Cell cultureHeLa cells (631183, Takara) were cultured in Dulbecco’s modified Eagle’s medium (043-30085, Wako) supplemented with 10% fetal bovine serum at 37 °C in a humidified incubator with 5% CO2.Antibodies and reagentsAnti-MTR4 (A5576, Lot 00202580201, 1:1,000 dilution) antibody was obtained from Abclonal. Anti-hnRNPK (11426-1-AP, Lot 00022887, 1:1,000 dilution) and anti-PAPD5 (55197-1-AP, Lot 09000346, 1:1,000 dilution) antibodies were obtained from Proteintech. Anti-GAPDH (MAB374, clone 6C5, Lot 3725986, 1:1,000 dilution) antibody was obtained from Merck Millipore. Anti- ZFC3H1 (NB100-68267, Lot R33689, 1:1,000 dilution) antibody was obtained from Novus Biologicals. Anti-EXOSC5 (ab168804, Lot GR138334-1, 1:1,000 dilution) and anti-PABPN1 (ab75855, Lot GR32937-10, 1:1,000 dilution) antibodies were obtained from Abcam. Anti-DIS3 (HPA039281, Lot A95805, 1:1,000 dilution), anti-RBM7 (HPA013993, Lot F119176, 1:1,000 dilution), anti-ZCCHC7 (HPA021088, Lot A117087, 1:1,000 dilution), anti-ZCCHC8 (SAB1102086, Lot 080M2750, 1:1,000 dilution), and anti-FLAG (F3165, clone M2, Lot 058K6113, 1:1,000 dilution) antibodies were obtained from Sigma. Anti-UHRF1 (612264, clone 28, Lot 83172, 1:1,000 dilution) antibody was obtained from BD Biosciences. Anti-α-tubulin (CP-06, clone DM1A, Lot D00175772, 1:1,000 dilution) antibody was obtained from CALBIOCHEM. Secondary antibodies were purchased from GE Healthcare (NA934, Lot 9599347, 1:5,000 dilution; NA931, Lot 384924, 1:5,000 dilution) and Cell Signaling (7074S, Lot 29, 1:5,000 dilution). Normal rabbit IgG (PM035, Lot 007) was obtained from MBL. Antisera against human KCTD13 3XT protein were raised in rabbits by immunization with a chemically synthesized peptide of KCTD13 3XR (LGELPKQVLEQKKGDMGYG) by Eurofins Genomics (Hokudo, 1:500 dilution for IF). ECL-plus was purchased from GE Healthcare. VeriBlot for IP Detection Reagent (HRP) (ab131366, Abcam) was used to avoid the detection of heavy and light chains.qRT-PCR analysisTotal RNA was isolated using a NucleoSpin RNA mini kit (740955, MACHEREY-NAGEL) with the EconoSpin column for RNA (EP-21201, Epoch Life Science) and reverse transcribed using the PrimeScript RT Reagent Kit with gDNA eraser (RR047B, Takara Bio). qRT-PCR analysis of cDNA was performed on a Thermal Cycler Dice Real-Time System (Takara Bio) using TB Green Premix Ex Taq II (RR820L, Takara Bio). Prior to fold-change calculation, the values were normalized to the signal generated from GAPDH mRNA. Primer sequences are listed in Supplementary Data 9.RNA interferencesiRNA duplexes targeting MTR4 were purchased from Hokkaido System Science (HSS). Stealth siRNA duplexes targeting ZCCHC8 and RBM7 were purchased from Invitrogen. siRNA duplexes targeting hnRNPK, PCBP2, EXOSC5, ZFC3H1, PABPN1, ZCCHC7, and PAPD5 were purchased from Ambion. siRNA duplexes targeting DIS3 were purchased from Ambion and GeneDesign. Cells were transfected with RNA duplexes using Lipofectamine RNAiMAX (13778150, Invitrogen). For DRB chase experiments, a final concentration of 100 µM 5,6-dichloro-1-β-D-ribofuranosyl-1H-benzimidazole (DRB, 10010302, Cayman Chemical) in DMSO (047-29353, FUJIFILM) was added to cells 72 h after siRNA transfection. Sequences or IDs of siRNAs are shown in Supplementary Data 10.Constructs and transfectionFlag-tagged hnRNPK and its deletion mutants were amplified by PCR and cloned into pcDNA3.1(+). Flag-tagged MTR4 was amplified by PCR and cloned into pcDNA3.1(+). Lambda N-HA tagged hnRNPK was cloned into pCAGGS using the In-Fusion HD cloning kit (Takara). The 3XRs of HECTD2, SPRED2 and KCTD13 were amplified by PCR and cloned into pTet-BBB. The flag-tagged CDS region of KCTD13 3XT and its deletion mutants were synthesized and cloned into pcDNA3.1(+) using GeneArt (Thermo Fisher Scientific). Primer sequences are listed in Supplementary Data 11. Plasmids were transfected into cells using Lipofectamine 2000 (11668019, Invitrogen).Immunoblot analysisCells were lysed with Lysis buffer (50 mM HEPES pH 7.5, 150 mM KCl, 0.5% NP-40, 2 mM EDTA, 1 mM NaF) containing Protease Inhibitor Cocktail (P8340, Sigma) for 20 min on ice. Following centrifugation at 13,000 × g, an equal volume of SDS sample buffer (100 mM Tris-HCl pH 6.8, 20% glycerol, 4% SDS, 12% β-mercaptoethanol, 0.1% bromophenol blue) was added to the supernatant. After heating to 95 °C for 5 min, samples were resolved by SDS-PAGE, transferred to polyvinylidene difluoride (PVDF) membranes (Immobilon-P, IPVH00010, Merck Millipore), and analyzed by immunoblotting using HRP-conjugated secondary antibodies. Membranes were blocked with 5% bovine serum albumin (BSA, 011-27055, Wako) in TBST buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.1% Tween-20) at room temperature for one hour or with Blocking One (03953-95, Nacalai Tesque) for 5 min at room temperature before probing with antibodies. After each incubation, membranes were washed three times with TBST buffer. Visualization was performed using HRP Substrate (WBKLS0500, Merck Millipore) and the LAS-4000UVmini Luminescent Image Analyzer (FUJIFILM).ImmunoprecipitationHeLa cells (107 cells) were lysed with Lysis buffer (50 mM HEPES pH 7.5, 150 mM KCl, 0.5% NP-40, 2 mM EDTA, 1 mM NaF) containing Protease Inhibitor Cocktail (P8340, Sigma) for 20 min on ice. Following centrifugation at 13,000 × g for 20 min, the supernatants were treated with RNase A/T1 (1 µg/µL RNase A, 40 U/µl RNase T1, 50 mM Tris HCl, 50 mM NaCl, 50% glycerol) for 15 min on ice. RNase-treated lysates were incubated with anti-hnRNPK (1.375 µg) antibody, anti-MTR4 (2 µg) antibody, or normal rabbit IgG for 2 h at 4 °C with gentle rotation. Twenty microliters of Dynabeads Protein G (10009D, Invitrogen) were added and incubated for 1 h at 4 °C with gentle rotation. The beads were washed three times with Wash buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.05% NP-40) containing Protease Inhibitor Cocktail (P8340, Sigma) and then twice with TBST (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.1% Tween-20) containing Protease Inhibitor Cocktail (P8340, Sigma). After washing, immunocomplexes were eluted with 1× SDS sample buffer (50 mM Tris-HCl pH 6.8, 10% glycerol, 2% SDS, 6% β-mercaptoethanol, 0.05% bromophenol blue) containing Protease Inhibitor Cocktail (P8340, Sigma) and immunoblotted with antibodies against target proteins.RNA ImmunoprecipitationRIP assays was performed as described previously65,66,67 with minor modifications. HeLa cells (1 × 107 cells, for targeting endogenous proteins; 1 × 106 cells, for targeting overexpressed proteins) were lysed with Lysis buffer (50 mM HEPES pH 7.5, 150 mM KCl, 0.5% NP-40, 2 mM EDTA, 1 mM NaF) containing Protease Inhibitor Cocktail (P8340, Sigma) and Recombinant RNase Inhibitor (2313 A, Takara) for 20 min on ice. Following centrifugation at 13,000 × g, the supernatants were incubated with anti-hnRNPK (1.375 µg) or anti-MTR4 (2 µg) antibodies, or normal rabbit IgG for 2 h at 4 °C with gentle rotation. Thirty microliters of Dynabeads Protein G (10009D, Invitrogen) were added and incubated for one hour at 4 °C with gentle rotation. For Flag-IP analysis, the supernatants were incubated with 10 µL of Anti-FLAG M2 Magnetic Beads (M8823, Sigma, Lot 091M6291V) for one hour at 4 °C with gentle rotation. The beads were washed three times with Wash buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.05% NP-40) and then twice with TBST (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.1% Tween-20) containing Protease Inhibitor Cocktail (P8340, Sigma) and Recombinant RNase Inhibitor (2313 A, Takara). After washing, RNA was isolated using a NucleoSpin RNA mini kit (740955, MACHEREY-NAGEL), treated with Recombinant DNase I (2270 A, Takara) and Recombinant RNase Inhibitor (2313 A, Takara) for 20 min at 37 °C, and purified using a NucleoSpin RNA clean-up kit (740948, MACHEREY-NAGEL). qRT-PCR was performed as described above. Primer sequences for qRT-PCR are shown in Supplementary Data 9.ImmunofluorescenceHeLa Cells were seeded on coverslips in 12-well plates 24 h before transfection with the Flag-tagged hnRNPK, KCTD13 3XT coding sequence or their mutants. At 24 h after transfection, cells were washed twice with PBS plus 0.1% Tween20 and fixed with cold 100% methanol for 5 min at room temperature. The coverslips were then washed three times with PBS, permeabilized in 0.1% Triton-X100/PBS for 5 min at 4 °C, and washed three times with PBS. After blocking with Blocking One Histo (06349-64, Nacalai Tesque) for 5 min at room temperature, the cells were incubated in a 20-fold dilution of Blocking One Histo containing anti-Flag or anti-KCTD13 3XT protein antibody at 4 °C overnight. After washing three times with PBS plus 0.1% Tween20, a 20-fold dilution of Blocking One Histo with Alexa 488- (ThermoFisher Scientific, A11029 Lot 56649 A, 1: 1000 dilution) or 555-conjugated (ThermoFisher Scientific, A31570, Lot 2387458, 1: 1000 dilution) secondary antibody was added to the cells, and the samples were incubated for 1 h at room temperature. After washing three times with PBS plus 0.1% Tween20, the cells were incubated in DAPI (340-07971, Dojindo)/PBS or Hoechst33258 (H1343, TCI)/PBS for 10 min at room temperature. The coverslips were mounted with ProLong gold (P36934, Thermo Fisher Scientific), and images were acquired with a ZEISS LSM 980 with Airyscan 2. Airyscan processing and Z-axis projection were performed using Zeiss ZEN 3.4 (blue) (version 3.4.91.00000).For the evaluation of the LLPS properties, cells were treated with 1,6-hexanediol (087-00432, Wako) or 2,5-hexanediol (11904-10 G, Sigma) dissolved in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum at room temperature for 5 min. After these treatments, the cells were fixed with cold 100% methanol (137-01823, Wako) for 5 min at room temperature, followed by the immunofluorescence procedure described above.Tethering assayHeLa cells were transfected with the β-globin-5BoxB reporter or the β-globin control, a plasmid expressing EGFP, Flag-tagged MTR4, and plasmids expressing λN-HA-hnRNPK or λN-HA control. At 24 h after transfection, RNA was isolated from the cells using a NucleoSpin RNA mini kit (740955, MACHEREY-NAGEL) with the EconoSpin column for RNA (EP-21201, Epoch Life Science), treated with Recombinant DNase I (2270 A, Takara) and Recombinant RNase Inhibitor (2313 A, Takara) for 20 min at 37 °C, and purified using a NucleoSpin RNA Clean-up Kit (740948, MACHEREY-NAGEL). qRT-PCR was performed as described above. Prior to fold-change calculation, the values were normalized to the signal generated from EGFP mRNA. Primer sequences for qRT-PCR are shown in Supplementary Data 9.Subcellular fractionationCell pellets were resuspended in 1 packed cell volume of Hypotonic buffer (10 mM HEPES pH 7.5, 10 mM KCl, 1.5 mM MgCl2). After incubation on ice for 15 min, cells were disrupted by 10 passages through a 25-gauge needle. Cells were centrifuged for 10 min at 1000 g at 4 °C and the supernatant containing the cytoplasmic fraction was collected by further centrifugation at 15,000 × g for 20 min. The remaining pellets were washed thrice with Hypotonic buffer, resuspended in Hypotonic-NP-40 buffer (10 mM HEPES pH 7.5, 10 mM KCl, 1.5 mM MgCl2, 0.5% NP-40) and centrifuged for 10 min at 6000 × g at 4 °C. The remaining pellets were washed once with Hypotonic buffer, resuspended in Hypertonic buffer (20 mM HEPES pH 7.5, 250 mM KCl, 1.5 mM MgCl2, 0.5% NP-40) and incubated at 4 °C for 30 min with gentle rotation. Cells were centrifuged for 3 min at 6800 × g at 4 °C and the supernatant containing the nucleoplasm fraction was collected by centrifugation at 15,000 × g for 15 min.Sequence analysisRNA was isolated using a NucleoSpin RNA mini kit (740955, MACHEREY-NAGEL), treated with Recombinant DNase I (2270 A, Takara) and Recombinant RNase Inhibitor (2313 A, Takara) for 20 min at 37 °C, and purified using a Nucleospin RNA Clean-up Kit (740948, MACHEREY-NAGEL). For RNA sequencing analyzes of MTR4-depleted HeLa cells, libraries were prepared with polyA-selected RNA using a KAPA mRNA Capture Kit (KK8440, KAPA Biosystems) and an MGIEasy RNA Directional Library Prep Set (1000006385, MGI). Libraries were sequenced on the DNBSEQ-T7RS platform with 150-bp paired-end reads by GenomeLead Co., Ltd. Direct RNA sequencing was performed on the PromethION platform (Oxford Nanopore Technologies). 3′-Tag-sequencing (3′-Tag-seq) was performed on the NextSeq system by the DNA Technologies & Expression Analysis Core at the UC Davis Genome Center.The classification of aberrant RNAs with APA eventsAll last exons from alternative isoforms were extracted and compared with basic gene annotations (Fig. 1b). The last exons were classified into the following six categories in descending order of priority, according to their relative position to the annotated exons. (a) Intergenic: the last exon is included in the intergenic region. (b) Intron-exonized: the last exon is included in or identical to the intronic region. (c) Known exon: the last exon is identical to an annotated exon. (d) 5′-Extended: the 5′ end extends to an intronic/intergenic region, while the 3′ end is included in an annotated exon. (e) Both extended: the last exon overlaps with annotated exon(s), with its 5′/3′ ends extending to intronic/intergenic regions. (f) 3′-Extended: the opposite of 5′-extended. The above classification was conducted by custom scripts.Transcript and alternative isoform detectionRNA-seq data (DNBSEQ) were used to mark reliable splicing junctions. The human genome (hg38) and basic gene annotations (GENCODE v32) were downloaded from GENCODE68. The reads were mapped to the genome using STAR with the following parameters: –outFilterMultimapNmax 1 –outFilterMismatchNmax 10 –alignSJoverhangMin 20. Junctions and supporting reads were extracted from the mapped reads of all samples by junctionsfromsam.py in FLAIR (v1.4.0)69. Junctions with fewer than 50 supporting reads were removed.Long-read direct RNA sequencing data were used to identify alternative isoforms. First, the reads were mapped to the human genome using Minimap2 (v2.17-r954-dirty)70 with the following parameters: -ax splice -uf -k14. For subsequent processing, the mapped reads were converted to sorted, indexed BAM format files using Samtools (v1.11)71 and then further converted to BED format files by bam2Bed12.py in FLAIR. Second, the FLAIR correct subcommand corrected misaligned splicing junctions in these mapped reads using gene annotations and reliable splicing junctions. The corrected reads in all samples were merged, and the redundancies were eliminated. Finally, the FLAIR collapse subcommand built high-confidence alternative isoforms from the genome, gene annotations, the reads from PromethION, and corrected reads with the following parameter: -s 10. The FLAIR quantify subcommand calculated the expression of alternative isoforms in each sample based on the reads from PromethION. A custom script normalized the read counts to CPM (counters per million mapped reads) which were calculated by excluding reads that did not map to annotated transcripts from the total mapped reads in the denominator. Genes that were upregulated by ≥ 1.5-fold due to MTR4 knockdown, using two different siRNAs, were considered differentially expressed. RNA isoforms from long-read direct RNA sequencing data were visualized using NanoBlot72.RNA-seq data were retrieved from NCBI’s GEO or SRA databases (siPABPN1 and siRBM7, GSE8417221; si-hnRNPK, SRP11175647; siPCBP2, GSE13056473) using SRA Run Selector. After removing low-quality reads and adaptor sequences using fastp v0.23.474, the remaining reads were mapped to the genome using STAR v2.7.10b75. Expression count estimates and transcripts per million (TPM) were generated using RSEM v1.3.376 with the transcript reference generated from the above FLAIR analysis.3′-Tag-seq analysis3′-Tag-seq reads were trimmed using Cutadapt (v2.6)77 with the following parameters: -m 30 -a AGATCGGAAGAG. Trimmed reads were mapped to the human genome (hg38) using STAR, and uniquely mapped reads were kept. PCR duplicates were removed with Picard (v2.18.1, http://broadinstitute.github.io/picard/). Given that 3′-Tag-seq data are expected to have read coverage only at the 3′ ends of transcripts, the 100 bp-long 3′ ends from the above alternative isoforms were taken for transcript quantification. Read counting was performed using featureCounts (v2.0.1)78 with the following parameters: -p -M -O –fraction. Then, a custom script normalized the read counts to CPM (counts per million mapped reads) which were calculated by excluding reads that did not map to annotated transcripts from the total mapped reads in the denominator.Sequence motif analysis and quantitative motif comparisonMotif analysis for MTR4-target multi-exon 3XRs was carried out using MEME v5.4.179 with the following parameters: -mod anr -nmotifs 5 -minw 6 -maxw 7 -objfun classic -markov_order 0. To compare known motifs with the motif derived from MEME analysis, a pair-wise motif comparison was performed using Tomtom motif comparison search tool30 and the known motifs in the human known RBP motifs dataset from the CISBP-RNA database80.Protein annotationCoding sequences within the multi-exon 3XTs were predicted and extracted using the TransDecoder tool (version-5.5.0) (https://github.com/TransDecoder/TransDecoder) with the default parameters. BLASTp programs (version-2.12.0 + ) were used for homology-based similarity searches of the transcripts and to predict proteins against the latest human protein sequence dataset from the Ensemble database (https://asia.ensembl.org/info/data/ftp/index.html, Last modified, 2022-12-13) with a maximum e-value of 1e−10. The coding sequences and lengths of multi-exon 3XTs are listed in Supplementary Data 7.Prediction of phase separation regiondSCOPE81 was used to predict the potential phase separation regions of the predicted proteins from multi-exon 3XTs.AlphaFold2 structure predictionWe predicted structures for the KCTD13 3XT coding sequence using AlphaFold Colab v2.3.282. The wild-type hnRNPK structure was retrieved from the AlphaFold Protein Structure Database (AlphaFold DB)82,83.Statistical analysisStatistical analysis was performed using a paired two-tailed Student’s t-test (Fig. 1d; Fig. 2a–d; Fig. 3c, d; Fig. 4a, b, d; Supplementary Fig. 1k, l; Supplementary Fig. 2a–g; Supplementary Fig. 3b–e, g, h, j–n; Supplementary Fig. 4a, b, f–h; Supplementary Fig. 5b, c, e), an unpaired two-tailed Student’s t-test (Supplementary Fig. 3f), or the Brunner-Munzel test (Supplementary Fig. 2h). A p-value < 0.05 was considered to be statistically significant. All experiments were independently repeated three times, as stated in figure legends. Statistical analysis was performed with R 4.3.1 or Microsoft Excel.Reporting summaryFurther information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
