Skip to main content
NCBI home page
Science Advances logoLink to Science Advances
. 2025 Oct 17;11(42):eady3894. doi: 10.1126/sciadv.ady3894

Nuclear 2′-O-methylation regulates RNA splicing through its binding protein FUBP1

Boyang Gao 1,2, Bochen Jiang 2,3, Zhongyu Zou 2,3, Bei Liu 2,3, Weijin Liu 4, Li Chen 3, Lisheng Zhang 4,5, Chuan He 2,3,*
PMCID: PMC12533644  PMID: 41105785

Abstract

2′-O-methylation (Nm) is an abundant RNA modification exists on different mammalian RNA species. However, potential Nm recognition by proteins has not been extensively explored. Here, we used RNA affinity purification, followed by mass spectrometry to identify Nm-binding proteins. The Nm-binding protein candidates exhibit enriched binding at known Nm sites. Some candidates display nuclear localization and functions. We focused on the splicing factor FUBP1. Electrophoretic mobility shift assay validated preference of FUBP1 to Nm-modified RNA. As FUBP1 predominantly binds intronic regions, we profiled Nm sites in chromatin-associated RNA (caRNA) and found Nm enrichment within introns. Depletion of Nm led to skipped exons, suggesting Nm-dependent splicing regulation. The caRNA Nm sites overlap with FUBP1-binding sites, and Nm depletion reduced FUBP1 occupancy on modified regions. Furthermore, FUBP1 depletion induced exon skipping in Nm-modified genes, supporting its role in mediating Nm-dependent splicing regulation. Overall, our findings identify FUBP1 as an Nm-binding protein and uncover previously unrecognized nuclear functions for RNA Nm modification.

FUBP1 is identified as an Nm-binding protein that regulates Nm-dependent splicing via intronic sites on chromatin-associated RNA.

INTRODUCTION

Throughout the life cycles of RNA, chemical modifications play critical roles in regulating RNA processing, metabolism, and function (1, 2). These modifications can alter the physical properties of the RNA molecule (3, 4) or recruit specific binding proteins (readers) to modulate RNA function and subsequent cellular pathways (5–9). While changes in physical properties primarily influence RNA structure–dependent regulation, the recruitment of reader proteins can affect diverse downstream processes, including splicing (10), degradation (5), translation (6, 9), and RNA transport to specialized cellular locations (11–13). In addition, recruitment of binding proteins may alter the surrounding state of the modified RNA; examples include chromatin state regulation by N6-methyladenosine (m6A) methylation of chromatin-associated RNA (caRNA) (7, 8, 14).

2′-O-methylation (Nm) is one of the most abundant modifications (15). It can be found in almost all RNA species, including ribosomal RNA (rRNA), tRNA, small nuclear RNA, small nucleolar RNA (snoRNA), microRNA, and mRNA. It has been shown to regulate ribosome biogenesis (16, 17), gene expression (18, 19), innate immune sensing (20, 21), and cell fate decisions (22). Nm entails methylation at the 2′-OH position of the ribose on the RNA backbone. Consequently, it can occur on any of the four ribonucleotide residues, namely, 2′-O-methyladenosine (Am), 2′-O-methylguanosine (Gm), 2′-O-methylcytidine (Cm), and 2′-O-methyluridine (Um). The methylation further stabilizes the ribose 3′-endo conformation, favoring the A-type RNA helix and restricting strand flexibility (23–25). Consequently, Nm installation alters physical properties of the modified RNA. This effect orchestrates Nm-dependent cellular functions of different RNA species, such as stabilizing ribosome structure through modified rRNA (26) and controlling translational efficiency through internal mRNA Nm modification (18). However, whether Nm could be recognized by binding proteins (readers) remains largely unexplored, particularly within the internal regions of mRNA or premature mRNA. While previous studies have identified over a thousand internal Nm sites (27), the corresponding function remains elusive. The processing, metabolism, or function of Nm-modified RNA could be unveiled by the identification of Nm-binding proteins. A previous study reported a functional relationship between cleavage and polyadenylation specificity factor subunit 7 (CPSF7) binding and Nm-mediated alternative polyadenylation, suggesting CPSF7 as a potential Nm-binding protein (28). However, more comprehensive identification of Nm-binding proteins is still lacking.

In this work, we conducted RNA affinity purification, followed by mass spectrometry (MS) to identify Nm-binding protein candidates, among which we focused on the splicing factor, far upstream element binding protein 1 (FUBP1). Up-regulation of FUBP1 has been suggested to promote proliferation of multiple types of cancers (29, 30). Initially characterized as a transcription factor modulating MYC expression, FUBP1 is now recognized as an RNA binding protein (RBP) involved in pre-mRNA splicing (31, 32). Previous studies have indicated that FUBP1 stabilizes splicing machineries at 3′ splice sites, including U2AF2 and SF1. In addition, it interacts with components of U1 small nuclear ribonucleoproteins, potentially facilitating splice sites pairing in long introns. Understanding the binding specificity of FUBP1 could elucidate the mechanism underlying alternative splicing (AS) events that could be affected by Nm modification through FUBP1.

Using electrophoretic mobility shift assay (EMSA), we confirmed the binding preference of FUBP1 to Nm-modified oligos, supporting its role as an Nm-binding protein. Nm-mut-seq (27) analysis of caRNA in HepG2 cells identified 5575 Nm sites, with more than half of the intragenic Nm sites localized in introns. Disruption of Nm installation led to altered splicing patterns, especially increased exon skipping. These Nm sites were bound by FUBP1 in a manner responsive to Nm depletion. Last, FUBP1 preferentially regulates exon skipping at Nm-modified regions. Together, our study identifies FUBP1 as one of the first examples of Nm-binding proteins and highlights a previously unrecognized role of Nm in splicing regulation.

RESULTS

RNA affinity purification followed by MS identified Nm-binding protein candidates

To identify proteins that may preferentially recognize internal RNA Nm modification, we designed biotinylated RNA probes based on published internal Nm sites in HepG2 (27). Given that Nm can occur on four distinct types of ribonucleotides (Am, Gm, Cm, and Um), probes with a single type of Nm may not fully capture the binding proteins recognizing four different Nm modifications. Considering the predominant presence of Gm and Am in mRNA Nm modifications (27), we designed two probes: one with Gm modification (Gm-1) and the other with Am modification (Am-2) (Fig. 1A). Each probe was accompanied by a corresponding control oligo lacking the Nm modification, denoted as Ctrl-1 and Ctrl-2. These probes were selected on the basis of two reported Nm sites, with relatively high mutation ratio and responsiveness to the knockdown (KD) of their methyltransferase fibrillarin (FBL; Fig. 1) (27). The Gm-1 sequence is in the 3′ untranslated region (3′UTR) of RPL13, while the Am-2 sequence situates in the coding sequence (CDS) of UQCRC2.

Fig. 1. RNA affinity purification followed by proteomics identified Nm-binding protein candidates.

Fig. 1.

Open in a new tab

(A) Design of Gm-1 and Am-2 probes based on known Nm sites in HepG2 mRNA. Integrative genomics viewer (IGV) tracks of published Nm-mut-seq showed read depth of mutated (red) and unmutated reads (G in orange and A in green). (B) Proteins that may preferentially bind to RNA Gm-1 (left) and Am-2 (right) probes over control unmodified probes. Tentative Nm-binding proteins (red) have a log2 fold change of >1.58, an adjusted P < 0.05, and a peptide number of ≥4. (C) Gene ontology (GO) analysis of tentative Nm-binding proteins. (D to F) Immunoblotting of Nm-binding protein candidate enrichment by oligo pull-down. The ratio of the corresponding lysate amount of input versus pull-down samples are 1:100.

We performed affinity purification with Gm-1 and Am-2 using HepG2 cell lysate, respectively, and used MS for protein identification of the pull-down fraction. Excess probes were used in these experiments (fig. S1A). Peptide abundance was normalized to the total peptide abundance, and fold changes of Nm-modified versus control groups were computed. With a cutoff of a log2 fold change of >1.58, an adjusted P < 0.05, and a peptide number ≥ 4, we identified 37 enriched proteins in the Gm-1 group and 33 enriched proteins in Am-2 group (Fig. 1B and fig. S1B). These enriched proteins are tentative binding proteins for Gm and Am modifications. The identified protein candidates include mRNA binding proteins and proteins involved in RNA-related metabolic pathways, based on gene ontology (GO) analysis (Fig. 1C), validating the reliability of our assay.

In the candidate protein list, we observed proteins present in both Gm- and Am-enriched groups, indicating them as potential Nm binders without base specificity. One example is FUBP1; another homolog within the same protein family, FUBP3, is also weakly enriched by both oligo pairs. The structural similarity within the protein family suggests that both two proteins may preferentially bind to Nm. To validate this speculation, we detected the protein enrichment by immunoblotting after RNA affinity purification using the two pairs of probes (Fig. 1A). As expected, both FUBP1 and FUBP3 were enriched by the two Nm probes compared to their respective controls (Fig. 1D). This observation indicates that FUBP1 and FUBP3 are tentative candidates recognizing both Gm and Am modifications on RNA.

In addition, insulin-like growth factor 2 mRNA binding protein 1 (IGF2BP1) and IGF2BP3 emerged as candidate binders of Gm and Am. IGF2BPs were initially identified as m6A-binding proteins that stabilize target mRNAs and facilitates their storage under stress conditions (33). Our previous work also revealed their role in recognizing 7-methylguanosine (m7G) (34). While IGF2BP1 primarily binds m6A and stabilizes associated mRNAs, IGF2BP3 preferentially recognizes m7G and promotes mRNA degradation. Immunoblotting following affinity purification of both probe pairs confirmed the preferentially binding of IGF2BP1 and IGF2BP3 to Nm-modified RNA (Fig. 1E). These findings may suggest a complex mechanism of RNA modification recognition by the IGF2BP proteins.

In addition to the FUBP and IGF2BP families, we also validated enrichment of fragile X messenger ribonucleoprotein 1 (FMRP) and nuclear receptor coactivator 5 (NCOA5). FMRP and NCOA5 were enriched by both Gm-1/Ctrl-1 and Am-2/Ctrl-2 probe pairs through affinity purification, followed by immunoblotting (Fig. 1F). The RBP FMRP has previously been shown to colocalize with internal Nm sites (28). Our validation by RNA affinity purification further supports its tentative role as an Nm-binding protein.

Notably, NCOA5 is a transcriptional coactivator that interacts with estrogen receptors (35). The enrichment of NCOA5 by RNA affinity purification suggests it as another example of a transcription factor that binds to and is potentially regulated by RNA. We analyzed a published RNA binding region identification dataset (36) and observed enriched NCOA5 peptides, indicating its involvement in RNA binding as an RBP (fig. S1C). Its preferential binding to Nm may suggest a possible role for Nm in transcriptional regulation. Overall, we validated six proteins with enriched binding to Nm-modified probes: FUBP1, FUBP3, IGF2BP1, IGF2BP3, FMRP, and NCOA5.

Nm-binding protein candidates are enriched at internal mRNA Nm sites

To provide further cellular evidence of the association between binding of these proteins and Nm modifications, we investigated their RNA binding profiles using publicly available eCLIP datasets from the ENCODE project (37, 38). Among the six identified tentative Nm binders, FUBP3, IGF2BP1, and IGF2BP3 have eCLIP data generated in HepG2 cells. Metagene plots generated from these datasets revealed enriched binding of all four proteins around the reported confident Nm sites (Fig. 2, A to C) (27), supporting the association of binding by these proteins with Nm modification. This is consistent with previous reports, where IGF2BP1 and IGF2BP3 have also been suggested for their colocalization with internal Nm sites in other cell lines (28). In vitro validation of the direct binding preference is still lacking for these three proteins, so their roles as Nm-binding proteins require further validations.

Fig. 2. RNA binding sites of Nm-binding protein candidates were enriched at Nm sites.

Fig. 2.

Open in a new tab

(A to C) Metagene plots of eCLIP signals at HepG2 mRNA Nm sites (left) and representative IGV tracks (right) from binding sites of Nm binding protein candidates FUBP3 (A), IGF2BP1 (B), and IGF2BP3 (C). Ranges of the signal intensity are annotated in the brackets.

We examined the sequence contexts of the three proteins at their Nm-associated binding sites. The enriched Nm motifs from the overlapped binding sites closely resembled their own RNA binding motifs (fig. S2, A to C), suggesting that the binding selectivity of these Nm-binding protein candidates are determined by their canonical sequence contexts but may be further enhanced by the Nm modification. To further investigate how RNA binding preferences affect potential Nm site selectivity, we analyzed the distribution of their binding targets and associated Nm sites across mRNA regions. While Nm sites were generally enriched in the CDS compared to the 3′UTR and 5′UTR, Nm-binding protein with 3′UTR preferences, FUBP3, predominantly bound to Nm sites within the 3′UTR, highlighting its preferences to its canonical binding sites (fig. S2D). Conversely, IGF2BP1 and IGF2BP3 exhibited a stronger enrichment of bound Nm sites in the CDS than their overall binding sites (fig. S2D), suggesting a potentially important contribution of selective Nm installation to RNA binding by these RBPs.

The reported function of IGF2BP1 in stabilizing mRNA aligns with the general effect of Nm on mRNA levels as previously described (19, 27, 28). Accordingly, we analyzed RNA level changes using published KD datasets for IGF2BP1 in HepG2 cells (33). Nm-modified mRNA transcripts showed a more notable decrease in expression following IGF2BP1 KD compared to unmodified transcripts (fig. S2E), supporting the association between IGF2BP1 binding and Nm modifications. Further investigation into IGF2BP1 binding to methylated RNA is needed to elucidate the interplay among m6A, m7G, and Nm modifications.

FUBP1 preferentially binds internal mRNA Nm sites

The enrichment of nuclear proteins in the Nm binder candidate list suggests intriguing functions of Nm in the cell nucleus. We chose FUBP1 as an example for more detailed investigations, as FUBP1 is a well-characterized splicing factor (29, 32). Its preference for Nm has been validated using the two probe pairs in both proteomics and immunoblotting analyses. Nevertheless, the RNA affinity purification approach leaves a possibility of indirect FUBP1 binding mediated by an Nm-recognizing partner protein. To confirm the direct interaction between FUBP1 and the Nm modification, we recombinantly expressed and purified FUBP1 with a C-terminal strep-tag from the Expi293 expression system (fig. S3A). We then assessed the binding affinity of the purified FUBP1 toward two probe pairs by EMSA. We observed preferential binding of FUBP1 to both Nm-modified RNA probes compared to their respective controls (Fig. 3A). As a sugar 2′-OH modification, Nm recognition by FUBP1 could be more subtle, when compared with other well-recognized RNA modifications such as m6A by the reader YTH N6-methyladenosine RNA binding proteins (YTHDF proteins) (5). Our EMSA results support that the preferential binding of FUBP1 to Nm is mediated by a direct interaction. Future structural characterization may reveal details of this recognition.

Fig. 3. Splicing regulator FUBP1 prefers Nm-modified RNA.

Fig. 3.

Open in a new tab

(A) EMSA of FUBP1 binding toward Gm-1 and Am-2 probes versus their controls. Probe concentration, 20 nM; protein concentration, starting from 400 nM with twofold dilution. (B) Distribution of FUBP1 PAR-CLIP binding clusters across transcript elements. TTS, transcription termination sites. (C) Metagene plot of FUBP1 PAR-CLIP signals at HepG2 mRNA Nm sites. (D) FUBP1-binding motifs identified in PAR-CLIP. (E) Enrichment of Nm motifs at the FUBP1-binding sites. (F) FUBP1-binding motifs identified in PAR-CLIP clusters overlapping with mRNA Nm sites.

In addition to biochemical evidence, we examined whether FUBP1 binding occurs at endogenous Nm sites within various RNA sequence contexts. We conducted photoactivatable ribonucleoside-enhanced cross-linking and immunoprecipitation (PAR-CLIP) of FUBP1 in HepG2 cells. Genome-wide mutation analysis revealed a T-to-C mutation ratio of >55% (fig. S3B). We identified 3913 FUBP1-binding sites (fig. S3C), with ~59% of these located within intronic regions (Fig. 3B), consistent with the reported intron preference of FUBP1 binding (32). With the published mRNA Nm sites at base resolution, we first investigated FUBP1 binding at reported mature mRNA Nm sites. A metagene plot revealed a clear enrichment of PAR-CLIP reads at the confident Nm sites (Fig. 3C). FUBP1-binding motifs were found to be U-rich (Fig. 3D), consistent with the U-rich motifs associated with mRNA Nm. Notably, Nm motifs enriched at FUBP1-binding sites closely resembled the RNA binding motifs shared across all its binding sites (Fig. 3, E and F). These results indicate an intrinsic preference of FUBP1 for Nm-modified regions, further supporting its role as an Nm-binding protein.

Abundant Nm modifications on caRNA regulate splicing events

The preference of splice factor FUBP1 for Nm modifications indicates an unrecognized intron-dependent function of Nm. To explore whether Nm modifications are also present in intronic regions, we measured the overall levels of Nm in rRNA-depleted HepG2 caRNA by ultrahigh-performance liquid chromatography coupled with triple quadrupole MS (UHPLC-QQQ-MS/MS), with quantitative polymerase chain reaction (qPCR) validation of the rRNA contamination comparable to or lower than mRNA (fig. S4A). The intensity ratios of Am/A and Cm/C are notably higher in caRNA compared to mRNA (Fig. 4A). The Gm/G ratio in caRNA is about one-third of that in mRNA; however, the considerably longer length of introns likely compensates, resulting in an overall more Gm sites in introns than those in exons. Our findings suggest enrichment of Nm modifications in caRNA.

Fig. 4. Nm modifications on HepG2 caRNA affect splicing events.

Fig. 4.

Open in a new tab

(A) UHPLC-QQQ-MS/MS of Nm abundances in HepG2 mRNA and caRNA. (B) Composition of Gm, Am, and Cm among the 5575 caRNA Nm sites. (C) Transcript element distribution of 2228 caRNA Nm sites in protein-coding genes. (D) Averaged caRNA Nm mutation ratio across intronic regions in a 200-bp binning. Intronic sequences at the boundary of splice sites were marked in blue. (E) Profile of modified protein-coding gene transcripts with the corresponding numbers of Nm sites. (F) Number and average mutation ratio of all five-base Nm motifs. UNmU was enriched in more frequent and highly modified motifs. (G) Number of interactions with various snoRNA detected by snoKARR-seq at all caRNA Nm sites. (H) Profile of FBL-dependent AS events, with a significance threshold false discovery rate (FDR) of <0.1. (I) Inclusion level differences of FBL/U3 or FBL-only SE events.

To examine caRNA Nm distribution profiles, we performed Nm-mut-seq on HepG2 caRNA. Nm-mut-seq uses an imbalanced deoxynucleotide triphosphate (dNTP) supply in reverse transcription (RT) reactions to induce mutations of Gm, Am, and Cm into T (27). During analysis, the treated samples could be controlled by RT reactions using balanced dNTP supply (input), as well as a set of spike-in that does not harbor Nm modifications but undergoes RT with imbalanced dNTP (background). We analyzed the mutations of caRNA Nm-mut-seq by JACUSA (39) and designated sites as Nm-modified when all three replicates showed a sequencing depth of ≥10, a mutated read depth of ≥3, mutation ratios of >3-fold of input mutation ratios, and >1.5-fold of background mutation ratios. On the basis of these criteria, we identified 5575 Nm sites (fig. S4B), far more than the reported 1051 Nm sites or 494 confident Nm sites found in HepG2 mRNA (27).

Similar to mRNA, Gm sites are the most abundantly modified Nm sites in caRNA (Fig. 4B), despite the relatively higher abundances of Am and Cm detected by UHPLC-QQQ-MS/MS (Fig. 4A). The numbers of confident Am and Cm sites in caRNA are comparable. Most of caRNA Nm sites are located within protein-coding genes (fig. S4C), with ~53.9% of them residing in introns (Fig. 4C). This is consistent with our hypothesis of intron-dependent function of Nm. These intronic Nm sites show modest enrichment near 5′ and 3′ splice sites (Fig. 4D), suggesting a potential role of Nm in splicing regulation. Our previous work has shown that internal Nm sites on mRNA often appear densely clustered along a stretch of RNA (27). We hypothesized that this clustering may amplify the relatively weak binding preferences of Nm-binding proteins toward modified RNA. We observed that 2228 caRNA Nm sites on protein-coding genes were distributed across only 323 genes, with more than half of these genes harboring multiple Nm sites (Fig. 4E). In addition to protein-coding genes, several long noncoding RNAs (lncRNAs) also accumulate hundreds of Nm sites (fig. S4D), suggesting a previously unrecognized layer of Nm-dependent regulation in their respective functions.

Nm installation in HepG2 caRNA is enriched in the UNmU sequence context (Fig. 4F), similar to the motifs observed in HepG2 mRNA (27). The consistency should be expected, as caRNA modifications in intronic regions and noncoding RNA (ncRNA) are generally installed by the same methyltransferases responsible for mRNA modification, despite their different fates after RNA processing and nuclear export (7). The observed motifs also overlap with U-rich elements recognized by key splicing factors, such as the 5′ splice sites GU, the branch sites, the UGCAUG hexanucleotides (40), the polypyrimidine tract (41), etc. This further indicates a role of Nm in splicing regulation. Notably, the U-rich motif is consistent with the known RNA binding preference of FUBP1 (32), further suggesting its role as an Nm-binding protein that recognizes intronic Nm to potentially modulate splicing.

To understand the regulation of Nm installation on caRNA, we analyzed snoRNA interactions with Nm-modified regions using the published snoRNA enriched and kethoxal assisted RNA-RNA interaction sequencing (snoKARR-seq) dataset (42). Most of snoRNA interaction with caRNA Nm sites were mediated through U3 (Fig. 4G and fig. S4E), highlighting the critical role of U3 in directing Nm installation on caRNA. Our finding suggests that U3 KD may serve as an effective strategy to perturb caRNA Nm installation in addition to depletion of the Nm methyltransferase FBL.

The preference of Nm binding by the splicing factor FUBP1 may suggest the potential role of Nm in splicing regulation. It has been reported that ectopic overexpression of snoRNA that installs Nm in pre-mRNA branch sites could manipulate splicing of certain RNA transcripts (43, 44). To investigate the role of caRNA Nm in splicing regulation, we depleted FBL or U3 in HepG2 cells (fig. S4F). We examined five types of AS events: alternative 3′ splice sites (A3SSs), alternative 5′ splice sites (A5SSs), mutually exclusive exons (MXEs), retained introns (RIs), and skipped exons (SEs). AS events dependent on FBL or U3 exhibited similar distributions, with SE events being the most prevalent (Fig. 4H and fig. S4G). In both FBL depletion and U3 KD, more SE events were up-regulated than down-regulated, while other AS types showed nonsignificant or inconsistent changes (fig. S4H). This indicates that Nm installation may play a role in preventing exon skipping. To further examine this, we compared SE events that consistently occurred following both FBL and U3 KD (FBL/U3) with those that occurred only or inconsistently after FBL KD (FBL-only) or U3 KD (U3-only). The FBL/U3 SE events exhibited a higher degree of up-regulation compared to FBL-only and U3-only groups (Fig. 4I and fig. S4I), further supporting a role for Nm in suppressing SE events. Overall, our findings reveal abundant Nm modifications in caRNA, with intronic installation affecting splicing regulation.

FUBP1 mediates Nm-dependent splicing regulation

Our Nm-mut-seq analysis of HepG2 caRNA revealed a substantial presence of Nm modifications within intronic regions, potentially involved in splicing regulation. Given that FUBP1 is a known splicing factor implicated in splice site recognition (32), we hypothesized that it may mediate Nm-dependent splicing regulation. To investigate this, we first validated the association of FUBP1 with caRNA Nm sites. Of the 5575 caRNA Nm sites, 2030 were bound by FUBP1, as determined by PAR-CLIP analysis (Fig. 5A). Correspondingly, FUBP1 PAR-CLIP signals were enriched around caRNA Nm sites (Fig. 5B). Although FUBP1 primarily targets protein-coding genes (fig. S3C), it also binds to the heavily modified lncRNAs MALAT1 and NEAT1 in multiple clusters (fig. S5A). Together, these findings demonstrate that FUBP1 preferentially binds to Nm sites in both caRNA and mRNA, consolidating its role as an Nm-binding protein.

Fig. 5. FUBP1 binds to Nm-modified caRNA to affect splicing.

Fig. 5.

Open in a new tab

(A) Overlap between caRNA Nm sites and FUBP1 PAR-CLIP clusters extending 200 bp on each side. (B) Metagene plot of FUBP1 PAR-CLIP signals of ±0.3 kb around HepG2 caRNA Nm sites. (C) FUBP1 CLIP signal changes at peaks overlapping with Nm sites (Nm peaks) versus others after FBL KD. (D) FUBP1 RIP signal changes at gene transcripts with Nm modifications (Nm genes) versus others after FBL KD. (E) EMSA of FUBP1 binding toward Nm-modified or control probes originated from SEC14L1 intron. Probe concentration, 10 nM; protein concentration, starting from 400 nM with twofold dilution. (F) Profile of the FUBP1-dependent AS events. (G) Inclusion level differences of SE events after FUBP1 depletion in genes with down-regulated FUBP1 CLIP-seq peaks after siFBL (FUBP1 down genes) versus other genes (others).

If FUBP1 acts as a binding protein for Nm, then its binding should be affected by the depletion of Nm. To test this, we conducted FBL KD. As PAR-CLIP provides mutation-based identification with less reliable measurement of differential peaks, we performed cross-linking immunoprecipitation sequencing (CLIP-seq) of FUBP1 following FBL depletion. Consistent with our PAR-CLIP results, FUBP1 CLIP-seq peaks were predominantly located within intronic regions and protein-coding genes (fig. S5, B and C). Differential peak analysis revealed a decrease in FUBP1 peak signals overlapping with Nm sites, while the overall peak intensity remained largely unchanged (Fig. 5C). Similarly, gene-based differential binding measured by FUBP1 RNA immunoprecipitation sequencing (RIP-seq) also demonstrated a significant decrease in Nm-modified genes compared to unmodified ones after FBL KD (Fig. 5D). These findings collectively confirmed that FUBP1 binding to Nm-modified RNA is affected by the presence of Nm modification.

With FUBP1 CLIP-seq identifying valid Nm-binding sites, we further designed Nm-modified oligos based on FUBP1 target sequences responsive to Nm depletion. EMSA revealed enhanced binding of FUBP1 to Nm-modified oligos compared to their controls (Fig. 5E and fig. S5D), consolidating FUBP1 as an Nm-binding protein. Notably, both oligos harboring multiple Nm sites showed more pronounced differences in binding affinity compared to single-modified probes (Fig. 3A), suggesting that densely clustered Nm may facilitate the recruitment of Nm-binding proteins.

After confirming FUBP1 as a caRNA Nm-binding protein, we aimed to explain the Nm-dependent splicing regulation by FUBP1. FUBP1 facilitates bridging of 5′ and 3′ splice sites (32), and its depletion primarily induces exon skipping (29). Consistent with this, we knocked down FUBP1 in HepG2 cells (fig. S5E) and observed that SE was the main AS event (Fig. 5F). This aligns with the Nm-dependent splicing regulation observed after FBL and U3 KD (Fig. 4H and fig. S4G). Among the SE events, up-regulation occurred more frequently than down-regulation (fig. S5F), consistent with the biased up-regulation induced by FBL or U3 depletion (fig. S4H). To further elucidate the relationship between Nm modification, FUBP1 binding, and splicing changes, we examined SE events in genes with caRNA Nm modifications. We found that Nm-modified genes showed greater up-regulation of SE events after FUBP1 depletion than nonmodified genes, supporting that FUBP1 mediates Nm-dependent inhibition of exon skipping. (fig. S5, G and H). Similarly, genes with reduced FUBP1 binding following FBL depletion exhibited higher SE up-regulation after FUBP1 KD (Fig. 5G), further reinforcing the causal connection between Nm modification, FUBP1 binding, and splicing changes.

DISCUSSION

Nm modifications affect the physical properties of mRNA, exerting crucial regulation on the translation efficiency of modified genes (18). However, the “reader” proteins of Nm modifications on protein-coding genes remain elusive. Characterization of Nm-binding proteins could reveal Nm-dependent function. In this study, we used RNA affinity purification, followed by LC-MS/MS to identify Nm-binding proteins. Immunoblotting validated the enrichment of FUBP1, FUBP3, IGF2BP1, IGF2BP3, FMRP, and NCOA5. This is consistent with previous reports that IGF2BP1, IGF2BP3, and FMRP showed colocalized RNA binding with Nm modifications (28). Published eCLIP datasets of FUBP3, IGF2BP1, and IGF2BP3 further supported their enrichment around endogenous mRNA Nm sites. While functions of IGF2BP proteins align with the effect of mRNA Nm on RNA expression levels, the identified FUBP family proteins and NCOA5 suggest unrecognized realms of nuclear Nm regulation. However, without quantitative biochemical validation, the enrichment of these proteins could be indirect, and their roles as Nm-binding proteins need further validations.

We focused on FUBP1 and validated its role as an Nm-binding protein. EMSA confirmed the direct interaction between FUBP1 and Nm-modified RNA, while PAR-CLIP confirmed enrichment of FUBP1 around internal mRNA Nm sites. To further examine nuclear functions of Nm mediated through FUBP1, we profiled Nm modification sites on caRNA by Nm-mut-seq. We identified 5575 caRNA Nm sites, with more than half of caRNA Nm sites in protein-coding genes localized in introns. We found that intronic Nm displayed enrichment around 5′ and 3′ splice sites, suggesting Nm-mediated splicing regulation. Previous snoKARR-seq data revealed that U3 is the predominant snoRNA responsible for Nm installation on caRNA. Depletion of U3, as well as methyltransferase FBL, led to up-regulation of SE events, supporting a role of Nm in splicing regulation. Prior studies have shown that ectopic installation of Nm at splicing branch sites by snoRNA can alter the splicing of certain transcripts (43, 44). Our findings support the Nm-mediated splicing regulation, pointing to the potential of snoRNA mimics as tools for targeted splicing modulation.

With the profiled Nm sites on caRNA, we validated the overlap of FUBP1 binding with caRNA Nm sites. To further explore the causal relationship, we disrupted Nm by FBL depletion and detected impaired FUBP1 binding at Nm-modified regions by CLIP-seq and RIP-seq. This further supports FUBP1 as the caRNA Nm-binding protein. FUBP1 depletion up-regulates SE events preferentially in Nm-modified genes and genes with down-regulated FUBP1 binding after FBL depletion, confirming its role in mediated Nm-dependent splicing regulation. Overall, our findings identify FUBP1 as a caRNA Nm-binding protein and uncover a new function of Nm in splicing modulation through FUBP1. Nevertheless, it is likely that FUBP1 also contributes to other regulatory processes upon Nm binding, given that it was initially characterized as a transcriptional regulator.

A total of 2228 identified Nm sites were densely populated on 323 protein-coding genes, suggesting cooperation among Nm sites in recruiting binding proteins. This may compensate for the modest preference of FUBP1 for a single Nm modification observed in EMSA experiments, contributing to notable enrichment of FUBP1 at Nm-modified regions. EMSA using densely modified oligos demonstrated a stronger preference of FUBP1 for Nm modifications. These observations highlight the importance of quantitative analysis in studies of Nm-mediated protein recruitment, as a single Nm may exert weaker effect compared to densely clustered Nm sites. Further investigation is needed to elucidate the underlying mechanism of the weak but additive role of Nm in recruiting its potential binding proteins.

Limitations of the study

The Nm-binding protein candidates we identified, aside from FUBP1, were not fully validated. More quantitative biochemical evidence supporting their interactions with Nm are required in the future. Moreover, the binding affinity of FUBP1 to individual Nm sites appears relatively weak compared to that of canonical “readers” of RNA modifications such as m6A. Therefore, additional caution and context are needed when interpreting FUBP1 as an Nm-binding protein.

MATERIALS AND METHODS

Cell culture

HepG2 cells (American Type Culture Collection, HB-8065) were cultured with medium containing Dulbecco’s modified Eagle’s medium (Gibco, 11995040), 10% fetal bovine serum (Gibco, 2614079), and 1% penicillin-streptomycin (Gibco, 15140122) at 37°C with 5% CO2 in the environment. Cells were passaged when reaching ~90% confluency at 1:4 ratio. Mycoplasma were tested by PCR with primers gggagcaaacaggattagataccct and tgcaccatctgtcactctgttaacctc every half a year.

In the small interfering RNA (siRNA)–mediated KD assay, cells at ~90% confluency were passaged at 1:4 ratio into 15-cm plates. Within 16 hours after passaging, 60 μl of Lipofectamine RNAiMAX Transfection Reagent (Invitrogen, 13778150) and 200 pmol of siRNA were diluted in 1 ml of Opti-MEM I Reduced Serum Medium (Gibco, 31985070), respectively. The solutions were mixed together and incubated at room temperature for 5 min before being added into the cell culture. KD reactions for fewer cells were scaled down on the basis of the bottom area of culture plates. The siRNA used in this paper include siControl (QIAGEN, 1027310), siFBL (QIAGEN, SI04164951), siControl2 (Invitrogen, 4390846), and siFUBP1 (Invitrogen, s16966). Cells were harvested 48 hours posttransfection. KD efficiency was measured by qPCR with primers from Origene (HP205317).

U3 KD was conducted with U3 allele-specific oligonucleotide (ASO) (mU*mC*mA*mC*mC*T*T*C*A*C*C*C*T*C*T*mC*mC*mA*mC*mU) controlled by green fluorescent protein ASO (mC*mU*mG*mC*mC*A*T*C*C*A*G*A*T*C*G*mU*mU*mA*mU*mC). Transfection was done by Lipofectamine 3000 transfection reagents (Invitrogen, L3000015), where 600 pmol of ASO and 8 μl of P3000 reagents were added to 125 μl of Opti-MEM medium. In parallel, 6 μl of Lipofectamine 3000 reagents were diluted in 125 μl of Opti-MEM medium. The two mixtures were mixed together, incubated at room temperature for 10 min, and then added to 2 ml of medium in a six-well plate. HepG2 cells corresponding to one-third of a 10-cm plate at ~90% confluency were then added to the transfected medium. Cells were cultured for 72 hours before harvest. KD efficiency was measured by qPCR primer pair CGTGTAGAGCACCGAAAACC and CACTCAGACCGCGTTCTCTC.

Immunoblotting

Samples were lysed in 2× NuPAGE LDS sample buffer (Invitrogen, NP0007) supplemented with 1:20 (v/v) 2-mercaptoethanol (Sigma-Aldrich, M6250-1L). After incubation at 95°C for 10 min, the denatured samples were loaded into 4 to 12% NuPAGE bis-tris gels (Invitrogen, NP0322BOX) and transferred onto nitrocellulose membranes (Bio-Rad, 1620115). Samples were blocked by 5% bovine serum albumin (BSA) (Fisher Scientific, BP1600-1) in PBS Tween-20 (PBST; Thermo Fisher Scientific, 28352) for 30 min, followed by overnight incubation at 4°C in primary antibodies with designated dilution ratios in 3% BSA diluted by PBST. Membranes were washed three times and incubated in the secondary antibody conjugated to HRP at room temperature for 1 hour. Protein signals were developed by SuperSignal West Dura Extended Duration Substrate (Thermo Fisher Scientific, 34075). Antibodies used in this study and their dilution ratios are as follows: anti-FUBP1 (Abcam, ab192867; 1:1000), anti-FUBP3 (Abcam, ab181025; 1:1000), anti-IGF2BP1 (MBL International, RN007P; 1:1000), anti-IGF2BP3 (MBL International, RN009P; 1:1000), anti-NCOA5 (Proteintech, 20175-1-AP; 1:500), anti-FMRP (Abcam, ab17722; 1:1000), and anti-rabbit immunoglobulin G (IgG) linked with HRP (Cell Signaling Technology, 7074S; 1:2000).

RNA affinity purification

The experiment followed the protocol in the previous publication (45) with adjustment. A total of 30 μl of Dynabeads MyOne Streptavidin C1 bead suspension was washed with RNA binding buffer [50 mM Hepes (pH 7.5), 150 mM NaCl, 0.5% NP-40 substitute, and 10 mM MgCl2] and incubated in RNA binding buffer supplemented with yeast tRNA (100 μg/ml; Invitrogen, AM7119) for 1 hour at 4°C with rotation. After two rounds of washing, 400 pmol of Nm-modified or control probes were incubated with beads suspended in RNA binding buffer for 30 min at 4°C with rotation. Beads conjugated with oligos were washed with RNA wash buffer [50 mM Hepes (pH 7.5), 250 mM NaCl, 0.5% NP-40 substitute, and 10 mM MgCl2] and then with protein incubation buffer [10 mM tris-HCl (pH 7.5), 150 mM KCl, 1.5 mM MgCl2, 0.1% NP-40 substitute, and 0.5 mM dithiothreitol (DTT)] twice. HepG2 cell pellets in the volume of 45 μl were lysed in 400 μl of lysis buffer [50 mM tris (pH 7.5), 100 mM NaCl, 1% NP-40 substitute, 0.5% sodium deoxycholate, 100× proteinase inhibitor cocktail (Sigma-Aldrich, P8340), and 100× SUPERase·In RNase (ribonuclease) Inhibitor (Invitrogen, AM2696)] for 30 min at 4°C with rotation. The supernatant of the lysate was harvested by centrifugation at 12,000g for 15 min and separated to two equal halves after saving 5% as input. The lysate was incubated with beads conjugated with oligos, supplemented with tRNA (50 μg/ml), 0.5 mM DTT, and 100× SUPERase·In, and incubated at room temperature for 30 min and 4°C for 90 min with rotation. The beads were washed three times with protein incubation buffer before removal of all supernatant.

LC-MS/MS analysis

Samples were prepared with three replicates, harvested on dry beads, and frozen on dry ice when shipped to the MS center. The beads were heated in 3× reducing LDS sample buffer with 15 mM DTT and 2 M biotin for 10 min at 95°C, and the supernatant was loaded on 12% bis-tris propane SDS–polyacrylamide gel electrophoresis gel for removal of detergent. The gel was run shortly and stained with colloidal Coomassie blue for gel cut of the whole lane. Gel pieces were reduced with DTT, alkylated with iodoacetamide, washed properly, and digested with trypsin overnight at 37°C. The extracted peptides were dried down, redissolved in 2.5% acetonitrile-water solution with 0.1% formic acid, and then run by nano–LC-MS/MS using a 2-hour gradient on a 0.075 mm–by–250 mm C18 column feeding into an Orbitrap Eclipse mass spectrometer.

The quantitation was done by Proteome Discoverer (Thermo Fisher Scientific; version 2.4). All MS/MS samples were searched with Mascot (Matrix Science, London, UK; version 2.6.2) using cRAP_20150130.fasta (124 entries) and uniprot-human_20201207 database (75777 entries) assuming trypsin digestion, with a fragment ion mass tolerance of 0.02 Da and a parent ion tolerance of 10.0 parts per million. The specified variable modifications included asparagine and glutamine deamidation, methionine oxidation, lysine and arginine methylation, and cysteine carbamidomethylation. Peptides were validated by Percolator with a 0.01 posterior error probability threshold. The data were searched with a decoy database to set the false discovery rate (FDR) to 1%. The peptides were quantified using the precursor abundance based on intensity, with the peak abundance normalized by total peptide amount. The sum of peptide group abundances for each sample were normalized to the maximum sum of the analyzed samples. The protein ratios were calculated using summed abundance for each replicate separately and the geometric median of the resulting ratios was used as the protein ratios. To compensate for missing values in some of the replicates, the replicate-based resampling imputation mode was used. The significance of differential expression was generated by analysis of variance (ANOVA) test, and the P values were adjusted by the Benjamini-Hochberg method.

Protein purification

FUBP1-strep were expressed in 100 ml of Expi293F cells (Gibco, A14527) for 48 hours, lysed in lysis buffer [20 mM tris (pH 8.0), 150 mM KCl, 2 mM EDTA, 100× phenylmethylsulfonyl fluoride (PMSF), 1% NP-40 substitute, and 0.5 mM DTT] at 4°C for 30 min. Supernatant was harvested by centrifugation and diluted to threefold volume before passing through 0.22-μm filter. Samples were loaded to 200 μl of Strep-Tactin Sepharose resin (IBA, 21201010) and incubated at 4°C for 1 hour. The resin was flowed through by lysate twice and washed with 25 ml of wash buffer [50 mM tris (pH 8.0), 250 mM NaCl, 0.05% NP-40 substitute, 0.5 mM DTT, and 100× PMSF]. Protein was eluted six times with elution buffer [50 mM tris, (pH 8.0), 250 mM NaCl, 0.01% NP-40 substitute, 0.5 mM DTT, and 2.5 mM desthiobiotin] with 100 μl each time, concentrated with 30-kDa Amicon Ultra Centrifugal Filter (Millipore, UFC503008), and exchanged to storage buffer [50 mM tris, (pH 8.0), 150 mM KCl, 0.1 mM EDTA, and 20% glycerol]. The purified protein was aliquoted, snap frozen in liquid nitrogen, and stored at −80°C.

Electrophoretic mobility shift assay

Probes for EMSA were labeled with 6-carboxyfluorescein (FAM) fluorophore at the 5′ end, with sequence noted in the figures. Probes were refolded by incubation at 75°C for 2 min and natural cooling to room temperature for 10 min. FUBP1 proteins of designated concentrations were incubated with FAM-labeled oligos in binding buffer [10 mM tris (pH 7.5), 140 mM KCl, 10 mM NaCl, 1 mM MgCl2, 10% glycerol, 1 mM DTT, and SUPERase·In RNase Inhibitor (1 U/μl)] at room temperature for 30 min. The mixtures were loaded to 4 to 20% Novex tris-borate EDTA (TBE) gels (EC62255BOX) that have been rerun at 90 V for 30 min at 4°C in 0.5× TBE. Gels were run for 2 hours before imaging. The fluorescence signal intensity was quantified by ImageJ, and Kd value was calculated by GraphPad.

UHPLC-QQQ-MS/MS quantification

HepG2 mRNA were purified by two rounds of polyadenylate [poly(A)] selection following the commercial protocol of Dynabeads mRNA DIRECT Purification Kit (Invitrogen, 61011). HepG2 chromatin were purified following the published protocol reported previously (7), with adjustment of lysis buffer to 10 mM tris-HCl (pH 7.5), 0.15% NP-40 substitute, and 150 mM NaCl. The caRNA were purified from fractionated chromatin, followed by two rounds of RiboMinus reactions according to the commercial instruction of RiboMinus Eukaryote System v2 (Invitrogen, A15026).

The verification of rRNA contamination was performed by qPCR for 18S (with primer pair CGGACAGGATTGACAGATTG and CAAATCGCTCCACCAACTAA) and 45S (with primer pair TCGCTGCGATCTATTGAAAG and AGGAAGACGAACGGAAGGAC) rRNA. The same amount of RNA was used for RT. The absolute amount of cDNA product was quantitated by the control of PCR product of the two primer pairs with known amount of DNA template.

For UHPLC-QQQ-MS/MS detection, 100 ng of RNA was digested with Nuclease P1 [New England Biolabs (NEB), M0660S] at 37°C overnight, followed by digestion with shrimp alkaline phosphatase (rSAP, NEB, M0371S) in rCutSmart buffer 2 hours at 37°C. The samples were diluted and filtered by 0.22-μm polyvinylidene difluoride filter (Millipore, SLGVR33RB) and then injected into a C18 reversed phase column coupled online to Agilent 6460 LC-MS/MS spectrometer in positive electrospray ionization mode. Nucleosides were quantified using nucleoside-to-base transitions of Am (282 to 136), A (268 to 136), Gm (198.1 to 152.1), G (284 to 152), Cm (258.2 to 112), and C (244 to 112). The signal intensity of Nm nucleotides was normalized to the corresponding unmodified nucleotides to enable comparison of samples with different length.

Photoactivatable ribonucleoside-enhanced cross-linking and immunoprecipitation

The experiment was performed with an adapted protocol from a previous report (46). Two biological replicates of 150 million HepG2 cells were cultured with 200 μM 4-Thiouridine for 14 hours. Cells were cross-linked by 365-nm ultraviolet (UV) at 1500 J/m2 twice, harvested, and lysed by iCLIP lysis buffer [50 mM tris (pH 7.5), 100 mM NaCl, 1% NP-40 substitute, 0.1% SDS, 0.5% sodium deoxycholate, 100× protease inhibitor cocktail, and 100× SUPERase·In RNase Inhibitor] at 4°C for 15 min with rotation. To release FUBP1 proteins associated with the chromatin, cell lysate was supplemented with 1% SDS, sonicated at 30% amplitude with 2-s:4-s cycles for 1 min on ice. The lysate was 10-fold diluted by iCLIP lysis buffer without SDS, centrifuged to save the supernatant, and treated with RNase T1 (0.2 U/μl; Thermo Fisher Scientific, EN0642) at 22°C for 15 min before being quenched on ice for 5 min. Protein G beads (Invitrogen, 10009D) were conjugated with 20 μg of FUBP1 antibody (Abcam, ab192867) or rabbit IgG (Cell Signaling Technology, 2729) by incubation at 4°C for 1 hour, then washed, and mixed with RNase T1–treated lysate to rotate at 4°C for 4 hours. Beads were washed three times with CLIP wash buffer [50 mM tris (pH 7.5), 300 mM KCl, 0.05% NP-40 substitute, 1000× proteinase inhibitor cocktail, and 1000× SUPERase·In RNase Inhibitor], digested with RNase T1 (10 U/μl) at 22°C for 8 min, washed with CLIP high-salt buffer [50 mM tris (pH 7.5), 500 mM KCl, 0.05% NP-40 substitute, 1000× protease inhibitor cocktail, and 1000× SUPERase·In RNase Inhibitor] and T4 polynucleotide kinase (T4 PNK) buffer without DTT three times each, and underwent end repair by T4 PNK (NEB, M0201L). The immunoprecipitation was validated by biotinylation, eluted with proteinase K digestion (Thermo Fisher Scientific, EO0491), with purified RNA used to construct libraries with NEBNext Small RNA Library Prep Set (NEB, E7330S). Sequencing was performed by Illumina NovaSeq 6000 paired-end 50-bp reads.

Bioinformatic analysis of PAR-CLIP

For both FUBP1 PAR-CLIP and its IgG control, adapters were trimmed by Cutadapt (47), and reads were mapped to hg38 human genome by HISAT2 (48), with parameter “--reorder --no-unal --pen-noncansplice 12.” Duplicates were filtered by Picard MarkDuplicates. R1 reads with mapping quality higher than 30 were used for identification of FUBP1-binding clusters by wavClusteR (49, 50), with removal of reads containing “^” in the MD tag. The identified clusters were overlapped and then annotated by Homer (51) annotatepeaks. The binding clusters identified by IgG were supposed to be removed from the FUBP1-binding clusters. Nevertheless, we identified no clusters from IgG control. Metaplots centering at Nm sites were generated by deepTools (52). Motifs were identified by Homer findMotifsGenome with parameter “-rna -size 200 -len 6”.

Cross-linking immunoprecipitation sequencing

The experiment followed the procedure in the previous publication (53). Three biological replicates of 150 million HepG2 cells were harvested after 48 hours of FBL KD, with the KD efficiency validated by RT-qPCR with primers from Origene (HP205317). Samples were cross-linked three times by UV (1500 J/m2) at 254 nm, lysed, centrifuged, and digested with the same condition as PAR-CLIP. After saving 2% of the lysate as input, lysates were incubated with beads treated the same as PAR-CLIP. The following steps were exactly the same, with the exception that input samples were treated separately with RNase T1 (10 U/μl) at 22°C for 8 min, followed by proteinase K digestion and end repair in the next. Sequencing by Illumina NovaSeq X was conducted for single-end 100-bp reads.

Bioinformatic analysis of CLIP-seq

Cutadapt (47) was used to trim the adapters, and HISAT2 (48) mapping to the reference genome (hg38) was performed, with parameter “--reorder --no-unal --pen-noncansplice 12.” Peaks were identified by Piranha (54) with input samples as the covariate, and bin size designated as 50 bp. The identified peaks of siControl and siFBL were merged and annotated by Homer (51) annotatepeaks. Differential binding was analyzed by DiffBind (55), with relative log expression normalization considering the background (normalize = DBA_NORM_RLE, background = TRUE). Down-regulated peaks were defined as a log2 fold change of <−0.58 and an FDR < 0.1. Peaks within 2 kb of Nm sites were considered as Nm modified peaks for computing binding fold change.

RNA sequencing

HepG2 cell RNA was harvested by TRIzol reagent (Invitrogen, 15596026) after 48 hours of siRNA-mediated KD or 72 hours of ASO-mediated KD with three biological replicates. RNA was purified with the manufacturer’s procedures, and the KD was measured by RT-qPCR with primers from Origene (HP207343). After poly(A) RNA selection with Dynabeads mRNA DIRECT Purification Kit and the samples were constructed to libraries with SMARTer Stranded Total RNA-Seq Kit v2–Pico Input Mammalian (Takara, 634412). Next-generation sequencing was performed with Illumina NovaSeq 6000 paired-end 150-bp reads.

RNA immunoprecipitation sequencing

Two biological replicates of HepG2 cells were harvested after 48 hours of FBL KD. The KD efficiency was validated by RT-qPCR. Cells were lysed by iCLIP lysis buffer at 4°C for 15 min with rotation, followed by centrifugation at 13,000g for 15 min. Protein G beads were conjugated with 16 μg of FUBP1 antibody by incubation at 4°C for 1 hour, then washed, and mixed with lysate supernatant to rotate at 4°C for 4 hours. Beads were washed three times with CLIP wash buffer and eluted with proteinase K digestion (Thermo Fisher Scientific, EO0491) Input samples harvested before immunoprecipitation was also digested with proteinase K for RNA recovery. The purified RNA was constructed for library with SMARTer Stranded Total RNA-Seq Kit v2–Pico Input Mammalian. Next-generation sequencing was performed with Illumina NovaSeq 6000 paired-end 150-bp reads.

Bioinformatic analysis for RNA sequencing and RIP-seq

Adapters were trimmed by Cutadapt (47), and reads were mapped to hg38 human genome by HISAT2 (48), with parameter “--reorder --no-unal --pen-noncansplice 12.” Reads were counted for each gene by HTSeq (56), and differential expression analysis by DESeq2 (57) was conducted. For RIP-seq samples, the differential expression analysis compares immunoprecipitation versus input samples, where the fold change was considered as enrichment of each gene. The enrichment was further compared between control and FBL KD samples to measure the differential binding of FUBP1 upon Nm depletion.

The AS events were detected by rMATS (58), where we used the input samples of RIP-seq for analysis upon FBL KD. Only junction reads were used for calculation of AS scores. Events with an FDR of <0.1 were considered as events.

Nm-mut-seq

Three biological replicates of HepG2 caRNA were fractionated and purified in the same procedures as UHPLC-QQQ-MS/MS analysis. rRNA was removed by RiboMinus Eukaryote System v2, and the residual RNA was cleaned up by RNA Clean & Concentrator-5 (Zymo Research, R1014) with removal of a short RNA of <200 nt. The construction of Nm-mut-seq followed the published procedures (27), using 3′ linker 5’rApp-NNNNNATCACGAGATCGGAAGAGCACACGTCT-3SpC3 and 5′ SR adapter supplied in the NEBNext Small RNA Library Prep Set. Sequencing by Illumina NovaSeq X was conducted with single-end 100-bp reads.

Bioinformatic analysis for Nm-mut-seq

Adapters were trimmed by Cutadapt (47), and duplicates marked by unique molecular identifier (UMI) was removed using BBMap (59). Reads were mapped to hg38 human genome by TopHat2 (60), with parameters “-g 4 -N 3 --read-edit-dist 3.” Mutations to T were identified by JACUSA (39), with further selection of mutated sites in all three replicates having a sequencing depth of ≥10, a mutated read depth of ≥3, mutation ratios of >3-fold of input mutation ratios, and mutation ratios of >1.5-fold of background mutation ratios. Nm sites were annotated by Homer (51) annotatepeaks.

Statistical analysis

P values annotated in figures were quantified on the basis of two-tailed Student’s t test. Chi-squared test was used to statistically test the preferential occurrence of different splicing events with increased or decreased inclusion level differences. The Kd value and SD for EMSA were fitted by GraphPad Prism under the mode of “one site–specific binding with Hill slope.”

Acknowledgments

We are grateful to the Genomics Facility at University of Chicago for high-throughput sequencing, the Mass Spectrometry Facility at University of Chicago for UHPLC-QQQ-MS/MS, and the Proteomics & Metabolomics Facility at University of Nebraska–Lincoln for LC-MS/MS analysis.

Funding: This work is supported by the National Institute of Health RM1 HG008935 and R01 HL155909 (to C.H.). C.H. is an investigator of the Howard Hughes Medical Institute.

Author contributions: Conceptualization: C.H. and L.Z. Methodology: B.G., L.C., L.Z., W.L., and B.L. Investigation: B.G., L.Z., and B.L. Visualization: B.G., Z.Z., L.Z., and W.L. Resources: C.H., L.Z., and L.C. Funding acquisition: C.H. and L.Z. Formal analysis: B.G., L.Z., and W.L. Software: B.G. and L.Z. Validation: L.Z., B.J., L.C., and B.L. Data curation: L.Z. Project administration: C.H. and L.Z. Supervision: C.H. and L.Z. Writing—original draft: C.H., B.G., and L.Z. Writing—review and editing: C.H., B.G., Z.Z., and L.Z.

Competing interests: C.H. is a scientific founder; a member of the scientific advisory board and equity holder of Aferna Bio, AllyRNA, and Ellis Bio; a scientific cofounder and equity holder of Accent Therapeutics; and a member of the scientific advisory board of Rona Therapeutics and Element Biosciences. All other authors declare that they have no competing interests.

Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. The sequencing data are accessible in the Gene Expression Omnibus through accession numbers GSE294792, GSE294793, and GSE294794. Published Nm-mut-seq dataset of internal mRNA (GSE174518); ENCODE datasets of eCLIP experiments targeting FUBP3 (ENCSR486YGP), IGF2BP1 (ENCSR744GEU), and IGF2BP3 (ENCSR993OLA); and their IgG control (GSE77629) were used for analysis.

Supplementary Materials

The PDF file includes:

Figs. S1 to S5

Legends for tables S1 to S10

sciadv.ady3894_sm.pdf (2.2MB, pdf)

Other Supplementary Material for this manuscript includes the following:

Tables S1 to S10

sciadv.ady3894_tables_s1_to_s10.zip (25.3MB, zip)

REFERENCES AND NOTES

  • 1.Zhao B. S., Roundtree I. A., He C., Post-transcriptional gene regulation by mRNA modifications. Nat. Rev. Mol. Cell Biol. 18, 31–42 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Shi H., Wei J., He C., Where, when, and how: Context-dependent functions of RNA methylation writers, readers, and erasers. Mol. Cell 74, 640–650 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Liu N., Dai Q., Zheng G., He C., Parisien M., Pan T., N6-methyladenosine-dependent RNA structural switches regulate RNA-protein interactions. Nature 518, 560–564 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Zhou H., Kimsey I. J., Nikolova E. N., Sathyamoorthy B., Grazioli G., McSally J., Bai T., Wunderlich C. H., Kreutz C., Andricioaei I., Al-Hashimi H. M., m1A and m1G disrupt A-RNA structure through the intrinsic instability of Hoogsteen base pairs. Nat. Struct. Mol. Biol. 23, 803–810 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Wang X., Lu Z., Gomez A., Hon G. C., Yue Y., Han D., Fu Y., Parisien M., Dai Q., Jia G., Ren B., Pan T., He C., N6-methyladenosine-dependent regulation of messenger RNA stability. Nature 505, 117–120 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Wang X., Zhao B. S., Roundtree I. A., Lu Z., Han D., Ma H., Weng X., Chen K., Shi H., He C., N6-methyladenosine modulates messenger RNA translation efficiency. Cell 161, 1388–1399 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Liu J., Dou X., Chen C., Chen C., Liu C., Xu M. M., Zhao S., Shen B., Gao Y., Han D., He C., N6-methyladenosine of chromosome-associated regulatory RNA regulates chromatin state and transcription. Science 367, 580–586 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Dou X., Xiao Y., Shen C., Wang K., Wu T., Liu C., Li Y., Yu X., Liu J., Dai Q., Pajdzik K., Ye C., Ge R., Gao B., Yu J., Sun S., Chen M., Chen J., He C., RBFOX2 recognizes N6-methyladenosine to suppress transcription and block myeloid leukaemia differentiation. Nat. Cell Biol. 25, 1359–1368 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Zou Z., Sepich-Poore C., Zhou X., Wei J., He C., The mechanism underlying redundant functions of the YTHDF proteins. Genome Biol. 24, 17 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Xiao W., Adhikari S., Dahal U., Chen Y. S., Hao Y. J., Sun B. F., Sun H. Y., Li A., Ping X. L., Lai W. Y., Wang X., Ma H. L., Huang C. M., Yang Y., Huang N., Jiang G. B., Wang H. L., Zhou Q., Wang X. J., Zhao Y. L., Yang Y. G., Nuclear m6A reader YTHDC1 regulates mRNA splicing. Mol. Cell 61, 507–519 (2016). [DOI] [PubMed] [Google Scholar]
  • 11.Roundtree I. A., Luo G. Z., Zhang Z., Wang X., Zhou T., Cui Y., Sha J., Huang X., Guerrero L., Xie P., He E., Shen B., He C., YTHDC1 mediates nuclear export of N6-methyladenosine methylated mRNAs. eLife 6, e31311 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Fu Y., Zhuang X., m6A-binding YTHDF proteins promote stress granule formation. Nat. Chem. Biol. 16, 955–963 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Zou Z., Wei J., Chen Y., Kang Y., Shi H., Yang F., Shi Z., Chen S., Zhou Y., Sepich-Poore C., Zhuang X., Zhou X., Jiang H., Wen Z., Jin P., Luo C., He C., FMRP phosphorylation modulates neuronal translation through YTHDF1. Mol. Cell 83, 4304–4317.e8 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Chen C., Liu W., Guo J., Liu Y., Liu X., Liu J., Dou X., Le R., Huang Y., Li C., Yang L., Kou X., Zhao Y., Wu Y., Chen J., Wang H., Shen B., Gao Y., Gao S., Nuclear m6A reader YTHDC1 regulates the scaffold function of LINE1 RNA in mouse ESCs and early embryos. Protein Cell 12, 455–474 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Höfler S., Carlomagno T., Structural and functional roles of 2′-O-ribose methylations and their enzymatic machinery across multiple classes of RNAs. Curr. Opin. Struct. Biol. 65, 42–50 (2020). [DOI] [PubMed] [Google Scholar]
  • 16.Yelland J. N., Bravo J. P. K., Black J. J., Taylor D. W., Johnson A. W., A single 2′-O-methylation of ribosomal RNA gates assembly of a functional ribosome. Nat. Struct. Mol. Biol. 30, 91–98 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Krogh N., Jansson M. D., Häfner S. J., Tehler D., Birkedal U., Christensen-Dalsgaard M., Lund A. H., Nielsen H., Profiling of 2′-O-Me in human rRNA reveals a subset of fractionally modified positions and provides evidence for ribosome heterogeneity. Nucleic Acids Res. 44, 7884–7895 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Choi J., Indrisiunaite G., DeMirci H., Ieong K.-W., Wang J., Petrov A., Prabhakar A., Rechavi G., Dominissini D., He C., Ehrenberg M., Puglisi J. D., 2′-O-methylation in mRNA disrupts tRNA decoding during translation elongation. Nat. Struct. Mol. Biol. 25, 208–216 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Elliott B. A., Ho H.-T., Ranganathan S. V., Vangaveti S., Ilkayeva O., Abou Assi H., Choi A. K., Agris P. F., Holley C. L., Modification of messenger RNA by 2′-O-methylation regulates gene expression in vivo. Nat. Commun. 10, 3401 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Züst R., Cervantes-Barragan L., Habjan M., Maier R., Neuman B. W., Ziebuhr J., Szretter K. J., Baker S. C., Barchet W., Diamond M. S., Siddell S. G., Ludewig B., Thiel V., Ribose 2′-O-methylation provides a molecular signature for the distinction of self and non-self mRNA dependent on the RNA sensor Mda5. Nat. Immunol. 12, 137–143 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Ringeard M., Marchand V., Decroly E., Motorin Y., Bennasser Y., FTSJ3 is an RNA 2′-O-methyltransferase recruited by HIV to avoid innate immune sensing. Nature 565, 500–504 (2019). [DOI] [PubMed] [Google Scholar]
  • 22.Häfner S. J., Jansson M. D., Altinel K., Andersen K. L., Abay-Nørgaard Z., Ménard P., Fontenas M., Sørensen D. M., Gay D. M., Arendrup F. S., Tehler D., Krogh N., Nielsen H., Kraushar M. L., Kirkeby A., Lund A. H., Ribosomal RNA 2′-O-methylation dynamics impact cell fate decisions. Dev. Cell 58, 1593–1609.e9 (2023). [DOI] [PubMed] [Google Scholar]
  • 23.Prusiner P., Yathindra N., Sundaralingam M., Effect of ribose O(2′)-methylation on the conformation of nucleosides and nucleotides. Biochim. Biophys. Acta 366, 115–123 (1974). [DOI] [PubMed] [Google Scholar]
  • 24.He C. C., Hamlow L. A., Devereaux Z. J., Zhu Y., Nei Y. w., Fan L., McNary C. P., Maitre P., Steinmetz V., Schindler B., Compagnon I., Armentrout P. B., Rodgers M. T., Structural and energetic effects of O2′-ribose methylation of protonated purine nucleosides. J. Phys. Chem. B. 122, 9147–9160 (2018). [DOI] [PubMed] [Google Scholar]
  • 25.Abou Assi H., Rangadurai A. K., Shi H., Liu B., Clay M. C., Erharter K., Kreutz C., Holley C. L., Al-Hashimi H. M., 2′-O-methylation can increase the abundance and lifetime of alternative RNA conformational states. Nucleic Acids Res. 48, 12365–12379 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Natchiar S. K., Myasnikov A. G., Kratzat H., Hazemann I., Klaholz B. P., Visualization of chemical modifications in the human 80S ribosome structure. Nature 551, 472–477 (2017). [DOI] [PubMed] [Google Scholar]
  • 27.Chen L., Zhang L.-S., Ye C., Zhou H., Liu B., Gao B., Deng Z., Zhao C., He C., Dickinson B. C., Nm-Mut-seq: A base-resolution quantitative method for mapping transcriptome-wide 2′-O-methylation. Cell Res. 33, 727–730 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Li Y., Yi Y., Gao X., Wang X., Zhao D., Wang R., Zhang L. S., Gao B., Zhang Y., Zhang L., Cao Q., Chen K., 2′-O-methylation at internal sites on mRNA promotes mRNA stability. Mol. Cell 84, 2320–2336.e2326 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Elman J. S., Ni T. K., Mengwasser K. E., Jin D., Wronski A., Elledge S. J., Kuperwasser C., Identification of FUBP1 as a long tail cancer driver and widespread regulator of tumor suppressor and oncogene alternative splicing. Cell Rep. 28, 3435–3449.e5 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Jiang P., Huang M., Qi W., Wang F., Yang T., Gao T., Luo C., Deng J., Yang Z., Zhou T., Zou Y., Gao G., Yang X., FUBP1 promotes neuroblastoma proliferation via enhancing glycolysis-a new possible marker of malignancy for neuroblastoma. J. Exp. Clin. Cancer Res. 38, 400 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Jacob A. G., Singh R. K., Mohammad F., Bebee T. W., Chandler D. S., The splicing factor FUBP1 is required for the efficient splicing of oncogene MDM2 pre-mRNA. J. Biol. Chem. 289, 17350–17364 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Ebersberger S., Hipp C., Mulorz M. M., Buchbender A., Hubrich D., Kang H. S., Martínez-Lumbreras S., Kristofori P., Sutandy F. X. R., Llacsahuanga Allcca L., Schönfeld J., Bakisoglu C., Busch A., Hänel H., Tretow K., Welzel M., Di Liddo A., Möckel M. M., Zarnack K., Ebersberger I., Legewie S., Luck K., Sattler M., König J., FUBP1 is a general splicing factor facilitating 3′ splice site recognition and splicing of long introns. Mol. Cell 83, 2653–2672.e15 (2023). [DOI] [PubMed] [Google Scholar]
  • 33.Huang H., Weng H., Sun W., Qin X., Shi H., Wu H., Zhao B. S., Mesquita A., Liu C., Yuan C. L., Hu Y.-C., Hüttelmaier S., Skibbe J. R., Su R., Deng X., Dong L., Sun M., Li C., Nachtergaele S., Wang Y., Hu C., Ferchen K., Greis K. D., Jiang X., Wei M., Qu L., Guan J.-L., He C., Yang J., Chen J., Recognition of RNA N6-methyladenosine by IGF2BP proteins enhances mRNA stability and translation. Nat. Cell Biol. 20, 285–295 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Liu C., Dou X., Zhao Y., Zhang L., Zhang L., Dai Q., Liu J., Wu T., Xiao Y., He C., IGF2BP3 promotes mRNA degradation through internal m7G modification. Nat. Commun. 15, 7421 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Sauvé F., McBroom L. D., Gallant J., Moraitis A. N., Labrie F., Giguère V., CIA, a novel estrogen receptor coactivator with a bifunctional nuclear receptor interacting determinant. Mol. Cell. Biol. 21, 343–353 (2001). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Oksuz O., Henninger J. E., Warneford-Thomson R., Zheng M. M., Erb H., Vancura A., Overholt K. J., Hawken S. W., Banani S. F., Lauman R., Reich L. N., Robertson A. L., Hannett N. M., Lee T. I., Zon L. I., Bonasio R., Young R. A., Transcription factors interact with RNA to regulate genes. Mol. Cell 83, 2449–2463.e13 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.The ENCODE Project Consortium , An integrated encyclopedia of DNA elements in the human genome. Nature 489, 57–74 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Luo Y., Hitz B. C., Gabdank I., Hilton J. A., Kagda M. S., Lam B., Myers Z., Sud P., Jou J., Lin K., Baymuradov U. K., Graham K., Litton C., Miyasato S. R., Strattan J. S., Jolanki O., Lee J. W., Tanaka F. Y., Adenekan P., O'Neill E., Cherry J. M., New developments on the encyclopedia of DNA elements (ENCODE) data portal. Nucleic Acids Res. 48, D882–D889 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Piechotta M., Wyler E., Ohler U., Landthaler M., Dieterich C., JACUSA: Site-specific identification of RNA editing events from replicate sequencing data. BMC Bioinformatics 18, 7 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Wang Z., Burge C. B., Splicing regulation: From a parts list of regulatory elements to an integrated splicing code. RNA 14, 802–813 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Sharma S., Kohlstaedt L. A., Damianov A., Rio D. C., Black D. L., Polypyrimidine tract binding protein controls the transition from exon definition to an intron defined spliceosome. Nat. Struct. Mol. Biol. 15, 183–191 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Liu B., Wu T., Miao B. A., Ji F., Liu S., Wang P., Zhao Y., Zhong Y., Sundaram A., Zeng T. B., Majcherska-Agrawal M., Keenan R. J., Pan T., He C., snoRNA-facilitated protein secretion revealed by transcriptome-wide snoRNA target identification. Cell 188, 465–483.e22 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Zhao X., Yu Y.-T., Targeted pre-mRNA modification for gene silencing and regulation. Nat. Methods 5, 95–100 (2008). [DOI] [PubMed] [Google Scholar]
  • 44.Ge J., Liu H., Yu Y. T., Regulation of pre-mRNA splicing in Xenopus oocytes by targeted 2′-O-methylation. RNA 16, 1078–1085 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Edupuganti R. R., Geiger S., Lindeboom R. G. H., Shi H., Hsu P. J., Lu Z., Wang S. Y., Baltissen M. P. A., Jansen P., Rossa M., Müller M., Stunnenberg H. G., He C., Carell T., Vermeulen M., N6-methyladenosine (m6A) recruits and repels proteins to regulate mRNA homeostasis. Nat. Struct. Mol. Biol. 24, 870–878 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Shi H., Wang X., Lu Z., Zhao B. S., Ma H., Hsu P. J., Liu C., He C., YTHDF3 facilitates translation and decay of N6-methyladenosine-modified RNA. Cell Res. 27, 315–328 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Martin M., Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet J 17, 3 (2011). [Google Scholar]
  • 48.Kim D., Paggi J. M., Park C., Bennett C., Salzberg S. L., Graph-based genome alignment and genotyping with HISAT2 and HISAT-genotype. Nat. Biotechnol. 37, 907–915 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Sievers C., Schlumpf T., Sawarkar R., Comoglio F., Paro R., Mixture models and wavelet transforms reveal high confidence RNA-protein interaction sites in MOV10 PAR-CLIP data. Nucleic Acids Res. 40, e160 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Comoglio F., Sievers C., Paro R., Sensitive and highly resolved identification of RNA-protein interaction sites in PAR-CLIP data. BMC Bioinformatics 16, 32 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Heinz S., Benner C., Spann N., Bertolino E., Lin Y. C., Laslo P., Cheng J. X., Murre C., Singh H., Glass C. K., Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol. Cell 38, 576–589 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Ramírez F., Ryan D. P., Grüning B., Bhardwaj V., Kilpert F., Richter A. S., Heyne S., Dündar F., Manke T., deepTools2: A next generation web server for deep-sequencing data analysis. Nucleic Acids Res. 44, W160–W165 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.He P. C., Wei J., Dou X., Harada B. T., Zhang Z., Ge R., Liu C., Zhang L.-S., Yu X., Wang S., Lyu R., Zou Z., Chen M., He C., Exon architecture controls mRNA m6A suppression and gene expression. Science 379, 677–682 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Uren P. J., Bahrami-Samani E., Burns S. C., Qiao M., Karginov F. V., Hodges E., Hannon G. J., Sanford J. R., Penalva L. O., Smith A. D., Site identification in high-throughput RNA-protein interaction data. Bioinformatics 28, 3013–3020 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Stark R., Brown G., DiffBind: Differential binding analysis of ChIP-seq peak data. In R package version 100, (2011). [Google Scholar]
  • 56.Anders S., Pyl P. T., Huber W., HTSeq—A Python framework to work with high-throughput sequencing data. Bioinformatics 31, 166–169 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Love M. I., Huber W., Anders S., Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Shen S., Park J. W., Lu Z. X., Lin L., Henry M. D., Wu Y. N., Zhou Q., Xing Y., rMATS: Robust and flexible detection of differential alternative splicing from replicate RNA-seq data. Proc. Natl. Acad. Sci. U.S.A. 111, E5593–E5601 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.B. Bushnell. (United States, 2014).
  • 60.Kim D., Pertea G., Trapnell C., Pimentel H., Kelley R., Salzberg S. L., TopHat2: Accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol. 14, R36 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Figs. S1 to S5

Legends for tables S1 to S10

sciadv.ady3894_sm.pdf (2.2MB, pdf)

Tables S1 to S10

sciadv.ady3894_tables_s1_to_s10.zip (25.3MB, zip)

Articles from Science Advances are provided here courtesy of American Association for the Advancement of Science

RESOURCES