Ethical statements

All the mouse procedures were carried out in compliance with the guidelines of the Animal Care and Use Committee of the Institute of Zoology, Chinese Academy of Sciences, and the Animal Care and Use Committee of the Northeast Agricultural University.

Antibodies

The following antibodies were used for immunoblotting, immunostaining and/or immunoprecipitation: mouse monoclonal (G-2) anti-CARM1 (Santa Cruz, sc-393381), chicken polyclonal anti-GFP (Abcam, ab13970), rat monoclonal (YL1/2) anti-Tubulin (Abcam, ab6160), goat polyclonal anti-PCBP1 (Abcam, ab109577), rat monoclonal (EPR14859(2)) anti-PCBP2 (Abcam, ab200835), rat monoclonal (EPR8239) anti-SRSF1 (Abcam, ab129108), rabbit polyclonal anti-HA (Abcam, ab9110), rabbit polyclonal anti-NONO (Abcam, ab70335), rabbit polyclonal anti-PSPC1 (Abcam, ab104238), rabbit polyclonal anti-HNRNPU (Abcam, ab20666), rabbit polyclonal anti-U2AF2 (Abcam, ab37530), rabbit polyclonal anti-H3R26me2 (Abcam, ab127095), FITC-conjugated donkey anti-Chicken IgY polyclonal secondary antibody (Invitrogen, SA1-72000), Alexa Fluor Plus 555-conjugated goat anti-Mouse IgG polyclonal secondary antibody (Invitrogen, A32727), Alexa Fluor Plus 555-conjugated donkey anti-Goat IgG polyclonal secondary antibody (Invitrogen, A32816), Alexa Fluor Plus 555-conjugated donkey anti-Rabbit IgG polyclonal secondary antibody (Invitrogen, A32794), Alexa Fluor Plus 555-conjugated donkey anti-Rat IgG polyclonal secondary antibody (Invitrogen, A48270), Alexa Fluor 488-conjugated rabbit anti-Rat IgG polyclonal secondary antibody (Invitrogen, A-21210), HRP-conjugated goat anti-Rabbit IgG polyclonal secondary antibody (Invitrogen, A27036), HRP-conjugated mouse anti-Goat IgG polyclonal secondary antibody (Invitrogen, 31400) and HRP-conjugated goat anti-Rat IgG polyclonal secondary antibody (Invitrogen, 31470).

Mouse embryo collection

The CD1 (ICR) mice were purchased from Vital River company. All mice used for experiments were 7–8 weeks old. All mice were housed in the animal care facilities under specific pathogen-free conditions, with a 12 h dark–light cycle, ambient temperature ranging from 21 °C to 26 °C and a humidity level of 50% to 60%. To obtain pre-implantation embryos, female mice were injected with 10 U of pregnant mare serum gonadotropin (PMSG, Prospec, HOR-272) and 10 U of human chorionic gonadotropin (hCG, Prospec, HOR-250) at 46–48 h intervals and then crossed with 7–8-week-old CD1 (ICR) male mice. Embryos were collected at the following times post hCG injection: early one-cell stage (phCG 19 h), late one-cell stage (phCG 30 h), early two-cell stage (phCG 39 h), late two-cell stage (phCG 48 h), early four-cell stage (phCG 54 h), late four-cell stage (phCG 62 h), early eight-cell stage (phCG 68 h), late eight-cell stage (phCG 74 h), 16-cell stage (phCG 80 h), 32-cell stage (phCG 90 h), early blastocyst stage (phCG 98 h) and late blastocyst stage (phCG 114 h).

Culture cells

mEpiSCs were established in Q. Zhou’s lab in the State Key Laboratory of Stem Cell and Reproductive Biology, Institute of Zoology, Chinese Academy of Sciences, and cultured in a fibronectin-coated dish in N2B27 medium plus 12 ng ml−1 bFGF (R&D, 233-FB-001MG/CF) and 10 ng ml−1 activin A (R&D, 338-AC-01M) at 37 °C in a 5% CO2 incubator. The culture medium was changed every day, and the mEpiSCs were passaged every 2–3 days and then digested to single cells by 0.05% trypsin/EDTA (GIBCO, 25300062). The MCF-7 cells were purchased from ATCC (ATCC, HTB-22). MCF-7 cells were cultured in DMEM/F-12 (1:1) medium supplemented with 10% FBS at 37 °C in a 5% CO2 incubator. For cell transfection, the cells were passaged and seeded at a density of 1–1.5 × 104 cells per cm2. After 2 days (60–70% confluence), plasmid DNA (MS2-LincGET and MS2P-HA for mEpiSCs; PCBP-shRNA for MCF-7 cells) was transferred into cells using Lipofectamine 3000 transfection reagent (GIBCO, L3000015) according to the manufacturer’s instructions. mEpiSCs and MCF-7 cells were collected for further analysis 36 and 72 h after transfection, respectively.

RNA extraction, reverse transcription and qPCR

RNA was extracted using the RNeasy Mini Kit (QIAGEN, 74104), and the RNase-Free DNase Set (QIAGEN, 79254) was used to prevent DNA contamination. Reverse transcription was performed using a High-Capacity cDNA Reverse Transcription Kit (ABI, 4368814). SYBR-qPCR was performed using a Power SYBR Green PCR Master Mix (ABI, 4367659). TM-qPCR was performed using TaqMan Universal Master Mix II (Life, 4440048). All kits were used following the manufacturers’ guidelines.

Construction of plasmid vectors

To create Alexa Fluor 488-labeled RNA probes for LincGET RNA-FISH, the specific LincGET region (2,574–2,763) was amplified using the 2× Vazyme Lamp Master Mix (Dye Plus) (Vazyme, P312). Primers are shown in Supplementary Table 3. These sequences were sub-cloned into the plasmid pEASY-T3 cloning vector (TransGen, CT301-02), which contains the T7 promoter. For co-IP experiments, the MS2 coat protein (MS2P), MS2-labeled LincGET and HA-labeled MS2P were cloned into the PB533A vector (SBI, PB533A-2) digested with EcoRI or SalI, respectively.

For in vitro transcription, the GFP-KASH sequence with T7 promoter was synthesized by BGI company and sub-cloned into the pUC57 vector; LincGET mutants were generated by PCR, 5′-phosphorylation and ligation; plasmid containing NONO-BFP or NONO-mCherry with T7 promoter was purchased from YouBio company (YouBio, L3774 and L2772).

For coding potential analysis for exon-skipping variants of Carm1, Carm1 (NM_021531) cDNA for non-skipping splicing (noS) and Carm1 (NM_153141) cDNA for the E15S variant were purchased from YouBio company (YouBio, G156971 and G156972). The EF1α promoter from Addgene plasmid no. 61422, Carm1 cDNA and bGH poly(A) signal element from Addgene plasmid no. 61422 were sub-cloned into plasmid pEGFP-C1, purchased from YouBio company (YouBio, VT1118), using NEBuilder HiFi DNA Assembly Cloning Kit (NEB, E5520S). Carm1 mutants of E3S, E5S, E6S, E56S, E3456S and E11S were generated by PCR, 5′-phosphorylation and ligation.

For short hairpin RNA (shRNA) expression, primers containing sense-loop–antisense structure (Supplementary Table 3) were synthesized and annealed to generate double-strand oligonucleotides with a re-annealing process: 95 °C for 10 min; 95–85 °C ramping at −2 °C s−1; 85–25 °C at −0.25 °C s−1; and 25 °C hold for 1 min. Then the double-strand oligonucleotides were sub-cloned into the pLL3.7 vector (Addgene, 11795).

For protein expression and purification, the NONO, PCBP1 or PCBP2 sequence followed by a 14 amino acid linker sequence ‘GAPGSAGSAAGGSG’ and mCherry or GFP was cloned into a modified version of a T7 pET expression vector (YouBio) containing a 5′ MBP tag and a 3′ 6×His tag.

For minigene analysis of ESS of Carm1 pre-mRNAs, wild-type Carm1 exon 3 flanked by a 539 bp 5′ splicing acceptor sequence and 258 bp 3′ splicing donor sequence, wild-type Carm1 exon 5 flanked by a 531 bp 5′ splicing acceptor sequence and 506 bp 3′ splicing donor sequence, and wild-type Carm1 exon 6 flanked by a 550 bp 5′ splicing acceptor sequence and 418 bp 3′ splicing donor sequence were amplified by PCR with the primers listed in Supplementary Table 3 using Q5U Hot Start High-Fidelity DNA Polymerase (NEB, M0515) and then cloned into the KpnI-digested pSpliceExpress (Addgene, 32485) splicing minigene plasmid45 using the In-Fusion HD Cloning Kit (Clontech, 639636) following the manufacturers’ guidelines. Plasmids for E3-C26 and E3-Mut were generated by PCR with primers containing mutations using Q5U Hot Start High-Fidelity DNA Polymerase (NEB, M0515), phosphorylation using T4 polynucleotide kinase (NEB, M0201L) and ligation using T4 DNA ligase (NEB, M0202L).

Micro-injection

Mouse embryos at the one-cell stage were collected and microinjected with about 1–2 pL RNA at 150 ng μl−1 concentration or plasmid at 10 ng μl−1 concentration into the pronucleus at phCG 25 h, using an Eppendorf micromanipulator on a Nikon inverted microscope. The LincGET and mRNAs for NONO and PCBP1 were in vitro transcribed with HiScribe T7 ARCA pre-mRNAs Kit (with tailing) (NEB, E2060), while the single-guide RNAs (sgRNAs) were in vitro transcribed with HiScribe T7 Quick High Yield RNA Synthesis Kit (NEB, E2050) following the manufacturers’ guidelines. DNA templates with T7 promoter were amplified using the 2× Vazyme Lamp Master Mix (Dye Plus) (Vazyme, P312). Primers are shown in Supplementary Table 3. The fluorescently labeled LincGET or sgRNAs were generated by Poly(U) Polymerase (NEB, M0337S) with modified nucleotide 5-(3-aminoallyl)-UTP (Ambion, AM8437) and then labeled with Alexa Fluor 488 NHS Ester (Succinimidyl Ester) (Invitrogen, A20000) or Alexa Fluor 546 NHS Ester (Succinimidyl Ester) (Invitrogen, A20002) following the manufacturers’ guidelines. Locked nucleic acid (LNA) or short interfering RNA (siRNA) was used in co-injection, in the following concentrations: 10 μM LNA for control; 1 μM LNA for LincGET; and 10 μM siRNA for Nono, Pspc1, Hnrnpu or control.

Western blot

For each lane, 200 single blastomeres from embryos were lysed with RIPA lysis buffer (10 μl per lane; Beyotime, P00138) containing 1 mM phenylmethyl sulfonyl fluoride (PMSF; Beyotime, ST506), mixed with 30 μl of sample buffer (Beyotime, P0283) and incubated for 5 min in boiling water bath. The samples were then separated by SDS–PAGE with a 5% stacking gel (10 ml; 5.7 ml ddH2O, 1.7 ml 30% acrylamide (29:1), 100 μl of 10% SDS, 2.5 ml 1.5 M pH 6.8 Tris-HCl, 50 μl of 10% ammonium persulfate and 10 μl of TEMED) and a 10% separating gel (10 ml; 4.1 ml ddH2O, 2.5 ml 1.5 M pH 8.8 Tris-HCl, 3.3 ml 30% acrylamide (29:1), 100 μl of 10% SDS, 50 μl of 10% ammonium persulfate and 5 μl of TEMED) at 100 V for 1 h. Separated proteins were then electrophoretically transferred onto a nitrocellulose membrane at 250 mA for 1 h at 4 °C. Membranes were then blocked in TBST buffer (150 mM NaCl, 10 mM Tris, 0.1% Tween-20, pH 7.4) containing 3% BSA (Sigma, B2064), for 1 h at room temperature and then incubated with primary antibody, diluted (mass:volume ratio, 1:200) in TBST containing 1% BSA, overnight at 4 °C. After three washes for 10 min each in TBST, membranes were incubated for 1 h at room temperature with the fluorescence-labeled secondary antibody diluted (mass:volume ratio, 1:5,000) in TBST containing 1% BSA. After three washes for 10 min each, the membrane was exposed to a Bio-Rad GelDoc XR+ Gel imaging system for the acquisition of signals.

Immunoprecipitation

Magnetic Beads Protein G were coated with 5 μg of primary antibody in RIP wash buffer (150 mM sodium chloride, 50 mM Tris-HCl, pH 7.4, 1 mM MgCl2 and 0.05% NP-40) containing 5% BSA overnight with rotation at 4 °C. Then, we collected approximately 1 × 106 mEpiSCs expressing HA-MS2P, with or without MS2-LincGET, added to 100 μl of RIP lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA pH 8.0 (Invitrogen, AM9260G), 1% NP-40, 0.1% SDS, 0.5% sodium deoxycholate, 0.5 mM DTT, 1 mM PMSF/cocktail) and 10 μl of RNase inhibitor (Ambion, AM2694) and incubated on ice for 10 min. For embryos injected with MS2-LincGET or MS2-anti-LincGET, pre-mRNAs for HA-MS2P and pre-mRNAs for NONO, about 400 embryos were used for each group. Next, we centrifuged the RIP lysate at 10,000×g for 10 min at 4 °C, removed 100 μl of the supernatant and added this to 900 μl of the beads–antibody complex (mass:volume ratio, 1:200) in RIP Immunoprecipitation Buffer (860 μl RIP wash buffer, 35 μl 0.5 M EDTA pH 8.0 and 5 μl RNase inhibitor) and incubated this with rotation overnight at 4 °C. The residual 10 μl of the supernatant of RIP lysate was treated as input. After washing, the immunoprecipitate was divided into two parts. One part was mixed with 15 μl of western blot sample buffer and incubated for 5 min in boiling water; this blot was used to detect HA-MS2P and NONO/PSPC1/HNRNPU (according to the antibody). The other part was treated with proteinase K at 55 °C for 30 min with shaking to digest the protein, followed by RNA extraction from the supernatant; reverse transcription polymerase chain reaction (RT–PCR) was then performed to detect LincGET.

Immunofluorescence staining

Mouse embryos were permeabilized in 1% Triton X-100 (Sigma, T9284) in 1× PBS (Invitrogen, AM9625) and fixed in ice-cold 100% methanol for 30 min at −20 °C, followed by permeabilization again in 70% ethanol for 30 min at 4 °C. Then, embryos were blocked in blocking solution (1% BSA in 1× PBS) for 1 h at room temperature after three washes for 5 min each in washing solution (0.1% Tween-20, 0.01% Triton X-100 in 1× PBS), followed by incubation with primary antibody diluted (mass:volume ratio, 1:50) with blocking solution overnight at 4 °C. After three washes, embryos were incubated with Alexa series fluorescent tag-conjugated secondary antibody diluted (mass:volume ratio, 1:1,000) with washing solution for 1 h at room temperature. After three washes, embryos were mounted with DAPI-Vectashield solution (Vector laboratories, H1200). Notably, when the embryos were injected with Alexa Fluor 488-labeled LincGET and pre-mRNAs for NONO-mCherry, the embryos were directly mounted with DAPI-Vectashield solution (Vector laboratories, H1200) after permeabilization. When the embryos were injected with Alexa Fluor 488-labeled LincGET, pre-mRNAs for NONO-BFP and Alexa Fluor 546-labeled sgRNAs, the embryos were directly stained with SYTOX Deep Red Nucleic Acid Stain (Invitrogen, S11380) and then mounted after permeabilization. Imaging of embryos was then performed using a laser-scanning confocal microscope (Leica, TCS SP8). IMARIS software (Bitplane) was then used to calculate the number of speckles in each picture.

Immunofluorescence combined with RNA-FISH

Probes targeting LincGET used in RNA-FISH were labeled by in vitro transcription using the MEGAshortscript Kit (Ambion, AM1354) with ATP, CTP, GTP, UTP and ChromaTide Alexa Fluor 488-5-UTP (Invitrogen, C11403) solution (4:4:4:1:4) in which 80% of uracil was labeled by Alexa Fluor 488. The zona pellucida of the mouse embryos was removed with incubation in acidic Tyrode’s Solution (Sigma, T1788). Then the embryos were incubated in PBSA (1× PBS containing 6 mg ml−1 BSA) for 3 min and transferred onto Superfrost/Plus microscope slides and dried as quickly as possible (less than 5 min). Embryos were permeabilized in 1% Triton X-100 in 1× PBS and fixed in ice-cold 100% methanol for 30 min at −20 °C. Then, slides were transferred into 70% ethanol on ice for 20 min. To perform IF, slides were rinsed in PBS for 5 min, blocked in blocking buffer (1× PBS with 0.1% Tween-20, 1% BSA and 0.4 U μl−1 of Ribonuclease Inhibitor (Invitrogen, 10777019)) for 20 min at room temperature, and then incubated with the primary antibody (mass:volume ratio, 1:50) in blocking buffer for 1 h at room temperature. After three washes with 0.1% Tween-20 in PBS, slides were incubated with secondary antibody (mass:volume ratio, 1:1,000) in blocking buffer for 1 h at room temperature. After three washes with 0.1% Tween-20 in PBS, slides were post-fixed in ice-cold 100% methanol for 30 min at −20 °C. The slides were then transferred into 70% ethanol on ice. Dehydration was performed in 80%, 95% and 100% ethanol (×2), with each incubation lasting for 5 min at room temperature. Slides were then dried for 5 min. Embryos were then hybridized in the hybridization solution (50% Formamide (Sigma, F9037), 1% Dextran Sulfate (Sigma, 30915), 2× SSC (Sigma, S6639-1L), 10 mM VRC (Sigma, 94742), 2 mg ml−1 BSA) containing 5 μg of Alexa Fluor 488-labeled RNA probes per slide at 37 °C overnight (14–15 h). After three washes for 5 min each in hybridization washing solution (50% formamide, 2× SSC) at 42 °C and four washes for 5 min each in 2× SSC, embryos were mounted with DAPI-Vectashield solution (Vector laboratories, H1200). Fluorescence staining was imaged using a laser-scanning confocal microscope (Leica, TCS SP8). IMARIS software (Bitplane) was then used to calculate the number of speckles for each picture.

DNA-FISH combined with RNA-FISH

For the preparation of probes in DNA-FISH, 17 DNA fragments containing T7 promoter (2,100–2,700 bp) were amplified from the mouse genome using LongAmp Taq DNA Polymerase (NEB, M0534L). Primers are given in Supplementary Table 3. Then, in vitro transcription was performed with mixed 17 DNA fragments using the MEGAshortscript Kit (Ambion, AM1354) with ATP, CTP, GTP, UTP and ChromaTide Alexa Fluor 488-5-UTP (Invitrogen, C11403) solution (4:4:4:1:4), in which 80% of uracil was labeled by Alexa Fluor 488. The labeled RNA was then fragmented by adding 1× Ambion RNA fragmentation reagent (Ambion, AM8740) with incubation at 70 °C for 2 min. After adding the stop solution, the labeled RNA was purified and used as probes for DNA-FISH.

The zona pellucida of the mouse embryos was removed with incubation in acidic Tyrode’s solution. Then the embryos were incubated in PBSA for 3 min and transferred onto Superfrost/Plus microscope slides and dried as quickly as possible (less than 5 min). Embryos were permeabilized in 1% Triton X-100 in 1× PBS and fixed in ice-cold 100% methanol for 30 min at −20 °C. Then, slides were transferred into 70% ethanol on ice for 20 min. To perform RNA-FISH, the procedures for dehydration and hybridization were performed as described in the ‘IF combined with the RNA-FISH’ section. After three washes for 5 min each in hybridization washing solution at 42 °C and four washes for 5 min each in 2× SSC, slides were post-fixed in 3% paraformaldehyde (PFA; Sigma, 158127) in PBS for 10 min at room temperature.

To perform DNA-FISH, the embryos were incubated with RNase Cocktail (Invitrogen, AM2288) in 1× PBS for 1 h at 37 °C. After three washes with 1× PBS, embryos were permeabilized in permeabilization solution II (0.7% Triton X-100 and 0.1 M HCl in 1× PBS) for 15 min on ice. Then, slides were transferred into 70% ethanol on ice for 20 min. Dehydration was performed in 80%, 95% and 100% ethanol (×2), with each incubation lasting for 5 min at room temperature. Slides were then dried for 5 min. Then the slides were denatured in the hybridization washing solution for 30 min at 80 °C. After dehydration in cold ethanol, the embryos were hybridized in the hybridization solution containing 5 μg of Alexa Fluor 546-labeled DNA probes per slide at 37 °C overnight (14–15 h). After three washes for 5 min each in hybridization washing solution (50% Formamide, 2× SSC) at 42 °C and four washes for 5 min each in 2× SSC, embryos were mounted with DAPI-Vectashield solution (Vector laboratories, H1200). Fluorescence staining was imaged using a laser-scanning confocal microscope (Leica, TCS SP8). IMARIS software (Bitplane) was then used to calculate the number of speckles for each picture.

Immunofluorescence combined with RNA-FISH and DNA-FISH

Immunofluorescence, RNA-FISH and DNA-FISH were performed as described above. For analysis of the heterogenous occupancy of LincGET speckles and PCBP1 around the Carm1 gene loci, the embryos were injected with Alexa Fluor 488-labeled LincGET, pre-mRNAs for PCBP1-BFP and Alexa Fluor 546-labeled sgRNAs; then, the embryos were directly stained with SYTOX Deep Red Nucleic Acid Stain (Invitrogen, S11380) and Alexa Fluor 647 Phalloidin (Invitrogen, A22287) and then mounted after permeabilization.

ESS percentage analysis from published data

The raw data were downloaded from the GEO database. To trim the original data, the trim_galore (v.0.6.7) software was used with the default parameters. Next, STAR software was used to align reads to mouse genome sequences (https://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_mouse/release_M29/GRCm39.primary_assembly.genome.fa.gz) with default parameters. RMATS (v.4.1.0) software was used to analyze the alternative splicing events and Rmats2sashimiplot (v.2.0.4) software was used for visualization. The levels of all types of alternative splicing events on Carm1 pre-mRNAs were analyzed, including exon-skipping, retained introns, alternative 5′-splicing site, alternative 3′-splicing site, mutually exclusive exons, alternative first exons and alternative last exons.

The percentage of ESS events on exons 3 to 6 relative to noS splicing events is calculated based on Psi values (Ψ). The ESS events on exons 3 to 6 conclude E3S, E5S, E6S, E56S and E3456S; thus, the calculation formulas are as follows:

$$P\rm})=\frac\rm)}}\rm\,E7)}}}\rm)}}\rm\,E7)}}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}}$$

$$P\rm=\frac\rm}\rm}}\rm)}}\rm\,E7)}}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}}$$

$$P\rm=\frac\rm}\rm}}\rm)}}\rm\,E7)}}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}}$$

$$P\rm=\frac\rm}\rm}}\rm)}}\rm\,E7)}}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}}$$

$$P\rm=\frac\rm}\rm}}\rm)}}\rm\,E7)}}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}}$$

$$P\rm=\frac\rm}\rm}}\rm)}}\rm\,E7)}}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}+\frac\rm}\rm}}$$

where P is the percentage of the splicing event; R is the number of reads with the splicing event; and L is the length of exons.

For other alternative splicing events, including the alternative 3′-splicing site of exon 4 (E4A3SS), retained intron 14 (RI14), retained intron 15 (RI15), exon 2 skipping splicing (E2S), exon 3a inclusive (E3a), E11S and E15S, the percentage of these alternative splicing (AS) relative to non-alternative splicing (noAS) is calculated based on Ψ, and the formulas are as follows:

$$P\rm})=\frac\rm()}\rm)}+\it\rm)}}$$

$$P(\rm})=\frac\rm)}}\rm)}+\it\rm)}}$$

where P is the percentage of the splicing event and R is the number of reads with the splicing event.

SPAR-seq and data analysis

The zona pellucida of embryos was removed using Tyrode’s solution (Sigma, T1788). Zona-free embryos were incubated for 5 min in Ca2+-free and Mg2+-free M2 medium before disaggregation by careful pipetting. For single-cell SPAR-seq, each blastomere of two-cell or four-cell embryos was transferred to individual tubes containing 2.3 μl of lysis buffer (0.2% Triton X-100 supplemented with 1 U ml−1 ribonuclease inhibitor (Invitrogen, 10777019)). For bulk SPAR-seq, blastomeres of 50 embryos at the four-cell stage of each group were placed into two individual tubes containing 2.3 μl of lysis buffer separated by GFP+ or GFP−. Reverse transcription was performed with a Single Cell-to-CT qRT–PCR Kit (Invitrogen, 4458237). PCR was performed using primers targeting exon 2 and exon 7 of Carm1 mRNA (Supplementary Table 3).

The samples were sequenced using an Illumina MiSeq sequencer with 300 bp paired-end sequencing reactions (PE300) at BGI company (https://www.genomics.cn). Clean reads were merged using pear (v.0.9.6)55 software with default parameters. Given that the noS and E3a (with longer exon 4, from XM_030244479.2) sequences are longer than 600 bp, the noS and E3a reads were separated into unassembled.forward.fastq and fq.unassembled.reverse.fastq files, whereas the AS reads (E3S, E4S, E5S, E6S, E34S, E56S, E345S and E3456S) and the Gfp spike-ins were separated into assembled.fastq files. To count the AS reads, reads in the assembled.fastq files were mapped to the reference transcripts, including all the exon-skipping candidates, with bowtie2 (v.2.2.5) software56 and transferred into bam files with samtools (v.1.9)57 software. To count the noS and E3a reads, Carm1_E3a_analysis.py (https://github.com/NEAU-Wang-lab/SPAR-seq) was used to note reads in unassembled.forward.fastq and fq.unassembled.reverse.fastq files.

Some samples were sequenced on a PacBio Sequel System according to the standard PacBio Iso-Seq procedures at Annoroad company (http://www.annoroad.com) mixed with samples from other experiments. Then, circular consensus sequencing (CCS) reads were generated using ccs (v.5.0.0) software and converted to fastq format using bam2fastq software in the pbbam (v.1.0.6) package. The number of passes for each of the raw CCS reads was generated using GetCCSpass.pl (https://github.com/Lulab-IGDB/polyA_analysis/blob/main/bin). Then the clean reads of Carm1 were extracted from the CCS reads using Carm1_SPAR_reads_extract.py. Carm1_AS_analysis.py (https://github.com/NEAU-Wang-lab/SPAR-seq) was used to count the AS reads, and Carm1_E3a_analysis.py was used to count the noS and E3a reads.

SHAPE-MaP assays and data analysis

The in vitro transcribed LincGET was re-folded in folding buffer (100 mM NaCl, 100 mM HEPES pH 8.0 and 10 mM MgCl2 in water) at 37 °C for 20 min with or without NONO, PSPC1 or HNRNPU. As a control, one group of LincGET was denatured in denaturing control buffer (50% formamide, 50 mM HEPES pH 8.0 and 4 mM EDTA pH 8.0 in water) at 95 °C for 1 min. For each group, approximately 5 µg RNA was added to the one-ninth volume of NMIA (Invitrogen, M25) at 100 mM in neat DMSO (10 mM final concentration) and incubated at 37 °C for 22 min. The background was assessed by performing no-reagent and denaturing controls.

After fragmentation with RNA fragmentation reagent (Ambion, AM8740), modified LincGET was subjected to MaP reverse transcription27, with SuperScript II Reverse Transcriptase (Invitrogen, 18064014) under Mn2+ conditions (50 mM Tris-HCl pH 8.0, 75 mM KCl, 10 mM DTT, 2 mM dNTPs and 15 mM MnCl2 in water) using random nonamer primers (200 ng μl−1; NEB, S1254S). After synthesizing the second strand by NEBNext pre-mRNAs Second Strand Synthesis Module (NEB, E6111S), the resulting cDNAs were constructed for high-throughput sequencing libraries and sequenced by BGI company.

The deep sequencing datasets were analyzed by ShapeMapper2 (v.2.1.5)58 software with default parameters. The reads were mapped to target sequences by bowtie2 (v.2.2.5)56 software with default parameters as recommended by ShapeMapper2. RNA secondary structures were modeled by Superfold (v.1.0)59 with map files produced by ShapeMapper2. The RNA stem-loop structures for specific fragments were produced by VARNA (v.3.93)60. The ct files from Superfold were used for visualization and varna_colors.txt files from ShapeMapper2 output were used for the reactivity coloring.

Protein purification

Plasmids expressing the NONO, PCBP1 or PCBP2 fused to EGFP tagged with MBP and 6×His (MBP–NONO-EGFP–6×His) were transformed into Transetta (DE3) Chemically Competent Cells (TransGen, CD801). A fresh bacterial colony was inoculated into 4 ml LB media containing ampicillin and grown for 6–8 h at 37 °C until the A600 reached about 0.5. Cells were diluted into 600 ml (1:150) room temperature LB with freshly added ampicillin and grown for 8–12 h at 37 °C. IPTG (TransGen, GF101-01) was added to 1 mM and growth continued for 18 h at 16 °C. Pellets from 600 ml cells were resuspended in 30 ml of Buffer A (50 mM Tris pH 7.5, 500 mM NaCl, 1% Triton X-100, 10 mM imidazole and 1× cOmplete protease inhibitors (Roche, 11873580001)) and divided equally into three 50 ml tubes. For each tube containing 10 ml suspended cells, 15 min of sonication (90 cycles of 5 s on, 5 s off) at 300 W was suitable. The lysate was mixed and cleared by centrifugation at 8,000g for 5 min at 4 °C followed by filtration with a 0.45 μm filter.

Next, the MBP-tagged proteins were purified using a PurKine MBP-Tag Dextrin Packed Column (Dextrin, BMC20206), according to the manufacturer’s instructions. In brief, the top and bottom stoppers on the column were removed to let the stored buffer drain away. Two aliquots of resin-bed volume binding/washing buffer (2 mM Tris-HCl, 20 mM NaCl, 0.1 mM EDTA pH 8.0 and 1 mM DTT) were added to the column and drained away to equilibrate the column. The equilibration step was repeated three times. The cleared lysate was mixed with equal binding/washing buffer and added to the column. For maximal binding, the sample was incubated for 30 min at room temperature or 4 °C. After the sample was drained away, two aliquots of resin-bed volume binding/washing buffer were added to the column and drained to remove the non-specifically adsorbed protein. The washing step was repeated six times. Then, 10 ml of elution buffer (2 mM Tris-HCl, 0.1 mM EDTA pH 8.0, 1 mM Maltose pH 7.4 and 1 mM DTT) was added to the column to wash the target protein. The wash liquid was collected and the content was analyzed using a Coomassie-stained SDS–PAGE gel.

After testing the protein concentration using the Super-Bradford Protein Assay Kit (CWBio, CW0013S), the MBP tag was removed from the purified protein using Factor Xa Protease (NEB, P8010L), in which 1 μg of Factor Xa was added to 50 μg of MBP fusion protein for 24 h at 4 °C.

Next, the sample containing NONO-EGFP–6×His, PCBP1-EGFP–6×His or PCBP2-EGFP–6×His was purified by ProteinIso Ni-NTA Resin (TransGen, DP101-02) combined with an Affinity Chromatography Column (12 ml; TransGen, GP101-03), according to the manufacturer’s instructions. In brief, 6 ml Ni-NTA was added to the column and equilibrated with 60 ml Buffer A. The sample was poured into the column and then washed with 15 volumes of Buffer B (20 mM Tris pH 7.5, 500 mM NaCl, 1% NP-40, 1 mM DTT, 10 mM imidazole and 1× cOmplete protease inhibitors). Protein was eluted with 4 ml Buffer C (20 mM Tris pH 7.5, 500 mM NaCl, 1% NP-40, 1 mM DTT, 200 mM imidazole, and 1× cOmplete protease inhibitors). The wash liquid was collected and the contents were analyzed using a Coomassie-stained SDS–PAGE gel.

Next, the sample (4 ml) was dialyzed in Slide-A-Lyzer 20K Dialysis Cassettes (Thermo, 66012) against 1 l of Buffer D (50 mM Tris-HCl pH 7.5, 125 mM NaCl, 10% glycerol, 1 mM DTT) twice at 4 °C for 8 h each time. Then the sample was concentrated using 30K MWCO Amicon ΜLtra Centrifugal Filters (Millipore, UFC803024) to 200–1,000 μl via centrifuge at 7,500×g at 4 °C for 30 min. The protein concentration was measured using the Super-Bradford Protein Assay Kit (CWBio, CW0013S).

ChIP and data analysis

For chromatin immunoprecipitation to analyze LincGET binding DNA sites, embryos were injected with the LincGET-MS2 and mRNAs for HA-tagged MS2P at the pronuclear stage and collected at the early four-cell stage. For each batch, embryos were washed three times with 1× PBS and were crosslinked for 40 min in a droop containing 1.5 mM freshly prepared ethylene glycol bis(succinimidyl succinate) (EGS; Thermo Scientific, 21566) in a chemical fume hood at room temperature with rotation. Then the embryos were dual-crosslinked in a droop containing 1% formaldehyde (Sigma, 8187081000) for 20 min in a chemical fume hood at room temperature with rotation. To quench the crosslinkers, the embryos were moved to a droop containing 200 mM glycine and were incubated for 10 min at room temperature with rotation in a chemical fume hood. Then embryos were washed three times with 1× PBS and stored at −80 °C with as little liquid as possible until use.

Some batches were combined to collect about 1,000 embryos for each replicate using 20 μl nuclei extraction buffer (10 mM Tris-HCl pH 8.5 (Sigma, 87772), 140 mM NaCl (Sigma, S6546), 5 mM MgCl2 (Invitrogen, AM9530G), 0.6% NP-40 (Abcam, ab142227), 1 mM PMSF, 1× protease unhibitors complex (PIC, CST, 7012)) and incubated for 2 min on ice. For chromatin digestion, the samples were added with 80 μl chromatin digestion buffer (1× MNase buffer (NEB, M0247S), 2 mM DTT, 5% PEG6000 (Avantor, 1008060), 60 U μl−1 MNase (NEB, M0247S)) and incubated at room temperature for 5 min. Then the samples were added with 11 μl MNase stop solution and incubated at room temperature for 1 min. Next, the samples were added with 13 μl nuclear break buffer (1% Triton X-100, 1% odium deoxycholate (Sigma, 30970), 1× PIC) and 86 μl 10× ChIP buffer (CST, 7008) to a final volume of 210 μl, 10 μl of which was treated as input and the other 200 μl of which was used for immunoprecipitation. Before immunoprecipitation, the magnetic beads (CST, 9006) were washed three times with 1× ChIP buffer.

For immunoprecipitation, each sample was added with 1 μg anti-HA antibody (mass:volume ratio, 1:200) and incubated overnight at 4 °C with rotation. Then the samples were added with 12 μl magnetic beads (CST, 9006) and incubated for 5 h at 4 °C with rotation. Next, the beads were washed with 1× ChIP buffer four times for 5 min each time at 4 °C with rotation and with high salt wash buffer (300 μl 10× ChIP buffer, 2,700 μl H2O, 210 μl 5 M NaCl) three times for 5 min at 4 °C with rotation. After washing, chromatin was eluted from the beads with 100 μl 1× ChIP elution buffer (CST, 7009) with incubation at 65 °C for 30 min. The reverse crosslink was performed by adding 4 μl 5 M NaCl and 1 μl proteinase K (20 mg ml−1; Beyotime, ST533) and incubation at 65 °C for 4 h. Then the DNA was extracted by 100 μl chloroform and was precipitated by adding 10 μl of 3 M sodium acetate pH 5.5 (Invitrogen, AM9740), 300 μl isopropanol and 1 μl glycogen (15 mg ml−1; Invitrogen, AM9516) and incubation for 30 min at −80 °C. After washing twice with cold 75% ethanol, the DNA pellet was dissolved with 10 μl of nuclease-free water. Finally, DNA library preparation and sequencing on Hiseq 2500 were performed by BGI company.

For H3R26me2 ChIP–qPCR, anti-H3R26me2 antibody (Abcam, ab127095) was used for immunoprecipitation (mass:volume ratio, 1:200), and qPCR was performed instead of DNA library preparation and sequencing.

The deep sequencing datasets were trimmed by trim_galore (v.0.6.7)61 software with default parameters. The reads were mapped to mouse genome sequences (https://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_mouse/release_M29/GRCm39.primary_assembly.genome.fa.gz) by bwa (v.0.7.17-r1188)62 software with default parameters and transferred into bam files with samtools (v.1.9)57 software. Finally, the bam files were loaded into the IGV (v.2.16.2) for visualization. Peak calling was performed by macs2 (v.2.1.1.20160309) software after duplex removing by picard (v.2.18.29-0) software.

RNA electrophoretic mobility shift assays

RNA oligonucleotides were in vitro transcribed using HiScribe T7 Quick High Yield RNA Synthesis Kit (NEB, E2050) following the manufacturers’ guidelines. DNA templates with a T7 promoter were generated by annealing with primers listed in Supplementary Table 3. RNA oligonucleotides were then biotinylated using the Pierce RNA 3′ End Desthiobiotinylation Kit (Thermo Scientific, 20163) according to the manufacturer’s instructions. The labeled oligonucleotides were gel-purified on 12% denaturing gels before use. The gel shift assay was carried out using the LightShift Chemiluminescent RNA EMSA Kit (Thermo Scientific, 20158). In brief, 5 ng biotinylated wild-type or mutant RNA probe was mixed with 50 μg of 6×His-purified PCBP1 and PCBP2 mix (6×His–PCBP1) (and anti-PCBP1 or anti-PCBP2 antibody for super-shifts (mass:volume ratio, 1:500)) and incubated at room temperature for 30 min in a 20 μl binding reaction containing 1× binding buffer, 5% glycerol and 0.1 mg ml−1 tRNA. The samples were electrophoresed on a 5% native PAGE in 0.5× Tris Borate EDTA (Thermo Scientific, B52), transferred to a positively charged BrightStar-Plus Nylon membrane (Invitrogen, AM10100) and crosslinked in a UV Stratalinker 1800 (Stratagene). To block the membrane, the membrane was incubated in 20 ml nucleic acid detection blocking buffer for 15 min with gentle shaking and then was incubated in 20 ml conjugate/blocking solution (20 ml nucleic acid detection blocking buffer containing 66.7 μl stabilized streptavidin-horseradish peroxidase conjugate) for 15 min with gentle shaking. After washing four times for 5 min each in 20 ml 1× wash buffer with gentle shaking, the membrane was incubated in 30 ml substrate equilibration buffer for 5 min with gentle shaking. Then the membrane was incubated in 12 ml substrate working solution (6 ml luminol/enhancer solution and 6 ml stable peroxide solution) for 5 min without shaking. Finally, the membrane was exposed to a Bio-Rad GelDoc XR+ Gel imaging system for the acquisition of signals.

Minigene analysis of ESS

The minigene plasmids and plasmids encoding shRNAs targeting Pcbp1/2 or control shRNAs were co-transfected into MCF-7 cells. Cells were collected 72 h post transcription, and total RNA was extracted. Then RT–PCR was performed with the primers listed in Supplementary Table 3.

Dot blotting

Single blastomeres were separated from late two-cell and early four-cell embryos and transferred into PCR tubes containing 10 μl of 0.2% (vol/vol) Triton X-100 (Sigma, T9284) and 0.5 μl of RNase inhibitor (Clontech, 2313A). The lysate was split into three equal portions, which were used for dot plotting for CARM1, qPCR for the ESS isoform Carm1_E56S and qPCR for the total expression level of Carm1. For dot blotting, lysates from a series of single blastomeres were spotted onto a nitrocellulose membrane and allowed to absorb. Then the membrane was blocked, incubated sequentially with anti-CARM1 antibody (mass:volume ratio, 1:100) and fluorescence-labeled secondary antibody (mass:volume ratio, 1:5,000), and then exposed to a Bio-Rad GelDoc XR+ Gel imaging system in the same way as those in the 'Western blot' section above.

Secondary structure comparison between LincGET and Neat1

The RNA secondary structures of LincGET and Neat1 were predicted and compared using the ViennaRNA package63. Firstly, sequences of LincGET and Neat1 were split into 500 nt fragments, with the two adjacent fragments overlapped by 100 nt. Then the RNAfold function was used to predict the RNA secondary structures and the RNAdistance function was used to measure the dissimilarity of the RNA secondary structures in a bracket format. The resulting f values are shown in Supplementary Table 5. The smaller the f value, the more similar the structure.

To determine the f values that represent high enough structural similarity, the distribution of f values from the comparison between 52 Neat1 fragments and 10,000 random mRNA fragments (500 nt each) was plotted. We found that the distribution curve is approximately normal, and the boundary of the left 2.5% (P < 0.05) is 320 (Supplementary Fig. 2c). Therefore, we considered structural similarity to be high enough if the f value is less than 320.

Two comparisons with f values less than 320 (Supplementary Fig. 2d and Supplementary Table 5) were found. The secondary structures of these fragments were predicted using the RNAfold web server (http://rna.tbi.univie.ac.at/cgi-bin/RNAWebSuite/RNAfold.cgi).

Absolute qPCR for LincGET and Neat1_2

Firstly, PCR using Q5U Hot Start High-Fidelity DNA Polymerase (NEB, M0515) with primers for TaqMan quantitative real-time PCR (Supplementary Table 3) was performed, and the PCR products were purified using the Zymoclean Gel DNA Recovery Kit (ZYMO, D4007). Standard curves were made using the purified PCR results for LincGET and Neat1. Serial diluted PCR products were used. Standard curves with threshold cycle (Ct) values on the x axis and the logarithmic value with base 2 (log2) of molecular concentration (number of molecules/μl) on the y axis were generated for quantification (Supplementary Table 6). Next, three groups of late two-cell embryos (200 embryos each), three groups of early four-cell embryos (200 embryos each) and three groups of mEpiSCs using FACS (10,000 cells each) were collected. After total RNA extraction and reverse transcription, real-time PCR was performed using TaqMan Universal Master Mix II. The copy numbers of LincGET and Neat1 in late two-cell and early four-cell embryos as well as in mEpiSCs were calculated in Excel.

Statistical analyses

Statistical analyses (mean ± s.e.m.) for differential gene expression, differential fluorescence intensity and differential abundance on gels were performed in Excel. Levels of significance were calculated with Student’s t-tests. Isoform abundance on SDS–PAGE gels or agarose gels was measured in Fiji/ImageJ. The co-localization analysis of LincGET signals with NONO, PSPC1, PCBP1, U2AF2, SRSF1 or Carm1 gene locus in live imaging assays, RNA-FISH combined with immunofluorescence assays and RNA-FISH combined with DNA-FISH assays were calculated by the two (green and red) or three (green, red and blue) separate channel signals per nucleus using Pearson’s correlation coefficient with Coloc 2 plugins in Fiji/ImageJ. Line scans of the relative fluorescence intensity of signals were drawn by separate channel signals with the Plot Profile plugins in Fiji/ImageJ. Particle numbers of LincGET speckles, NONO speckles, PSPC1 speckles and PCBP1 speckles were calculated by Imaris with the particle building system. The R2 value for linear regression was calculated in Excel.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.