PDGFRα signaling regulates Srsf3 transcript binding to affect PI3K signaling and endosomal trafficking

20 Signaling through the platelet-derived growth factor receptor alpha (PDGFRa) plays a 21 critical role in craniofacial development, as mutations in PDGFRA are associated with cleft 22 lip/palate in humans and Pdgfra mutant mouse models display varying degrees of facial clefting. 23 Phosphatidylinositol 3-kinase (PI3K)/Akt is the primary effector of PDGFRa signaling during 24 skeletal development in the mouse. We previously demonstrated that Akt phosphorylates the 25 RNA-binding protein serine/arginine-rich splicing factor 3 (Srsf3) downstream of PI3K-mediated 26 PDGFRa signaling in mouse embryonic palatal mesenchyme (MEPM) cells, leading to its 27 nuclear translocation. We further showed that ablation of Srsf3 in the murine neural crest 28 lineage results in severe midline facial clefting, due to defects in proliferation and survival of 29 cranial neural crest cells, and widespread alternative RNA splicing (AS) changes. Here, we 30 sought to determine the molecular mechanisms by which Srsf3 activity is regulated downstream 31 of PDGFRa signaling to control AS of transcripts necessary for craniofacial development. We 32 demonstrated via enhanced UV-crosslinking and immunoprecipitation (eCLIP) of MEPM cells 33 that PDGF-AA stimulation leads to preferential binding of Srsf3 to exons and loss of binding to 34 canonical Srsf3 CA-rich motifs. Through the analysis of complementary RNA-seq data, we 35 showed that Srsf3 activity results in the preferential inclusion of exons with increased GC 36 content and lower intron to exon length ratio. Moreover, we found that the subset of transcripts 37 that are bound by Srsf3 and undergo AS upon PDGFRa signaling commonly encode regulators 38 of PI3K signaling and early endosomal trafficking. Functional validation studies further 39 confirmed that Srsf3 activity downstream of PDGFRa signaling leads to retention of the receptor 40 in early endosomes and increases in downstream PI3K-mediated Akt signaling. Taken together, 41 our findings reveal that growth factor-mediated phosphorylation of an RNA-binding protein 42 underlies gene expression regulation necessary for mammalian craniofacial development. 43 44 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 3

45 Craniofacial development is a complex morphogenetic process that requires a precise 46 interplay of multiple cell and tissue types to generate the frontonasal skeleton. Disruption of this 47 process can result in some of the most common birth defects in humans, such as cleft lip and 48 palate (Mai et al., 2019). Signaling through the platelet derived growth factor receptor alpha 49 (PDGFRa) receptor tyrosine kinase (RTK) is essential for human craniofacial development. 50 Heterozygous missense mutations in the coding region of PDGFRA that alter amino acids within 51 the extracellular, transmembrane or cytoplasmic domains of the receptor, in addition to single 52 base-pair substitutions in the 3’ untranslated region (3’ UTR), are associated with nonsyndromic 53 cleft palate (Rattanasopha et al., 2012). Further, single-nucleotide polymorphisms that repress 54 transcriptional activity of the promoter upstream of PDGFC, which encodes one of two PDGFRa 55 ligands, are associated with cleft lip and palate (Choi et al., 2009). This role of PDGFRa 56 signaling in craniofacial development is conserved in mice, as Pdgfra mutant mouse models 57 display a variety of defects that range from cleft palate to complete facial clefting (Klinghoffer et 58 al., 2002; Soriano, 1997; Tallquist & Soriano, 2003; Fantauzzo & Soriano, 2014; He & Soriano, 59 2013). These phenotypes are recapitulated in embryos lacking both Pdgfa and Pdgfc (Ding et 60 al., 2004). Phosphatidylinositol 3-kinase (PI3K) is the primary effector of PDGFRa signaling 61 during skeletal development in the mouse (Klinghoffer et al., 2002). Following activation, PI3K 62 increases phosphatidylinositol-3,4,5-trisphosphate (PIP3) levels at the cell membrane, leading to 63 the recruitment and subsequent phosphorylation of the serine/threonine kinase Akt. Akt 64 subsequently dissociates from the membrane to phosphorylate an array of target proteins that 65 are involved in wide-ranging cellular processes (Manning & Cantley, 2007). We previously 66 identified proteins phosphorylated by Akt downstream of PI3K-mediated PDGFRa signaling in 67 primary mouse embryonic mesenchyme (MEPM) cells (Fantauzzo & Soriano, 2014). Gene 68 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 4 ontology analysis revealed that 25% of the 56 proteins were involved in RNA processing, with a 69 particular enrichment for RNA splicing (Fantauzzo & Soriano, 2014). 70 Alternative RNA splicing (AS) is a process by which different combinations of exons from 71 the same gene are incorporated into mature RNA transcripts, thereby contributing to gene 72 expression regulation and enhancing the diversity of protein isoforms (Licatalosi & Darnell, 73 2010). AS occurs in approximately 95% of multi-exon human genes, frequently in a tissue-74 specific manner (Pan et al., 2008; Wang et al., 2008). Dysregulation of AS causes a number of 75 diseases, due to mutations in precursor RNA sequences, mutations in core components of the 76 spliceosome complex and/or mutations in auxiliary RNA-binding proteins (RBPs) (Scotti & 77 Swanson, 2016). These trans-acting auxiliary RBPs bind to specific sequence and/or structural 78 motifs in a target RNA via one or more RNA-binding domains to promote or inhibit exon 79 inclusion (Fu & Ares, 2014; Licatalosi & Darnell, 2010). The phenotypes resulting from global 80 and/or tissue-specific knockout of multiple RBPs have established that RBP-mediated AS is an 81 essential process during mouse craniofacial development (Bebee et al., 2015; Cibi et al., 2019; 82 Dennison et al., 2021; Forman et al., 2021; Lee et al., 2020). We previously demonstrated that 83 ablation of Srsf3 in the murine neural crest lineage results in severe midline facial clefting and 84 facial bone hypoplasia, due to defects in proliferation and survival of cranial neural crest cells, 85 and widespread AS changes (Dennison et al., 2021). 86 Srsf3 belongs to the serine/arginine-rich (SR) protein family of splicing factors that 87 generally promote exon inclusion by binding to exonic and intronic splicing enhancers and by 88 recruiting spliceosome components to the 5’ and 3’ splice sites (Fu & Ares, 2014; Licatalosi & 89 Darnell, 2010). Srsf3 specifically was shown to bind pyrimidine-rich motifs, with a preference for 90 exonic regions (Akerman et al., 2009; Änkö et al., 2012). Srsf3 is phosphorylated downstream of 91 PDGF and EGF stimulation and all-trans retinoic acid treatment, and by the kinases Akt and 92 SRPK2 (Bavelloni et al., 2014; Dennison et al., 2021; Fantauzzo & Soriano, 2014; Long et al., 93 2019; Zhou et al., 2012). Phosphorylation of the Akt consensus sites within the C-terminal 94 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 5 arginine/serine-rich (RS) domain of Srsf3 drives its translocation to the nucleus, where AS takes 95 place (Bavelloni et al., 2014; Dennison et al., 2021; Long et al., 2019). Moreover, 96 phosphorylation of the RS domain can alter the ability of SR proteins to interact with other 97 proteins, such as the U1 small nuclear ribonucleoprotein (snRNP) component of the 98 spliceosome and additional RBPs, and affect the ability of SR proteins to bind RNA (Huang et 99 al., 2004; Shen & Green, 2006; Shin et al., 2004; Xiao & Manley, 1997). However, the molecular 100 mechanisms by which Srsf3 activity is regulated downstream of specific signaling inputs in a 101 context-specific manner to regulate RNA binding and/or sequence specificity remain 102 undetermined. 103 Here, we identified changes in Srsf3-dependent AS and Srsf3 transcript binding in the 104 absence or presence of PDGF-AA ligand in MEPM cells. RNA-sequencing (RNA-seq) analysis 105 revealed that Srsf3 activity and PDGFRa signaling have more pronounced effects on AS than 106 gene expression, as well as a significant dependence in regulating AS. Using enhanced UV-107 crosslinking and immunoprecipitation (eCLIP), we found a shift from intronic to exonic Srsf3 108 binding and loss of CA-rich sites upon PDGF-AA stimulation. Further, we demonstrated that the 109 subset of transcripts that are bound by Srsf3 and undergo AS upon PDGFRa signaling 110 commonly encode regulators of PI3K signaling and early endosomal trafficking, ultimately 111 serving as a feedback mechanism to affect trafficking of the receptors. Combined, our findings 112 provide significant insight into the mechanisms underlying RBP-mediated gene expression 113 regulation in response to growth factor stimulation within the embryonic mesenchyme. 114 115

116 PDGFRa signaling for one hour minimally affects gene expression 117 To determine the Srsf3-dependent changes in gene expression and AS downstream of 118 PDGFRa signaling, we stably integrated a scramble shRNA (scramble) or an shRNA targeting 119 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 6 the 3’ UTR of Srsf3 (shSrsf3) into immortalized MEPM (iMEPM) cells via lentiviral transduction 120 (Fig. 1A). Western blotting revealed a 66% decrease in Srsf3 protein levels in the shSrsf3 cell 121 line (Fig. 1B). We previously demonstrated that phosphorylated Srsf3 levels peaked in the 122 nucleus of iMEPM cells following 1 hour of PDGF-AA ligand treatment (Dennison et al., 2021). 123 As such, cells were left unstimulated (-PDGF-AA) or treated with 10 ng/mL PDGF-AA for 1 hour 124 (+PDGF-AA). RNA was isolated and sequenced from three biological replicates across each of 125 the four conditions (Fig. 1A; Table S1). We first examined Srsf3-dependent differentially-126 expressed (DE) genes by comparing scramble versus shSrsf3 samples across the same PDGF-127 AA stimulation condition (-PDGF-AA or +PDGF-AA). We detected 827 DE genes in the absence 128 of ligand treatment and 802 DE genes upon ligand stimulation (Fig. 1C, Table S2). There was 129 high overlap (1,042 out of 1,629; 64.0%) of DE genes across ligand treatment conditions (Fig. 130 1C,D; Fig. S1A). Of these 521 shared DE genes, 514 (98.7%) had the same directionality, 131 including 273 (52%) with shared increases in expression in the shSrsf3 samples and 241 (46%) 132 with shared decreases in expression in the shSrsf3 samples (Fig. 1C,D). These findings 133 demonstrate that expression of a set of genes (521) depends on Srsf3 activity independent of 134 PDGFRa signaling, while a similarly sized set of genes (587) is differentially expressed in 135 response to both Srsf3 activity and PDGFRa signaling (Fig. 1C,D). Gene ontology (GO) 136 analysis of the Srsf3-dependent DE genes revealed that the most significant terms for biological 137 process commonly involved regulation of osteoblast differentiation, calcium-dependent cell-cell 138 adhesion, regulation of cell migration and canonical Wnt signaling, while only a handful of 139 significant terms for molecular function were detected in unstimulated cells, relating to cation 140 channel activity and phosphatase activity (Fig. S2A,B). 141 We next examined PDGF-AA-dependent DE genes by comparing -PDGF-AA versus 142 +PDGF-AA across the same Srsf3 condition (scramble or shSrsf3). We detected only 37 DE 143 genes in scramble cells and 14 DE genes in shSrsf3 cells (Fig. 1E, Table S2). There was low 144 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 7 overlap (8 out of 51; 15.7%) of DE genes across Srsf3 conditions (Fig. 1E,F; Fig. S1B). Of these 145 4 shared DE genes, 3 (75%) had shared increases in expression upon PDGF-AA stimulation 146 (Fig. 1E,F). These findings demonstrate that, unlike Srsf3 activity, PDGFRa signaling minimally 147 affects gene expression at one hour of ligand stimulation, consistent with our previous findings 148 in mouse embryonic facial mesenchyme (Dennison et al., 2021). Further, these results show 149 that expression of a set of genes (4) depends on PDGFRa signaling independent of Srsf3 150 activity, while a larger set of genes (43) is differentially expressed in response to both PDGFRa 151 signaling and Srsf3 activity (Fig. 1E,F). GO analysis of the PDGF-AA-dependent DE genes 152 revealed significant terms for biological process in the scramble cells relating to cell migration, 153 response to growth factor stimulation and regulation of transcription (Fig. S2A). The most 154 significant terms for molecular function commonly involved DNA binding (Fig. S2B). 155 156 PDGFRa signaling for one hour has a more pronounced effect on alternative RNA splicing 157 We previously demonstrated that AS is an important mechanism of gene expression 158 regulation downstream of PI3K/Akt-mediated PDGFRa signaling in the murine mid-gestation 159 palatal shelves (Dennison et al., 2021). Accordingly, we next assessed AS in our same RNA-160 seq dataset. In examining Srsf3-dependent alternatively-spliced transcripts, we detected 1,354 161 differential AS events in the absence of ligand treatment and 1,071 differential AS events upon 162 ligand stimulation (Fig. 2A). When filtered to include events detected in at least 10 reads in 163 either condition, we obtained a list of 1,113 differential AS events in the absence of ligand 164 treatment and 795 differential AS events upon ligand stimulation (Fig. 2B, Tables S3 and S4). 165 There was low overlap (406 out of 1,908; 21.3%) of alternatively-spliced transcripts across 166 ligand treatment conditions. Of these 203 shared alternatively-spliced transcripts, 100% had the 167 same directionality, including 81 (40%) with shared negative changes in percent spliced in (PSI) 168 (exon included more often in shSrsf3 samples) and 122 (60%) with shared positive changes in 169 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 8 PSI (exon skipped more often in shSrsf3 samples) (Fig. 2A,B). We confirmed the differential AS 170 of two transcripts, Arhgap12 and Cep55, between scramble and shSrsf3 samples by qPCR 171 using primers within constitutively-expressed exons flanking the alternatively-spliced exon (Fig. 172 S3A,B). These findings demonstrate that AS of a set of transcripts (203) depends on Srsf3 173 activity independent of PDGFRa signaling, while a much larger set of transcripts (1,502) is 174 alternatively spliced in response to both Srsf3 activity and PDGFRa signaling (Fig. 2A,B). GO 175 analysis of these Srsf3-dependent alternatively-spliced transcripts revealed that the most 176 significant terms for biological process and molecular function commonly involved regulation of 177 RNA splicing, and RNA binding and cadherin binding, respectively (Fig. S4A,B). 178 In examining PDGF-AA-dependent alternatively-spliced transcripts, we detected 595 179 differential AS events in scramble cells and 398 differential AS events in shSrsf3 cells (Fig. 2C). 180 When filtered to include events detected in at least 10 reads in either condition, we obtained a 181 list of 375 differential AS events in scramble cells and 256 differential AS events in shSrsf3 cells 182 (Fig. 2D, Tables S5 and S6). There was extremely low overlap (18 out of 631; 2.85%) of 183 alternatively-spliced transcripts across Srsf3 conditions (Fig. 2C,D). Of these 9 shared 184 alternatively-spliced transcripts, 100% had the same directionality, including 5 (56%) with 185 shared negative changes in PSI (exon included more often in +PDGF-AA samples) and 4 (44%) 186 with shared positive changes in PSI (exon skipped more often in +PDGF-AA samples) (Fig. 187 2C,D). Taken together, these findings demonstrate that both Srsf3 activity and PDGFRa 188 signaling have more pronounced effects on AS than gene expression. While more transcripts 189 and genes are subject to these regulatory mechanisms upon loss of Srsf3 activity, the 190 magnitude of events is more greatly skewed towards AS upon PDGF-AA stimulation. Further, 191 these results show that AS of a set of transcripts (9) depends on PDGFRa signaling 192 independent of Srsf3 activity, while a much larger set of transcripts (613) is alternatively spliced 193 in response to both PDGFRa signaling and Srsf3 activity (Fig. 2C,D). When combined with the 194 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 9 data above (Fig. 2B), this points to a profound dependence between PDGFRa signaling and 195 Srsf3 in regulating AS of transcripts in the facial mesenchyme. GO analysis of these PDGF-AA-196 dependent alternatively-spliced transcripts demonstrated only a handful of significant terms for 197 molecular function in the scramble cells, all relating to protein kinase activity (Fig. S4B). 198 The vast majority of Srsf3-dependent (70.2-73.7%) and PDGF-AA dependent (62.1-199 65.1%) AS events involved skipped exons, with minimal contribution from retained introns, 200 mutually exclusive exons, alternative 5’ splice sites, or alternative 3’ splice sites (Fig. 2E). For 201 the Srsf3-dependent skipped exon events, there were more transcripts with +DPSI (exon 202 skipped more often in shSrsf3 samples) (44.8%) as opposed to -DPSI (exon included more 203 often in shSrsf3 samples) (25.4%) in the absence of PDGF-AA stimulation (Fig. 2E), consistent 204 with previous results that SR proteins tend to promote exon inclusion (Fu & Ares, 2014; 205 Licatalosi & Darnell, 2010). However, PDGF-AA ligand treatment led to an increase in the 206 percentage of transcripts with -DPSI (36.5%) (Fig. 2E), indicating that PDGFRa signaling 207 promotes exon skipping in the presence of Srsf3. Among the PDGF-AA-dependent skipped 208 exon events, there was a significant shift in transcripts with -DPSI (exon included more often in 209 +PDGF-AA samples) in the absence (41.0%) versus presence (19.5%) of Srsf3, and a 210 corresponding shift in transcripts with +DPSI (exon skipped more often in +PDGF-AA samples) 211 (21.1% and 45.6%, respectively) (Fig. 2E), again suggesting that PDGF-AA stimulation causes 212 increased exon skipping when Srsf3 is present. 213 214 Srsf3 exhibits differential transcript binding upon PDGFRa signaling 215 To determine direct binding targets of Srsf3 downstream of PDGFRa signaling, we 216 conducted eCLIP (Van Nostrand et al., 2016, 2017) of iMEPM cells that were left unstimulated (-217 PDGF-AA) or treated with 10 ng/mL PDGF-AA for 1 hour (+PDGF-AA) as above (Fig. 3A; Table 218 S7). Immunoprecipitation with a previously validated Srsf3 antibody (Dennison et al., 2021) 219 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 10 successfully enriched for Srsf3 in UV-crosslinked cells (Fig. 3B). We detected 6,555 total eCLIP 220 peaks in protein-coding genes in the -PDGF-AA samples and 8,584 total peaks in the +PDGF-221 AA samples (Table S8). Among the -PDGF-AA peaks, 3,727 (56.9%) were located in exons 222 (CDS) and 1,690 (25.8%) were located within introns, with the rest binding within 5’ UTRs (768, 223 11.7%) and 3’ UTRs (367, 5.60%) (Fig. 3C,D; Table S9). We observed substantial shifts in 224 Srsf3-bound regions upon PDGF-AA stimulation. Many more +PDGF-AA peaks were located 225 within exons (7,139, 83.2%), and less were located within introns (765, 8.91%), 5’ UTRs (389, 226 4.53%) and 3’ UTRs (287, 3.34%) (Fig. 3C,D; Table S9). Given that SR proteins are crucial for 227 exon definition and bind to exonic splicing enhancers to recruit and stabilize core splicing 228 machinery (Fu & Maniatis, 1990; Krainer et al., 1991; Zahler et al., 1993), we investigated Srsf3 229 binding around 5’ and 3’ splice sites in response to PDGFRa signaling. Consistent with the 230

above and previous findings (Änkö et al., 2012), Srsf3 binding was enriched in exonic 231 regions, as opposed to intronic regions, surrounding the splice sites (Fig. 3E). There was 232 greater mean coverage of Srsf3 peaks in the +PDGF-AA condition versus the -PDGF-AA 233 condition within 100 nucleotides upstream of the 5’ splice site and within 100 nucleotides 234 downstream of the 3’ splice site (Fig. 3E). Additionally, we detected decreased mean coverage 235 in the +PDGF-AA condition within 25 nucleotides downstream of the 5’ splice site boundary 236 (Fig. 3E). Taken together, these data show that PDGFRa signaling leads to increased binding of 237 Srsf3 to exons. Finally, we performed unbiased motif enrichment analysis of Srsf3 peaks in 238 unstimulated and PDGF-AA-treated samples, revealing that the mostly highly enriched motifs in 239 -PDGF-AA samples were CACACA and AAGAAG (Fig. 3F). Of note, these CA-rich motifs have 240 previously been identified as canonical Srsf3 motifs in a CLIP study utilizing a stably integrated 241 SRSF3-EGFP transgene (Änkö et al., 2012). However, in PDGF-AA-stimulated samples, the 242 most highly enriched motifs were GAAGCG, GAAGAA, and AGAAGA (Fig. 3G), suggesting that 243 PDGFRa signaling influences Srsf3 binding specificity. 244 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 11 245 Srsf3 and PDGFRa signaling are associated with differential GC content and length of 246 alternatively-spliced exons 247 We next probed our RNA-seq dataset to determine whether specific transcript features 248 were associated with Srsf3-dependent AS. When comparing significant AS events between 249 scramble and shSrsf3 samples in the absence of PDGF-AA stimulation, we found that included 250 exons had a significantly higher GC content (median value of 53.4%) than both skipped exons 251 (50.0%) and exons that were not differentially alternatively spliced (51.2%) when Srsf3 is 252 present (Fig. 4A, Table S10). Additionally, we observed that included exons had a significantly 253 lower ratio of downstream intron to exon GC content (0.856) than both skipped exons (0.901) 254 and exons that were not differentially alternatively spliced (0.888) in the presence of Srsf3 (Fig. 255 4B, Table S10). The same comparisons revealed that the ratio of upstream and downstream 256 intron to exon length was significantly decreased in included exons (median values of 12.4 and 257 10.6, respectively) as compared to both skipped exons (19.5 and 20.0) and exons that were not 258 differentially alternatively spliced (14.2 and 13.3) in the presence of Srsf3 (Fig. 4C,D, Table 259 S10). Taken together, our data demonstrate that exons which are included in the presence of 260 Srsf3 tend to have a higher GC content and lower intron to exon length ratio. 261 To determine whether PDGFRa signaling had an effect on the transcript features to 262 which Srsf3 bound, we subsequently examined our eCLIP dataset. We found that the exon GC 263 content was significantly increased in exons bound by Srsf3 in the absence of ligand treatment 264 (median value of 57.9%) as compared to unbound exons (51.5%) (Fig. 4E, Table S10). 265 However, exon GC content was similar between unbound exons and those bound by Srsf3 266 upon PDGF-AA stimulation (53.8%) (Fig. 4E, Table S10). These findings indicate that 267 PDGFRa signaling mediates binding of Srsf3 to exons with a lower GC content. 268 269 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 12 Transcripts bound by Srsf3 that undergo alternative splicing upon PDGFRa signaling encode 270 regulators of PI3K signaling 271 To determine which transcripts are directly bound by Srsf3 and subject to DE and/or AS, 272 we cross-referenced the eCLIP and RNA-seq datasets. We collated transcripts with an Srsf3 273 eCLIP peak that were uniquely detected in the -PDGF-AA or +PDGF-AA samples (2,660 274 transcripts across 3,388 peaks) (Tables S8 and S11). Similarly, we gathered differentially-275 expressed genes (596) or differentially alternatively-spliced transcripts (985) uniquely found in 276 one of the four treatment comparisons (Fig. 5A, Table S11). Only 32 (5.4%) of the DE genes 277 were directly bound by Srsf3, while 233 (23.7%) of the alternatively-spliced transcripts had an 278 Srsf3 eCLIP peak, with very little overlap (1 transcript) between all three categories (Fig. 5A). 279 We next correlated the eCLIP peaks with AS events across all four treatment 280 comparisons by identifying transcripts in which Srsf3 bound within an alternatively-spliced exon 281 or within 250 bp of the neighboring introns (Tables S12-S15). In agreement with our findings 282 above for the entire eCLIP dataset, Srsf3 exhibited differential binding in exons and surrounding 283 the 5’ and 3’ splice sites upon PDGF-AA stimulation (Fig. S5A,B). We performed an unbiased 284 motif enrichment analysis of Srsf3 peaks within the overlapping datasets, again revealing 285 different motifs between ligand treatment conditions and an enrichment of GA-rich motifs in the 286 +PDGF-AA samples (Fig. S5C,D). 287 To determine whether transcripts that are differentially bound by Srsf3 and undergo 288 differential AS downstream of PDGFRa signaling contribute to shared biological outputs, we 289 conducted GO analysis of the 149 unique transcripts from the overlapping datasets. The most 290 significant terms for biological process involved protein phosphorylation and deacetylation, and 291 RNA metabolism (Fig. 5B). Given that PI3K-mediated PDGFRα signaling is critical for 292 craniofacial development and regulates AS in this context (Dennison et al., 2021; Fantauzzo & 293 Soriano, 2014; Klinghoffer et al., 2002), we turned our focus toward GO terms associated with 294 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 13 this signaling pathway. We noted enrichment of PI3K-related GO terms (Fig. 5C), which 295 encompassed the genes Becn1, Wdr81 and Acap3 (Fig. 5D). Related to their roles in PI3K 296 signaling, each of these gene products has been shown to regulate membrane and/or endocytic 297 trafficking. Phosphatidylinositol 3-phosphate (PI(3)P) is a critical component of early endosomes 298 and is mainly generated by conversion of phosphatidylinositol (PI) by the class III PI3K complex, 299 which includes Beclin-1 (encoded by Becn1) (Wallroth & Haucke, 2018). WDR81 and Beclin-1 300 have been shown to interact, resulting in decreased endosomal PI(3)P synthesis, PI(3)P 301 turnover and early endosome conversion to late endosomes (Liu et al., 2016). Importantly, this 302 role of WDR81 contributes to RTK degradation (Rapiteanu et al., 2016). 303 Within our data, Srsf3 binding was increased in Becn1 exon 7 upon PDGF-AA 304 stimulation, at an enriched motif within the overlapping datasets, and we observed a 305 corresponding increase in retention of adjacent intron 7 (Fig. 5D,E). As Becn1 intron 7 contains 306 a premature termination codon (PTC), this event is predicted to result in nonsense-mediated 307 decay (NMD) in the absence of Srsf3 (Fig. 5E). Srsf3 binding was also increased in Wdr81 exon 308 8 in response to PDGF-AA treatment, and our analyses revealed a corresponding increase in 309 excision of adjacent exon 9 (Fig. 5D,E). Because Wdr81 exon 9 encodes two WD-repeat 310 domains, which are generally believed to form a b propellor structure required for protein 311 interactions (Li & Roberts, 2001), this AS event is predicted to result in a protein missing a 312 functional domain (Fig. 5E). These splicing patterns predict increased levels of Beclin-1 and 313 decreased levels of functional Wdr81 in the presence of Srsf3 and PDGF-AA stimulation, 314 resulting in augmented early endosome formation. 315 316 Srsf3 regulates early endosome size and phosphorylation of Akt downstream of PDGFRα 317 signaling 318 Given that Srsf3 differentially binds to transcripts that encode proteins involved in early 319 endosomal trafficking downstream of PDGFRa signaling, we first examined the formation of 320 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 14 Rab5-positive early endosomes (Zerial & McBride, 2001) in response to a time course of PDGF-321 AA ligand stimulation in scramble versus shSrsf3 cells. As expected, Rab5 puncta size 322 increased from 0 min (9.79 x 10-4 ± 1.13 x 10-4 arbitrary units; mean ± s.e.m.) to 15 min of ligand 323 stimulation (2.01 x 10-3 ± 7.20 x 10-4 arbitrary units) in scramble cells, and significantly so by 60 324 min (2.58 x 10-3 ± 9.20 x 10-4 arbitrary units) (Fig. 6A, C-C”, E-E”, G-G”). However, this increase 325 was not observed in the absence of Srsf3 (9.21 x 10-4 ± 1.61 x 10-4 arbitrary units at 60 min) 326 (Fig. 6A, D-D”, F-F”, H-H”), demonstrating that Srsf3-mediated PDGFRa signaling leads to 327 enlarged early endosomes. 328 We next examined colocalization of PDGFRa with Rab5, as an estimate of receptor 329 levels in the early endosome. Colocalization levels increased from 0 min (0.332 ± 0.0832 330 Pearson’s correlation coefficient (PCC); mean ± s.e.m.) to 15 min (0.429 ± 0.108 PCC) of 331 PDGF-AA treatment in scramble cells, then decreased to near baseline levels by 60 min (0.348 332 ± 0.0885 PCC) (Fig. 6B, C-C”, E-E”, G-G”) as a subset of PDGFRa homodimers are likely 333 trafficked to late endosomes (Rogers et al., 2022). Strikingly, shSrsf3 cells exhibited a 334 significant decrease in colocalization between PDGFRa and Rab5 by 60 min of ligand treatment 335 (0.186 ± 0.0102 PCC) (Fig. 6B, D-D”, F-F”, H-H”), indicating that Srsf3 activity downstream of 336 PDGFRa signaling results in retention of PDGFRa in early endosomes. 337 Finally, we previously demonstrated that rapid internalization of PDGFRa homodimers 338 following PDGF-AA ligand stimulation is critical for downstream AKT phosphorylation (Rogers et 339 al., 2022). As such, we examined phospho-Akt levels as a readout of PDGFRa activation in 340 early endosomes. While PDGF-AA treatment for 15 min induced a robust phospho-Akt 341 response in scramble cells (13.5 ± 7.39 relative induction, mean ± s.e.m.) this response was 342 muted in shSrsf3 cells at the same timepoint (5.73 ± 1.91) (Fig. 6I). These findings further 343 suggest that retention of PDGFRa in early endosomes leads to increases in downstream PI3K-344 mediated Akt signaling. Collectively, our data point to a feedback loop in which PI3K/Akt-345 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 15 mediated PDGFRa signaling results in the nuclear translocation of Srsf3 and the subsequent 346 AS of transcripts to decrease levels of proteins that promote PDGFRa trafficking out of early 347 endosomes (Fig. 6J). 348 349

350 In this study, we confirmed our prior in vivo results in the mouse embryonic facial 351 mesenchyme (Dennison et al., 2021) that PDGFRa signaling primary regulates gene expression 352 via AS. PDGF-AA-dependent differential gene expression was minimal following one hour of 353 ligand treatment and led to increased expression of immediate early genes Klf10, Egf3 and 354 Egr1, consistent with previous findings (Vasudevan et al., 2015) and in line with the enriched 355 GO terms of regulation of transcription and DNA binding. Alternatively, PDGF-AA-dependent 356 alternatively-spliced transcripts were enriched for protein kinase activity, consistent with our 357 prior AS findings upon disruption of PI3K-mediated PDGFRa signaling in the palatal shelf 358 mesenchyme (Dennison et al., 2021). Importantly, our results demonstrated a significant 359 dependence on the RBP Srsf3 for AS downstream of PDGFRa activation. In fact, we found that 360 79% of Srsf3-dependent and 97% of PDGF-AA-dependent alternatively-spliced transcripts were 361 responsive to both Srsf3 activity and PDGFRa signaling. As discussed above, Srsf3 is 362 phosphorylated downstream of multiple stimuli (Bavelloni et al., 2014; Dennison et al., 2021; 363 Fantauzzo & Soriano, 2014; Long et al., 2019; Zhou et al., 2012), and it is likely that these 364 additional inputs contributed to Srsf3-dependent AS that was non-responsive to PDGF-AA 365 treatment. Further, our previous mass spectrometry screen identified 11 additional RBPs 366 involved in AS that are phosphorylated by Akt downstream of PI3K-mediated PDGFRa signaling 367 in primary MEPM cells (Fantauzzo & Soriano, 2014), which may account for the small fraction of 368 PDGF-AA-dependent AS that was non-responsive to Srsf3 knockdown. However, our RNA-seq 369

together with the phenotypic overlap of embryos with neural crest-specific ablation of 370 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 16 Srsf3 and mutant mouse models of Pdgfra and/or its ligands (Andrae et al., 2016; Dennison et 371 al., 2021; Ding et al., 2004; Fantauzzo & Soriano, 2014; Fredriksson et al., 2012; Klinghoffer et 372 al., 2002; Soriano, 1997) make a compelling case for Srsf3 serving as a critical effector of 373 PDGFRa signaling in the facial mesenchyme. 374 Here, we observed that Srsf3-dependent skipped exon events were enriched for 375 transcripts with a +DPSI (exon skipped more often in shSrsf3 samples) in the absence of PDGF-376 AA stimulation, consistent with Srsf3 promoting exon inclusion. Alternatively, PDGF-AA ligand 377 treatment led to an increase in the percentage of Srsf3-dependent transcripts with a -DPSI 378 (exon included more often in shSrsf3 samples), suggesting that PDGFRa signaling causes 379 decreased exon inclusion in the presence of Srsf3. Interestingly, a recent paper found that 380 tethering Srsf3 downstream of the 5’ splice site or upstream of the 3’ splice site using MS2 stem 381 loops did not lead to AS of a splicing reporter (Schmok et al., 2024). However, the assay did not 382 test tethering within the exon and used a single minigene sequence context, and thus had the 383 potential to lead to false negative results (Schmok et al., 2024). Whether phosphorylation of 384 Srsf3 directly influences its binding to target RNAs or acts to modulate Srsf3 protein-protein 385 interactions which then contribute to differential RNA binding remains to be determined, though 386 findings from Schmok et al., 2024 may argue for the latter mechanism. Studies identifying 387 proteins that differentially interact with Srsf3 in response to PDGF-AA ligand stimulation are 388 ongoing and will shed light on these mechanisms. 389 This study represents the first endogenous CLIP analysis of Srsf3 in the absence of 390 protein tagging, and thus circumvents potential limitations with prior approaches in which 391 assayed RBPs were overexpressed and/or fused to another protein. Our eCLIP analyses 392 revealed several changes in Srsf3 transcript binding downstream of PDGF-AA stimulation, 393 including increased Srsf3 binding to exons and loss of Srsf3 binding to canonical CA-rich motifs. 394 A previous CLIP study using a stably integrated SRSF3-EGFP transgene in mouse P19 cells 395 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 17 determined that SRSF3 binding was enriched in exons, particularly within 100 nucleotides of the 396 5’ and, to a lesser extent, 3’ splice sites (Änkö et al., 2012), consistent with our results. This 397 same study identified a CA-rich canonical SRSF3 motif (Änkö et al., 2012). While such motifs 398 were identified here in the absence of PDGF-AA treatment, they were lost upon ligand 399 stimulation. Again, this shift could be due to loss of RNA binding owing to electrostatic repulsion 400 and/or changes in ribonucleoprotein composition and will be the subject of future studies. 401 Our findings additionally pointed to novel properties of exons whose inclusion is 402 dependent on Srsf3 in the absence of PDGFRa signaling. We demonstrated that these included 403 exons had a higher GC content, a lower ratio of downstream intron to exon GC content and a 404 decreased ratio of upstream and downstream intron to exon length. These findings are 405 consistent with previous results demonstrating that included exons tend to have higher GC 406 content than the flanking introns (Amit et al., 2012). Of note, PDGF-AA ligand stimulation 407 resulted in binding of Srsf3 to exons with a lower GC content, again suggesting that 408 phosphorylation of the RBP downstream of this signaling axis promotes exon skipping. 409 By cross-referencing our RNA-seq and eCLIP datasets, we showed that 24% of the 410 alternatively-spliced transcripts across our four treatment comparisons had an Srsf3 eCLIP 411 peak. As Srsf3 also has functions in the cytoplasm, such as RNA trafficking, translation and 412 degradation (Howard & Sanford, 2015), the additional eCLIP peaks may reflect alternate roles 413 for Srsf3 in RNA metabolism. Conversely, Srsf3-mediated AS may be delayed following 414 transcript binding in the short timeframe of our experimental design. However, the extent of 415 overlap that we observed is in line with previous studies correlating alternatively-spliced 416 transcripts upon knockdown of an RBP with endogenous eCLIP results for that same RBP, 417 including Rbfox2 (10%) (Moss et al., 2023) and LUC7L2 (18-26%) (Jourdain et al., 2021). The 418 degree to which our RNA-seq and eCLIP datasets overlapped here points to the robustness and 419 biological significance of our findings. 420 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 18 Our combined analyses demonstrated that Srsf3 binds and mediates the AS of 421 transcripts that encode proteins that regulate PI3K signaling and early endosomal trafficking 422 downstream of PDGFRa activation, including Becn1, Wdr81 and Acap3. Homozygous missense 423 mutations in WDR81 cause Cerebellar ataxia, impaired intellectual development, and 424 dysequilibrium syndrome 2 (OMIM 610185) in humans (Gulsuner et al., 2011), with some 425 patients exhibiting coarse facial features and strabismus (Garcias & Roth, 2007), pointing to a 426 critical role for this gene product in craniofacial development. Consistently, we found that related 427 GO terms, such as phosphatidylinositol phosphate binding and endosome to lysosome 428 transport, were significantly enriched among alternatively-spliced transcripts in murine 429 embryonic facial mesenchyme upon loss of PI3K binding to PDGFRa and/or knockdown of 430 Srsf3 (Dennison et al., 2021). These data further confirm that our iMEPM model system served 431 as a powerful platform to uncover mechanisms that are utilized during craniofacial development 432 in vivo. 433 Our subsequent in vitro validation studies showed that Srsf3-mediated PDGFRa 434 signaling results in enlarged early endosomes, retention of the receptor in these early 435 endosomes and increased downstream PI3K-mediated Akt signaling. Relatedly, we and others 436 have linked spatial organization with the propagation of PDGFRa signaling, such that rapid 437 internalization of the receptors into early endosomes or autophagy of the receptors are required 438 for maximal phosphorylation of AKT downstream of PDGFRa activation (Rogers et al., 2022; 439 Simpson et al., 2024). Together, these results indicate a feedback loop at play in the 440 craniofacial mesenchyme in which stimulation of PDGFRa homodimer signaling leads to Srsf3-441 dependent AS of transcripts, a subsequent increase in the levels of proteins that maintain the 442 receptor in early endosomes and a corresponding decrease in the levels of proteins that 443 promote trafficking of the receptor to late endosomes for eventual degradation. These findings 444 thus represent a novel mechanism by which PDGFRa activity is maintained and propagated 445 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 19 within the cell. Whether similar mechanisms exist downstream of alternate RTKs or contribute to 446 increases in the phosphorylation of effector proteins other than Akt remain to be determined. 447 Taken together, our findings significantly enhance our understanding of the molecular 448 mechanisms by which Srsf3 activity is regulated downstream of growth factor stimulation. 449 Interestingly, a recent study demonstrated that retention of another shuttling SR protein, Srsf1, 450 exclusively in the nucleus resulted in widespread ciliary defects in mice (Haward et al., 2021). 451 This finding indicates that dissecting nuclear from cytoplasmic functions of SR proteins can 452 provide powerful insight into the physiological relevance of each. Going forward, it will be critical 453 to explore the in vivo consequences of abrogating Akt-mediated phosphorylation of Srsf3 and 454 comparing the resulting phenotype to those of embryos with constitutive or conditional ablation 455 of Srsf3 in the neural crest lineage (Dennison et al., 2021; Jumaa et al., 1999). These 456 experiments are ongoing and should shed considerable light on the roles of RBP post-457 translational modifications during development. 458 459

460 Generation of scramble and Srsf3 shRNA iMEPM cell lines 461 Immortalized mouse embryonic palatal mesenchyme (iMEPM) cells were derived from a 462 male Cdkn2a-/- embryo as previously described (Fantauzzo & Soriano, 2017). iMEPM cells were 463 cultured in growth medium [Dulbecco’s modified Eagle’s medium (Gibco, Thermo Fisher 464 Scientific, Waltham, MA, USA) supplemented with 50 U/mL penicillin (Gibco), 50 µg/mL 465 streptomycin (Gibco) and 2 mM L-glutamine (Gibco) containing 10% fetal bovine serum (FBS) 466 (Hyclone Laboratories Inc., Logan, UT, USA)] and grown at 37°C in 5% carbon dioxide. iMEPM 467 cells were tested for mycoplasma contamination using the MycoAlert Mycoplasma Detection Kit 468 (Lonza Group Ltd, Basel, Switzerland). Packaged lentiviruses containing pLV[shRNA]-469 EGFP:T2A:Puro-U6>Scramble_shRNA (vectorID: VB010000-0009mxc) with sequence 470 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 20 CCTAAGGTTAAGTCGCCCTCGCTCGAGCGAGGGCGACTTAACCTTAGG or pLV[shRNA]-471 EGFP:T2A:Puro-U6>mSrsf3[shRNA#1] (vectorID: VB900060-7699yyh) with sequence 472 GAATGATAAAGCGGTGTTTACTCGAGTAAACACCGCTTTATCATTCC were purchased from 473 VectorBuilder (Chicago, IL, USA). Medium containing lentivirus for a multiplicity of infection of 474 10 for 200,000 cells with the addition of 10 ug/mL polybrene was added to iMEPM cells for 16 h, 475 and cells were subsequently grown in the presence of 4 ug/mL puromycin for 10 days. Cells 476 with the highest GFP expression (top 20%) were isolated on a Moflo XDP 100 cell sorter 477 (Beckman Coulter Inc., Brea, CA, USA) and expanded. Srsf3 expression in scramble and Srsf3 478 shRNA cell lines was confirmed by western blotting. Once the stable cell lines were established, 479 they were split at a ratio of 1:4 for maintenance. Scramble and Srsf3 shRNA cells were used for 480 experiments at passages 9-20. 481 482 Immunoprecipitation and western blotting 483 To induce PDGFRa signaling, cells at ~70% confluence were serum starved for 24 h in 484 starvation medium [Dulbecco’s modified Eagle’s medium (Gibco) supplemented with 50 U/mL 485 penicillin (Gibco), 50 µg/mL streptomycin (Gibco) and 2 mM L-glutamine (Gibco) containing 486 0.1% FBS (Hyclone Laboratories Inc.)] and stimulated with 10 ng/mL rat PDGF-AA ligand (R&D 487 Systems, Minneapolis, MN, USA) diluted from a 1.5 µg/mL working solution in 40 nM HCl 488 containing 0.1% BSA for the indicated length of time. When applicable, UV-crosslinking was 489 performed at 254 nm and 400 mJ/cm2 using a Vari-X-Link system (UVO3 Ltd, Cambridgeshire, 490 UK). For immunoprecipitation of Srsf3, cells were resuspended in ice-cold CLIP lysis buffer [50 491 mM Tris-HCl pH 7.4, 100 mM NaCl, 1% NP-40, 0.1% SDS, 0.5% sodium deoxycholate, 1x 492 complete Mini protease inhibitor cocktail (Roche, MilliporeSigma, Burlington, MA, USA), 1 mM 493 PMSF, 10 mM NaF, 1 mM Na3VO4, 25 mM β-glycerophosphate]. Cleared lysates were collected 494 by centrifugation at 18,000 g for 20 min at 4°C. Anti-Srsf3 antibody (10 µg/sample) (ab73891, 495 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 21 Abcam, Waltham, MA, USA) was added to protein A Dynabeads (125 µL/sample) (Thermo 496 Fisher Scientific) washed twice in ice-cold CLIP lysis buffer and incubated for 45 min at room 497 temperature. Cells lysates were incubated with antibody-conjugated Dynabeads or Dynabeads 498 M-280 sheep anti-rabbit IgG (Thermo Fisher Scientific) washed twice in ice-cold CLIP lysis 499 buffer overnight at 4°C. The following day, Dynabeads were washed twice each with ice-cold 500 high salt wash buffer [50 mM Tris-HCl pH 7.4, 1M NaCl, 1 mM EDTA, 1% NP-40, 0.1% SDS, 501 0.5% sodium deoxycholate] followed by ice-cold wash buffer [20 mM Tris-HCl pH 7.4, 10 mM 502 MgCl2, 0.2% Tween-20]. The precipitated proteins were eluted with 1x NuPAGE LDS buffer 503 (Thermo Fisher Scientific) containing 100 mM dithiothreitol, heated for 10 minutes at 70°C, and 504 separated by SDS-PAGE. For western blotting of whole-cell lysates, protein lysates were 505 generated by resuspending cells in ice-cold NP-40 lysis buffer (20 mM Tris-HCl pH 8, 150 mM 506 NaCl, 10% glycerol, 1% Nonidet P-40, 2 mM EDTA, 1x complete Mini protease inhibitor cocktail 507 (Roche), 1 mM PMSM, 10 mM NaF, 1 mM Na3VO4, 25 mM β-glycerophosphate) and collecting 508 cleared lysates by centrifugation at 13,400 g at 4°C for 20 min. Laemmli buffer containing 10% 509 β-mercaptoethanol was added to the lysates, which were heated for 5 min at 100°C. Proteins 510 were subsequently separated by SDS-PAGE. Western blot analysis was performed according to 511 standard protocols using horseradish peroxidase-conjugated secondary antibodies. Blots were 512 imaged using a ChemiDoc XRS+ (Bio-Rad Laboratories, Inc., Hercules, CA, USA) or a 513 ChemiDoc (Bio-Rad Laboratories, Inc.). The following primary antibodies were used for western 514 blotting: Srsf3 (1:1,000, ab73891, Abcam), Gapdh (1:50,000, 60004, Proteintech Group, Inc., 515 Rosemont, IL, USA), phospho-Akt (Ser473) (1:1,000, 9271, Cell Signaling Technology, Inc., 516 Danvers, MA, USA), Akt (1:1,000, 9272, Cell Signaling Technology, Inc.), horseradish 517 peroxidase-conjugated goat anti-mouse IgG (1:20,000, 115035003, Jackson ImmunoResearch 518 Inc., West Grove, PA, USA), horseradish peroxidase-conjugated goat anti-rabbit IgG (1:20,000, 519 111035003;, Jackson ImmunoResearch Inc.). Quantifications of signal intensity were performed 520 with ImageJ software (version 1.53t, National Institutes of Health, Bethesda, MD, USA). Relative 521 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 22 phospho-Akt levels were determined by normalizing to total Akt levels. When applicable, 522 statistical analyses were performed with Prism 10 (GraphPad Software Inc., San Diego, CA, 523 USA) using a two-tailed, ratio paired t-test within each cell line and a two-tailed, unpaired t-test 524 with Welch’s correction between each cell line. Immunoprecipitation and western blotting 525 experiments were performed across three independent experiments. 526 527 RNA-sequencing and related bioinformatics analyses 528 8 x 105 cells obtained from each of three independent biological replicates per treatment 529 were frozen on liquid nitrogen and stored at -80°C. Following thawing, total RNA was 530 simultaneously isolated from all samples using the RNeasy Mini Kit (Qiagen, Inc., Germantown, 531 MD, USA) according to the manufacturer’s instructions. RNA was forwarded to the University of 532 Colorado Cancer Center Genomics Shared Resource for quality control, library preparation, and 533 sequencing. RNA purity, quantity and integrity were assessed with a NanoDrop (Thermo Fisher 534 Scientific) and a 4200 TapeStation System (Agilent Technologies, Inc., Santa Clara, CA, USA) 535 prior to library preparation. Total RNA (200 ng) was used for input into the Universal Plus 536 mRNA-Seq kit with NuQuant (Tecan Group Ltd., Männedorf, Switzerland). Dual index, stranded 537 libraries were prepared and sequenced on a NovaSeq 6000 Sequencing System (Illumina, San 538 Diego, CA, USA) to an average depth of ~54 million read pairs (2x150 bp reads). 539 Raw sequencing reads were de-multiplexed using bcl2fastq (Illumina). Trimming, filtering 540 and adapter contamination removal was performed using BBDuk (from the BBmap v35.85 tool 541 suite) (Bushnell, 2015). For differential expression analysis, transcript abundance was quantified 542 using Salmon (v1.4.0) (Patro et al., 2017) and a decoy-aware transcriptome index prepared 543 using GENCODE (Frankish et al., 2019) GRCm39 M26. Gene level summaries were calculated 544 using tximport (Soneson et al., 2016) in R and differential expression was measured using 545 DESeq2 (v.1.32.0) (Love et al., 2014). Significant changes in gene-level expression are 546 reported for cases with adjusted P £ 0.05 and fold change |FC| ³ 2. Spearman correlation was 547 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 23 computed between conditions for differentially-expressed genes. For alternative splicing 548 analysis, raw FASTQ were trimmed to a uniform length of 125 bp. Reads were aligned to the 549 mouse genome (GRCm39 Gencode M26) using STAR (v.2.7.9a) (Dobin et al., 2013). Additional 550 parameters for STAR: --outFilterType BySJout --outFilterMismatchNmax 10 --551 outFilterMismatchNoverLmax 0.04 --alignEndsType EndToEnd --runThreadN 16 --552 alignSJDBoverhangMin 4 --alignIntronMax 300000 --alignSJoverhangMin 8 --alignIntronMin 20. 553 All splice junctions detected in at least 1 read from the first pass alignment were used in a 554 second pass alignment, per software documentation. Alternative splicing events were detected 555 using rMATS (v4.0.2, default parameters plus ‘—cstat 0.0001’) (Shen et al. 2014). Reads 556 mapping to the splice junction as well as those mapping to the exon body were used in 557 downstream analyses. Detected events were compared between treatment groups and 558 considered significant with false discovery rate (FDR) £ 0.05, a difference in percent spliced in 559 (|ΔPSI|) ³ 0.05 and event detection in at least 10 reads in either condition. Raw read pairs, 560 trimmed read pairs for Salmon input, Salmon mapping rate per sample, trimmed read pairs (125 561 bp) for STAR input and STAR unique mapping rate can be found in Table S1. Gene ontology 562 analysis was performed with various libraries from the Enrichr gene list enrichment analysis tool 563 (Chen et al., 2013; Kuleshov et al., 2016) and terms with P < 0.05 were considered significant. 564 565 qPCR 566 Total RNA was isolated using the RNeasy mini kit (Qiagen, Germantown, MD, USA) 567 according to the manufacturer’s instructions. First-strand cDNA was synthesized using a ratio of 568 2:1 random primers:oligo (dT) primer and SuperScript II RT (Invitrogen, Thermo Fisher 569 Scientific) according to the manufacturer’s instructions. All reactions were performed with 1× 570 ThermoPol buffer [0.02 M Tris (pH 8.8), 0.01 M KCl, 0.01 M (NH4)2SO4, 2 mM MgSO4 and 0.1% 571 Triton X-100], 200 μM dNTPs, 200 nM primers (Integrated DNA Technologies, Inc., Coralville, 572 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 24 IA, USA), 0.6 U Taq polymerase and 1 μg cDNA in a 25 μL reaction volume. The primers used 573 can be found in Table S16. The following PCR protocol was used for Arhgap12: step1, 3 min at 574 94°C; step 2, 30 s at 94°C; step 3, 30 s at 47°C; step 4, 30 s at 72°C; repeat steps 2-4 for 34 575 cycles; and step 5, 5 min at 72°C. The following PCR protocol was used for Cep55: step1, 3 min 576 at 94°C; step 2, 30 s at 94°C; step 3, 30 s at 48°C; step 4, 30 s at 72°C; repeat steps 2-4 for 34 577 cycles; and step 5, 5 min at 72°C. Two-thirds of total PCR products were electrophoresed on a 578 2% agarose/TBE gel containing ethidium bromide and photographed on an Aplegen Omega 579 Fluor Gel Documentation System (Aplegen Inc., Pleasanton, CA, USA). Quantifications of band 580 intensity were performed with ImageJ software (version 1.53t, National Institutes of Health). The 581 PSI was calculated independently for each sample as the percentage of the larger isoform 582 divided by the total abundance of all isoforms within the given gel lane. Statistical analyses were 583 performed with Prism 10 (GraphPad Software) using a two-tailed, unpaired t-test with Welch’s 584 correction. qPCR reactions were performed using three biological replicates. 585 586 Enhanced UV-crosslinking and immunoprecipitation and related bioinformatics analyses 587 Experiments were performed as previously described in biological duplicates (Van 588 Nostrand et al., 2016, 2017). Briefly, 2 million cells per treatment were serum starved and 589 treated with 10 ng/mL PDGF-AA as described above. Cells were subsequently UV-crosslinked 590 at 254 nm and 400 mJ/cm2, scraped in 1x phosphate buffered saline (PBS) and transferred to 591 1.5 mL Eppendorf tubes, at which point excess PBS was removed and cells were frozen on 592 liquid nitrogen and stored at -80°C. Following thawing, cells were lysed in ice-cold CLIP lysis 593 buffer, sonicated by BioRuptor (Diagenode, Denville, NJ, USA) and treated with RNase I 594 (Thermo Fisher Scientific). 2% of lysates were set aside as size-matched input samples. Srsf3-595 RNA complexes were immunoprecipitated with anti-Srsf3 antibody (10 µg per sample) 596 (ab73891, Abcam) conjugated to protein A Dynabeads (Thermo Fisher Scientific). IP samples 597 were washed and dephosphorylated with FastAP (New England Biolabs, Ipswich, MA, USA) 598 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 25 and T4 PNK (New England Biolabs). IP samples underwent on-bead ligation of barcoded RNA 599 adapters (/5phos/rArGrArUrCrGrGrArArGrArGrCrGrUrCrGrUrG/3SpC3/) to the 3’ end using T4 600 RNA ligase (New England Biolabs). Following elution, protein-RNA complexes were run on 4-601 12% Bis-Tris 1.5 mm gels (Thermo Fisher Scientific) and transferred onto nitrocellulose 602 membranes. The 20-75 kDa region was excised and digested with proteinase K (New England 603 Biolabs). RNA was isolated with acid phenol/chloroform/isoamyl alcohol (pH 6.5) (Thermo 604 Fisher Scientific), reverse transcribed with Superscript III (Thermo Fisher Scientific) and treated 605 with ExoSAP-IT (Affymetrix, Thermo Fisher Scientific) to remove excess primers and 606 unincorporated nucleotides. Samples underwent 3’ ligation of barcoded DNA adapters 607 (/5Phos/NNNNNNNNNNAGATCGGAAGAGCACACGTCTG/3SpC3/), clean-up with Dynabeads 608 MyOne Silane (Thermo Fisher Scientific) and qPCR to determine the appropriate number of 609 PCR cycles. Libraries were then amplified with Q5 PCR mix (New England Biolabs) for a total of 610 16-25 cycles. Libraries were forwarded to the University of Colorado Cancer Center Genomics 611 Shared Resource for quality control and sequencing. Sample integrity was assessed with a 612 D1000 ScreenTape System (Agilent Technologies, Inc.) and sequenced on a NovaSeq 6000 613 Sequencing System (Illumina) to an average depth of ~20 million read pairs (2x150 bp reads). 614 Raw sequencing reads were de-multiplexed using bcl2fastq (Illumina). Adapters were 615 trimmed using cutadapt (v.1.18) (Martin, 2011). Trimmed reads were quality filtered and 616 collapsed using a combination of FASTX-Toolkit (v.0.0.14) 617 (http://hannonlab.cshl.edu/fastx_toolkit), seqtk (v.1.3-r106) (https://github.com/lh3/seqtk) and 618 custom scripts. After collapsing the reads, unique molecular identifiers were removed using 619 seqtk. STAR index for repetitive elements was created using repetitive sequences from 620 msRepDB (Liao et al., 2022). Reads ≥ 18 nt were mapped to the repetitive elements using 621 STAR (v.2.7.9a) (Dobin et al. 2013). Reads unmapped to the repetitive elements were mapped 622 to the mouse genome (GRCm39 Gencode M26) using STAR (v.2.7.9a) with parameters 623 alignEndsType: EndtoEnd and outFilterMismatchNoverReadLmax: 0.04. Peaks were called 624 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 26 using omniCLIP (v.0.20) (Drewe-Boss et al., 2018) with the foreground penalty (--fg_pen) 625 parameter set to 5. Peaks were annotated and motif analyses performed using RCAS (v.1.19.0) 626 (Uyar et al., 2017) and custom R script. For visualization purposes, bigWig files were created 627 from bam files using deepTools (v.3.5.5) (Ramírez et al., 2016). Peaks were visualized in 628 Integrative Genomics Viewer (v.2.13.0) (Robinson et al., 2011). Intron and exon features were 629 calculated using Matt (v.1.3.1) (Gohr & Irimia, 2019), and statistical analyses were performed 630 using a Mann-Whitney U test. For overlap of eCLIP peaks and alternative splicing events, peak 631 coordinates were taken from omniCLIP bed files and alternative splicing coordinates were taken 632 from rMATS output. Overlapping coordinates from alternative splicing events were defined 633 following the rMAPS default values (Park et al., 2016). Overlap was calculated using valr 634 (v.0.6.4) (Riemondy et al., 2017) and custom R scripts. Raw read pairs, trimmed read pairs, 635 collapsed reads, reads after removing repetitive elements, mapped reads, peaks and annotated 636 peaks can be found in Table S7. Gene ontology analysis was performed with various libraries 637 from the Enrichr gene list enrichment analysis tool (Chen et al., 2013; Kuleshov et al., 2016) and 638 terms with P < 0.05 were considered significant. 639 640 Immunofluorescence analysis 641 Cells were seeded onto glass coverslips at ~40% confluency per 24-well plate well in 642 iMEPM growth medium. After 24 h, cells were serum starved and treated with 10 ng/mL PDGF-643 AA as described above. Cells were fixed in 4% paraformaldehyde (PFA) in PBS with 0.1% 644 Triton X-100 for 10 min and washed in PBS. Cells were blocked for 1 h in 5% normal donkey 645 serum (Jackson ImmunoResearch Inc.) in PBS and incubated overnight at 4°C in primary 646 antibody diluted in 1% normal donkey serum in PBS. After washing in PBS, cells were 647 incubated in Alexa Fluor 488-conjugated donkey anti-rabbit secondary antibody (1:1,000; 648 A21206; Invitrogen) or Alexa Fluor 546-conjugated donkey anti-mouse secondary antibody 649 (1:1,000; A10036; Invitrogen) diluted in 1% normal donkey serum in PBS with 2 μg/ml DAPI 650 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 27 (Sigma-Aldrich, St. Louis, MO, USA) for 1 h. Cells were mounted in VECTASHIELD HardSet 651 Antifade Mounting Medium (Vector Laboratories, Inc., Burlingame, CA, USA) and photographed 652 using an Axiocam 506 mono digital camera (Carl Zeiss Microscopy LLC, White Plains, NY, 653 USA) fitted onto an Axio Observer 7 fluorescence microscope (Carl Zeiss Microscopy LLC) with 654 the 63x oil objective with a numerical aperture of 1.4 at room temperature. The following 655 antibodies were used for immunofluorescence analysis: Rab5 (1:200, C8B1, 3547, Cell 656 Signaling Technology Inc.), PDGFRa (1:20, AF1062, R&D Systems). For assessment of Rab5 657 puncta size and colocalization experiments, three independent trials, or biological replicates, 658 were performed. For each biological replicate, 20 technical replicates consisting of individual 659 cells were imaged with Z-stacks (0.24 μm between Z-stacks with a range of 1–6 Z-stacks) per 660 timepoint. Images were deconvoluted using ZEN Blue software (Carl Zeiss Microscopy LLC) 661 using the ‘Better, fast (Regularized Inverse Filter)’ setting. Extended depth of focus was applied 662 to Z-stacks using ZEN Blue software (Carl Zeiss Microscopy LLC) to generate images with the 663 maximum depth of field. For assessment of Rab5 puncta size, images were converted to 8-bit 664 using Fiji software (version 2.14.0/1.54f). Images were subsequently converted to a mask and 665 watershed separation was applied. A region of interest (ROI) was drawn around each Rab5-666 positive cell and particles were analyzed per cell using the “analyze particles” function. For 667 colocalization measurements, an ROI was drawn around each PDGFRa-positive cell in the 668 corresponding Cy3 (marker) channel using Fiji. For each image with a given ROI, the Cy3 669 channel and the EGFP channel were converted to 8-bit images. Colocalization was measured 670 using the Colocalization Threshold function, where the rcoloc value [Pearson’s correlation 671 coefficient (PCC)] was used in statistical analysis. Statistical analyses were performed on the 672 average values from each biological replicate with Prism 10 (GraphPad Software Inc.) using a 673 two-way ANOVA followed by uncorrected Fisher’s LSD test. 674 675 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 28

676 We are grateful to Jessica Johnston and Erin Binne for technical assistance, and Drs. Allison 677 Swain, Justin Roberts and Aaron Johnson at the University of Colorado Anschutz Medical 678 Campus for advice on eCLIP experiments. Cell sorting was performed at the University of 679 Colorado Cancer Center Flow Cytometry Shared Resource with assistance from Dr. Dmitry 680 Baturin. RNA-seq and eCLIP sequencing experiments were performed at University of Colorado 681 Cancer Center Genomics Shared Resource. We thank members of the Fantauzzo laboratory for 682 their critical comments on the manuscript. 683 684 Competing Interests 685 The authors declare no competing or financial interests. 686 687 Author contributions 688 Conceptualization: TEF, MS, NM, KAF; Methodology: TEF, MS, NM, KAF; Formal analysis: 689 TEF, MS, EDL, KAF; Investigation: TEF, MS, EDL; Writing – Original Draft: TEF, KAF; Writing – 690 Review & Editing: MS, EDL, NM; Visualization: TEF, MS, KAF; Supervision: NM, KAF; Project 691 administration: KAF; Funding acquisition: TEF, KAF. 692 693 Funding 694 This work was supported by National Institutes of Health grants R01DE030864 (to K.A.F.), 695 R35GM147025 (to N.M.), the University of Colorado Anschutz Medical Campus RNA 696 Bioscience Initiative (to N.M. and M.P.S.) and F31DE032252 (to T.E.F.). The Flow Cytometry 697 Shared Resource and Genomics Shared Resource are supported by National Institutes of 698 Health grant P30CA046934. 699 700 Data Availability 701 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 29 The eCLIP and RNA-sequencing datasets generated during this study have been deposited in 702 GEO under SuperSeries accession number GSE263170. Custom analysis scripts will be 703 provided by Dr. Larson through GitHub. 704 705

706 Akerman, M., David-Eden, H., Pinter, R. Y., & Mandel-Gutfreund, Y. (2009). A computational 707 approach for genome-wide mapping of splicing factor binding sites. Genome Biology, 708 10(3), R30. https://doi.org/10.1186/gb-2009-10-3-r30 709 Amit, M., Donyo, M., Hollander, D., Goren, A., Kim, E., Gelfman, S., Lev-Maor, G., Burstein, D., 710 Schwartz, S., Postolsky, B., Pupko, T., & Ast, G. (2012). Differential GC Content 711 between Exons and Introns Establishes Distinct Strategies of Splice-Site Recognition. 712 Cell Reports, 1(5), 543–556. https://doi.org/10.1016/j.celrep.2012.03.013 713 Andrae, J., Gouveia, L., Gallini, R., He, L., Fredriksson, L., Nilsson, I., Johansson, B. R., 714 Eriksson, U., & Betsholtz, C. (2016). A role for PDGF-C/PDGFRα signaling in the 715 formation of the meningeal basement membranes surrounding the cerebral cortex. 716 Biology Open, 5(4), 461–474. https://doi.org/10.1242/bio.017368 717 Änkö, M.-L., Müller-McNicoll, M., Brandl, H., Curk, T., Gorup, C., Henry, I., Ule, J., & 718 Neugebauer, K. M. (2012). The RNA-binding landscapes of two SR proteins reveal 719 unique functions and binding to diverse RNA classes. Genome Biology, 13(3), R17. 720 https://doi.org/10.1186/gb-2012-13-3-r17 721 Bavelloni, A., Piazzi, M., Faenza, I., Raffini, M., D’Angelo, A., Cattini, L., Cocco, L., & Blalock, 722 W. L. (2014). Prohibitin 2 represents a novel nuclear AKT substrate during all- trans 723 retinoic acid–induced differentiation of acute promyelocytic leukemia cells. The FASEB 724 Journal, 28(5), 2009–2019. https://doi.org/10.1096/fj.13-244368 725 Bebee, T. W., Park, J. W., Sheridan, K. I., Warzecha, C. C., Cieply, B. W., Rohacek, A. M., 726 Xing, Y., & Carstens, R. P. (2015). The splicing regulators Esrp1 and Esrp2 direct an 727 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 30 epithelial splicing program essential for mammalian development. eLife, 4, e08954. 728 https://doi.org/10.7554/eLife.08954 729 Bushnell, B. (2015). BBMap. Https://Sourceforge.Net/Projects/Bbmap/. 730 https://sourceforge.net/projects/bbmap/ 731 Chen, E. Y., Tan, C. M., Kou, Y., Duan, Q., Wang, Z., Meirelles, G. V., Clark, N. R., & Ma’ayan, 732 A. (2013). Enrichr: Interactive and collaborative HTML5 gene list enrichment analysis 733 tool. BMC Bioinformatics, 14, 128. https://doi.org/10.1186/1471-2105-14-128 734 Choi, S. J., Marazita, M. L., Hart, P. S., Sulima, P. P., Field, L. L., McHenry, T. G., Govil, M., 735 Cooper, M. E., Letra, A., Menezes, R., Narayanan, S., Mansilla, M. A., Granjeiro, J. M., 736 Vieira, A. R., Lidral, A. C., Murray, J. C., & Hart, T. C. (2009). The PDGF-C regulatory 737 region SNP rs28999109 decreases promoter transcriptional activity and is associated 738 with CL/P. European Journal of Human Genetics, 17(6), 774–784. 739 https://doi.org/10.1038/ejhg.2008.245 740 Cibi, D. M., Mia, M. M., Guna Shekeran, S., Yun, L. S., Sandireddy, R., Gupta, P., Hota, M., 741 Sun, L., Ghosh, S., & Singh, M. K. (2019). Neural crest-specific deletion of Rbfox2 in 742 mice leads to craniofacial abnormalities including cleft palate. eLife, 8, e45418. 743 https://doi.org/10.7554/eLife.45418 744 Dennison, B. J. C., Larson, E. D., Fu, R., Mo, J., & Fantauzzo, K. A. (2021). Srsf3 mediates 745 alternative RNA splicing downstream of PDGFRα signaling in the facial mesenchyme. 746 Development, 148(14), dev199448. https://doi.org/10.1242/dev.199448 747 Ding, H., Wu, X., Boström, H., Kim, I., Wong, N., Tsoi, B., O’Rourke, M., Koh, G. Y., Soriano, P., 748 Betsholtz, C., Hart, T. C., Marazita, M. L., Field, L. L., Tam, P. P. L., & Nagy, A. (2004). 749 A specific requirement for PDGF-C in palate formation and PDGFR-α signaling. Nature 750 Genetics, 36(10), 1111–1116. https://doi.org/10.1038/ng1415 751 Dobin, A., Davis, C. A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., Batut, P., Chaisson, 752 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 31 M., & Gingeras, T. R. (2013). STAR: ultrafast universal RNA-seq aligner. Bioinformatics, 753 29(1), 15-21. https://doi.org/10.1093/bioinformatics/bts635 754 Drewe-Boss, P., Wessels, H.-H., & Ohler, U. (2018). omniCLIP: Probabilistic identification of 755 protein-RNA interactions from CLIP-seq data. Genome Biology, 19(1), 183. 756 https://doi.org/10.1186/s13059-018-1521-2 757 Fantauzzo, K. A., & Soriano, P. (2014). PI3K-mediated PDGFR signaling regulates survival and 758 proliferation in skeletal development through p53-dependent intracellular pathways. 759 Genes & Development, 28(9), 1005–1017. https://doi.org/10.1101/gad.238709.114 760 Fantauzzo, K. A., & Soriano, P. (2017). Generation of an immortalized mouse embryonic palatal 761 mesenchyme cell line. PLOS ONE, 12(6), e0179078. 762 https://doi.org/10.1371/journal.pone.0179078 763 Forman, T. E., Dennison, B. J. C., & Fantauzzo, K. A. (2021). The Role of RNA-Binding Proteins 764 in Vertebrate Neural Crest and Craniofacial Development. Journal of Developmental 765 Biology, 19. 766 Frankish, A., Diekhans, M., Ferreira, A. M., Johnson, R., Jungreis, I., Loveland, J., Mudge, J. 767 M., Sisu, C., Wright, J., Armstrong, J., Barnes, I., Berry, A., Bignell, A., Carbonell Sala, 768 S., Chrast, J., Cunningham, F., Di Domenico, T., Donaldson, S., Fiddes, I. T., … Flicek, 769 P. (2019). GENCODE reference annotation for the human and mouse genomes. Nucleic 770 Acids Research, 47(D1), D766-763. https://doi.org/10.1093/nar/gky955 771 Fredriksson, L., Nilsson, I., Su, E. J., Andrae, J., Ding, H., Betsholtz, C., Eriksson, U., & 772 Lawrence, D. A. (2012). Platelet-Derived Growth Factor C Deficiency in C57BL/6 Mice 773 Leads to Abnormal Cerebral Vascularization, Loss of Neuroependymal Integrity, and 774 Ventricular Abnormalities. The American Journal of Pathology, 180(3), 1136–1144. 775 https://doi.org/10.1016/j.ajpath.2011.12.006 776 Fu, X.-D., & Ares, M. (2014). Context-dependent control of alternative splicing by RNA-binding 777 proteins. Nature Reviews Genetics, 15(10), 689–701. https://doi.org/10.1038/nrg3778 778 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 32 Fu, X.-D., & Maniatis, T. (1990). Factor required for mammalian spliceosome assembly is 779 localized to discrete regions in the nucleus. Nature, 343(6257), 437–441. 780 https://doi.org/10.1038/343437a0 781 Garcias, G. D. L., & Roth, M. D. G. M. (2007). A Brazilian family with quadrupedal gait, severe 782 mental retardation, coarse facial characteristics, and hirsutism. International Journal of 783 Neuroscience, 117(7), 927–933. https://doi.org/10.1080/00207450600910721 784 Gohr, A., & Irimia, M. (2019). Matt: Unix tools for alternative splicing analysis. Bioinformatics, 785 35(1), 130–132. https://doi.org/10.1093/bioinformatics/bty606 786 Gulsuner, S., Tekinay, A. B., Doerschner, K., Boyaci, H., Bilguvar, K., Unal, H., Ors, A., Onat, O. 787 E., Atalar, E., Basak, A. N., Topaloglu, H., Kansu, T., Tan, M., Tan, U., Gunel, M., & 788 Ozcelik, T. (2011). Homozygosity mapping and targeted genomic sequencing reveal the 789 gene responsible for cerebellar hypoplasia and quadrupedal locomotion in a 790 consanguineous kindred. Genome Research, 21(12), 1995-2003. 791 https://doi.org/10.1101/gr.126110.111. 792 Haward, F., Maslon, M. M., Yeyati, P. L., Bellora, N., Hansen, J. N., Aitken, S., Lawson, J., von 793 Kriegsheim, A., Wachten, D., Mill, P., Adams, I. R., & Caceres, J. F. (2021). Nucleo-794 cytoplasmic shuttling of splicing factor SRSF1 is required for development and cilia 795 function. eLife, 10, e65104. https://doi.org/10.7554/eLife.65104 796 He, F., & Soriano, P. (2013). A critical role for PDGFRalpha signaling in medial nasal process 797 development. PLoS Genetics, 9(9), e1003851. 798 https://doi.org/10.1371/journal.pgen.1003851 799 Howard, J. M., & Sanford, J. R. (2015). The RNAissance family: SR proteins as multifaceted 800 regulators of gene expression: Wiley Interdisciplinary Reviews: RNA, 6(1), 93–110. 801 https://doi.org/10.1002/wrna.1260 802 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 33 Huang, Y., Yario, T. A., & Steitz, J. A. (2004). A molecular link between SR protein 803 dephosphorylation and mRNA export. Proceedings of the National Academy of 804 Sciences, 101(26), 9666–9670. https://doi.org/10.1073/pnas.0403533101 805 Jourdain, A. A., Begg, B. E., Mick, E., Shah, H., Calvo, S. E., Skinner, O. S., Sharma, R., Blue, 806 S. M., Yeo, G. W., Burge, C. B., & Mootha, V. K. (2021). Loss of LUC7L2 and U1 snRNP 807 subunits shifts energy metabolism from glycolysis to OXPHOS. Molecular Cell, 81(9), 808 1905-1919.e12. https://doi.org/10.1016/j.molcel.2021.02.033 809 Jumaa, H., Wei, G., & Nielsen, P. J. (1999). Blastocyst formation is blocked in mouse embryos 810 lacking the splicing factor SRp20. Current Biology, 9(16), 899–902. 811 https://doi.org/10.1016/S0960-9822(99)80394-7 812 Klinghoffer, R. A., Hamilton, T. G., Hoch, R., & Soriano, P. (2002). An Allelic Series at the 813 PDGFaR Locus Indicates Unequal Contributions of Distinct Signaling Pathways During 814 Development. Developmental Cell, 2, 103–113. 815 Krainer, R., Mayeda, A., & Kozak, D. (1991). Functional Expression of Cloned Human Splicing 816 Factor SF2: Homology to RNA-Binding Proteins, Ul 70K, and Drosophila Splicing 817 Regulators. Cell, 66, 383–394. 818 Kuleshov, M. V., Jones, M. R., Rouillard, A. D., Fernandez, N. F., Duan, Q., Wang, Z., Koplev, 819 S., Jenkins, S. L., Jagodnik, K. M., Lachmann, A., McDermott, M. G., Monteiro, C. D., 820 Gundersen, G. W., & Ma’ayan, A. (2016). Enrichr: A comprehensive gene set 821 enrichment analysis web server 2016 update. Nucleic Acids Research, 44(W1), W90–822 W97. https://doi.org/10.1093/nar/gkw377 823 Lee, S., Sears, M. J., Zhang, Z., Li, H., Salhab, I., Krebs, P., Xing, Y., Nah, H.-D., Williams, T., & 824 Carstens, R. P. (2020). Cleft lip and cleft palate in Esrp1 knockout mice is associated 825 with alterations in epithelial-mesenchymal crosstalk. Development, 147(21), dev187369. 826 https://doi.org/10.1242/dev.187369 827 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 34 Li, D., & Roberts, R. (2001). WD-repeat proteins: Structure characteristics, biological function, 828 and their involvement in human diseases. Cellular and Molecular Life Sciences, 58. 829 Liao, X., Hu, K., Salhi, A., Zou, Y., Wang, J., & Gao, X. (2022). msRepDB: A comprehensive 830 repetitive sequence database of over 80 000 species. Nucleic Acids Research, 50(D1), 831 D236–D245. https://doi.org/10.1093/nar/gkab1089 832 Licatalosi, D. D., & Darnell, R. B. (2010). RNA processing and its regulation: Global insights into 833 biological networks. Nature Reviews Genetics, 11(1), 75–87. 834 https://doi.org/10.1038/nrg2673 835 Liu, K., Jian, Y., Sun, X., Yang, C., Gao, Z., Zhang, Z., Liu, X., Li, Y., Xu, J., Jing, Y., Mitani, S., 836 He, S., & Yang, C. (2016). Negative regulation of phosphatidylinositol 3-phosphate 837 levels in early-to-late endosome conversion. Journal of Cell Biology, 212(2), 181-198. 838 https://doi.org/10.1083/jcb.201506081. 839 Long, Y., Sou, W. H., Yung, K. W. Y., Liu, H., Wan, S. W. C., Li, Q., Zeng, C., Law, C. O. K., 840 Chan, G. H. C., Lau, T. C. K., & Ngo, J. C. K. (2019). Distinct mechanisms govern the 841 phosphorylation of different SR protein splicing factors. Journal of Biological Chemistry, 842 294(4), 1312–1327. https://doi.org/10.1074/jbc.RA118.003392 843 Love, M. I., Huber, W., & Anders, S. (2014). Moderated estimation of fold change and 844 dispersion for RNA-seq data with DESeq2. Genome Biology, 15(12), 550. 845 https://doi.org/10.1186/s13059-014-0550-8 846 Mai, C. T., Isenburg, J. L., Canfield, M. A., Meyer, R. E., Correa, A., Alverson, C. J., Lupo, P. J., 847 Riehle-Colarusso, T., Cho, S. J., Aggarwal, D., Kirby, R. S., & National Birth Defects 848 Prevention Network. (2019). National population-based estimates for major birth defects, 849 2010–2014. Birth Defects Research, 111(18), 1420–1435. 850 https://doi.org/10.1002/bdr2.1589 851 Manning, B. D., & Cantley, L. C. (2007). AKT/PKB Signaling: Navigating Downstream. Cell, 852 129(7), 1261–1274. https://doi.org/10.1016/j.cell.2007.06.009 853 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 35 Martin, M. (2011). Cutadapt removes adapter sequences from high-throughput sequencing 854 reads. EMBnet.Journal, 17(1), 10. https://doi.org/10.14806/ej.17.1.200 855 Moss, N. D., Wells, K. L., Theis, A., Kim, Y.-K., Spigelman, A. F., Liu, X., MacDonald, P. E., & 856 Sussel, L. (2023). Modulation of insulin secretion by RBFOX2-mediated alternative 857 splicing. Nature Communications, 14(1), 7732. https://doi.org/10.1038/s41467-023-858 43605-4 859 Pan, Q., Shai, O., Lee, L. J., Frey, B. J., & Blencowe, B. J. (2008). Deep surveying of alternative 860 splicing complexity in the human transcriptome by high-throughput sequencing. Nature 861 Genetics, 40(12), 1413–1415. https://doi.org/10.1038/ng.259 862 Park, J. W., Jung, S., Rouchka, E. C., Tseng, Y-T., & Xing, Y. (2016). rMAPS: RNA map 863 analysis and plotting server for alternative exon regulation. Nucleic Acids Research 864 44(W1), 333-338. https://doi.org/10.1093/nar/gkw410 865 Patro, R., Duggal, G., Love, M. I., Irizarry, R. A., & Kingsford, C. (2017). Salmon provides fast 866 and bias-aware quantification of transcript expression. Nature Methods, 14(4), 417–419. 867 https://doi.org/10.1038/nmeth.4197 868 Ramírez, F., Ryan, D. P., Grüning, B., Bhardwaj, V., Kilpert, F., Richter, A. S., Heyne, S., 869 Dündar, F., & Manke, T. (2016). deepTools2: A next generation web server for deep-870 sequencing data analysis. Nucleic Acids Research, 44(W1), W160–W165. 871 https://doi.org/10.1093/nar/gkw257 872 Rapiteanu, R., Davis, L. J., Williamson, J. C., Timms, R. T., Paul Luzio, J., & Lehner, P. J. 873 (2016). A Genetic Screen Identifies a Critical Role for the WDR81-WDR91 Complex in 874 the Trafficking and Degradation of Tetherin. Traffic, 17(8), 940–958. 875 https://doi.org/10.1111/tra.12409 876 Rattanasopha, S., Tongkobpetch, S., Srichomthong, C., Siriwan, P., Suphapeetiporn, K., & 877 Shotelersuk, V. (2012). PDGFRa mutations in humans with isolated cleft palate. 878 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 36 European Journal of Human Genetics, 20(10), 1058–1062. 879 https://doi.org/10.1038/ejhg.2012.55 880 Riemondy, K. A., Sheridan, R. M., Gillen, A., Yu, Y., Bennett, C. G., & Hesselberth, J. R. (2017). 881 valr: Reproducible genome interval analysis in R. F1000Research, 6, 1025. 882 https://doi.org/10.12688/f1000research.11997.1 883 Robinson, J. T., Thorvaldsdóttir, H., Winckler, W., Guttman, M., Lander, E. S., Getz, G., & 884 Mesirov, J. P. (2011). Integrative genomics viewer. Nature Biotechnology, 29(1), 24–26. 885 https://doi.org/10.1038/nbt.1754 886 Rogers, M. A., Campaña, M. B., Long, R., & Fantauzzo, K. A. (2022). PDGFR dimer-specific 887 activation, trafficking and downstream signaling dynamics. Journal of Cell Science, 888 135(17), jcs259686. https://doi.org/10.1242/jcs.259686 889 Schmok, J. C., Jain, M., Street, L. A., Tankka, A. T., Schafer, D., Her, H.-L., Elmsaouri, S., 890 Gosztyla, M. L., Boyle, E. A., Jagannatha, P., Luo, E.-C., Kwon, E. J., Jovanovic, M., & 891 Yeo, G. W. (2024). Large-scale evaluation of the ability of RNA-binding proteins to 892 activate exon inclusion. Nature Biotechnology. https://doi.org/10.1038/s41587-023-893 02014-0 894 Scotti, M. M., & Swanson, M. S. (2016). RNA mis-splicing in disease. Nature Reviews Genetics, 895 17(1), 19–32. https://doi.org/10.1038/nrg.2015.3 896 Shen, H., & Green, M. R. (2006). RS domains contact splicing signals and promote splicing by a 897 common mechanism in yeast through humans. Genes & Development, 20(13), 1755–898 1765. https://doi.org/10.1101/gad.1422106 899 Shen, S., Park, J. W., Lu, Z., Lin, L., Henry, M. D., Wu, Y. N., Zhou, Q., & Xing, Y. (2014). 900 rMATS: Robust and flexible detection of differential alternative splicing from replicate 901 RNA-Seq data. Proceedings of the National Academy of Sciences, 111(51), E5593–902 E5601. https://doi.org/10.1073/pnas.1419161111 903 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 37 Shin, C., Feng, Y., & Manley, J. L. (2004). Dephosphorylated SRp38 acts as a splicing 904 repressor in response to heat shock. Nature, 427(6974), 553–558. 905 https://doi.org/10.1038/nature02288 906 Simpson, J. E., Muir, M. T., Lee, M., Naughton, C., Gilbert, N., Pollard, S. M., & Gammoh, N. 907 (2024). Autophagy supports PDGFRA-dependent brain tumor development by 908 enhancing oncogenic signaling. Developmental Cell, S1534580723006214. 909 https://doi.org/10.1016/j.devcel.2023.11.023 910 Soneson, C., Love, M. I., & Robinson, M. D. (2016). Differential analyses for RNA-seq: 911 Transcript-level estimates improve gene-level inferences [version 2; referees: 2 912 approved]. F1000Research, 4, 1521. https://doi.org/10.12688/F1000RESEARCH.7563.2 913 Soriano, P. (1997). The PDGFα receptor is required for neural crest cell development and for 914 normal patterning of the somites. Development, 124, 2691–2700. 915 Tallquist, M. D., & Soriano, P. (2003). Cell autonomous requirement for PDGFRα in populations 916 of cranial and cardiac neural crest cells. Development, 130(3), 507–518. 917 https://doi.org/10.1242/dev.00241 918 Uyar, B., Yusuf, D., Wurmus, R., Rajewsky, N., Ohler, U., & Akalin, A. (2017). RCAS: An RNA 919 centric annotation system for transcriptome-wide regions of interest. Nucleic Acids 920 Research, 45(10), e91–e91. https://doi.org/10.1093/nar/gkx120 921 Van Nostrand, E. L., Nguyen, T. B., Gelboin-Burkhart, C., Wang, R., Blue, S. M., Pratt, G. A., 922 Louie, A. L., & Yeo, G. W. (2017). Robust, Cost-Effective Profiling of RNA Binding 923 Protein Targets with Single-end Enhanced Crosslinking and Immunoprecipitation 924 (seCLIP). In Y. Shi (Ed.), mRNA Processing (Vol. 1648, pp. 177–200). Springer New 925 York. https://doi.org/10.1007/978-1-4939-7204-3_14 926 Van Nostrand, E. L., Pratt, G. A., Shishkin, A. A., Gelboin-Burkhart, C., Fang, M. Y., 927 Sundararaman, B., Blue, S. M., Nguyen, T. B., Surka, C., Elkins, K., Stanton, R., Rigo, 928 F., Guttman, M., & Yeo, G. W. (2016). Robust transcriptome-wide discovery of RNA-929 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 38 binding protein binding sites with enhanced CLIP (eCLIP). Nature Methods, 13(6), 508–930 514. https://doi.org/10.1038/nmeth.3810 931 Vasudevan, H. N., Mazot, P., He, F., & Soriano, P. (2015). Receptor tyrosine kinases modulate 932 distinct transcriptional programs by differential usage of intracellular pathways. eLife, 4, 933 e07186. https://doi.org/10.7554/eLife.07186 934 Wallroth, A., & Haucke, V. (2018). Phosphoinositide conversion in endocytosis and the 935 endolysosomal system. Journal of Biological Chemistry, 293(5), 1526–1535. 936 https://doi.org/10.1074/jbc.R117.000629 937 Wang, E. T., Sandberg, R., Luo, S., Khrebtukova, I., Zhang, L., Mayr, C., Kingsmore, S. F., 938 Schroth, G. P., & Burge, C. B. (2008). Alternative isoform regulation in human tissue 939 transcriptomes. Nature, 456(7221), 470–476. https://doi.org/10.1038/nature07509 940 Xiao, S.-H., & Manley, J. (1997). Phosphorylation of the ASF/SF2 RS domain affects both 941 protein-protein and protein-RNA interactions and is necessary for splicing. Genes & 942 Development, 12. 943 Zahler, A. M., Neugebauer, K. M., Lane, W. S., & Roth, M. B. (1993). Distinct Functions of SR 944 Proteins in Alternative pre-mRNA Splicing. Science, 260(5105), 219–222. 945 https://doi.org/10.1126/science.8385799 946 Zerial, M., & McBride, H. (2001). Rab proteins as membrane organizers. Nature Reviews 947 Molecular Cell Biology, 2(2), 107–117. https://doi.org/10.1038/35052055 948 Zhou, Z., & Fu, X.-D. (2013). Regulation of splicing by SR proteins and SR protein-specific 949 kinases. Chromosoma, 122(3), 191–207. https://doi.org/10.1007/s00412-013-0407-z 950 Zhou, Z., Qiu, J., Liu, W., Zhou, Y., Plocinik, R. M., Li, H., Hu, Q., Ghosh, G., Adams, J. A., 951 Rosenfeld, M. G., & Fu, X.-D. (2012). The Akt-SRPK-SR Axis Constitutes a Major 952 Pathway in Transducing EGF Signaling to Regulate Alternative Splicing in the Nucleus. 953 Molecular Cell, 47(3), 422–433. https://doi.org/10.1016/j.molcel.2012.05.014 954 955 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 39 956 Figure 1: PDGFRa signaling for one hour minimally affects gene expression. (A) 957 Schematic of RNA-seq experimental design. iMEPM cells were transduced to stably express a 958 scramble shRNA (scramble) or shRNA targeting the 3’ UTR of Srsf3 (shSrsf3). iMEPM cells 959 expressing either scramble or shSrsf3 were left unstimulated or stimulated with 10 ng/mL 960 PDGF-AA for 1 hour and RNA was isolated for RNA-seq analysis. (B) Western blot (WB) 961 analysis of whole-cell lysates from scramble and shSrsf3 cell lines with anti-Srsf3 and anti-962 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 40 Gapdh antibodies. The percentage of Srsf3 expression normalized to Gapdh expression is 963 indicated below. (C) Volcano plots depicting differentially-expressed genes in scramble versus 964 shSrsf3 cell lines in the absence (left) or presence (right) of PDGF-AA stimulation. Log2(fold 965 change) (FC) values represent log2(shSrsf3 normalized counts/scramble normalized counts). 966 Significant changes in gene-level expression are reported for genes with adjusted P (padj) < 967 0.05 and fold change |FC| ³ 2. (D) Venn diagram of significant genes from C. (E) Volcano plots 968 depicting differentially-expressed genes in the absence versus presence of PDGF-AA ligand in 969 scramble (left) or shSrsf3 (right) cell lines. Log2(FC) values represent log2(+PDGF-AA 970 normalized counts/-PDGF-AA normalized counts). (F) Venn diagram of significant genes from 971 E. 972 973 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 41 974 Figure 2: PDGFRa signaling for one hour has a more pronounced effect on alternative 975 RNA splicing. (A) Volcano plots depicting alternatively-spliced transcripts in scramble versus 976 shSrsf3 cell lines in the absence (left) or presence (right) of PDGF-AA stimulation. Difference in 977 percent spliced in (ΔPSI) values represent scramble PSI – shSrsf3 PSI. Significant changes in 978 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 42 alternative RNA splicing are reported for events with a false discovery rate (FDR) £ 0.05 and a 979 difference in percent spliced in (|ΔPSI|) ³ 0.05. (B) Venn diagram of significant transcripts from 980 A, filtered to include events detected in at least 10 reads in either condition. (C) Volcano plots 981 depicting alternatively-spliced transcripts in the absence versus presence of PDGF-AA ligand in 982 scramble (left) or shSrsf3 (right) cell lines. Difference in percent spliced in (ΔPSI) values 983 represent -PDGF-AA PSI – +PDGF-AA PSI. (D) Venn diagram of significant transcripts from C, 984 filtered to include events detected in at least 10 reads in either condition. (E) Bar graph 985 depicting alternative RNA splicing events in scramble versus shSrsf3 cell lines in the absence or 986 presence of PDGF-AA stimulation (left) or in the absence versus presence of PDGF-AA ligand 987 in scramble or shSrsf3 cell lines (right). 988 989 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 43 990 Figure 3: Srsf3 exhibits differential transcript binding upon PDGFRa signaling. (A) 991 Schematic of eCLIP experimental design. iMEPM cells were left unstimulated or stimulated with 992 10 ng/mL PDGF-AA for 1 hour and processed for eCLIP analysis. (B) Immunoprecipitation (IP) 993 of Srsf3 from cells that were UV-crosslinked or not UV-crosslinked with IgG or an anti-Srsf3 994 antibody followed by western blotting (WB) of input, supernatant (Sup), and IP samples with an 995 anti-Srsf3 antibody. (C) Mapping of eCLIP peaks to various transcript locations in the absence 996 or presence of PDGF-AA stimulation. 5’ UTR, 5’ untranslated region; CDS, coding sequence; 3’ 997 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 44 UTR, 3’ untranslated region. (D,E) Mean coverage of eCLIP peaks across various transcript 998 locations (D) and surrounding the 5’ and 3’ splice sites (E) in the absence or presence of PDGF-999 AA stimulation. (F,G) Top three motifs enriched in eCLIP peaks in the absence (F) or presence 1000 (G) of PDGF-AA stimulation. 1001 1002 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 45 1003 Figure 4: Srsf3 and PDGFRa signaling are associated with differential GC content and 1004 length of alternatively-spliced exons. (A) Box and whisker plot depicting the percentage of 1005 exon GC content in exons that are not differentially alternatively spliced, and exons that are 1006 included or skipped when Srsf3 is present from the rMATS analysis. (B) Box and whisker plot 1007 depicting the ratio of downstream intron to exon GC content in exons that are not differentially 1008 alternatively spliced, and exons that are included or skipped when Srsf3 is present from the 1009 rMATS analysis. (C,D) Box and whisker plots depicting the ratio of upstream intron to exon 1010 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 46 length (C) and downstream intron to exon length (D) in exons that are not differentially 1011 alternatively spliced, and exons that are included or skipped when Srsf3 is present from the 1012 rMATS analysis. (E) Violin and box and whisker (inset) plots depicting the percentage of exon 1013 GC content in exons that are not bound by Srsf3, and exons that are bound in the absence 1014 and/or presence of PDGF-AA stimulation from the eCLIP analysis. *, P < 0.05; **, P < 0.01; ***, 1015 P < 0.001. 1016 1017 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 47 1018 Figure 5: Transcripts bound by Srsf3 that undergo alternative splicing upon PDGFRa 1019 signaling encode regulators of PI3K signaling. (A) Venn diagram of genes with differential 1020 expression (DE) or transcripts subject to alternative RNA splicing (AS) across the four treatment 1021 comparisons that overlap with transcripts with Srsf3 eCLIP peaks in the absence or presence of 1022 PDGF-AA stimulation. (B,C) Top ten (B) and PI3K-related (C) biological process gene ontology 1023 (GO) terms for transcripts from the overlapping datasets. p.val, P. (D) Difference in percent 1024 spliced in (DPSI) values for PI3K/endosome-related transcripts of interest. FDR, false detection 1025 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 48 rate. (E) Peak visualization for input and eCLIP samples in the absence or presence of PDGF-1026 AA stimulation from Integrative Genomics Viewer (left) with location of motifs from Figure S5 1027 indicated below for PI3K/endosome-related transcripts of interest. Predicted alternative splicing 1028 outcomes for PI3K/endosome-related transcripts of interest (right). 1029 1030 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 49 1031 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint 50 Figure 6: Srsf3 regulates early endosome size and phosphorylation of Akt downstream of 1032 PDGFRa signaling. (A,B) Scatter dot plots depicting average size of Rab5 puncta per cell (A) 1033 and Pearson’s correlation coefficient of signals from anti-Rab5 and anti-PDGFRa antibodies (B) 1034 in scramble and shSrsf3 cell lines in the absence or presence (15-60 min) of PDGF-AA 1035 stimulation. Data are mean ± s.e.m. *, P < 0.05. Shaded shapes correspond to independent 1036 experiments. Summary statistics from biological replicates consisting of independent 1037 experiments (large shapes) are superimposed on top of data from all cells; n = 20 technical 1038 replicates across each of three biological replicates. (C-H”) PDGFRa antibody signal (white or 1039 magenta) and Rab5 antibody signal (white or green) as assessed by immunofluorescence 1040 analysis of scramble and shSrsf3 cells in the absence or presence (15-60 min) of PDGF-AA 1041 stimulation. Nuclei were stained with DAPI (blue). White arrows denote colocalization. Scale 1042 bars: 20 µm (main images), 3 µm (insets). (I) Western blot (WB) analysis of whole-cell lysates 1043 (WCL) from scramble (left) and shSrsf3 (right) cell lines following a time course of PDGF-AA 1044 stimulation from 15 min to 4 h, with anti-phospho-(p)-Akt and anti-Akt antibodies. Line graphs 1045 depicting quantification of band intensities from n = 3 biological replicates as above. Data are 1046 mean ± s.e.m. *, P < 0.05; **, P < 0.01. (J) Model of experimental results in which PI3K/Akt-1047 mediated PDGFRa signaling results in the nuclear translocation of Srsf3 and the subsequent 1048 AS of transcripts to decrease levels of proteins that promote PDGFRa trafficking out of early 1049 endosomes. 1050 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted April 3, 2024. ; https://doi.org/10.1101/2024.04.03.587975doi: bioRxiv preprint