expression is reduced in several tumour types, and its lower expression associates with poor prognosis (Supplementary Fig. S6E and F, and Supplementary Table S2). C1QTNF1–AS1 was downregulated in some cancers, including adrenal cortical carcinoma, consistent with its tissue expression pattern [9] (Supplementary Fig. S6G and H, and Supplementary Table S2). We also investigated the relationship between CIN [measured via the weighted genome instability index (wGII) [49]] and the expression of RSRC2 and C1QTNF1–AS1 in a panel of 936 cancer cell lines from the Depmap portal [67]. A modest negative correlation was observed between RSRC2 and wGII, while no significant correlation was observed for C1QTNF1–AS1 (Supplementary Fig. S6I and J). As expected, AURKA positively correlated with CIN status [68] (Supplementary Fig. S6K). In colorectal cancer cell lines with defined CIN status [68], RSRC2 tended to be lower in CIN + cells, while C1QTNF1–AS1 was slightly elevated (Supplementary Fig. S6L–N), potentially reflecting compensatory upregulation.
Together, our data demonstrate that loss of C1QTNF1–AS1 and RSRC2 induces mitotic defects across multiple cell lines and is associated with SAC activation, suggesting a potential role for dysregulation of this pathway in CIN.
C1QTNF1 is not involved in cell division and is independent of C1QTNF1–AS1 regulation
Because antisense lncRNAs frequently regulate the expression or processing of their overlapping protein-coding genes [56], we first investigated whether C1QTNF1–AS1 modulates the expression of the adjacent C1QTNF1–CTRP1 gene. Depletion of C1QTNF1–AS1 using LNA gapmers did not alter C1QTNF1–CTRP1 mRNA levels in HCT116 cells (Supplementary Fig. S7A). This finding was further confirmed using CRISPR–Cas13-mediated knockdown of C1QTNF1–AS1 (Supplementary Fig. S7B). Furthermore, C1QTNF1–AS1 depletion also did not affect C1QTNF1–CTRP1 protein levels (Supplementary Fig. S7C and D). Since antisense lncRNAs can regulate splicing of their sense pre-mRNA partners [69], we additionally examined all annotated C1QTNF1–CTRP1 isoforms and found no changes in splice-variant expression (Supplementary Fig. S7E). Thus, C1QTNF1–AS1 does not regulate C1QTNF1–CTRP1 expression or splicing.
We next asked whether C1QTNF1–CTRP1 contributes directly to mitosis. Knockdown of C1QTNF1–CTRP1 (Supplementary Fig. S7F–H) did not induce chromosome congression defects (Supplementary Fig. S7I and J) or mitotic delay in RPE1 H2B–GFP cells (Supplementary Fig. S7K and L). Together, these results demonstrate that C1QTNF1–CTRP1 is neither regulated by its neighbouring antisense lncRNA nor required for mitotic progression, and therefore the mitotic functions of C1QTNF1–AS1 are independent of its neighbouring protein-coding gene.
RSRC2 localizes to the nuclear speckles through its N-terminal IDR domain
RSRC2 (arginine- and serine-rich coiled-coil 2) contains an arginine- and serine-rich region, two coiled-coil domains, and a small acidic protein-like (SMAP) domain (Fig. 4A). Although RSRC2 has been identified as an mRNA-interacting protein [54] and reported to inhibit the proliferation of cancer cells [57, 58], its molecular function remains poorly characterized. Notably, RSRC2 harbours extensive N-terminal stretches of amino acids predicted to form intrinsically disordered regions (IDRs) (Fig. 4B), a feature commonly associated with non-canonical RNA binding [14]. Since the serine- and arginine-rich (SR) protein class of SFs contains arginine and serine residues, we tested whether RSRC2 localizes to the nuclear speckles, which are enriched in RNA and SFs [70].
RSRC2 localization to nuclear speckles is driven by its N-terminal intrinsically disordered region. (A) Schematic representation of the annotated domain structure of the human RSRC2 protein. RSRC2 contains an N-terminal disordered region, two coiled-coil domains and a C-terminal SMAP domain. (B) Computational disorder prediction for the human RSRC2 protein sequence using IUPred2A (https://iupred2a.elte.hu/) to predict the likelihood of disorder for RSRC2 through the ANCHOR2 or IUPred2 algorithms. Higher disorder scores indicate a greater likelihood that the protein region is intrinsically disordered. (C) Representative IF images of interphase HCT116 cells stained with RSRC2 (green) and DNA (Hoescht, magenta) after siRNA-mediated depletion of RSRC2. (D) Quantification of RSRC2 nuclear intensity in HCT116 cells after siRNA-mediated depletion of RSRC2. Results are presented relative to control siRNA (Ctl) treated cells; N = 3 (n(Ctl si) = 202; n(RSRC2 si) = 179). (E) Representative images of interphase HCT116 cells stained for nuclear speckles (SC35, cyan) and RSRC2 (green) after siRNA-mediated depletion of RSRC2. DAPI is shown in magenta. Threshold masks used for colocalization analysis are shown. The orange box indicates the section taken for the inset images shown in panel (F). The correlation plot shows the level of correlation between thresholded SC35 and thresholded RSRC2 signal, with average Manders correlation coefficient for each condition: RCtlsi = 0.5347; RRSRC2si = 0.1391. (F) Insets show examples of SC35 and RSRC2 staining from RSRC2-depleted cells, as in panel (E); scale bar: 0.5 µm. (G) Quantitative colocalization analysis of SC35 and RSRC2 in HCT116 cells following siRNA-mediated depletion of RSRC2 (as shown in panel E); N = 3 (n(Ctl si) =154; n(RSRC2si) =153). (H) Schematic representation of RSRC2 and its deletion constructs, all containing an eGFP tag at the N-terminus. Full-length RSRC2 protein shown (RSRC2-Full) along with deletion constructs without the IDR (RSRC2-ΔIDR), CC1 (RSRC2-ΔCC1), CC2 (RSRC2-ΔCC2), and SMAP (RSRC2-ΔSMAP) domains. (I) Western blot showing GFP protein expression (indicating exogenous RSRC2) in HCT116 cells transfected with the RSRC2 and its deletion constructs as outlined in panel (H). GAPDH and Ponceau staining were used as loading controls. (J) Representative images of HCT116 cells transfected with RSRC2 and its deletion constructs, as outlined in panel (H). The cells were stained for GFP (exogenous RSRC2, green) and nuclear speckles (SC35, cyan). DAPI is shown in magenta. Error bars in all panels are shown as mean ± S.E.M.; scale bar: 5 μm. N = number of cells analysed. The following statistics were applied: a one-sample t-test (comparing to a hypothetical mean of 1) was used in panel (D), and an unpaired t-test in panel (G).**<0.01.
Endogenous RSRC2 was predominantly nuclear (Fig. 4C and D) and formed nuclear puncta resembling speckles. Co-immunostaining with the speckle marker SC35 [70] showed substantial colocalization, which was reduced upon RSRC2 knockdown (Fig. 4E–G). To determine which domains mediate this localization, we generated RSRC2 mutants lacking the IDR, coiled-coil domain 1 (CC1), coiled-coil domain 2 (CC2), or SMAP, and expressed them in HCT116 cells (Fig. 4H and I). Full-length RSRC2 localized to nuclear speckles, whereas deletion of the IDR domain, but not CC1, CC2, or SMAP, abolished RSRC2’s speckle localization without affecting the nuclear speckle structure (Fig. 4J).
Thus, although RSRC2 has been noted in the RBP datasets [71], our results reveal how it is targeted to nuclear speckles and identify the IDR as the structural determinant of this localization. Given that many speckle-associated IDR-containing proteins regulate pre-mRNA processing [72–75] and that RSRC2 binds mRNA [54], we propose that RSRC2 acts as a previously uncharacterized trans-acting factor that facilitates splicing.
RSRC2 protein, but not C1QTNF1–AS1, is a splicing regulator
Given the established roles of RSRC2 and C1QTNF1–AS1 in cell division, we investigated whether their interaction influences gene expression and alternative splicing (AS) during this process. We performed RNA sequencing (RNA-seq) following LNA-mediated depletion of C1QTNF1–AS1 or siRNA-mediated depletion of RSRC2 in HCT116 cells. We selected DEGs (adjusted P < 0.01 and logFC > |1|) that consistently changed in the same direction across three independent LNA gapmers targeting C1QTNF1–AS1, a strategy previously shown to minimize off-target effects [9]. Depletion of C1QTNF1–AS1 with LNA 1, 2, and 3 gapmers led to 78, 5, and 96 DEGs, respectively (Fig. 5A and Supplementary Table S3). The small number of DEGs in LNA gapmer 2 likely underlies the absence of overlap across all three sets (Fig. 5B). Because off-target effects of individual gapmers have been reported [76], we assessed if the overlap between LNA gapmers 1 and 3 exceeded chance. A hypergeometric test confirmed that the 10 shared DEGs are significantly enriched (P = 2.14 × 10^−^¹⁰). Additionally, correlation analysis of the unfiltered and filtered DEG lists also showed positive correlation (Fig. 5C and D). Thus, the transcriptomic effects result from on-target perturbation of C1QTNF1–AS1 rather than non-specific gapmer artefacts, consistent with previous lncRNA studies [77].
RSRC2, but not C1QTNF1–AS1, regulates AS of genes enriched for organelle and actin cytoskeleton functions. (A) Volcano plot of transcriptional differences induced by LNA-mediated depletion of C1QTNF1–AS1 in HCT116 cells compared to negative control LNA A (Ctl LNA A). The black horizontal line represents the significance threshold corresponding to a padj of 0.01. Black vertical lines are log2-fold change thresholds of |logFC| > 1. The genes that do not show significant changes in expression are shown in black. (B) Venn diagram illustrating overlap of the sets of DEGs following depletion of C1QTNF1–AS1 in HCT116 cells with LNA 1, LNA 2, and LNA 3 targeting C1QTNF1–AS1. Only 10 DEGs were in common between LNA 1 and LNA 3 targeting C1QTNF1–AS1 using the log2-fold change threshold of |logFC| > 1 and FDR of 1%. (C) Correlation plot on an unfiltered list of transcriptional differences induced by LNA-mediated depletion of C1QTNF1–AS1 in HCT116 cells, comparing LNA gapmer 1 to 3. Both Spearman (R(Sp) = 0.507, P < 0.0001) and Pearson (R(Pe) = 0.589, P < 0.0001) correlations were performed. (D) Correlation plot of transcriptional differences induced by LNA-mediated depletion of C1QTNF1–AS1 in HCT116 cells comparing LNA gapmer 1 to 3, following filtering by P-value and fold change thresholds defined in panel (A). Both Spearman (R(Sp) = 0.723, P < 0.0001) and Pearson (R(Pe) = 0.885, P < 0.0001) correlations were performed. (E) Heat map of 10 common DEGs from LNA 1- and LNA 3-mediated depletion of C1QTNF1-AS1 (as shown in panel B). (F) Volcano plot of transcriptional differences induced by siRNA-mediated depletion of RSRC2 in HCT116 cells compared to negative control siRNA. The black horizontal line represents the significance threshold corresponding to a padj of 0.01. Black vertical lines are log2-fold change thresholds of |logFC| > 1. The genes that do not show significant changes in expression are shown in black. (G) Venn diagram illustrating no overlap between the sets of DEGs identified after siRNA- and LNA-mediated depletion of RSRC2 and C1QTNF1–AS1 in HCT116 cells. LNA 2 targeting C1QTNF1–AS1 was omitted in this analysis due to the low number of DEGs. (H) Volcano plot of FDR (log10) versus dPSI of all AS events upon C1QTNF1–AS1 depletion in HCT116 cells. Differential splicing analysis of RNA-seq data were performed with the rMATS (FDR < 0.05 and |dPSI| > 0.1). The genes that do not show significant changes in AS are shown in black. (I) Venn diagram showing no overlap of AS events following LNA-mediated depletion of C1QTNF1–AS1 in HCT116 cells. (J) Volcano plot of all AS events upon RSRC2 silencing in HCT116 cells (as in panel F). Key genes of interest are highlighted: PCNT (green), CDK5RAP2 (blue), and CENPE (red). (K) Pie chart showing the distribution of types of AS events enriched in RSRC2-depleted HCT116 cells based on rMATS analysis. A3SS, alternative 3′ splice site; A5SS, alternative 5′ splice site; MXE, mutually exclusive exons; RI, retained intron; SE, skipped exon. (L) Table categorizing the number of total significant AS events following C1QTNF1–AS1- and RSRC2 depletion in HCT116 cells (FDR < 0.05 and |dPSI| > 0.1).
Gene ontology (GO) enrichment analysis following C1QTNF1–AS1 depletion identified enrichment for developmental and cell division-associated pathways. (Supplementary Fig. S8A and Supplementary Table S4). Among the 10 candidate targets, DIAPH2, PLK2, and PTPRG have known mitotic functions [28, 78] (Fig. 5E). qPCR validation showed that DIAPH2 was reduced following C1QTNF1–AS1 depletion with two independent LNAs (Supplementary Fig. S8B), which was also confirmed at the protein level (Supplementary Fig. S8C and D). DIAPH2 did not respond to RSRC2 knockdown (Supplementary Fig. S8E–G), indicating it is not a shared target. Given the modest number of DEGs, these results suggest that C1QTNF1–AS1 does not exert broad transcriptional control in HCT116 cells, consistent with our earlier finding [9]. In contrast, RSRC2 depletion resulted in 259 DEGs (Fig. 5F and Supplementary Table S5) enriched for developmental and differentiation pathways (Supplementary Fig. S8H and Supplementary Table S6). The DEG list from C1QTNF1–AS1 and RSRC2 did not intersect (Fig. 5G), indicating that the two factors do not co-regulate gene expression at the mRNA abundance level.
Since RSRC2 was present in the nuclear speckles, we considered that RSRC2 and C1QTNF1–AS1 could be involved in gene regulation at the level of splicing and, therefore, co-regulate a common set of transcripts with functions in mitosis. Differential splicing analysis using rMATS [79] (FDR < 0.05, |ΔPSI| > 0.1, where PSI is the Percentage Splicing Inclusion) identified 10, 3, and 15 AS events after depletion of C1QTNF1–AS1 using LNA 1, 2, and 3 gapmer, respectively (Fig. 5H). No mis-splicing event was shared across all three LNA gapmers (Fig. 5I and Supplementary Table S7), and the low number of events indicates that C1QTNF1–AS1 does not participate broadly in splicing.
In contrast, RSRC2 loss resulted in 1105 AS events (Fig. 5J and Supplementary Table S8) dominated by exon skipping and alternative 3′ splice sites usage (Fig. 5K). These 1105 AS events map to 876 differentially spliced genes. GO analysis of these genes revealed strong enrichment for pathways related to cytoskeletal organization, organelle organization, such as centrosome and cell cycle regulation (Supplementary Fig. S8I and J, and Supplementary Table S9). Notably, several RSRC2-dependent splicing events occurred in key mitotic regulators, including pericentrin (PCNT), cyclin-dependent kinase 5 regulatory subunit-associated protein 2 (CDK5RAP2/CEP215), and the kinesin-like motor protein centrosome-associated protein E (CENPE) (Fig. 5J), suggesting a link between RSRC2-mediated splicing and centrosome integrity. By comparing the number of differential AS events between RSRC2- and C1QTNF1–AS1-depleted cells, we further confirmed that the loss of C1QTNF1–AS1 did not lead to splicing changes, in contrast to RSRC2 depletion (Fig. 5L). Altogether, these data demonstrate that RSRC2 is the major splicing regulator, whereas C1QTNF1–AS1 exerts its mitotic function independently of AS.
RSRC2 interacts with splicing and centrosomal proteins and is required for efficient splicing of mitotic genes
To investigate how RSRC2 regulates AS, we conducted an immunoprecipitation coupled with mass-spectrometry (IP-MS) to identify either RNA-mediated protein–protein or direct protein–protein interactions of RSRC2. As shown for other RBPs using similar approach [80], RSRC2 was efficiently and specifically enriched relative to IgG controls in HCT116 lysates, both in the presence and absence of RNase A/T1 (Supplementary Fig. S9A). We confirmed that RNase A/T1 was effective in degrading the RNA (Supplementary Fig. S9B). The analysis of the RSRC2 interactome without RNase A/T1 identified 123 proteins (Fig. 6A and Supplementary Table S10) that were significantly enriched in the RSRC2 IP. GO analysis showed enrichment for splicing proteins, centrosome components, ribosome biogenesis, and cell adhesion pathway (Fig. 6B). Interactions included U5 snRNP [81] components (e.g. PRPF8) and nuclear speckle proteins (SRRM2, SON) [70, 82]. Notably, RSRC2 interacts with the PCNT and CDK5RAP2, key pericentriolar material (PCM) proteins that, together with the centrioles, form a centrosome, a major microtubule-organizing centre that contributes to mitotic fidelity [83, 84]. Co-IP confirmed the RSRC2 interactions with PCNT, CDK5RAP2, and PRPF8, a pre-mRNA processing factor 8, a U5 snRNP component (Supplementary Fig. S9C). These data suggest that RSRC2 assembles multiple ribonucleoprotein complexes, consistent with its association with nuclear speckles and centrosomes. To determine whether these interactions are RNA-dependent, we repeated the IP-MS with RNase A/T1. The interaction profile of RSRC2 remained largely unchanged (Supplementary Fig. S9D and Supplementary Table S10) and GO enrichment was similar to untreated conditions (Supplementary Fig. S9E), indicating that most RSRC2 protein–protein interactions are RNA-independent.
RSRC2 interacts with spliceosome and centrosome proteins and is required for efficient splicing of mitotic genes. (A) Volcano plot of protein–protein interactions for RSRC2 comparing enriched proteins of the RSRC2 IP versus IgG IP in HCT116 cells without RNase A/T1 treatment. Curved lines mark the significance boundary (FDR = 1%, at least two peptides detected in each biological replicate). Each dot represents a protein. Selected splicing and centrosome proteins are indicated as one of the top RSRC2 interactors; N = 4. Statistically significant differences were detected using a two-tailed, two-sample t-test with permutation-based FDR. (B) Fisher’s exact test analysis of known protein categories that are over-represented among the RSRC2 interacting proteins (Benjamini–Hochberg FDR < 0.05). Each circle represents an enriched category from the Gene Ontology Cellular Compartments (GOCC) database, with the circle size representing the number of shared proteins. (C) Sashimi plot showing the skipping of CENPE exon 17 (the skipped exon is highlighted with a grey rectangle) in RSRC2-depleted HCT116 cells. The tracks were constructed from averages of four biological replicates for each condition (RSRC2 si in red; Ctl si in grey). (D) PSI of CENPE exon 17, as calculated from RNA-Seq data of RSRC2-depleted HCT116 cells (as shown in panel C), was -0.1 (dPSI); N = 4. (E) RT-PCR analysis of exon skipping in CENPE after RSRC2-mediated depletion compared to control siRNA. PCR products were amplified using primers targeting the CENPE long and short isoforms, with and without exon 17, and were visualized by capillary gel electrophoresis. (F) Quantification of PSI for CENPE exon 17 based on capillary gel electrophoresis (shown in panel E) following RSRC2 depletion; N = 3. (G) Sashimi plot showing skipping of CDK5RAP2 exon 19 (skipped exon is highlighted with a grey rectangle) in RSRC2-depleted HCT116 cells. The tracks were constructed from averages of four biological replicates for each condition (RSRC2 si in red; Ctl si in grey). (H) PSI of CDK5RAP2 exon 19 was calculated from RNA-seq data of RSRC2-depleted cells (as shown in panel G) and was −0.12747 (dPSI); N = 4. (I) RT-PCR analysis of exon skipping in CDK5RAP2 after RSRC2-mediated depletion compared to control siRNA. PCR products were amplified with primers targeting CDK5RAP2 and visualized by capillary gel electrophoresis. (J) Quantification of PSI for CDK5RAP2 exon 19 based on capillary gel electrophoresis (shown in panel I) following RSRC2 depletion; N = 3. (K) Sashimi plot showing skipping of PCNT exon 19 (skipped exon is highlighted with a grey rectangle) in RSRC2-depleted HCT116 cells. The tracks were constructed from averages of four biological replicates for each condition (RSRC2 si in red; Ctl si in grey). (L) PSI of PCNT exon 19 was calculated from RNA-seq data of RSRC2-depleted HCT116 cells (as shown in panel K), and the difference was −0.17421 (dPSI); N = 4. (M) RT-PCR analysis of exon skipping in PCNT after RSRC2-mediated depletion compared to control siRNA. PCR products were amplified using primers against PCNT and were visualized using capillary gel electrophoresis. (N) Quantification of PSI for PCNT exon 19 based on capillary gel electrophoresis (shown in panel M) following RSRC2 depletion; N = 3. (O) Right panel: qPCR analysis of GAPDH, CENPE, PCNT, CDK5RAP2, and C1QTNF1-AS1 in UV-RIP samples after the RSRC2 IP from HCT116 extracts. GAPDH was used as a negative control RNA for the RSRC2. UV-RIP enrichments are presented as % of input RNA (normalized to IgG); N = 4. Left panel: Western blot to show RSRC2 IP efficiency compared to IgG. (P) Capillary gel electrophoresis analysis reveals the splicing pattern of the PCNT and CENPE minigenes from HCT116 cells co-transfected with either empty vector [1] or RSRC2 full-length expression vector [2]. PCR products were amplified with primers specific to the pxJ41 backbone vector and visualized by capillary gel electrophoresis. (Q) Quantification of PSI for PCNT exon 19 and CENPE exon 17 based on capillary gel electrophoresis (shown in panel P) following overexpression of RSRC2; N = 3. Error bars in all panels are shown as mean ± S.E.M. An unpaired t-test was used in panels (D), (F), (H), (J), (L), (N), (O), and (Q). <0.05, **<0.01, and **<0.001.
Several of the top RSRC2 interactors were also identified as RSRC2-dependent splicing targets. Since aberrant CENPE splicing is associated with chromosome congression defects [85], we validated the exon-skipping events in our splicing analysis. The sashimi plot confirmed skipping of exon 17 of CENPE following RSRC2 depletion (Fig. 6C and D). Additionally, RT-PCR analysis validated the CENPE exon splicing by capillary gel electrophoresis (Fig. 6E and F). Interestingly, PCNT and CDK5RAP2, the same proteins with which RSRC2 interacts, are also RSRC2 splicing targets. RSRC2 knockdown resulted in the skipping of CDK5RAP2 exon 19 (Fig. 6G and H), a finding further confirmed by capillary gel electrophoresis in a separate set of experiments (Fig. 6I and J). Our splicing analysis also revealed exon skipping of PCNT in RSRC2-depleted cells, introducing a premature stop codon (Fig. 6K and L). Capillary gel electrophoresis verified that PCNT exon 19 is skipped in RSRC2-depleted cells (Fig. 6M and N).
To test whether RSRC2 directly binds these transcripts, we performed UV crosslinking and immunoprecipitation (UV-RIP) of endogenous RSRC2, followed by qPCR. RSRC2 bound PCNT, CENPE, and CDK5RAP2 transcripts but not control RNA (Fig. 6O), and also C1QTNF1–AS1, as shown in Fig. 1F. To further test functional involvement in splicing, we used minigene reporters: exon 19 of PCNT and exon 17 of CENPE, including flanking intronic sequences, which were cloned into the spliced β-globin splicing backbone [86]. RSRC2 overexpression promotes splicing activity. In contrast to its depletion, RSRC2 overexpression favoured inclusion of the respective exons (Fig. 6P, Q), the inverse of its depletion phenotype.
Overall, these data demonstrate that RSRC2 interacts with splicing and centrosome proteins, binds its target RNAs directly, and is required for efficient splicing of a defined group of mitotic regulators.
PCNT-targeting ASO induces exon 19 skipping and mitotic defects
To directly test whether RSRC2-dependent splicing contributes to the mitotic defects, we focused on PCNT for three reasons: (i) PCNT splicing is regulated by SON [22, 23], which we identified as an RSRC2 interactor (Fig. 6A), (ii) PCNT depletion induces chromosome congression defects [87] and (iii) PCNT and CDK5RAP2 were the strongest candidates linking RSRC2 to centrosome integrity (Fig 7). Consistent with previous findings [87], PCNT depletion reproduced the chromosome congression defects in HCT116 cells (Supplementary Fig. S10A and B, and Fig. 7N). Next, we used splice-switching ASOs [88] to specifically block the 3′ splice site (3′SS) of exon 19 and assessed their effect on PCNT splicing. ASO–3′SS, but not a scrambled control (ASO–Ctl), induced exon skipping (Supplementary Fig. S10C and D), reduced PCNT protein levels (Supplementary Fig. S10E and F) and increased chromosomal congression defects (Supplementary Fig. S10G and H). Thus, aberrant splicing of PCNT is sufficient to phenocopy the RSRC2-depletion phenotype, demonstrating a direct causal link between RSRC2-mediated splicing and mitotic fidelity.
RSRC2 is a PCM protein that functions in centriole structural integrity and proper localization of centrosomal proteins. (A) Representative confocal images of mitotic and interphase RPE1 cells stained for PCNT or RSRC2 (green), Centrin2/3 (yellow), β-tubulin (magenta), and Hoechst (blue). Scale bars: 10 and 1.5 μm, as indicated. (B) Left panel: PCNT area (μm2) in mitotic (n(mitotic) = 74) and interphase cells (n(interphase) = 116); Centre panel: RSRC2 area (μm2) in mitotic (n(mitotic) =80) and interphase cells (n(interphase) = 67); Right panel: Centrin2/3 area (μm2) in mitotic (n(mitotic) =66) and interphase cells (n(interphase) = 68); N = 3. (C) Confocal images of expanded HCT116 centrioles stained with acetylated tubulin (cyan) following siRNA-mediated depletion of RSRC2 (n(Ctl si) = 270; n(RSRC2si) = 163); scale bar: 15 μm. (D) Categories of centriole stability defects scored in HCT116 cells after siRNA-mediated depletion of RSRC2, as shown in panel (C); N = 3. (E) Representative images of mitotic HCT116 cells stained for RSRC2 (magenta), centrosomes (Cen2/3, orange), microtubules (β‐tubulin, grey), and DNA (Hoescht, cyan) after siRNA-mediated depletion of RSRC2. (F) Quantification of centrosomal RSRC2 intensity in HCT116 cells from sum projected images obtained from panel (E). Results are presented relative to control siRNA (Ctl) treated cells; N = 4 (n(Ctl si) = 221; n(RSRC2 si)) = 226). (G) Quantification of centrosomal Cen2/3 intensity in HCT116 cells from sum projected images obtained from panel (E). Results are presented relative to control siRNA (Ctl) treated cells; N = 4 (n(Ctl si) =220; n(RSRC2 si)) = 231). (H) Representative images of mitotic HCT116 cells stained for PCM (PCNT, magenta), centrioles (Cen2/3, orange), and DNA (Hoescht, cyan) after siRNA-mediated depletion of RSRC2. Centrosome contains a pair of centrioles and surrounding PCM matrix (PCNT, CDK5RAP2). (I) Quantification of PCNT signal intensity around the centrosome of mitotic HCT116 cells from the sum projected images obtained from panel (H). Results are presented relative to control siRNA (Ctl) treated cells. N = 3 (n(Ctl si) = 169; n(RSRC2 si)) = 161). (J) Representative images of mitotic HCT116 cells stained for PCM (CDK5RAP2, magenta), centrioles (Cen2/3, orange), and DNA (Hoescht, cyan) after siRNA-mediated depletion of RSRC2. (K) Quantification of CDK5RAP2 signal intensity around the centrosome of mitotic HCT116 cells from the sum projected images obtained from panel (J). Results are presented relative to control siRNA (Ctl) treated cells. N = 3 (n(Ctl si) = 176; n(RSRC2 si)) = 180). (L) Representative images of mitotic HCT116 cells stained for centrosomes (γ-tubulin, orange), microtubules (β-tubulin, magenta), and DNA (Hoescht, cyan) after siRNA-mediated depletion of RSRC2. (M) Quantification of γ-tubulin signal intensity around the centrosome of mitotic HCT116 from sum projected images obtained from panel (L). Results are presented relative to control siRNA (Ctl) treated cells; N = 3 (n(Ctl si) = 143; n(RSRC2 si)) = 137). (N) Representative western blot of HCT116 cells probed with RSRC2, PCNT, CDK5RAP2, and GAPDH antibodies after the RSRC2 and PCNT knockdown. GAPDH, β-tubulin, and Ponceau staining were used as loading controls. (O) Densitometric analysis of RSRC2, PCNT, and CDK5RAP2 levels from panel (N) relative to control siRNA (Ctl); N = 5. Error bars in all panels are shown as mean ± S.E.M. Scale bar: 5 μm unless otherwise indicated. N = number of cells analysed. For panel (B), the unpaired t-test was used. For panel (D), two-way ANOVA was used. For panels (F), (G), (I), (K), and (M), a one-sample t‐test (comparing to a hypothetical mean of 1) was used. An unpaired t-test with Welch’s correction was used in panel (O). Insets are magnifications of centrosome areas (A, E, H, J, and L) with the scale bar, 0.5 µm. <0.05, **<0.01, ***<0.001, and ***<0.0001.
Because RSRC2 interacts with both splicing and centrosome proteins, we next investigated whether its interactome is enriched for centrosome-associated SFs. Comparing the RSRC2 interactome with the centrosome and cilia database (CCDB) [89, 90], identified 26 shared proteins, including RSRC2 itself and SNRNP200, a component of the U5 snRNP (Supplementary Fig. S10I). Both RSRC2 and SNRNP200 localized to centrosomes (Fig. 7 and Supplementary Fig. S10J), supporting the emerging concept that splicing regulators contribute to centrosome function [30, 31, 91]. Although RSRC2 is not listed in the spliceosome database [92], a recent study has shown its enrichment in snRNP complexes isolated from cells depleted of CD2BP2, a U5 snRNP-binding protein [93]. Collectively, along with our findings that RSRC2 interacts with PRPF8, these data suggest that RSRC2 may serve as an assembly factor in U5 snRNP biogenesis.
Finally, to evaluate the specificity of RSRC2 as a splicing regulator, we compared its splicing targets to databases of established SFs using the SplicingLore database [94]. SNRNP200, SON, and PRPF8, three RSRC2-interactors also reported to exhibit mitotic defects [22, 28], shared several mitotic targets with RSRC2, including PCNT (Supplementary Fig. 10K). CENPE was jointly regulated by SNRNP200, RSRC2, and PRPF8, whereas CDK5RAP2 was co-regulated by RSRC2 and PRPF8. In contrast, SFs not linked to centrosomes (e.g. hnRNPK, U2AF2, and SNRNP70) did not co-regulate these mitotic genes (Supplementary Fig. S10L).
Collectively, these findings establish RSRC2 as a specific splicing regulator whose loss disrupts the splicing of mitotic genes. ASO-mediated PCNT exon 19 skipping mimics the mitotic phenotype, providing functional evidence that RSRC2-mediated splicing of its targets directly contributes to chromosome congression defects and reveals a regulatory module comprising RSRC2 and centrosome-associated SFs.
RSRC2 is a bona fide PCM protein required for centrosome integrity and the localization of centrosomal proteins
Having established that RSRC2 controls the splicing of mitotic genes, we next investigated the downstream consequences for centrosome structure and function. First, we confirmed that RSRC2 localizes to the centrosome similarly to PCNT, a core PCM scaffold protein [95] (Fig. 7A). The centrosomal RSRC2 signal expanded from interphase to mitosis (Fig. 7B), consistent with PCM expansion during centrosome maturation [96] and establishing RSRC2 as a PCM component. To our knowledge, RSRC2 has not previously been annotated as a PCM protein, highlighting this localization as a novel feature of its function rather than a secondary effect of a splicing defect.
To gain further insights into the role of RSRC2 in centrosome organization, we employed ultrastructure expansion microscopy (U-ExM). Staining for acetylated tubulin revealed broken or absent centrioles in RSRC2-depleted HCT116 (Fig. 7C and D) and RPE1 cells (Supplementary Fig. S11A and B). This phenotype resembles that reported for SON loss [23], a SF we identified in the RSRC2 interactome, and indicates that RSRC2 is required for centriole structural integrity. Indeed, colocalization of RSRC2 and centrin (Cen 2/3) was reduced in 30% of RSRC2-depleted mitotic HCT116 (Fig. 7E–G) and RPE1 cells (Supplementary Fig. S11C–E). Similar defects were also present in interphase cells (Supplementary Fig. S11F–I), despite unchanged overall RSRC2 protein levels (Supplementary Fig. S11J and K). Together, these results identify RSRC2 as a critical regulator of centrosome structure.
We next tested whether mis-splicing of PCNT and CDK5RAP2 underlies these defects. RSRC2-depleted cells exhibited reduced PCNT (Fig. 7H and I) and CDK5RAP2 (Fig. 7J and K) localization at mitotic centrosomes, and reduced PCNT was also observed in interphase cells (Supplementary Fig. S11L and M). Because PCNT recruits the γ-tubulin ring complex (γ-TuRC) to nucleate microtubules [95, 97], we detected a decrease in γ-tubulin at centrosomes in RSRC2-depleted mitotic cells (Fig. 7L and M). Western blot analyses confirmed reduced PCNT and CDK5RAP2 protein levels following RSRC2 depletion (Fig. 7N and O), consistent with RSCR2-dependent splicing. CDK5RAP2 levels were also reduced in PCNT-depleted cells, in agreement with the established interdependence of these proteins and the requirement of PCNT for targeting CDK5RAP2 to centrosomes [96, 98]. We further assessed spindle architecture by measuring pole-to-pole distance and spindle angle. Both parameters were unchanged in RSRC2-depleted RPE1 and HCT116 cells (Supplementary Fig. S11N and O), indicating that RSRC2 primarily affects PCM composition and centriole integrity rather than spindle geometry. Further research is needed to determine whether RSRC2 affects microtubule dynamics and thereby alters spindle architecture.
Together, our data support a model in which RSRC2 as a bona fide PCM protein and maintains centrosome integrity by regulating the splicing of genes encoding centrosomal scaffold proteins. Loss of RSRC2 disrupts the localization of PCNT, CDK5RAP2, and γ-tubulin, leading to broken centrioles and impaired centrosome organization. Finally, the centrosomal defects observed in both normal and cancer cells emphasize that RSRC2 contributes to conserved, physiologically relevant pathways that control centrosome function.
The IDR and CC1 domains of RSRC2 are required for its role in splicing and mitotic fidelity
To determine whether mitotic defects in RSRC2-depleted cells arise from aberrant splicing of mitotic genes, we attempted a CRISPR knockout (KO) in HCT116 cells. Only RSRC2-edited clones with partial gene loss were recovered, consistent with RSRC2 being essential. This is supported by large-scale functional datasets and genome-wide CRISPR screening, which classify RSRC2 as an essential gene (Supplementary Fig. S12A) [99]. Two RSRC2-edited clones exhibited reduced RSRC2 protein levels and chromosome congression defects (Supplementary Fig. S12B and C). Sanger sequencing revealed deletions accompanied by a single-base insertion introducing a premature stop codon (Supplementary Fig. S12D and E), likely producing an unstable or non-functional RSRC2 protein.
To identify the domains required for splicing, we used the PCNT minigene assay in RSRC2-edited clones. Co-transfection of full-length RSRC2 or deletion mutants (Fig. 4) with the PCNT minigene revealed that deletion of either IDR or CC1 domains abolished rescue of the splicing defects, whereas full-length RSRC2 and mutants lacking SMAP or CC2 domains restored PCNT splicing (Supplementary Fig. S13A–C). Thus, the IDR and CC1 domains are necessary for RSRC2’s ability to regulate the splicing of mitotic genes. Consistent with these results, only full-length RSRC2 or RSRC2ΔSMAP rescued the mitotic defects in RSRC2-edited cells, whereas RSRC2ΔIDR and RSRC2ΔCC1 failed to do so (Supplementary Fig. S13D). Overexpression of tagged RSRC2 did not result in its localization to the centrosome under these conditions. These data support the model in which RSRC2’s splicing function is the primary determinant of mitotic fidelity.
C1QTNF1–AS1 regulates RSRC2 localization to centrosomes without altering its abundance or interactome
We examined whether C1QTNF1–AS1 affects RSRC2 expression or nuclear localization. C1QTNF1–AS1 depletion did not affect RSRC2 mRNA levels (Supplementary Fig. S14A), total protein amounts (Supplementary Fig. S14B and C), or its nuclear localization (Supplementary Fig. S14D and E). Conversely, C1QTNF1–AS1 expression did not change upon RSRC2 depletion in HCT116 and RPE1 cells (Supplementary Fig. S14F and G). Thus, the interaction does not involve reciprocal transcriptional regulation.
We next asked whether C1QTNF1–AS1 might regulate RSRC2 mitotic localization. IF analysis of RSRC2 and centrin in C1QTNF1–AS1-depleted cells revealed a reduction of RSRC2 localization at mitotic centrosomes (Fig. 8A and B). Centrin levels did not change upon C1QTNF1–AS1 knockdown (Fig. 8C), indicating that C1QTNF1–AS1 is not involved in centriole biogenesis. C1QTNF1–AS1 depletion also did not alter the protein levels of PCNT and CDK5RAP2 (Supplementary Fig. S15A and B), nor their centrosome localization on the mitotic spindle (Supplementary Fig. S15C–F), demonstrating that C1QTNF1–AS1 specifically affects RSRC2, not the distribution of PCM proteins.
RSRC2, whose localization is mediated by C1QTNF1–AS1, regulates PCNT mRNA localization at the centrosomes. (A) Representative images of mitotic HCT116 cells stained for RSRC2 (magenta), centrosomes (Cen2/3, orange), microtubules (β-tubulin, grey), and DNA (Hoescht, cyan) after LNA-mediated depletion of C1QTNF1–AS1. (B) Quantification of RSRC2 signal intensity around the centrosome of mitotic HCT116 cells from sum projected images obtained from panel (A). Results are presented relative to control LNA A (Ctl A) treated cells. N = 4 (n(Ctl A) = 204; n(C1QTNF1–AS1LNA1) = 245; n(C1QTNF1–AS1LNA5) = 221). (C) Quantification of Cen2/3 signal intensity around the centrosome of mitotic HCT116 cells from sum projected images obtained from panel (A). Results are presented relative to control LNA A (Ctl A) treated cells; N = 4 (n(Ctl A) = 217; n(C1QTNF1–AS1LNA1) = 269; n(C1QTNF1–AS1LNA5) = 252). (D) Representative image of C1QTNF1–AS1 ViewRNA FISH (orange, exonic probes), centrosome (γ-tubulin, magenta), and DNA (DAPI, cyan) after DMSO, Nocodazole, or Ciliobrevin D treatment in HCT116 cells. 3D mask shown as used for quantification in Imaris. (E) Quantification of C1QTNF1–AS1 distance from γ-tubulin in HCT116 cells from 3D images obtained from panel (D). Blue shading represents a 3 µm average distance seen in DMSO-treated cells; N = 3 (n(Cells+DMSO) = 238; n(Nocodazole) = 186; n(Ciliobrevin D) =200). (F) Proportion of C1QTNF1–AS1 foci as obtained in panel (D) found within a 3 µm distance from γ-tubulin in HCT116 cells; N = 3 (n(DMSO) = 238; n(Nocodazole) = 186; n(Ciliobrevin D) = 200). (G) Representative image of PCNT ViewRNA FISH (orange), centrosomes (γ-tubulin, cyan), and DNA (DAPI, magenta) following siRNA-mediated depletion of PCNT or RSRC2 in HCT116 cells. (H) Quantification of PCNT mRNA-positive centrosomes based on proximity to γ-tubulin signal in HCT116 cells from maximum intensity projections obtained from panel (G). N = 3 (n(Ctl si) = 448; n(PCNT si) = 435; n(RSRC2 si) = 491). Error bars in all panels are shown as mean ± S.E.M. Scale bar: 5 μm. N = number of cells analysed. The following statistics were applied: one sample t-test (using a hypothetical mean of 1) in panels (B) and (C); and a one-way ANOVA in panels (E), (F), and (H). Insets are magnifications of centrosome areas (A, D, G) with the scale bar, 0.5 µm. <0.05, **<0.01, ***<0.001, and ***<0.0001.
To determine whether C1QTNF1–AS1 modifies the RSRC2 interactome, we performed IP-MS for RSRC2 in control- and C1QTNF1–AS1-depleted cells. The RSRC2 interactome, including PCNT and CDK5RAP2 was unchanged (Supplementary Fig. S14H and Supplementary Table S11), and most interactions were RNA-independent (Fig 6A and Supplementary Fig. S9E). Since serine/arginine-rich proteins are highly phosphorylated on serine residues, we reasoned that C1QTNF1–AS1 could influence the phosphorylation of serine (Ser) residues in RSRC2, thereby impacting its localization and activity, as shown for other SFs [100, 101]. C1QTNF1–AS1 loss also did not affect the phosphorylation of RSRC2 (Supplementary Fig. S14I and Supplementary Table S11). Thus, C1QTNF1–AS1 does not modulate RSRC2 abundance, protein interactome, or post-translational modification but instead controls its centrosomal localization.
These results indicate that C1QTNF1–AS1 does not alter the RSRC2’s interactome or its phosphorylation status.
C1QTNF1–AS1 is positioned near centrosomes and RSRC2 is necessary for the localization of PCNT mRNA
Although C1QTNF1–AS1 does not localize within centrosomes (Supplementary Fig. S16A), a ViewRNA FISH experiment showed that it resides within a 3 μm radius of the centrosome in mitotic cells (Fig. 8D and E, and Supplementary Movie S11), consistent with known spatial ranges of centrosome-associated RNAs and RBPs [102–104]. This localization was microtubule-dependent as brief nocodazole treatment reduced C1QTNF1–AS1 signal within the 3-μm zone (Fig. 8F) and caused C1QTNF1–AS1 to shift away from the centrosome (Fig. 8E). Since motor proteins like dynein transport cargo along microtubules towards the centrosome, we treated cells with ciliobrevin D, an inhibitor of dynein [105]. This treatment marginally reduced C1QTNF1–AS1 localization, although not significantly (Fig. 8F). We concluded that C1QTNF1–AS1 is in proximity to the centrosome and its localization depends on intact microtubules.
Centrosomes are sites of mRNA localization and local translation, and PCNT mRNA is one of the best-characterized examples [102, 106, 107]. We therefore asked whether RSRC2 is required for local mRNA targeting. RSRC2 depletion reduced the centrosomal localization of PCNT mRNA (Fig. 8G and H), despite unchanged transcript abundance (Supplementary Fig. S16B). These data indicate that RSRC2 is required for localization of PCNT mRNA at centrosomes.
Since RSRC2 localization at the centrosome was mediated by C1QTNF1–AS1, we tested whether C1QTNF1–AS1 overexpression could rescue mitotic defects. Overexpression of the mature C1QTNF1–AS1 transcript, but not a scrambled sequence, rescued chromosome congression defects in RSRC2-depleted cells (Supplementary Fig. S16C and D). However, it did not restore RSRC2 localization at the centrosome (Supplementary Fig. S16E), suggesting that additional structural motifs or co-factors are required for RSRC2 centrosomal targeting. Collectively, these data support a model in which C1QTNF1–AS1 guides RSRC2 to the centrosome, where RSRC2 then promotes PCNT mRNA localization, which is known to rely on polysomes, microtubules, and dynein activity [102].
Discussion
We identify RSRC2 as a dual-functional regulator of cell division. Its primary role is to regulate the splicing of mitotic genes and interact with splicing- and centrosome-associated proteins. RSRC2 also localizes to centrosomes, and this requires C1QTNF1–AS1 lncRNA. In addition to its nuclear splicing role, our data support an independent centrosomal function for RSRC2 in RNA localization.
Our findings place RSRC2 as a novel splicing regulator of a specific set of mitotic genes, including PCNT, CDK5RAP2, and CENPE. Although splicing changes were modest, they still had a biological impact because they occurred in genes important for centrosome structure or kinetochore function. This is further supported by splice-switching oligonucleotides that induce mis-splicing of PCNT, resulting in mitotic abnormalities. The roles of RSRC2 in splicing and mitotic fidelity are primarily linked to its IDR and CC1 domains, consistent with previous research showing that these domains facilitate splicing regulation [72–75]. While coiled-coil motifs are common in centrosomal proteins [108, 109], they are also prevalent in RBPs [110]. The modular organization of RSRC2, with the CC1 domain directly following the IDR, suggests it may facilitate interactions to form ribonucleoprotein condensates or dock them to motor proteins for transport towards centrosomes. We propose that interactions between RSRC2 and other splicing regulators, such as SON, PRPF8, and SNRNP200, identified in our interactome analysis, enable cooperation in the splicing of specific mitosis-associated genes. Dysregulation of RSRC2 and its splicing interactors has been linked to mitotic defects [22, 28, 111] and to centrosome and cilia disorders [91], highlighting shared regulatory mechanisms that maintain mitotic integrity.
What are potential clinical implications of our findings regarding RSRC2-mediated splicing and mitotic defects? Mutations in centrosome and splicing genes are associated with similar genetic disorders, including microcephaly and ciliopathies [91]. Many neuronal centrosome interactors are enriched for splicing regulators [112]. Given that RSRC2 is a candidate gene for neurodevelopmental disorders [113], mutations in its IDR domain may contribute to pathogenic effects by disrupting the splicing of mitotic genes. Our rescue experiments in RSRC2-edited clones with PCNT minigene support this hypothesis.
The concept that splicing regulators exert functions at multiple cellular locations [114] aligns with our findings and supports RSRC2 as a “moonlighting” protein in PCNT mRNA localization at the centrosome. Centrosomes are hubs for mRNA transport and spatial regulation [115], and several mRNAs encoding centrosome proteins, including PCNT [116–118] localize there. RSRC2 localizes to centrosomes and interacts with PCM proteins, such as PCNT and CDK5RAP2, suggesting that these interactions help anchor the PCNT mRNA–RSRC2 complex at the centrosome. As the PCM scaffold permits RNA localization [119], we posit that RSRC2–protein interactions help drive mRNA localization to centrosomes. While we focused on PCNT mRNA, RSRC2 may regulate other centrosomal transcripts during early mitosis, given cell-cycle-dependent variation in RNA localization [102, 106]. Centrosomal mRNAs, including PCNT, rely on ribosome-mediated localization [102, 104, 106], and we propose that RSRC2 orchestrates both RNA localization and local translation. In support of this model, ribosomes are positioned near the centrosome [102, 120, 121] and RSRC2 interacts with ribosomal proteins (Fig. 6A and B). Further mapping of RSRC2’s RNA-binding landscape will enhance our understanding of its multifunctional roles, a feature increasingly recognized among RBPs [80].
Although we could not definitively distinguish whether the RSRC2-mediated mitotic defects arise from disrupted splicing or centrosome-associated functions, we favour the explanation that both functions contribute to chromosome congression defects following RSRC2 loss. Altered localization of centrosomal mRNA can lead to spindle defects and mitotic delays [6, 118], reinforcing our hypothesis that local dosage of centrosomal mRNA is critical for centrosome function and mitotic fidelity.
In recent years, the understanding of RNA–protein interactions has shifted from a protein-centric to an RNA-centric perspective, emphasizing that RNA can regulate protein function, including conformation, localization, interaction, and enzymatic activity [14, 122–126]. The RSRC2 function is RNA-mediated, as its localization to the centrosome during mitosis is facilitated by C1QTNF1–AS1. This lncRNA may recruit RSRC2 through interactions with centriolar satellites, which cluster around the centrosome and interact with centrosomal mRNAs, including PCNT, to promote their translation [127]. We propose that C1QTNF1–AS1 functions as a spatiotemporal chaperone, guiding RSRC2 to the centrosome, where RSRC2, in turn, can coordinate RNA localization and local translation. The small diameter of the centrosome suggests that any stoichiometric imbalances between lncRNA and RSRC2 can be overcome locally.
Our observations raise questions about the role of C1QTNF1–AS1 in cell division. Although its overexpression rescues chromosome congression defects in RSRC2-depleted cells, it does not restore RSRC2 localization at centrosomes. As RSRC2 levels at centrosomes decrease by 30% after its depletion, some RSRC2 protein likely remains, and increased C1QTNF1–AS1 might not be sufficient to fully rescue its function. An incomplete RSRC2 perturbation may allow the remaining RSRC2 to bind excess C1QTNF1–AS1, partially restoring mitotic function. Because RSRC2 is essential, we were unable to test this hypothesis further. Alternatively, C1QTNF1–AS1 may be a splicing target of RSRC2 or may have RSRC2-independent roles in cell division. Downregulation of DIAPH2, a downstream target of C1QTNF1–AS1, but not RSRC2, that causes similar mitotic defects [78] supports this hypothesis. These mechanisms may coexist.
In summary, we have identified RSRC2 as a previously unrecognized regulator of cell division with parallel roles in splicing and centrosome function. Our study, together with others, highlights the importance of lncRNA-protein interactions and provides a rationale for further studies into RNA-based mechanisms in cell division.
Supplementary Material
gkag229_Supplemental_Files
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Walczak CE, Cai S, Khodjakov A. Mechanisms of chromosome behaviour during mitosis. Nat Rev Mol Cell Biol. 2010;11:91–102. 10.1038/nrm 2832.20068571 PMC 2893392 · doi ↗ · pubmed ↗
- 2Maiato H, Gomes AM, Sousa F et al. Mechanisms of chromosome congression during mitosis. Biology (Basel). 2017;6:13.28218637 10.3390/biology 6010013 PMC 5372006 · doi ↗ · pubmed ↗
- 3Thompson SL, Bakhoum SF, Compton DA. Mechanisms of chromosomal instability. Curr Biol. 2010;20:R 285–295. 10.1016/j.cub.2010.01.034.20334839 PMC 3781365 · doi ↗ · pubmed ↗
- 4Ito KK, Watanabe K, Kitagawa D. The emerging role of nc RN As and RNA-binding proteins in mitotic apparatus formation. Noncoding RNA. 2020;6:13.32245090 10.3390/ncrna 6010013 PMC 7151635 · doi ↗ · pubmed ↗
- 5Blower MD, Nachury M, Heald R et al. A Rae 1-containing ribonucleoprotein complex is required for mitotic spindle assembly. Cell. 2005;121:223–34. 10.1016/j.cell.2005.02.016.15851029 · doi ↗ · pubmed ↗
- 6Groisman I, Huang YS, Mendez R et al. CPEB, maskin, and cyclin B 1 m RNA at the mitotic apparatus: implications for local translational control of cell division. Cell. 2000;103:435–47. 10.1016/S 0092-8674(00)00135-5.11081630 · doi ↗ · pubmed ↗
- 7Sharp JA, Plant JJ, Ohsumi TK et al. Functional analysis of the microtubule-interacting transcriptome. Mol Biol Cell. 2011;22:4312–23. 10.1091/mbc.e 11-07-0629.21937723 PMC 3216657 · doi ↗ · pubmed ↗
- 8Grenfell AW, Heald R, Strzelecka M. Mitotic noncoding RNA processing promotes kinetochore and spindle assembly in Xenopus. J Cell Biol. 2016;214:133–41. 10.1083/jcb.201604029.27402954 PMC 4949451 · doi ↗ · pubmed ↗