HeLa-FlpIn-Trex cells were a kind gift from Ivan Dikic (Goethe University, Frankfurt) and HEK293T cells were from ATCC (CRL-3216, Molsheim Cedex). HeLa and HEK cells were maintained in DMEM supplemented with 10% FBS and 5% L-glutamine.
hiPSCs were obtained from the Allen Cell Collection of the Coriell Institute for Medical Research (GM25256, AICS‑0094‑024). hiPSCs were maintained according to the standard operating procedures from the Allen Cell Institute (SOP: WTC culture v1.7) with minor adaptations. In short, hiPSCs were maintained in mTeSR Plus (STEMCELL Technologies, 100-0276). For passaging, hiPSCs were washed with DPBS and treated with Accutase (Thermo Fisher, A1110501) until they detached. Accutase treatment was quenched by dilution in DPBS. Cells were seeded on Geltrex (Thermo Fisher, A1413301) coated dishes and the growth media was supplemented with 10 µM Y-27632 (Hello Bio, HB2297) for 24 h after replating.
Cells were grown in a humidified incubator at 37 °C under 5% CO2 unless otherwise noted.
Immunofluorescence
Cells were fixed with 4% PFA for 15-30 min and permeabilized with 0.5% Triton-X for 30 min at room temperature. Cells were then blocked for 1 h at room temperature and stained with primary antibody for 1 h (HeLa cells) or 2 h (hiPSCs), washed in PBS, and stained with secondary antibody for 1 h (HeLa cells) or 2 h (hiPSCs).
Antibodies
1H4/pSR (Merck MABE50, clone 1H4, 1:100 or 1:150), B23/NPM1 (Sigma B0556, clone FC82291, 1:600), G3BP (abcam ab56574, clone 2F3, 1:400), HIF-1α (abcam ab179483, clone EPR16897, 1:500), PPP1CA (abcam ab137512, polyclonal, 1:250), SC35 (Sigma S4045, clone SC35, monoclonal, 1:1000), SON (abcam ab121759, polyclonal, 1:500), 16H3/SR (Merck MABE126, clone 16H3, 1:100).
Plasmids
GFP-PP1-NIPP1 and GFP-PP1m-NIPP1 plasmids were a kind gift from the Bollen Lab22. The cargo-binding domain of TNPO3 (CBDT) plasmid was a kind gift from the Shav-Tal Lab37. GFP-phosphatase plasmids were a kind gift from the Bodenmiller Lab33. pCMV-hyPBase was a kind gift from the Wellcome Trust Sanger Institute80. pECFP(C1)-NIPP1 (Addgene plasmid # 44226; http://n2t.net/addgene:44226; RRID:Addgene_44226), pEGFP(C1)-PP1alpha (Addgene plasmid # 44224; http://n2t.net/addgene:44224; RRID:Addgene_44224), pEGFP(N3)-PP1beta (Addgene plasmid # 44223; http://n2t.net/addgene:44223; RRID:Addgene_44223), and pEYFP(C1)-PP1gamma (Addgene plasmid # 44230; http://n2t.net/addgene:44230; RRID:Addgene_44230) were gifts from Angus Lamond & Laura Trinkle-Mulcahy. AICSDP-82:SON-mEGFP was a gift from Allen Institute for Cell Science (Addgene plasmid # 133964; http://n2t.net/addgene:133964; RRID:Addgene_133964). AICSDP-42:AAVS1-mTagRFPT-CAAX was a gift from Allen Institute for Cell Science (Addgene plasmid # 107580; http://n2t.net/addgene:107580; RRID:Addgene_107580). APEX2-NLS was a gift from Alice Ting (Addgene plasmid # 124617; http://n2t.net/addgene:124617; RRID:Addgene_124617)81. XLone-GFP was a gift from Xiaojun Lian (Addgene plasmid # 96930; http://n2t.net/addgene:96930; RRID:Addgene_96930)82.
Fluorescence recovery after photobleaching (FRAP) experiments
All FRAP experiments were performed on a Leica SP8 Falcon microscope using a 63 × 1.3 NA, glycerol, Plan-Apochromat objective. SRRM2-mCh was transiently overexpressed in HeLa TREx cells in combination with various GFP- or YFP-tagged plasmids. For FRAP during drug treatments and environmental perturbations, hiPSCs were endogenously tagged with 2xGFP at the SON locus as described below. Photobleaching was performed during 50–70 min of DRB treatment, 90–110 min of GSK626616 treatment, 70–85 min of Pladienolide-B treatment, or after 40–85 min of arsenite (500 µM) treatment or 50–105 min of heat shock (43 °C).
Proteomics and phosphoproteomics
Hek293T cells were seeded in 10-cm dishes to reach 70% confluency at the time of transfection. Triplicate technical replicates were gathered per condition. Cells were transfected with 5 µg DNA (EGFP-C1, EGFP-DYRK3, GFP-PP1-NIPP1, or GFP-PP1m-NIPP1) using GeneJuice. 24 h post-transfection, cells were washed twice in PBS and lysed in 450 µl lysis buffer (25 mM Tris HCl pH 7.4, 125 mM NaCl, 1 mM MgCl2, 1 mM EGTA, 5% glycerol, 1% Triton-X, 2x protease inhibitor cocktail, and 2x phosphatase inhibitor cocktail in milliQ-H2O) for 30 min on ice, harvested by scraping, then centrifuged at 17,000 g for 10 min at 4 °C. 25 µl anti-GFP magnetic agarose beads (Chromotek) per sample were equilibrated by washing 3x with 500 µl lysis buffer. Supernatants from the cellular lysates were then added to the beads and rotated at 4 °C for 1 h. Beads were washed 2x with 500 µl lysis buffer, and once with 125 µl wash buffer (1 mM Tris HCl, 125 mM NaCl, 1 mM MgCl2 in milliQ H2O).
For each sample, the anti-GFP beads with 100 µl of 10 mM Tris/2 mM CaCl2, pH 8.2 and re-suspended in 45 µl digestion buffer (triethylammonium bicarbonate (TEAB), pH 8.2), reduced with 5 mM TCEP (tris(2-carboxyethyl) phosphine) and alkylated with 15 mM chloroacetamide. Proteins were on-bead digested using 500 ng Sequencing Grade Trypsin (Promega). The digestion was carried out at 37 °C overnight. The supernatants were transferred to new tubes and the beads were washed with 150 µl trifluoroacetic acid (TFA) buffer (0.1% TFA, 50% acetonitrile) and combined with the first supernatant. For controls without phosphoenrichment, 10% of the samples were dried to completeness and re-solubilized in 20 µl of MS sample buffer (3% acetonitrile, 0.1% formic acid).
For the enrichment of phosphopeptides, the residual 90% of the samples were dried almost to completeness ( ~ 5 µl). The phosphopeptide enrichment was performed using a KingFisher Flex System (Thermo Fisher Scientific) and Ti-IMAC MagBeads (ReSyn Biosciences). Beads were conditioned following the manufacturer’s instructions, consisting of 3 washes with 200 µl of binding buffer (80% acetonitrile, 0.1 M glycolic acid, 5% TFA). Each sample was dissolved in 200 µl binding buffer. The beads, wash solutions and samples were loaded into 96 well deep well plates and transferred to the KingFisher. Phosphopeptide enrichment was carried out using the following steps: washing of the magnetic beads in binding buffer (5 min), binding of the phosphopeptides to the beads (30 min), washing the beads in wash 1-3 (binding buffer, wash buffer 1 and 2, 3 min each) and eluting peptides from the beads (50 µl 1% NH4OH, 10 min). The phosphopeptides were dried to the completeness and re-solubilized with 10 µl of 3% acetonitrile, 0.1% formic acid for MS analysis.
LC-MS/MS analysis was performed on a Q Exactive mass spectrometer (Thermo Scientific) equipped with a Digital PicoView source (New Objective) and coupled to a nanoAcquity UPLC (Waters Inc.). Solvent composition at the two channels was 0.1% formic acid for channel A and 0.1% formic acid, 99.9% acetonitrile for channel B. Column temperature was 50 °C. For each sample, 4 μl of peptides were loaded on a commercial Symmetry C18 trap column (5 µm, 180 µm x 20 mm, Waters Inc.) connected to a BEH300 C18 column (1.7 µm, 75 µm x 150 m, Waters Inc.). The peptides were eluted at a flow rate of 300 nl/min with a gradient from 5 to 35% B in 60 min, 35 to 60% B in 5 min and the column was washed at 80% B for 10 min before equilibrating back to 5% B.
The mass spectrometer was operated in data-dependent mode (DDA) using Xcalibur, with spray voltage set to 2.5 kV and heated capillary temperature at 275 °C. Full-scan MS spectra (350 − 1500 m/z) were acquired at a resolution of 70’000 at 200 m/z after accumulation to a target value of 3’000’000 and a maximum injection time of 100 ms, followed by HCD (higher-energy collision dissociation) fragmentation on the ten most intense signals per cycle. Ions were isolated with a 1.2 m/z isolation window and fragmented by higher-energy collisional dissociation (HCD) using a normalized collision energy of 25 %. HCD spectra were acquired at a resolution of 35’000 or 70’000 and a maximum injection time of 125 or 250 ms for phospho-enriched and non-enriched samples, respectively. The automatic gain control (AGC) was set to 3000 ions. Charge state screening was enabled and singly and unassigned charge states were rejected. Only precursors with intensity above 25’000 or 12’000 for phospho-enriched and non-enriched samples, respectively, were selected for MS/MS. Precursor masses previously selected for MS/MS measurement were excluded from further selection for 40 s, and the exclusion window tolerance was set at 10 ppm. The samples were acquired using internal lock mass calibration on m/z 371.1010 and 445.1200.
The mass spectrometry proteomics data were handled using the local laboratory information management system (LIMS)83.
Proteomics analysis
The acquired raw MS data were processed by MaxQuant (version 1.6.2.3), followed by protein identification using the integrated Andromeda search engine84. Spectra were searched against the Uniprot Homo sapiens reference proteome (taxonomy 9606, canonical version from 2019-07-09), concatenated to its reversed decoyed fasta database and common protein contaminants. Carbamidomethylation of cysteine was set as fixed modification, while methionine oxidation, phosphor (STY) and N-terminal protein acetylation were set as variable. Enzyme specificity was set to trypsin/P allowing a minimal peptide length of 7 amino acids and a maximum of two missed cleavages. MaxQuant Orbitrap default search settings were used. The maximum false discovery rate (FDR) was set to 0.01 for peptides and 0.05 for proteins. Label-free quantification was enabled and a 2-minute window for match between runs was applied. In the MaxQuant experimental design template, each file is kept separate in the experimental design to obtain individual quantitative values.
Data was then processed using R (v3.6.3). Results were first filtered to exclude reverse database hits, potential contaminants, and proteins identified only by site. Protein groups were then filtered for entries for ≥ 2 replicates in any condition under comparison. Missing LFQ intensities were imputed with random noise simulating the detection limit of the mass spectrometer (a log-normal distribution with 0.25x the standard deviation of the measured, logarithmized values, down-shifted by 1.8 standard deviations). Sample differences were then tested with the t.test function in R.
For analysis of differential phosphorylation from phosphoproteomics data, significance was calculated for single sites (phosphopeptides and matched unphosphorylated peptides from input samples) by comparing generalized linear models with and without an interaction term for phosphorylation status and condition using a likelihood ratio test.
5-EU and pulse-chase
Induction of R-MCD ( + 0.2)-GFP and inducible GFP hiPSCs was started 24 h prior to the start of the experiment by supplementing the growth medium with 2 µg/ml doxycycline. Then, the cells were pulsed with fresh growth medium containing 20 µM CX5461 (MCE, HY-13323) and 1 mM 5-EU for 30 min. After the pulse, the cells were washed twice with a growth medium and then kept in a growth medium containing 1 mM Uridine for up to four hours. A new batch of cells was pulsed every hour and all cells were fixed at the same timepoint.
For 5-EU measurements under environmental perturbations, cells were similarly treated with 20 µM CX5461 and 1 mM 5-EU for 30 min, then fixed.
CLICK reactions to detect 5-EU were performed after fixation and permeabilization by washing cells twice in TBS, then adding a solution of 2 mM CuSO4, 10 µM AlexaFluor 647 Azide Triethylammonium Salt (ThermoFisher A10277), and freshly added 100 mM sodium ascorbate in TBS, and incubating in the dark at room temperature for 30 min before washing in PBS and proceeding with additional stainings.
PolyA FISH
Cells were fixed and permeabilized as for immunofluorescence imaging. Cells were then washed twice with FISH wash buffer (10% formamide, 2x saline-sodium citrate (SSC)), then incubated 1 h at 37 °C in pre-hybridization buffer (100 mg/ml dextran sulfate, 7.5% formamide, 1.5x SSC). Half the volume was then aspirated and 2x pre-warmed hybridization buffer (100 mg/ml dextran sulfate, 10% formamide, 2x SSC, 800 nM dT-30-Atto488, dT-30-Cy3, or dT-30-Cy5 oligomer) added in equal volume for overnight incubation at 37 °C. The following day, cells were washed twice for 30 min at 37 °C with pre-warmed FISH wash buffer, then washed with 2x SSC and finally PBS before proceeding with immunofluorescence staining and imaging.
Design of smFISH probes
Target selection for genes under baseline condition (DMSO) was performed using raw DESeq2 results without input normaliztion. Probes for smFISH were designed as branched DNA probes for signal amplification. Twelve primary probes were designed per target gene using PaintSHOP85, using the hg38 newBalance (isoform flattened) probe set with default settings or OligoMiner86,87. When more than twelve probes were found for a target, the probes with the highest on_target value (PaintSHOP) or lowest melting temperatures (OligoMiner) were chosen. For three gene targets, fewer than 12 probes were available (EXOSC1: 8, H2BC11: 8, POU5F1: 6). For pooled probes, around 47 or 48 gene targets were selected per pool (enriched pool: 48, depleted pool: 47, neither pool: 48) based on relaxed thresholds of significance (enriched and depleted pools: |log2 fold change | > 0.3, padj 0.99). The primary probes were then extended by color-specific barcodes to which four secondary probes could bind. Primer binding sequences for probe purification were also included. The signal was amplified in this manner up to quaternary probes. Finally, fluorophore labelled probes (label probes) were bound to quaternary probes. Color-specific barcodes were derived from orthogonal 25mer barcode sequences designed previously88. The labelled probes were based on sequences used in the amplification method for smFISH signals described previously89. Raw probe sequences as generated by PaintSHOP and OligoMiner, as well as sequences extended with barcodes, are listed in Supplementary Data 5. Probe sequences for amplification probes (secondary, tertiary, quaternary and label probes) are listed in Supplementary Data 6. Primary probes were ordered from Twist Bioscience or Integrated DNA Technologies (IDT) as oligo pools. Amplification probes were ordered from Microsynth.
Purification of oligo pools
FISH probes were purified from oligo pools in three steps. First, oligo pools were amplified by PCR using primers to introduce T7 RNA polymerase promoter sequences and the amplicon was cleaned up using Zymo DNA Clean & Concentrator-25 (Zymo Research, D4033). The resulting amplicon was used as starting material for in vitro transcription using HiScribe T7 High Yield RNA Synthesis Kit (NEB, E2040S). Finally, the RNA product was used for reverse transcription using Maxima H Minus Reverse Transcriptase (Thermo Fisher, #EP0752) and RNA was removed by alkaline lysis.
PCR primers for oligo pool amplification:
Pair 1:
Forward: GTTGGTCGGCACTTGGGTGC
Reverse: CCACCGGATGAACCGGCTTT
Pair 2:
Forward: CGATGCGCCAATTCCGGTTC
Reverse: CAACCCGCGAGCGATGATCA
smFISH
hiPSCs were grown on 96-well plates (Greiner, 655090) and fixed with 4% PFA (EMS, 15710) in PBS for 15 min at room temperature. When pooled probe sets were used, to ensure even entry of probes, cells were dissociated and seeded as single cells before the experiment. The sample was then washed with PBS and permeabilized with 0.5% Triton X-100 (Sigma Aldrich, X100) in PBS for 30 min at room temperature. After another PBS wash, if pooled probe sets were used, the sample was treated with Protease QS (Thermo Fisher, QVP0011) diluted 1:2000 in PBS for 10 min at room temperature while shaking. Protease treatment was stopped by washing the sample with protease stop buffer (Thermo Fisher, QVP0011) twice and once with PBS. The sample was then incubated with 40% wash buffer (40% Formamide, 2x SSC, 0.001% tween20 in RNase-free H2O) for one hour at 65 °C. The sample was then incubated with primary probes diluted to a final concentration of 2 nM per probe in primary probe hybridization solution (10% Dextran, 40% Formamide, 2x SSC, 0.01% yeast tRNA, murine RNase inhibitor, 0.001% tween20 in RNase free H2O) for 16 h at 37 °C. The sample was washed three times with 40% wash buffer and once with 30% wash buffer (30% Formamide, 2x SSC, 0.001% tween20 in RNase-free H2O) for 6 min at 37 °C. The sample was then incubated with secondary probes diluted to 5 nM in probe solution (10% Dextran, 40% Formamide, 2x SSC, 0.001% tween20 in RNase-free H2O) for 30 min at 37 °C. After incubation with secondary probes, the sample was washed three times with 30% wash buffer for 6 min at 37 °C. Probe addition and washing were repeated in the same manner for tertiary and quaternary branching probes. The sample was then washed with PBS and incubated with label probes diluted to 0.5 µM in PBS for 1 h at 37 °C protected from light.
Nucleofection of hiPSCs
Y-27632 was added to hiPSC growth media 1–6 h prior to nucleofection at a final concentration of 10 µM. Cells were then passaged as previously described and 8 × 105 cells were resuspended in 100 µl of nucleofector solution (Lonza, VPH-5012). Depending on the experiment, plasmids, sgRNA and Cas9 were added to the resuspended cells. The cell suspension was then transferred to a nucleofection cuvette and nucleofection was performed using Lonza Nucleofector 2b using program A-023. After nucleofection, 500 µl of warm growth medium was added to the nucleofection cuvette and cells were transferred to a Geltrex-coated well of a 6-well plate containing pre-equilibrated (37 °C, 5% CO2) mTeSR plus supplemented with 10 µM Y-27632. Y-27632 concentration in the growth medium was halved to 5 µM after 24 h, and it was removed completely after 48 h of plating.
Design of sgRNA and Donor Plasmids for APEX2-SON and APEX2-NLS hiPSCs
We used crRNA sequences specified by the Allen Cell Institute to guide Cas9 to the SON locus and the AAVS1 locus. We ordered sgRNAs incorporating the crRNA and tracrRNA sequences from Sigma Aldrich. We used plasmid AICSDP-82:SON-mEGFP as template for the APEX2-SON donor plasmid and AICSDP-42:AAVS1-mTagRFPT-CAAX as a template for the APEX2-NLS donor plasmid. These plasmids already contained the homology arms needed for downstream CRISPR/Cas9 genome editing. AICSDP-42:AAVS1-mTagRFPT-CAAX additionally contained the CAG promoter to allow stable and consistent transgene expression in hiPSCs90,91. The APEX2 enzyme sequence and NLS motif were PCR amplified from plasmid APEX2-NLS. The APEX2 sequence was inserted in front of mEGFP in AICSDP-82:SON-mEGFP to create the APEX2-mEGFP-SON donor plasmid. The APEX2-NLS sequence was inserted along with the mEGFP sequence from AICSDP-82:SON-mEGFP into AICSDP-42:AAVS1-mTagRFPT-CAAX, replacing mTagRFPT-CAAX, to create the mEGFP-APEX2-NLS donor plasmid. All donor plasmids were assembled by Gibson assembly using NEB Gibson Assembly Master Mix (NEB, E2611L).
crRNA sequences
SON: CTGCTCGATGTTGGTCGCCA
AAVS1: GGGGCCACTAGGGACAGGAT
Cell line generation
The protocols for CRISPR/Cas9 genome engineering in hiPSCs were adapted from published protocols92,93. To generate the 2xGFP-SON, APEX2-NLS, and APEX-SON cell lines, hiPSCs were nucleofected as described above with 2 µg of donor plasmid and 1.5 µl each of 10 µM sgRNA and 10 µM Cas9 (Sigma Aldrich, CAS9PROT-50UG). Once cells reached confluency, they were passaged, resuspended in phenol red free mTeSR1 (STEMCELL, 05876) and sorted by FACS for GFP positive cells.
To generate doxycycline inducible R-MCD ( + 0.2)-GFP and GFP cell lines, hiPSCs were nucleofected as described above with 1 µg of XLone-R-MCD ( + 0.2)-GFP or XLone-GFP plasmid along with 1 µg of PiggyBack transposase plasmid (pCMV-hyPBase). XLone-R-MCD ( + 0.2)-GFP plasmid was generated by PCR amplification of R-MCD ( + 0.2) domain from pmEGFP-N1-R-MCD( + 0.2) plasmid (kind gift from Dr. Gregory Jedd) and cloned into XLone-GFP plasmid using Gibson assembly kit (NEB, E2611S). Two days after nucleofection, selection was started by adding blasticidin (Santa Cruz, sc-495389) at a concentration of 10 μg/ml. Blasticidin selection was carried out for multiple passages before cells were frozen in liquid nitrogen for long-term storage.
Drug treatment
Drug treatment with GSK626616 (5 µM), DRB (75 µM) and Pladienolide-B (2 µM) was carried out in DMEM/F-12 medium without serum for two hours at 37 °C and 5% CO2. For DYRK3 and CLK1 inhibition in hypoxia, GSK626616 (5 µM), TG003 (100 µM), or DMSO were likewise diluted in serum-free medium (pre-equilibrated overnight under hypoxic conditions) and added to cells for two hours after 22 h in hypoxia (0.2% O2, 5% CO2, 37 °C). Inhibition with CLK-IN-T3 (1 µM) was carried out in complete medium for 8 h (OPP experiment) or 12 (FISH experiments) hours.
OPP
Translation was quantified based on the incorporation of O-propargyl puromycin (OPP) in nascent peptides94. Cells were incubated in media containing 30 µM OPP (with drugs) for 45 min before fixation and detection of OPP through a CLICK reaction, as described above for 5-EU.
Stresses and environmental perturbations
Heat shock was performed at 43 °C for 1 h in a cell culture incubator maintained at 5% CO2. Oxidative stress was induced for 1 h using 500 µM sodium meta-arsenite dissolved in media. To induce hypoxia, cell media was exchanged for media pre-equilibrated under hypoxic conditions (0.2% O2), and cells were maintained at 0.2% O2/5% CO2 in a humidified atmosphere at 37 °C in a hypoxia workstation (Baker-Ruskinn).
APEX2 enzymatic reaction
The protocol for the APEX2 enzymatic reaction was adapted from Fazal et al. 42,43 and Padron et al. 43. APEX2-tagged hiPSCs were washed with DPBS (Thermo Fisher, 14190136) and then incubated in DPBS containing 0.5 mM biotin-phenol (Iris Biotech, LS-3500.1000) and 0.005% digitonin (Sigma Aldrich, D141) for 3 min at room temperature. To trigger the enzymatic reaction, hydrogen peroxide (Sigma Aldrich, 1.07209) was added to a final concentration of 0.5 mM and the dish was tilted for 1 min at room temperature. To stop the reaction, cells were washed once with a quenching solution (5 mM Trolox, 10 mM sodium ascorbate, 10 mM sodium azide in DPBS) and three times with wash solution (5 mM Trolox, 10 mM sodium ascorbate in DPBS).
RNA extraction and enrichment of biotinylated RNA
The protocol for extraction and enrichment of biotinylated RNA was adapted from Fazal et al. 42,43 and Padron et al. 43 hiPSCs were lysed by adding RNA lysis buffer (Zymo Research, R1060-1-50) directly to the cell culture dish. The cells were scraped into a solution and total RNA was purified using a Zymo Quick-RNA Miniprep kit (Zymo Research, R1054). Of the isolated total RNA, 5 µl was set aside as an input sample. To purify biotinylated RNA, we used Pierce Streptavidin Magnetic Beads (Thermo Fisher, 88816). First, 30 µl of beads per sample were resuspended in Binding and Washing (B&W) buffer (5 mM Tris-HCL (pH 7.5), 1 mM EDTA, 2 M NaCl) by vortexing, and then washed three times with B&W buffer. The beads were then resuspended in Solution A (0.1 M NaOH, 0.05 M NaCl) and incubated for 2 min. Then the beads were washed twice with Solution B (0.1 M NaCl) and resuspended in 100 µl Solution B. An equal volume of total RNA sample was added, and the samples were incubated for 2 h at 4 °C on a rotator to allow the biotinylated RNA to bind to the beads. The beads were then washed three times with B&W buffer and resuspended in 54 µl of RNase-free H2O. A 3X proteinase buffer was prepared (For 1 ml: 300 µl PBS, 300 µl 20% N-Lauryl sarcosine sodium solution (Sigma Aldrich, L7414), 60 µl 0.5 M EDTA, 15 µl 1 M DTT, 225 µl RNase free H2O), and 33 µl of this buffer was added to the beads together with 10 µl Recombinant Proteinase K Solution (20 mg/ml, Thermo Fisher, AM2548) and 3 µl RiboLock RNase inhibitor (Thermo Fisher, EO0381). The beads were then incubated for 1 h at 42 °C and then for 1 h at 55 °C on a shaker. Finally, biotinylated RNA was purified using RNA Clean & Concentrator-5 kit (Zymo Research, R1013).
For biotinylated RNA, five biological replicates were collected per cell line and drug condition. For input RNA, three biological replicates were collected per cell line and drug condition.
Subcellular fractionation of hiPSCs
The protocol used for subcellular fractionation of hiPSCs was adapted from Mayer & Churchman, 2017. Throughout the protocol, samples were kept at 4 °C and handled under RNase-free conditions. Briefly, hiPSCs were grown to confluency on 10 cm dishes and were first washed with and then scraped into DPBS. Cells were then pelleted by centrifugation for 3 min at 211 g and resuspended in lysis buffer (0.15% NP-40, 150 mM NaCl, 25 µM α-amanitin, 10 U SUPERase.IN (Thermo Fisher, AM2696), 1x cOmplete protease inhibitor mix, EDTA free (Sigma Aldrich, 11873580001)). The lysate was layered onto a sucrose buffer (10 mM Tris-HCl (pH 7.0), 5 M NaCl, 25% (w/v) sucrose, 25 µM α-amanitin, 10 U SUPERase.IN, 1x cOmplete protease inhibitor mix, EDTA free) and centrifugated for 10 min at 16,000 g. The supernatant representing the cytoplasmic fraction was collected, and the remaining pellet was washed twice with nuclei wash buffer (1 mM EDTA, 25 µM α-amanitin, 40 U SUPERase.IN, 1x cOmplete protease inhibitor mix, EDTA free prepared in PBS). The nuclear pellet was then resuspended in glycerol buffer (20 mM Tris-HCl (pH 8.0), 75 mM NaCl, 0.5 mM EDTA, 50% (v/v) glycerol, 0.85 mM DTT, 25 µM α-amanitin, 10 U SUPERase.IN, 1x cOmplete protease inhibitor mix, EDTA free) to which nuclei lysis buffer (1% NP-40, 20 mM HEPES (pH 7.5), 300 mM NaCl, 1 M urea, 0.2 mM EDTA, 1 mM DTT, 25 µM α-amanitin, 10 U SUPERase.IN, 1x cOmplete protease inhibitor mix, EDTA free) was added. After 5 min of incubation, the suspension was centrifugated at 18500 g for 2 min. The supernatant representing the nuclear fraction was collected and the remaining pellet was washed with PBS and centrifugated for 1 min at 1150 g. The supernatant was discarded, and the pellet representing the chromatin fraction was resuspended in 50 µl of PBS. The chromatin fraction was then incubated with TRIzol (Life Technologies, 15596) and chloroform for 5 min at room temperature. The sample was centrifuged, and the upper aqueous phase was collected.
To isolate RNA, 3.5 sample volumes of RLT buffer (Qiagen, 79216) were added to chromatin, nuclei and cytoplasm fractions. After mixing, 2.5 volumes of ice-cold 75% ethanol was added. RNA was then cleaned up using the RNeasy Mini kit (Qiagen, 74104). A total of 5 biological replicates were collected.
Library preparation and RNA sequencing
For APEX2-sequencing samples, libraries were prepared using SMARTer Stranded Total RNA-Seq Kit v2 – Pico Input Mammalian (Takara, 634411). Single-end sequencing was performed on the Illumina NovaSeq platform with a sequencing depth of 80 million reads and a read length of 100 bp.
For subcellular fraction sequencing samples, libraries were prepared using the Truseq Stranded mRNA kit (Illumina, 20020594). Single-end sequencing was performed on the Illumina NovaSeq platform with a sequencing depth of 200 million reads and a read length of 100 bp.
Library preparation and RNA sequencing was performed by the Functional Genomics Center Zurich (FGCZ).
Processing of RNAseq data
Trimmomatic v0.3995 was used to trim adapter sequences from raw reads with the following settings: ILLUMINACLIP:TruSeq3-SE:2:30:10 LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:36. Quality checks were carried out before and after trimming with FastQC v0.11.9. Trimmed reads were mapped to the human reference genome (hg38, GRCh38.p14 primary genome assembly) using GENCODE v40 gene annotations with STAR v2.7.3a96. Count tables were generated using the featureCounts97 function of the R package Rsubread v2.10.598.
Differential gene expression analysis
Genes with less than 10 counts in any of the samples were removed prior to the analysis. Differential Gene Expression (DGE) analysis was performed with DESeq2 v1.36.099 using the default two-sided Wald test and Benjamini-Hochberg correction for multiple testing. To quantify enrichment in nuclear speckles vs. nucleoplasm in DMSO treatment, we tested for the influence of the cell line factor. Thresholds of |log2FoldChange | > 0.5 and padj 0.5 and padj 7 and Supplementary Data 3.
Differential transcript expression analysis
Transcript quantification was performed using Salmon v0.12.0100 with the –numGibbsSamples option set to 30 to generate Gibbs samples. Differential transcript expression analysis was performed using Fishpond v2.2.0101.
Normalization with input samples
Unless otherwise indicated, DESeq2 results with input normalization were used for all analysis. Log2FoldChanges for input sample nuclear speckle enrichment was quantified with DESeq2 as described above. DGE results for nuclear speckle enrichment were normalized by fitting a linear regression model with log2FoldChanges of input samples as the independent variable and log2FoldChanges of biotinylated samples as the dependent variable. The residuals of the linear model were used as corrected log2FoldChange.
Gene set enrichment analysis
Gene Set Enrichment analysis was performed with the GSEA software v4.2.3102,103. GSEA was run in pre-ranked mode with default settings and using KEGG, Reactome and GO:CC, GO:BP and GO:MF gene sets. As ranking metric, we used input normalized log2FoldChange in case of nuclear speckle enrichment in DMSO condition and -log10(padj) * sign(log2FoldChange) in case of nuclear speckle enrichment in drug vs. DMSO conditions.
Splicing analysis
Differences in splicing between the transcripts enriched in nuclear speckles or the nucleoplasm were assessed in DMSO control samples using three tools: iREAD104, VAST v2.5.1105,106, and MAJIQ v 2.4.dev3 + g85d0781.d20220721107.
For VAST, analysis was run starting with untrimmed reads, as recommended, with thresholds set for detection in ≥1 samples with ≥ 10 reads, with a minimum probability ≥ 0.95. Residual differences in retention were calculated using input samples from each cell line and a residual difference of >0.15 used for plotting. Control introns (n = 10,000) were selected at random from all detected introns for feature comparisons.
For MAJIQ, detection thresholds were set for observation of alternative splicing in ≥ 1 sequencing replicate with ≥ 5 reads per junction (prior-minreads) and ≥ 2 reads per experiment (minreads), along with the default probability threshold for local splice variation of 0.2. Control introns (n = 3633) were introns with absolute differential percent spliced in index 5G), splicing differences were calculated using MAJIQ for SON-enriched samples under drug treatments compared to DMSO.
Network visualization of GSEA results
Network visualizations of GSEA results were made using Cytoscape108 v3.9.1 and EnrichmentMap109. Only Nodes with gene set sizes between 29 and 496 and NES smaller than -2 or greater than 1.8 are displayed. Annotations were generated using the AutoAnnotate plugin110 and manually adapted. Annotated groups were positioned manually.
Prediction of transcript-level nuclear speckle enrichment
Feature categories used for the linear model were collected from the following sources. General sequence features: custom Python script using GENCODE v43 primary assembly gff3 annotations. As part of the general sequence features, we also included the GGACU m6A motif density and the AGCCC nuclear localization motif111 density. RNA binding protein features: oRNAment database112. Kinetic rates:113. Promoter motifs: The Eukaryotic Promoter Database EPD114. TPM: Salmon quantification (Supplementary Data 4).
Features for training were filtered to only include protein-coding transcripts. Features representing motif counts were transformed into density. Features that contained 0 values for more than 75% of the data were removed. All features were log transformed, missing values were imputed by the mean of the feature and features were z-scored. Principal Component Analysis (PCA)115 was performed on the resulting feature set and the number of principal components explaining 99% of the variance were retained (n = 414). All transcripts annotated to the same gene were only allowed to be in either the test or the train set. The reported coefficients of determination were calculated by averaging results of 10-fold cross-validation. Preprocessing and model training were done in Python (v3.9.7) using scikit-learn (v1.3.0).
Feature importance scoring
The linear regression model was trained for one hundred iterations with different test (15% of the data) and train (85% of the data) subsets. The top ten loadings of the ten principal components with the highest absolute coefficients were extracted for each iteration. The frequency at which a feature occurs in this subset of features was taken as a measure of feature importance. For example, a feature occurring in the top 10 loadings of every single top 10 PC will have a frequency of occurrence of 1.
UMAP representation
Transcript features were preprocessed as described above and were used as the input for the UMAP algorithm as implemented in Python58,59.
Immunofluorescence and FISH microscopy
Microscopy images were acquired on a CellVoyager 7000 microscope (Yokogawa) equipped with an enhanced CSU-W1 spinning disk (Microlens-enhanced dual Nipkow disk confocal scanner, wide view type) and Andor cSMOS cameras or on a CellVoyager 8000 microscope (Yokogawa) equipped with CSU-W1 spinning disk and ORCA-Flash4.0 V3 cameras. Images were acquired using a 60x Nikon water immersion objective (NA = 1.2) or 40x Nikon air objective (NA = 0.95) on the CellVoyager 7000, and using a UPLSAP60xW customized Yokogawa objective (NA = 1.2) on the CellVoyager 8000. Z-stacks with a 1 μm spacing were acquired per site, spanning the whole height of the sample (12–20 μm). For subsequent processing, maximum intensity projections (MIPs) were performed for each site, except where noted.
Image processing
Nuclear speckles were segmented based on SC35 or SON intensity images using the pixel classifier functionality of Ilastik v1.4.0116,117. Nuclei were segmented based on DAPI intensity images using Cellpose118. Image processing was performed on a computing cluster (ScienceCloud) (https://www.zi.uzh.ch/en/teaching-and-research/science-it/computing/sciencecloud.html) provided by the Service and Support for Science IT (S3IT) facility of the University of Zurich (UZH) using TissueMAPS (https://github.com/pelkmanslab/TissueMAPS), Fractal (https://fractal-analytics-platform.github.io/) and custom Python scripts. Pearson correlation coefficients were calculated in python using scipy.stats.
Pseudocoloring of smFISH images
Intensity values of smFISH images were mapped to different color gradients depending on the subcellular localization of the signal. Signal overlapping with nuclear speckle segmentation (FISH ∩ nuclear speckles, based on SON-mEGFP or SC35 antibody staining) was mapped to an orange gradient, while signal not overlapping with speckles (FISH \ nuclear speckles) was mapped to a blue gradient. Intensities for the same smFISH target were always scaled the same way across both gradients. Outlines of nuclei (segmented based on DAPI) and nuclear speckles (segmented based on SC35) are indicated by white overlays. Color gradients were applied in Python using the microfilm package v0.2.1.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.