Abstract

   Aberrant alternative splicing plays a crucial role in tumorigenesis.
   Here, splicing factor 3A subunit 3 (SF3A3) is   significantly
   upregulated in endometrial cancer (EC) tissues and associated with poor
   prognosis. Functionally, SF3A3 drives tumor progression by promoting
   cell proliferation, suppressing apoptosis, and enhancing cisplatin
   resistance in vitro and in vivo. Mechanistically, SF3A3 regulates
   alternative splicing of fos proto‐oncogene, AP‐1 transcription factor
   subunit (c‐FOS), an oncogene linked to chemoresistance, resulting in a
   ≈2‐fold increase in full‐length c‐FOS expression and activation of
   downstream anti‐apoptotic pathways. Notably, phenylethyl isothiocyanate
   (PEITC) as a direct inhibitor of SF3A3 through database screening and
   biophysical validation via surface plasmon resonance and mass
   spectrometry is identified. PEITC reduces c‐FOS expression and induces
   apoptosis in EC cells. Moreover, encapsulating PEITC in a hydrogel
   delivery system significantly enhances its therapeutic efficacy by
   enabling controlled release, reducing dosing frequency, and improving
   clinical applicability. The therapeutic potential of SF3A3 inhibition
   is further validated using patient‐derived tumor‐like cell clusters
   (PTCs), where PEITC and the c‐FOS inhibitor T‐5224 exhibit synergistic
   effects in suppressing EC. Collectively, our findings establish SF3A3
   as a novel oncogenic regulator in EC and highlight PEITC, particularly
   in its hydrogel formulation, as a promising therapeutic strategy for
   improving clinical outcomes in EC patients.

   Keywords: SF3A3, PEITC, c‐FOS, endometrial cancer, tumor progression
     __________________________________________________________________

   The role of splicing factor 3A subunit 3 (SF3A3) in driving endometrial
   cancer (EC) progression and potential therapeutic interventions. SF3A3,
   a core component of the U2 small nuclear ribonucleoprotein (U2 snRNP),
   is upregulated in endometrial cancer (EC) and promotes tumor growth and
   cisplatin resistance. RNA‐binding protein immunoprecipitation (RIP)
   pull‐down assays reveal that SF3A3 drives alternative splicing of fos
   proto‐oncogene, AP‐1 transcription factor subunit (c‐FOS), leading to
   sustained c‐FOS expression, increased B‐cell lymphoma‐2 (Bcl2)/Bcl‐2
   associated protein (Bax) ratio, and reduced apoptosis. Targeting SF3A3
   with phenylethyl isothiocyanate (PEITC) effectively suppresses tumor
   progression, induces apoptosis, and reduces c‐FOS expression in vitro,
   in vivo, and patient‐derived tumor‐like cell clusters (PTCs).
   Furthermore, hydrogel‐based PEITC delivery significantly enhances
   therapeutic efficacy, offering a promising, clinically feasible
   strategy for precision‐targeted treatment in EC.

   graphic file with name ADVS-12-e04184-g005.jpg

1. Introduction

   Endometrial cancer (EC) is one of the most common gynecologic
   malignancies, with a rising global burden. Approximately 320000 new
   cases and 76000 deaths are reported worldwide each year^[ [50]^1 ^]
   Despite advancements in surgical interventions and chemotherapeutic
   strategies,^[ [51]^2 , [52]^3 ^] a significant proportion of EC
   patients experience recurrence and develop resistance to standard
   treatments, leading to poor clinical outcomes.^[ [53]^4 , [54]^5 ,
   [55]^6 , [56]^7 , [57]^8 ^] Identifying novel molecular drivers
   contributing to EC progression and chemoresistance is essential to
   improving therapeutic strategies and patient prognosis.

   Alternative splicing (AS) is a strictly regulated process that removes
   noncoding introns and joins coding exons in different combinations,
   greatly expanding transcriptome and proteome diversity.^[ [58]^9 ,
   [59]^10 , [60]^11 ^] While AS is crucial for normal cellular function,
   its dysregulation has been implicated in cancer, promoting enhanced
   proliferation, evasion of apoptosis, and therapeutic resistance.^[
   [61]^12 , [62]^13 ^] Increasing evidence suggests that AS abnormalities
   are particularly prevalent in EC, where splicing factor imbalances
   promote oncogenesis. For example, CircRAPGEF5 and splicing factor 3B
   subunit 1 (SF3B1) respectively promote EC progression by modulating AS
   of downstream genes in EC.^[ [63]^14 , [64]^15 ^] However, the specific
   role of AS in EC pathogenesis remains incompletely understood,
   particularly regarding the oncogenic functions of individual splicing
   factors.

   The spliceosome, a multi‐subunit complex composed of small nuclear RNAs
   (snRNAs) and associated proteins, is central to AS process.^[ [65]^16 ,
   [66]^17 , [67]^18 ^] Among its components, splicing factor 3A subunit 3
   (SF3A3) is an essential member of the U2 small nuclear
   ribonucleoprotein (U2 snRNP), contributing to spliceosome assembly and
   pre‐mRNA recognition.^[ [68]^18 ^] Dysregulation of SF3A3 has been
   implicated in multiple cancers, including breast, bladder, and lung
   cancer, yet its role in EC remains unexplored.^[ [69]^19 , [70]^20 ,
   [71]^21 ^] Given its critical function in mRNA maturation, SF3A3 may
   act as a key oncogenic regulator in EC by altering the splicing
   landscape of cancer‐associated genes.

   Herein, we demonstrate that SF3A3 is frequently overexpressed in EC
   tissues and accelerates EC progression and cisplatin resistance.
   Mechanistically, SF3A3 regulates apoptosis and contributes to drug
   resistance, at least partially by modulating the AS of fos
   proto‐oncogene, AP‐1 transcription factor subunit (c‐FOS), a
   proto‐oncogene involved in cellular proliferation, drug resistance, and
   tumor growth. These findings identify SF3A3 as a promising target for
   novel therapeutic strategies in EC. Additionally, we evaluated the
   therapeutic efficacy of phenylethyl isothiocyanate (PEITC), a small
   molecule that directly inhibits SF3A3. Furthermore, encapsulating PEITC
   in a hydrogel delivery system significantly enhanced its therapeutic
   efficacy by enabling sustained release, reducing dosing frequency, and
   improving clinical feasibility. The cultivation conditions and drug
   concentrations determined in the PTCs model from EC patients support
   the potential for clinical translation. Our findings reveal a novel
   role of SF3A3 in EC progression and cisplatin resistance, and highlight
   PEITC, particularly in hydrogel formulation, as a promising strategy
   for improving treatment outcomes in EC patients.

2. Results

2.1. SF3A3 is Aberrantly Upregulated in EC Tissues and Promotes Poor
Prognosis in Patients with EC

   SF3A3 has been previously identified as a potential driver of tumor
   progression^[ [72]^19 , [73]^20 , [74]^21 , [75]^22 ^] (Figure [76]1a).
   To elucidate its role, we analyzed SF3A3 expression abnormalities in EC
   tissues. Higher SF3A3 expression levels were strongly correlated with
   poor overall survival in UCEC patients, as shown by Kaplan–Meier
   survival analysis (Figure [77]1b). Further investigation using TCGA and
   GTEx databases confirmed significant upregulation of SF3A3 in EC
   tissues compared to normal samples (Figure [78]1c). Notably, SF3A3
   expression varied slightly across tumor stages (Extended Data
   Figure [79]1a). High SF3A3 expression was most prominent in the POLE
   (ultramutated) and copy‐number high (serous‐like) subtypes, while lower
   expression was observed in the MSI (hypermutated) and copy‐number low
   (endometrioid) subtypes (Extended Data Figure [80]1a). Validation with
   our independent clinical samples dataset further confirmed that SF3A3
   expression was significantly upregulated in EC tissues compared to
   tumor‐adjacent non‐cancerous endometrial tissues (Figure [81]1d–f and
   Extended Data Figure [82]1b–c; Table [83]S7, Supporting Information).
   Collectively, these data indicated that elevated SF3A3 expression is
   strongly associated with worse clinical outcomes in EC patients and
   suggested that it may serve as a key biomarker for EC.

Figure 1.

   Figure 1
   [84]Open in a new tab

   SF3A3 is frequently overexpressed in EC tumors and correlates with poor
   disease prognosis in EC patients. a) Schematic representation of the
   screening process for variable splicing factor SF3A3 in EC and its
   impact on alternative splicing. b) Kaplan–Meier curves illustrating
   overall survival (OS) predictions based on PSI values. Data were
   obtained from the UCSC Xena ([85]https://xenabrowser.net/); p < 0.05,
   as compared with the low group; (n = 134 samples each group). c) SF3A3
   is differentially overexpressed in EC tissues compared to normal
   tissues, based on TCGA and GTEx databases ([86]https://gtexportal.org/
   and [87]https://cancergenome.nih.gov/); p < 0.05, as compared with the
   Normal group; (Normal = 104, tumor = 620 samples). d) SF3A3 RNA
   expression is significantly upregulated in EC tissues (n = 27 samples
   each group). p < 0.05, as compared with the Normal group. e) SF3A3
   protein expression is significantly elevated in EC tissues (n = 27
   samples each group). p < 0.05, as compared with the Normal group. f)
   Immunohistochemistry analysis of SF3A3 expression in human endometrial
   and EC specimens (n = 6 samples each group). p < 0.05, as compared with
   the Normal group; Scale bars = 50 µm. Data are presented as mean ± S.D.
   from at least three independent experiments.

2.2. SF3A3 Promotes Proliferation and Xenograft Tumor Growth in EC

   To investigate the functional role of SF3A3 in EC, we assessed its
   expression in the regular endometrial cell line (hEEC) and three EC
   cell lines. SF3A3 expression was markedly higher in EC cell lines,
   particularly KLE and Ishikawa cells, which were selected for further
   analysis (Extended Data Figure [88]2a–c). We then modulated SF3A3
   expression using lentiviral overexpression systems (pLV‐Puro‐SF3A3) in
   KLE and Ishikawa cells (Extended Data Figure [89]2d–h). In a mouse
   xenograft model, SF3A3‐overexpressing KLE cells were injected
   subcutaneously into nude mice. Tumor growth was monitored from day 14
   when tumors reached ∼100 mm^3. Overexpression of SF3A3 significantly
   increased tumor volume and weight after 35 days, without affecting the
   mice's body weight (Figure [90]2a–c, Extended Data Figure [91]2i,j).
   Immunofluorescence analysis revealed marked overexpression of SF3A3 and
   the proliferation marker Ki67 in tumors derived from
   SF3A3‐overexpressing cells compared to controls (Figure [92]2d and
   Extended Data Figure [93]2k). Consistent with in vivo findings, SF3A3
   overexpression enhanced cell viability and colony formation in both KLE
   and Ishikawa cells after 10 days (Figure [94]2e,g). Conversely,
   silencing SF3A3 significantly inhibited cell viability and colony
   formation in both cell lines, further supporting its pro‐tumorigenic
   role (Figure [95]2f,h).

Figure 2.

   Figure 2
   [96]Open in a new tab

   SF3A3 increases the oncogenic potential of EC cells. a) Schematic of
   the experimental protocol. Xenograft experiments by subcutaneous
   injection were conducted in KLE cells with vector or SF3A3
   overexpression vector. On day 14, tumor volumes and mice weight were
   monitored; Animals were sacrificed on day 35, and tumor samples were
   collected for analysis. b–d) (b) Tumor volumes over time. (c) Tumor
   weights at endpoint. p < 0.05, as compared with the NC group; (n = 5
   samples each group). (d) Representative images of SF3A3 and Ki67
   immunofluorescence and HE staining in sections of excised tumors. Scale
   bars = 50 µm. e–h) Cell proliferation assays (e,f) and colony formation
   assays (g,h) were performed in SF3A3‐overexpressing or si‐SF3A3 KLE and
   Ishikawa cells. p < 0.05, as compared with the NC or si‐NC group; (n =
   3 independent experiments). i,j) Changes in cisplatin sensitivity
   following SF3A3 overexpression in KLE and Ishikawa cells. (n = 3
   independent experiments). k,l) Apoptotic cells were detected via flow
   cytometry using Annexin V‐PE/7‐AAD staining in SF3A3‐overexpressing or
   si‐SF3A3 KLE and Ishikawa cells. Data represent the percentage of early
   and late apoptotic cells p < 0.05, as compared with the NC or si‐NC
   group; (n = 3 independent experiments). m) Apoptotic cells detected by
   TUNEL staining in SF3A3‐overexpressing or si‐SF3A3 KLE and Ishikawa
   cells. Scale bars = 20 µm. p < 0.05, as compared with the NC or si‐NC
   group; (n = 5 samples each group). n) Immunoblotting analysis of
   apoptotic markers (Bcl2 and Bax) in KLE and Ishikawa cells following
   SF3A3 overexpression or si‐SF3A3. p < 0.05, as compared with the NC or
   si‐NC group; (n = 3 independent experiments). Data are presented as
   mean ± S.D. from at least three independent experiments.

   Furthermore, after overexpression of SF3A3, the IC[50] of cisplatin in
   KLE cells changed from 3.09 to 8.10 µ[M], and in Ishikawa cells, it
   changed from 4.95 to 19.45 µm (Figure [97]2i,j). Apoptosis assays,
   including TUNEL staining, Annexin V‐PE/7‐AAD staining, and flow
   cytometry, demonstrated that SF3A3 overexpression significantly reduced
   apoptosis in KLE and Ishikawa cells (Figure [98]2k,m, and Extended Data
   Figure [99]2l). Conversely, SF3A3 knockdown induced apoptosis in EC
   cells (Figure [100]2l,m and Extended Data Figure [101]2m).
   Immunoblotting analysis further revealed that SF3A3 overexpression
   increased the anti‐apoptotic Bcl2/Bax ratio, while knockdown reduced it
   (Figure [102]2n and Extended Data Figure [103]2n,o). Together, these
   findings demonstrate that SF3A3 drives tumor progression in EC by
   promoting cell proliferation, enhanced cisplatin resistance, and
   reduced apoptosis in vitro and in vivo.

2.3. The Genome‐Wide Landscape of SF3A3‐Binding Sites on RNA

   As a splicing factor, SF3A3 may influence cancer progression through
   alternative splicing (AS) regulation. To identify its RNA targets, we
   performed RNA‐binding protein immunoprecipitation (RIP) followed by
   high‐throughput RNA sequencing (RNA‐seq) in KLE cells (Figure
   [104]3a,b). RIP is a technique for studying the binding of
   intracellular RNA to proteins. Compared to control cells, RIP‐Seq
   results revealed that 2451 AS events enriched in SF3A3‐expression cells
   (Figure [105]3c–e, Extended Data Figure [106]3a). These events were
   categorized into five AS types: skipped exons (SE), retained introns
   (RI), mutually exclusive exons (MXE), alternative 5′ splice sites
   (A5SS), and alternative 3′ splice sites (A3SS). SE and RI represented
   the most abundant categories (Figure [107]3c).

Figure 3.

   Figure 3
   [108]Open in a new tab

   Identification of the genome‐wide SF3A3‐binding sites on RNA. a)
   Genome‐wide distribution of RNA targets identified by RNA Binding
   Protein Immunoprecipitation (RIP) followed by sequencing (RIP‐seq). (n
   = 3 samples each group). b) Protein validation of SF3A3 RIP pulldown.
   (n = 3 independent experiment). c) Alternative splicing (AS) events
   identified in RIP‐seq, classified into five categories: skipped exons
   (SE), retained introns (RI), mutually exclusive exons (MXE),
   alternative 5′ splice sites (A5SS), and alternative 3′ splice sites
   (A3SS). d) De novo motif analysis of SF3A3 RIP clusters, showing the
   top four SF3A3‐binding motifs ranked by HOMER cluster values. Sequence
   elements enriched in the top motifs are shown. e) KEGG pathway
   enrichment analysis of SF3A3‐interacting genes in KLE cells. f) RT‐qPCR
   validation of the top four SF3A3‐binding motifs identified by RIP.
   p < 0.05, as compared with the pLJM1‐EGFP group; (n = 3 independent
   experiments). g) Semi‐quantitative RT‐qPCR validation of
   SF3A3‐regulated genes, c‐FOS. (n = 3 independent experiments). h)
   Schematic of the alternative splicing (AS) pattern and SF3A3‐binding
   sites in c‐FOS. Light blue, green, and red regions highlight the AS
   region and SF3A3‐binding sites. i,j) Hybridization chain reaction (HCR)
   analysis showing co‐localization of SF3A3 and c‐FOS in
   SF3A3‐overexpressing or si‐SF3A3 KLE and Ishikawa cells. Overexpression
   of SF3A3 alters c‐FOS gene expression. Scale bar = 10 µm. p < 0.05, as
   compared with the NC or si‐NC group; (n = 5 samples each group). k,m)
   RT‐qPCR analysis of SF3A3‐regulated c‐FOS expression in KLE and
   Ishikawa cells. p < 0.05, as compared with the NC or si‐NC group; (n =
   3 independent experiments). l,n) Immunoblotting analysis of
   SF3A3‐mediated changes in c‐FOS protein levels in KLE and Ishikawa
   cells. p < 0.05, as compared with the NC or si‐NC group; (n = 3
   independent experiments). o) Pattern diagram of constructing variable
   splicing mutant plasmids. p) Immunoblotting analysis of transfecting
   variable splicing mutant plasmids (n = 3 independent experiments). q)
   RT‐PCR assays validating SF3A3‐mediated alternative splicing in c‐FOS
   via transfecting with variable splicing mutant SF3A3 plasmids (n = 3
   independent experiments). Data are presented as mean ± S.D. from at
   least three independent experiments.

   To characterize the interaction of SF3A3 with RNA, we used the HOMER
   algorithm to identify RNA motifs recognized by SF3A3 (Figure [109]3d).
   Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG)
   enrichment analysis of SF3A3‐modulated AS targets revealed significant
   involvement in apoptosis, cisplatin resistance, cell cycle regulation,
   focal adhesion, and DNA repair pathways (Figure [110]3e). To validate
   the RIP‐seq results, we selected four representative alternative
   splicing (AS) targets—c‐FOS, CCND2, MDM4, and P53—based on their
   percent spliced‐in (PSI) values and false discovery rates (FDRs)
   between SF3A3‐overexpressing and control cells. RT‐qPCR confirmed that
   SF3A3 regulated all four targets, with c‐FOS showing the most prominent
   changes (Figure [111]3f). Among them, c‐FOS, a core component of the
   activator protein‐1 (AP‐1) complex, has been closely associated with
   tumor progression, chemoresistance, and apoptosis, which supported its
   selection for in‐depth investigation.^[ [112]^23 , [113]^24 , [114]^25
   ^]

   Semi‐quantitative PCR further validated these results, showing that
   SF3A3 overexpression significantly increased c‐FOS transcript levels
   (Figure [115]3g). Motif analysis revealed two high‐scoring
   SF3A3‐binding motifs near the exon 2–3 junctions, along with two
   additional motifs located within exonic regions (Figure [116]3h).
   Further studies showed that SF3A3 promoted the retention of introns 2
   and 3, but not intron 1, and primarily facilitated the production of
   full‐length c‐FOS mRNA spanning exons 1–4, whereas intron‐containing
   isoforms were rarely detected (Extended Data Figure [117]3b).
   Integration of RIP‐seq peak mapping and RT‐PCR data suggested that
   SF3A3 bound directly to the exon 2–3 region of c‐FOS pre‐mRNA.

   This binding interaction was further supported by hybridization chain
   reaction (HCR) analysis, which showed enhanced fluorescence intensity
   and co‐localization of SF3A3 with c‐FOS pre‐mRNA in
   SF3A3‐overexpressing cells (Figure [118]3i and Extended Data
   Figure [119]3c,d), with reciprocal signal reduction observed in SF3A3
   knockdown cells (Figure [120]3j and Extended Data Figure [121]3e,f).
   RT‐qPCR and immunoblotting analysis confirmed that SF3A3 overexpression
   elevated both RNA and protein levels of c‐FOS, increasing protein
   levels by 1.55‐fold in KLE cells and 2.13‐fold in Ishikawa cells
   (Figure [122]3k,l and Extended Data Figure [123]3g). Conversely, SF3A3
   knockdown significantly reduced c‐FOS expression (Figure [124]3m,n and
   Extended Data Figure [125]3h). To determine the functional domains of
   SF3A3 involved in c‐FOS regulation, we examined its protein structure.
   We identified the 129–178 amino acid (aa) region as the pre‐RNA binding
   domain and the 244–296 aa region as the splicing regulatory domain,
   based on InterPro annotations
   ([126]https://www.ebi.ac.uk/interpro/protein/UniProt/Q12874/) (Figure
   [127]3o). Truncation of the 129–296 aa impaired SF3A3's regulatory
   activity, which resulted in reduced c‐FOS mRNA (Figure [128]3p,q),
   indicating the importance of this region in modulating c‐FOS
   expression.

   To further expand the understanding of SF3A3's oncogenic network, we
   investigated additional targets identified by RIP‐seq. Notably, CCND2
   and MDM4, both involved in cell cycle progression, were also validated
   as SF3A3‐regulated genes. Overexpression of SF3A3 elevated both RNA and
   protein levels of CCND2 and MDM4 (Extended Data Figure [129]3i–m).
   These findings suggested that SF3A3 promoted endometrial cancer
   progression through the modulation of c‐FOS and by coordinating a
   broader oncogenic splicing network.

2.4. High SF3A3 Expression Induces Cisplatin Resistance and Oncogenic
Activity via c‐FOS Regulation

   Analysis of cell proliferation in vitro and EC growth in vivo strongly
   indicated that SF3A3 and c‐FOS have tumor‐suppressive effects. To
   further investigate the pathological significance of our findings, we
   examined SF3A3 and c‐FOS expression in 6 paired human EC lesions and
   adjacent noncancerous tissue samples. The results showed that SF3A3 and
   c‐FOS protein levels were significantly increased and Bcl2 was slightly
   increased in EC lesions compared to adjacent noncancerous tissue
   (Figure [130]4a) and showed a moderate correlation between the
   expression of SF3A3 and c‐FOS (Figure [131]4b). Similar trends were
   observed in immunohistochemistry (IHC) results of clinical EC tissues
   and adjacent normal samples and mouse xenograft tumors (Extended Data
   Figure [132]4a,c).

Figure 4.

   Figure 4
   [133]Open in a new tab

   The oncogenic roles of SF3A3 in EC are mediated through c‐FOS
   expression. a) Immunoblotting analysis of c‐FOS and SF3A3 protein
   levels in paired adjacent noncancerous endometrium tissues (N) and
   human EC tissues. (n = 6 samples each group). b) Correlation analysis
   of protein SF3A3 and c‐FOS using paired adjacent noncancerous
   endometrium tissues (N) and human EC tissues (n = 6 samples each
   group). c) Functional analysis of the relationship between SF3A3 and
   c‐FOS in EC cells. Following SF3A3 overexpression, siRNA‐mediated
   knockdown of c‐FOS was used to evaluate cell apoptosis, proliferation,
   and cisplatin resistance: d) Growth curves in KLE and Ishikawa cells.
   p < 0.05, SF3A3+si‐c‐FOS group as compared with the SF3A3+si‐NC group;
   (n = 3 independent experiments). e) colony formation assays in KLE and
   Ishikawa cells. p < 0.05, SF3A3+si‐c‐FOS group as compared with the
   SF3A3+si‐NC group; (n = 3 independent experiments). f) Changes in
   cisplatin sensitivity following SF3A3 overexpression in KLE and
   Ishikawa cells. p < 0.05, SF3A3+si‐c‐FOS group as compared with the
   SF3A3+si‐NC group; (n = 3 independent experiments). g) Percentage of
   apoptotic cells analyzed by flow cytometry. p < 0.05, SF3A3+si‐c‐FOS
   group as compared with the SF3A3+si‐NC group; (n = 3 independent
   experiments). h,i) Percentage of apoptotic cells analyzed by TUNEL.
   Scale bars = 20 µm. p < 0.05, SF3A3+si‐c‐FOS group as compared with the
   SF3A3+si‐NC group; (n = 5 samples each group). j) Immunoblotting
   analysis of c‐FOS, Bcl2, and Bax protein levels to investigate whether
   c‐FOS can rescue the loss of SF3A3 function in terms of apoptosis and
   proliferation. p < 0.05, SF3A3+si‐c‐FOS group as compared with the
   SF3A3+si‐NC group; (n = 3 independent experiments). Data are presented
   as mean ± S.D. from at least three independent experiments.

   To determine whether SF3A3's oncogenic effects rely on c‐FOS
   regulation, we performed rescue experiments. c‐FOS knockdown
   counteracted cell growth and apoptosis induced by SF3A3 overexpression
   (Figure [134]4c). Additionally, c‐FOS knockdown reversed SF3A3‐induced
   proliferation and clonogenic activity (Figure [135]4d,e). Importantly,
   cisplatin resistance and apoptosis mediated by SF3A3 overexpression
   were reduced when c‐FOS was silenced (Figure [136]4f–i). Immunoblotting
   analysis revealed that c‐FOS knockdown significantly decreased Bcl2
   levels and increased Bax levels in SF3A3‐overexpressing cells
   (Figure [137]4j and Extended Data Figure [138]4d,e). These findings
   revealed a previously uncharacterized SF3A3–c‐FOS axis that drove EC
   progression and chemoresistance.

2.5. PEITC Anti‐Tumor Effects are Mediated by SF3A3 Inhibition

   Given the oncogenic role of SF3A3 in EC, we sought to identify a
   small‐molecule inhibitor of SF3A3. Our previous research suggests that
   PEITC may be a potential inhibitor of SF3A3^[ [139]^22 ^]. Further
   verification of the binding between PEITC and SF3A3 was conducted via
   Surface plasmon resonance (SPR) and mass spectrometry (Figure [140]5a).
   SPR experiments confirmed a strong binding interaction between SF3A3
   and PEITC, with a K[D] value of 10.1 µm (Figure [141]5b,c). Mass
   spectrometry further revealed that SF3A3‐overexpressing samples
   exhibited significantly higher PEITC content than controls (Extended
   Data Figure [142]5a,b).

Figure 5.

   Figure 5
   [143]Open in a new tab

   PEITC inhibits EC cell proliferation by targeting SF3A3 and cell
   apoptosis. a) Schematic of the experimental protocol. verification of
   the binding between PEITC and SF3A3 via SRP and mass spectrometry. b)
   Coomassie brilliant staining and immunoblotting of SF3A3 protein. c)
   Surface plasmon resonance (SPR) analysis demonstrating the binding
   interaction between SF3A3 protein and the small molecule PEITC. K[D] =
   10.1 µm. d) Inhibitory effects of PEITC on human EC cell proliferation,
   measured using a CCK‐8 assay. (n = 3 independent experiments). e)
   Quantification of KLE and Ishikawa cell viability over multiple days of
   PEITC treatment, as assessed by a CCK‐8 assay. p < 0.05, as compared
   with the Control group; (n = 3 independent experiments). f) Colony
   formation assays showing the effects of PEITC treatment on KLE and
   Ishikawa cells cultured for 10 days. p < 0.05, as compared with the
   control group; (n = 3 independent experiments). g) The change of PEITC
   on EC proliferation with SF3A3 overexpression (PEITC 10 µm).
   p < 0.05, SF3A3 group as compared with the NC group; (n = 3 independent
   experiments). h) Changes in cisplatin efficacy after combined treatment
   with cisplatin (3 µm) and PEITC (10 µm) for 48 hours. p < 0.05, PEITC+
   cis group as compared with the cis group; (n = 3 independent
   experiments). i) Flow cytometry analysis of apoptosis in KLE and
   Ishikawa cells treated with PEITC, stained with Annexin V‐PE/7‐AAD to
   measure apoptosis, necrosis, and cell death. p < 0.05, as compared with
   the Control group; (n = 3 independent experiments). j,k) TUNEL staining
   of KLE and Ishikawa cells following PEITC treatment. Scale bars =
   20 µm. p < 0.05, as compared with the Control group; (n = 5 samples
   each group). l, Immunoblot analysis of c‐FOS, BAX, and Bcl2 expression
   levels in KLE and Ishikawa cells after PEITC treatment. p < 0.05, as
   compared with the Control group; (n = 3 independent experiments). Data
   are presented as mean ± S.D. from at least three independent
   experiments.

   The anti‐tumor activity of PEITC was evaluated in normal endometrial
   cells and EC cell lines, revealing that cells with high SF3A3
   expression (KLE and Ishikawa) were more sensitive to PEITC, as
   indicated by lower IC[50] values (Figure [144]5d). PEITC treatment
   induced a dose‐ and time‐dependent decrease in cell viability and
   colony formation in KLE and Ishikawa cells (Figure [145]5e,f and
   Extended Data Figure [146]5c). Cells overexpressing SF3A3 exhibited
   enhanced sensitivity to PEITC (Figure [147]5g and Extended Data
   Figure [148]5d). Moreover, the combination of PEITC and cisplatin
   effectively reduced cisplatin resistance induced by SF3A3
   overexpression (Figure [149]5h). PEITC treatment significantly
   increased apoptosis in KLE and Ishikawa cells, as demonstrated by
   Annexin V‐PE/7‐AAD staining, TUNEL assays, and reduced expression of
   c‐FOS and the anti‐apoptotic protein Bcl2, accompanied by an increased
   Bcl2/Bax ratio (Figure [150]5i–l and Extended Data Figure [151]5 e,f).
   These results indicated that the anti‐EC effects of PEITC were
   mediated, at least in partly, by SF3A3 inhibition.

2.6. PEITC shows Tumor Suppression in Xenograft Tumor Growth and Patient
Derived Tumor Like Cell Clusters (PTCs) in EC

   To confirm whether PEITC's in vitro anti‐tumor effects translated in
   vivo, we used a mouse xenograft model. Nude mice were subcutaneously
   injected with KLE cells, and when tumors reached ∼100 mm^3, the mice
   were treated with si‐SF3A3 or PEITC administered orally every 3 days
   until day 35. Tumor load analysis showed that PEITC and siRNA
   significantly reduced tumor volume and weight without affecting body
   weight (Figure [152]6a–c, Extended Data Figure [153]6a,b).
   Immunofluorescence assays revealed decreased expression of c‐FOS and
   the proliferation marker Ki67 in PEITC‐treated tumors compared to
   controls (Extended Data Figure [154]6c). These findings confirm that
   PEITC effectively suppresses tumor growth in vivo by targeting SF3A3.

Figure 6.

   Figure 6
   [155]Open in a new tab

   Effect of SF3A3 and PEITC on EC xenograft growth in mice in EC samples.
   a) Experimental protocol. Groups of athymic BALB/c nude mice were
   injected subcutaneously with 10^6 KLE cells. After the tumor volume
   reaches 100 mm, 5 nm si‐SF3A3, si‐NC or 16 mg kg^−1 PEITC was injected
   every 3 days via the intratumoral injection or intragastric
   administration route. On day 35, animals were sacrificed and tumors
   samples were collected for analysis. (n = 5 samples each group). b)
   Body weight change. c) Tumor photographs. d) % PEITC release from the
   hydrogel versus time (h) at 37 °C. e) Plasma concentration versus time
   profiles of PEITC following single administration. Red
   color‐PEITC‐loaded hydrogel (Hypo Hypodermic injection) green
   color‐Free PEITC (oral). (n = 6 samples each group). f) Schematic
   representation of the anti‐tumor mechanism of PEITC encapsulated in
   hydrogel. (n = 5 samples each group). Once tumors reached ∼100 mm^3,
   mice were treated with PEITC encapsulated in hydrogel (administered
   once weekly) or PEITC via oral gavage (every 3 days). g–n) (g,k) Body
   weight changes in mice during the experiment. (h,l) Tumor volumes
   measured at designated time points. (i,m) Photographs of excised tumor
   xenografts. (j,n) Tumor weights at endpoint. p < 0.05, as compared with
   the Control group; (n = 5 samples each group).

   To improve the therapeutic efficacy of PEITC, we encapsulated it in a
   hydrogel delivery system. Hydrogel‐based systems offer an effective
   platform for drug delivery due to their high water content, excellent
   biocompatibility, minimal cytotoxicity, and superior drug encapsulation
   capacity.^[ [156]^26 ^] Drug metabolism revealed almost 65% of PEITC
   was released within 12 h (Figure [157]6d). In vivo pharmacokinetic
   analysis and biodistribution studies demonstrated that
   hydrogel‐encapsulated PEITC (PEITC‐Hydrogel) resulted in more stable
   plasma concentrations (Figure [158]6e) and lower PEITC accumulation in
   major organs such as the liver and kidneys at 2 and 4 h post‐injection,
   compared to free PEITC formulations (Extended Data Figure [159]6e).
   These results demonstrate the feasibility of the PEITC‐Hydrogel for the
   sustained release of PEITC at EC tumor tissue.

   Using xenograft models with KLE cells (EC) and T24 cells (bladder
   cancer), we compared the effects of hydrogel‐encapsulated PEITC
   (PEITC‐Hydrogel) with oral administration. PEITC‐Hydrogel demonstrated
   superior tumor suppression in both xenograft models, significantly
   reducing tumor volume and weight with no observed difference
   (Figure [160]6f–n, Extended Data Figure [161]6a–d). Over time, the in
   situ hydrogels underwent gradual biodegradation and were fully resorbed
   within 14 days, accompanied by a mild inflammatory response (Extended
   Data Figure [162]6f), which was similar to previous report.^[ [163]^26
   ^] Toxicological assessments (16 mg kg^−1 PEITC‐Hydrogel) further
   confirmed that the PEITC‐Hydrogel group exhibited reduced hepatorenal
   toxicity (Extended Data Figure [164]6g,h). Collectively, these findings
   underscore the therapeutic potential of PEITC‐loaded hydrogels to
   enhance anti‐tumor efficacy while reducing systemic toxicity and dosing
   frequency.

   To better explore the prospects of PEITC in clinical treatment, we
   constructed PTCs, and the expression of SF3A3 showed higher in EC tumor
   tissues than normal tissues[165] ^26 (Figure [166]7a–c, Extended Data
   Figure [167]7a). To assess the genomic fidelity of PTCs, we isolated
   genomic DNA from both PTC cells and matched primary tumor tissues and
   performed whole exome sequencing (WES). The results demonstrated a high
   degree of genomic concordance, supporting the use of PTCs as a reliable
   preclinical model for downstream drug screening applications
   (Figure [168]7d, Extended Data Figure [169]7b,c). Interestingly, PEITC
   showed significant efficacy with a low concentration and c‐FOS
   inhibitor T‐5224 inhibited the survival rate of PTCs at high
   concentrations (Figure [170]7e,f). Then, the average cell area was
   measured at different PEITC and T‐5224 concentrations, and it was found
   that PEITC and T‐5224 inhibited cell size in a dose‐dependent manner
   (Figure [171]7e,f). We first plotted relative inhibition rates as a
   function of PEITC concentration with increasing T‐5224 dosage (Extended
   Data Figure [172]7d). We next compared the synergy summary scores from
   the different matrix sizes via Chou Talalay method (Figure [173]7g,
   Extended Data Figure [174]7d). The combination treatment of PEITC and
   T‐5224 showed significant synergistic effects at PEITC 3 µM in
   combination with T‐5224 30, 100, or 200 µm (Figure [175]7g,h).
   Immunofluorescence assays revealed decreased expression of c‐FOS and
   the proliferation marker Ki67 in PEITC‐treated PTCs compared to
   controls (Figure [176]7i). The above research indicates that there is a
   significant synergistic effect between SF3A3 inhibitors and downstream
   c‐FOS inhibitors, significantly increasing the drug use scenarios and
   efficacy of PEITC in EC. These findings suggest that targeting
   SF3A3–c‐FOS signaling with PEITC and T‐5224 may provide a promising
   therapeutic strategy for EC patients (Figure [177]7j).

Figure 7.

   Figure 7
   [178]Open in a new tab

   Effect of PEITC on PTCs constructed in EC samples. a) PTCs construction
   and drug treatment mode diagram. b) Immunoblotting analysis of SF3A3 in
   EC tissues. p < 0.05, as compared with the Control group; (n = 3
   independent experiments). (n = 3 samples each group). c) Images from
   different periods during the PTCs construction process. Scale bars =
   50 µm. (n = 3 samples each group). d) Scatter plots display the
   correlation between the DNA variant allele frequency (VAF) of EC
   organoids and their parental tumors. The figures show the linear
   regression line (black) and the corresponding r and P values. e,f) The
   therapeutic concentration gradient of drug c‐FOS inhibitor T‐5224 and
   PEITC in PTCs. p < 0.05, T‐5224 (200300400 µm) group as compared with
   the DMSO group; (n = 3 samples each group). g) Drug synergy plots for
   PEITC and the c‐FOS inhibitor‐T‐5224. Each data point indicates an
   average synergy score from a single dose–response matrix experiment
   performed. The Chou Talalay method was used to calculate the
   synergistic effect of drugs: the effects of drugs were divided into
   three parts: the effects of PEITC and T‐5224 drugs alone (E1 and E2)
   and the synergistic effect of PEITC and T‐5224 drugs (E1+E2); Calculate
   Composite Index (CI): Use the formula CI = (E1+E2) / (E1 * E2) to
   evaluate synergistic effects. If the CI is less than 1, it indicates
   that the combination of drugs has a synergistic effect; If CI is equal
   to 1, it means there is no synergistic effect; If CI is greater than 1,
   it indicates an antagonistic effect. h) Images of PEITC and T‐5224
   treatment for 48 h in PTCs from EC patient. Scale bar = 50 µm.
   p < 0.05, as compared with the (PEITC+T‐5224‐0 µM) group; (n = 3
   samples each group). i) Representative IHC images showing Ki67, c‐FOS
   expression and HE staining after PITEC treatment in PTCs from EC
   patient. p < 0.05, as compared with the Control group; (n = 3 samples
   each group). Scale bars = 50 µm. Data are presented as mean ± S.D. from
   at least three independent experiments. j) Schematic diagram
   illustrating the mechanism by which SF3A3 influences EC progression
   through c‐FOS regulation. Data are presented as mean ± S.D. from at
   least three independent experiments.

3. Discussion

   Endometrial cancer (EC) remains a significant clinical challenge, with
   a rising global incidence and limited effective treatments for advanced
   or recurrent cases. Although alternative splicing (AS) has emerged as a
   critical regulatory mechanism in cancer, its role in EC progression and
   chemoresistance remains poorly understood.^[ [179]^18 , [180]^27 ^] In
   this study, we identify SF3A3 as a key oncogenic splicing factor in EC
   and demonstrate that its overexpression drives tumor growth, enhances
   cisplatin resistance, and suppresses apoptosis. Mechanistically, SF3A3
   promotes the alternative splicing of c‐FOS, a proto‐oncogene implicated
   in cancer progression, leading to increased full‐length c‐FOS
   expression and upregulation of the anti‐apoptotic Bcl2/Bax ratio.
   Furthermore, we identify PEITC as a direct SF3A3 inhibitor, which
   effectively suppresses tumor growth and enhances cisplatin sensitivity.
   Importantly, the combination of PEITC and the c‐FOS inhibitor T‐5224
   demonstrated synergistic effects in patient‐derived tumor‐like cell
   clusters (PTCs). Our findings position SF3A3 as a novel therapeutic
   target and establish a clinically relevant strategy for improving EC
   treatment.

   Dysregulation of AS is a hallmark of cancer, enabling tumor cells to
   evade apoptosis, enhance proliferation, and develop drug resistance.^[
   [181]^9 , [182]^11 , [183]^27 , [184]^28 ^] In EC, AS has been
   increasingly recognized as a key contributor to tumor progression;
   however, the specific mechanisms and functional consequences of these
   aberrant splicing events remain poorly defined.^[ [185]^29 , [186]^30
   ^] Several core splicing factors—such as SF3B1, SRSF2, and U2AF1—have
   been implicated in the pathogenesis of both hematologic and solid
   tumors by altering the splicing of critical oncogenes and tumor
   suppressors. Among these, SF3B1, a well‐characterized component of the
   U2 snRNP complex, has been reported to recognize 3′ splice sites and
   promote EC progression by regulating KSR2 mRNA maturation.^[ [187]^30 ,
   [188]^31 ^] SF3B1 mutations are also frequently observed in uveal
   melanoma, with hotspot mutations detected in ≈20% of cases.^[ [189]^31
   ^]

   In contrast, SF3A3, another core member of the U2 snRNP complex, has
   remained largely underexplored in EC. Most existing studies on SF3A3
   have focused on its aberrant expression across various cancer types,
   but its functional role and mechanistic contributions to
   tumorigenesis—particularly in EC—have not been previously
   characterized. Our study fills this important gap by identifying SF3A3
   as a novel oncogenic regulator in EC, functioning through direct
   regulation of c‐FOS alternative splicing, which promotes tumor
   proliferation and chemoresistance. Notably, recent studies have begun
   to reveal broader oncogenic functions of SF3A3 in other cancers. For
   example, SF3A3 was shown to synergize with YTHDF2 to drive
   hepatocellular carcinoma progression,^[ [190]^32 ^] and to selectively
   regulate MYC‐driven splicing and metabolic reprogramming in breast
   cancer, where SF3A3 levels modulate MYC‐induced plasticity and
   oncogenesis.^[ [191]^21 ^] In non‐small cell lung cancer, circSCAP can
   interact with SF3A3 to suppress malignancy, and in bladder cancer,
   E2F6/KDM5C promotes SF3A3 expression through a hypomethylated promoter
   region.^[ [192]^19 , [193]^20 ^] More recently, structural studies by
   Shi et al. revealed that the separator helix of SF3A3 inserts into the
   gap between pre‐mRNA and U2 snRNA, stabilizing the pre‐mRNA–snRNA
   complex and supporting its core splicing function.^[ [194]^18 ,
   [195]^33 ^] Together, these findings, along with our work in EC,
   highlight SF3A3 not only as a structural component of the spliceosome,
   but also as a dynamic regulator of cancer‐specific splicing programs,
   with distinct oncogenic effects depending on tissue context. Our study
   is the first to demonstrate the direct functional role of SF3A3 in EC
   progression, thus providing novel mechanistic insight and laying the
   groundwork for future translational exploration.

   Through RIP‐seq, we identified c‐FOS, a proto‐oncogene and core
   component of the AP‐1 complex, as a direct target of SF3A3‐mediated
   alternative splicing.^[ [196]^34 ^] SF3A3 promoted the inclusion of
   exons 2 and 3 in c‐FOS pre‐mRNA, generating the full‐length isoform
   that activates the Bcl2/Bax signaling pathway, thereby enhancing tumor
   survival and cisplatin resistance.^[ [197]^23 , [198]^24 , [199]^34 ^]
   Quantitatively, SF3A3 overexpression led to a 1–2‐fold increase in
   c‐FOS protein, a 1–2 fold increase in the Bcl2/Bax ratio, and a
   2–3‐fold increase in cisplatin IC[50], consistent with enhanced
   chemoresistance. In contrast, SF3A3 knockdown reversed these effects,
   sensitizing cells to cisplatin.^[ [200]^35 ^] These findings
   demonstrate that SF3A3 exerts its oncogenic and chemoprotective effects
   primarily through c‐FOS regulation, positioning SF3A3 as a key driver
   of endometrial cancer progression and a potential therapeutic target to
   overcome chemotherapy resistance. c‐FOS knockdown confirmed that SF3A3
   exerted its oncogenic effects primarily through c‐FOS regulation,
   further solidifying its role as a key driver of EC progression.

   Given the oncogenic role of SF3A3 in EC, we sought to identify a
   small‐molecule inhibitor capable of targeting SF3A3‐driven
   tumorigenesis. Using bioinformatics screening and biophysical
   validation, we identify PEITC as a direct SF3A3 inhibitor. PEITC has
   established anti‐tumor activity in lung, breast, bladder, and
   colorectal cancers.^[ [201]^36 , [202]^37 , [203]^38 , [204]^39 ,
   [205]^40 ^] Surface plasmon resonance (SPR) and mass spectrometry
   analysis confirmed a strong binding interaction between PEITC and
   SF3A3. Importantly, PEITC restored cisplatin sensitivity in
   SF3A3‐overexpressing EC cells, underscoring its potential as a
   dual‐function therapy that both inhibited tumor growth and reversed
   drug resistance. Pharmacokinetic studies show that PEITC has high oral
   bioavailability, ranging from 90% to 114% at doses of 10–100 µm kg^−1
   in rats.^[ [206]^40 , [207]^41 , [208]^42 ^] However, frequent dosing
   and low solubility limit its clinical utility. To address these
   challenges, we developed a hydrogel‐based delivery system for sustained
   PEITC release. Hydrogel has a porous 3D network structure and can be
   used as the carrier for controlled‐release extracellular vesicle (EVs)
   to enable the sustained release of drug at injury sites.^[ [209]^43 ,
   [210]^44 ^] In this study, we used a hydrogel to encapsulate PEITC,
   enabling site‐specific delivery, reduced dosing frequency, and improved
   bioavailability. Pharmacokinetic and biodistribution studies
   demonstrated stable plasma PEITC concentrations and reduced hepatic and
   renal accumulation compared to free PEITC, thereby minimizing
   off‐target toxicity. Moreover, in vivo xenograft models using both KLE
   (endometrial cancer) and T24 (bladder cancer) cells confirmed that
   PEITC‐hydrogel achieved superior tumor suppression with reduced
   systemic toxicity.

   To further validate the therapeutic potential of SF3A3 inhibition, we
   employed a patient‐derived tumor‐like cell cluster (PTC) model, which
   recapitulates patient tumor heterogeneity and drug response.^[ [211]^44
   , [212]^45 ^] PEITC exhibited strong anti‐tumor activity in PTCs,
   particularly when combined with the c‐FOS inhibitor T‐5224,
   demonstrating synergistic effects in reducing tumor cell viability.
   This combination strategy offers a clinically relevant approach to
   improve treatment efficacy and overcome resistance in EC patients.^[
   [213]^45 ^] These results collectively demonstrate that PEITC's
   anti‐tumor efficacy is not solely due to general cytotoxicity, but is
   driven by its specific inhibition of the SF3A3–c‐FOS oncogenic axis.
   Our hydrogel‐based approach provides a clinically translatable strategy
   for targeted delivery, enhanced safety, and improved therapeutic index
   of SF3A3 inhibitors such as PEITC.

   While our study establishes SF3A3 as a key oncogenic driver in EC and
   identifies PEITC as a promising therapeutic inhibitor, several
   limitations should be acknowledged. First, we primarily focused on the
   SF3A3–c‐FOS axis, yet it is likely that other SF3A3‐regulated splicing
   events also contribute to EC progression. Comprehensive
   transcriptome‐wide analysis are needed to uncover additional downstream
   targets and delineate their functional roles. Second, although
   patient‐derived tumor‐like cell clusters (PTCs) offer a robust and
   physiologically relevant model, the number of PTC samples used in this
   study was limited due to stringent tissue quality requirements and
   extended culture timelines. This limitation may constrain the
   generalizability of our findings and warrants validation in larger
   patient cohorts. Third, while PEITC has entered phase II clinical
   trials ([214]NCT00691132), its anti‐tumor mechanisms are multifaceted.
   In addition to its direct binding to SF3A3, PEITC has been shown to
   modulate other signaling pathways, including HSF1 (heat shock
   transcription factor 1) and Nrf2 (nuclear factor erythroid 2‐related
   factor 2),^[ [215]^46 ^] both of which play important roles in cellular
   stress responses. Although our binding studies support a direct
   interaction between PEITC and SF3A3, potential off‐target effects
   cannot be fully excluded. Further investigation into PEITC's
   selectivity, as well as combination strategies that enhance specificity
   and therapeutic efficacy, will be critical for advancing SF3A3‐targeted
   therapy into clinical application.

   Moving forward, our findings suggest that targeting the SF3A3–c‐FOS
   axis may synergize with existing therapies, including immune checkpoint
   inhibitors and targeted agents. Optimizing PEITC‐based combination
   strategies and hydrogel‐mediated drug delivery will be crucial for
   advancing SF3A3‐targeted therapy toward clinical application. Beyond
   the scope of EC, SF3A3 dysregulation has been observed in multiple
   malignancies. These findings suggest that targeting SF3A3 may hold
   therapeutic value in other chemoresistant tumors, warranting further
   investigation. In summary, our study highlights SF3A3 as a potential
   biomarker and therapeutic target in EC. It provides a strong foundation
   for future research aimed at developing precision therapies for
   endometrial cancer.

4. Conclusion

   This study identifies SF3A3 as a critical oncogenic driver in EC. Its
   overexpression promotes tumor growth, cisplatin resistance, and
   apoptosis suppression through regulation the alternative splicing of
   c‐FOS. We further validate PEITC as a potent SF3A3 inhibitor that
   effectively suppresses tumor progression and restores cisplatin
   sensitivity. Hydrogel‐based delivery of PEITC significantly enhances
   its therapeutic efficacy by improving release control and reducing
   dosing frequency. PTC models further confirm the clinical potential of
   this approach. Together, these findings provide a strong preclinical
   rationale for targeting SF3A3 as a novel therapeutic strategy and pave
   the way for future translational research in EC.

5. Experimental Section

Human Tissue Samples

   EC specimens were collected from April 2022 to February 2024 February
   in the First Hospital of Lanzhou University. The EC samples were
   obtained from patients with EC who had not undergone any previous
   surgery or chemotherapy. Fresh tissue samples were collected within 2 h
   of surgery and were sliced to 5 mm^3 and respectively immersed in
   100 µL of TRIzol reagent (Thermofisher, #15 596 018) and protein lysate
   containing phosphatase inhibitors and protease inhibitors. The tissue
   samples were stored at −80 °C. All patients provided informed consent,
   and ethical approval was granted by the Ethics Committee for Clinical
   Research (Drugs and Devices) of Lanzhou University First Hospital
   (LDYYLL2024‐142).

   The inclusion criteria for clinical samples include histological
   diagnosis of endometrial cancer, age ≥ 18 years, sufficient tumor
   tissue for analysis, no prior chemotherapy or radiotherapy before
   tissue collection, and the ability to provide informed consent. The
   exclusion criteria mainly include other concurrent malignant tumors,
   insufficient quality or quantity of tissue samples, previous treatments
   that may affect tissue integrity (such as neoadjuvant therapy), and
   inability to provide informed consent.

Cell Culture and Reagents

   The HEC‐1A, hEEC, Ishikawa and KLE cell lines were purchased from the
   China Center of American Type Culture Collection (ATCC, Wuhan, China).
   HEC‐1A cells were maintained in McCoy's 5A, hEEC and Ishikawa cells
   were maintained in DMEM and KLE cells were maintained in DMEM/F12
   medium containing 10% fetal bovine serum (FBS) at 37 °C in a humidified
   atmosphere with 5% CO[2]. PEITC was obtained from Merck and PEITC's
   10 mM liquor configuration: PEITC was resuspended in 10% DMSO, 40%
   PEG300, 5% Tween‐80 and 45% phosphate‐buffered saline (PBS) for
   experiments. Dilute 10 mm PEITC with PBS to 5‐10‐15 µm for relevant
   experiments, and use a mixed solvent without PEITC with the same ratio
   as 15 µm for the control group The remaining reagents and antibodies
   are listed in the table (Tables [216]S1 and [217]S2, Supporting
   Information).

Construction of EC Cell Lines

   To achieve SF3A3 overexpression (OE), the aforementioned conditions
   were repeated with overexpression vectors (pLJM1‐ SF3A3 and pLJM1‐EGFP;
   Qizhen Lake, HangZhou, China). The viral infection diversity (MOI):
   Ishikawa = 50; KLE = 10; Next, the medium was exchanged for fresh DMEM
   or DMEM/F12 medium and the cells were cultured for an additional 24 h.
   Stable SF3A3 overexpression cell lines were selected by growth in a
   medium containing 10 µm puromycin. SF3A3 siRNA (si‐S F3A3 ‐1:
   5′–GACUGCUGCUUAUCAACAATT–3′ and si‐SF3A3 ‐2:
   5′–GCUGCUCGUGAUAUGCUGUTT–3′) and nontargeting siRNA (si‐NC) were
   purchased from Shanghai Gene Pharma Co., Ltd (Shanghai, China) (The
   relevant siRNA sequences are displayed in Table [218]S3, Supporting
   Information). Efficient si‐SF3A3 or overexpression of SF3A3 was
   verified by RT‐qPCR and immunoblotting. The efficiency percentages in
   siRNA are: qPCR KLE: si‐SF3A3‐1: 64.8% si‐SF3A3‐2: 69.5%; Ishikawa:
   si‐SF3A3‐1: 67.5% si‐SF3A3‐2: 55.2%. Immunoblotting: KLE: si‐SF3A3‐1:
   54.1% si‐SF3A3‐2: 70.3%; Ishikawa: si‐SF3A3‐1: 94,3% si‐SF3A3‐2: 88.0%.
   The efficiency percentages in overexpression are: RT‐qPCR: KLE: 455.7%
   Ishikawa: 741.6%; Immunoblotting: KLE: 228.4% Ishikawa: 482.4%.

Colony‐Forming Assay

   KLE and Ishikawa cells were seeded in 6‐well culture plates at 4000
   cells wel^l−1 and treated with PEITC (5‐10‐15 µm) for 10 days (wild
   type cells) or 14 days (si‐NC, si‐SF3A3‐1, si‐SF3A3‐2 or NC, SF3A3
   cells). Stain with 0.5% crystal violet (Solarbio Life Science, Beijing,
   China) for 15 min, wash off excess staining solution with PBS three
   times, take photos, and analyze the number of cell clones using Image
   J.

Colorimetric Cell Proliferation Assay

   Cells were seeded in 96‐well plates at 2000 cells well^−1 and incubated
   at 37 °C with 5% CO[2]. Cell proliferation was assessed using the Cell
   Counting Kit‐8 kit (CCK‐8; Beyotime, Jiangsu, China) according to the
   manufacturer's instructions. At 0, 24, 48, 72, 96, and 120 h, the
   medium was replaced with 100 µL of 10% CCK‐8 reagent in a serum‐free
   medium. Absorbance was measured after 1 h using a microplate reader
   (Molecular Devices, USA).

Flow Cytometry

   Flow cytometry analysis for cell apoptosis was performed using an
   Annexin V‐PE/7‐AAD Apoptosis Detection Kit (Yeasen Biotechnology,
   #40310ES20, Shanghai, China). 5 × 10^5 cells were digested with
   ethylenediaminetetraacetic acid (EDTA)‐free trypsin (Macgene, CC035)
   for 3 min, collected by centrifugation, washed with ice‐cold
   phosphate‐buffered saline (PBS), and resuspended at a density of 5 ×
   10^5 cells mL^−1 with 100 µL 1 × binding buffer. Then 5 µL Annexin V‐PE
   and 10 µL 7‐AAD were added and incubated for 10 min in the dark.
   Finally, cells were incubated with an additional 400 µL 1 × binding
   buffer and analyzed within 20 min by CytoFLEX S (Beckman Coulter Life
   Science). At least 1 × 10^4 cells were collected to determine the
   percentage of apoptotic cells.

   Annexin V‐PE and 7‐AAD are used with specific excitation and emission
   wavelengths: Annexin V‐PE (Ex = 488 nm, Em = 578 nm) emits orange‐red
   fluorescence and is detected via the FL2 channel, while 7‐AAD (Ex =
   546 nm, Em = 647 nm) emits red fluorescence and is detected via the FL3
   channel. Data analysis is performed using software to generate a
   two‐color dot plot with PE intensity on the x‐axis and 7‐AAD intensity
   on the y‐axis. A total of 10 000 cells are collected per sample. In
   typical experiments, three cell subpopulations are identified: live
   cells (low background fluorescence), early apoptotic cells (strong
   orange‐red fluorescence from PE), and late apoptotic cells (dual
   staining with PE and 7‐AAD). Cellular abundances were compared across
   distinct subgroups and validated subset differences using gating
   strategies in flow cytometry scatter plots, visualized with FlowJo
   software.

TUNEL Assays

   For TUNEL assays, KLE and Ishikawa cells were seeded in 12‐well culture
   plates at 1 × 10^6 cells well^−1 and incubated at 37 °C with 5% CO[2].
   Treated with PEITC (5‐10‐15 µm) for 2 days (wild type cells) or 2 days
   (si‐NC, si‐SF3A3‐1, si‐SF3A3‐2 or NC, SF3A3 cells) respectively. Fixed
   with 4% paraformaldehyde and the TUNEL Cell Apoptosis Detection Kit
   (Beyotime, C1098) was used according to the manufacturer's protocol.
   Cell nuclei were stained with DAPI. Confocal images of cells were
   sequentially acquired with a Nikon A1 Ti system. TUNEL positive cells
   are evaluated based on fluorescence intensity as a percentage of total
   cells.

Immunoblotting

   Si‐SF3A3 or overexpression of SF3A3 KLE and Ishikawa cells and cells
   were incubated with PEITC for 48 h were processed for immunoblotting.
   Primary antibodies against SF3A3 (ab157194) and a secondary
   FITC‐conjugated goat anti‐rabbit IgG (H+L) antibody (AS024) were
   purchased from Abclonal Biotechnology (Abclonal, Wuhan, China).
   Horseradish peroxidase (HRP)‐conjugated anti‐mouse IgG (AS061) was
   purchased from Abclonal Biotechnology (Abclonal, Wuhan, China) and
   anti‐rabbit IgG secondary antibodies (4414S) were purchased from Cell
   Signaling Technology (Beverly, MA, USA) (1 h). GoldBand 3‐color Low
   Range Protein Marker (Yeasen Biotechnology, #20350es, Shanghai, China
   and Vazyme Biotech, #MP102‐01, Nanjing, China). Protein bands were
   visualized using enhanced chemiluminescence reagents (Millipore,
   Burlington, MA).

RNA Binding Protein Immunoprecipitation Assay (RIP)‐seq

   RIP is a technique for studying the binding of intracellular RNA to
   proteins. 10 million KLE (pLJM1‐EGFP or pLJM1‐SF3A3) cells were
   collected and lysed in 500 µL RIP buffer (20 mM Tris‐HCl pH 7.4, 100 mm
   KCl, 1.5 mm MgCl[2], 0.2 mm EDTA, 1% NP‐40, 100U RNase inhibitor, 1 ×
   cocktail). Whole‐cell extracts were incubated with 30 µg GFP‐beads
   (AlpalifeBio, # ktsm1301) overnight for 3 h at 4 °C. Immunocomplexes
   were washed with washing buffer for 3 times (20 mM Tris‐HCl pH 7.4,
   100 mm KCl, 1.5 mm MgCl2, 0.2 mM EDTA), treated with proteinase K, and
   RNA was extracted using henol/chloroform/isoamyl alcohol (125:24:1).
   RNA library standard: Total RNA content>1 µg,
   concentration>20 µg µL^−1; OD[260/280]: 1.7‐2.5; OD[260/230]: 0.5‐2.5,
   RIN>6; 28s/18s:1‐5. RNA Clean reads were aligned to the hg38 genome
   using bowtie2 (v2.5.1). Both input and RIP samples were prepared for
   next‐generation sequencing by Genomic consistency analysis conducted by
   Beijing Biomarker Technologies Co. Ltd. (Beijing, China). RIP seq raw
   data are uploaded to the GEO database, GEO Submission (GSE287381) [NCBI
   tracking system #24 990 495].

RNA Sequencing Data Processing and Analysis

   Raw sequencing data were processed using fastpv0.23.4 to filter out
   low‐quality and low‐complexity reads and remove adapter sequences.^[
   [219]^47 ^] Cleaned reads were aligned to the hg38 reference genome in
   paired‐end mode using bowtie2 (v2.5.1).^[ [220]^48 ^] The resulting BAM
   files were converted to bigWig format using bam Coverage from deep
   tools (v3.5.3)^[ [221]^49 ^] for chromosome coverage visualization.
   Custom R scripts were utilized to generate chromosome coverage plots.

   Metagene plots were created using compute Matrix and plot Profile
   functions from deep tools (v3.5.3). Peak calling was performed with
   MACS2 (v2.2.9.1)^[ [222]^50 ^] using the call peak function, and highly
   reproducible peaks across replicates were identified using IDR
   (v2.0.4.2).^[ [223]^51 ^] After quality control checks, data from
   replicates were merged to generate a combined set of peaks for each
   group. The overlap and intersection of peaks between cell lines were
   assessed using Venn diagrams generated by the intersect function in bed
   tools (v2.27.1).^[ [224]^52 ^] Motif analysis was conducted with MEME
   (v5.0.2),^[ [225]^53 ^] restricting motif length to 6–10 bp and
   retaining motifs with significant E‐values (< 10^−3). Peak annotation,
   including overlapping gene locations, was performed with bed tools
   (v2.27.1) and reference gene annotation data from encode regions.
   Functional annotation of genes was done using the DAVID online database
   [226]https://david.ncifcrf.gov/; last accessed October 2024), focusing
   on cell cycle, apoptosis regulation‐related genes. Genomic distribution
   of peaks was visualized using IGV (v2.16.2).^[ [227]^54 ^]

   rMATS (RNA‐MATE Analysis Tool for Splicing) is a bioinformatics tool
   designed specifically for RNA splicing difference analysis
   ([228]http://rnaseg‐mats.sourceforge.net/index.html). There are five
   main splicing modes: exon skipping, mutually exclusive exons,
   alternative 5′splice site, alternative 3′ splice site, and intron
   retention. In terms of quantitative analysis, mats calculate the PSI
   value (percent spliced in) of each event based on transcript covered
   splice site reads and exon covered reads to quantify the proportion
   change of splicing isoforms. At the same time, the software uses
   nonparametric statistical tests (such as Wilcoxon rank sum test) to
   evaluate the significance of differences between conditions, and
   combines multiple hypothesis correction (such as Benjamini‐Hochberg
   method) to control the false discovery rate, so as to ensure the
   reliability of the results.

Hybridization Chain Reaction (HCR)

   On the first day, KLE and Ishikawa (pLJM1‐EGFP or pLJM1‐SF3A3) samples
   were fixed with 4% paraformaldehyde (Servicebio, G1101) and washed with
   PBST 5 times for 5 min. Hybridization buffer (50% formamide, 5 × sodium
   chloride sodium citrate, 9 mm citric acid pH = 6.0, 0.1% Tween‐20,
   50 µg mL^−1 heparin, 1 × Denhardt's solution, 10% dextran sulfate) and
   primers were added, primers were showed in Table [229]S6 (Supporting
   Information) and samples were incubated at 37 °C overnight. Second day:
   washing with 300 µL probe wash buffer (50% formamide, 5 × sodium
   chloride sodium citrate, 9 mm citric acid pH = 6.0, 0.1% Tween‐20,
   50 µg mL^−1 heparin) 3 times for 15 min at 37 °C, followed by washing
   with 1 mL5 × sodium chloride sodium citrate and 0.1% Tween‐20 at room
   temperature 3 times for 15 min. Then amplification was performed by
   incubating samples with amplification buffer (5 × sodium chloride
   sodium citrate, 0.1% Tween‐20, 10% dextran sulfate) and amplifier
   (SangonBiotech, China) at room temperature overnight. Third day:
   washing with 1 mL 5 × SSCT (0.75 m NaCl, 0.075 m Sodium citrate, 0.1%
   Tween‐20) 5 min for 2 times, then washing with 1 mL 5 × SSCT 30 min for
   2 times, then washing with 1 mL 5 × SSCT for 5 min. Then samples were
   imaged with a Nikon A1 Ti system. All assays were repeated at least
   three times.

Protein Mass Spectrometry Detection of PEITC and SF3A3

   10 million KLE (pLJM1‐EGFP or pLJM1‐SF3A3) cells were collected and
   lysed in 500 µL IP buffer (10 mm Tris‐HCl pH 7.5;150 mm NaCl;0.5 mm
   EDTA;0.1% SDS;1% Triton X‐100;1% Deoxycholate). Whole‐cell extracts
   were incubated with 30 µg GFP‐beads (AlpalifeBio, ktsm1301) for 3 h at
   4 °C. Immunocomplexes were washed with washing buffer (50 mM Tris‐HCl
   pH 7.5;150 mm NaCl;1 mM EDTA) for 3 times (20 mM Tris‐HCl pH = 7.4,
   100 mm KCl, 1.5 mm MgCl2, 0.2 mM EDTA), Then used for co‐incubation
   with PEITC (15 µM) respectively, followed by washing with washing
   buffer three times and validation using mass spectrometry experiments.
   Divided into a control group and an SF3A3 overexpression group.

   MS data were acquired using electrospray ionization in positive ion
   mode, m/z was 186‐130.1. Other MS settings include: sheath gas
   temperature 250 °C, sheath gas flow 11 L/min, Nebulizer 20 psi,
   Capillary 3000 V, Nozzle voltage 1500 V, gas temperature 200 °C. Raw
   data were processed using QQQ Quantitative Analysis (Quant‐My‐Way)
   (Agilent) for peak detection, alignment and integration.

Surface Plasmon Resonance

   The binding of PEITC to purified human SF3A3 protein (General biol,
   Anhui, China) was measured by SPR using a Biacore 8K system (Cytiva,
   America), which is from the Analysis Center of Agrobiology and
   Environmental Sciences, Zhejiang University. Briefly, the SF3A3 protein
   was immobilized on a CM7 sensor chip (Cytiva) to a level of 30 520.7
   response units (RUs) using Biacore (8K). For steady state affinity
   analysis, the PEITC was dissolved in 1 × PBS ‐P+ buffer containing 5%
   DMSO at concentrations of 0.39,0.78,1.56,3.125,6.25,12.5, 25 to 50 µm,
   and were run across the chip. Each sample that was bound to the surface
   was associated for 90s at a flow rate of 30 µL min^−1. Dissociation of
   sensor chips was performed for 90s. The dissociation constant (KD) was
   fitted and recorded by Biacore Insight Evaluation Software (Cytiva)
   using the One‐to‐One analysis model. All assays were repeated at least
   three times.

RNA Isolation, RT‐PCR, and RT‐qPCR

   Total RNA was isolated from cultured cells using TRIzol reagent
   (Thermofisher, #15 596 018), and cDNA was generated using PrimeScript
   RT Master Mix (Vazyme Biotech, #R333‐01, Nanjing, China) according to
   the manufacturer's protocol. The cDNA product was subjected to
   quantitative real‐time PCR (qPCR) using SYBR Premix Ex‐Taq (Tli RNaseH
   Plus) (Vazyme Biotech, #Q121‐02/03, Nanjing, China). Expression levels
   of target genes were normalized to those of the housekeeping gene
   actin, and data are present as mean ± SD. For RT‐PCR assays, cDNAs were
   amplified by 2 × Hieff Robust PCR Master Mix (Yeasen Biotechnology,
   #10106ES08, Shanghai, China), and the PCR products were then separated
   by DNA gel electrophoresis. The primer sequences used for RT‐qPCR and
   RT‐PCR are listed in Tables [230]S4 and [231]S5 (Supporting
   Information), respectively. RT‐qPCR and RT‐PCR primers were synthesized
   in Hangzhou Qingke Biotechnology and sequences are described in Tables
   [232]S4 and [233]S5 (Supporting Information). All assays were repeated
   at least three times.

PEITC(Hydrogel) Synthesis

   The PEG‐PLGA‐PEG triblock copolymer was synthesized via ring‐opening
   polymerization. Briefly, PEG (Mn = 1500, 4 g) was dried under vacuum at
   120 °C for 3 h to eliminate residual moisture. Subsequently, lactic
   acid (LA, 5 g), glycolic acid (GA, 0.5 g), and stannous octoate
   (0.05 g) were added as monomers and catalysts respectively. The mixture
   was stirred at 150 °C under an argon atmosphere for 24 h. The crude
   product was precipitated three times in deionized water, filtered,
   dialyzed, and lyophilized. Structural confirmation of the resulting
   copolymer was performed using ¹H NMR spectroscopy in CDCl₃ (Extended
   Data Figure 8a,b).

   Preparation of PLGA‐PEG‐PLGA‐@PEITC gel: To prepare the drug‐loaded
   hydrogel, 1 g of the synthesized PLGA‐PEG‐PLGA copolymer was dissolved
   in 5 mL of deionized water and dispersed via ultrasonication to obtain
   a homogeneous precursor solution. PEITC was then incorporated into this
   solution at a final concentration of 20 mg mL^−1. Upon warming to
   37 °C, the mixture underwent a sol–gel transition, forming the
   injectable PLGA‐PEG‐PLGA@PEITC hydrogel.

   The degradation of the PLGA‐PEG‐PLGA hydrogel is primarily governed by
   the hydrolysis of ester bonds within the PLGA segment. Water molecules
   cleave these bonds, progressively reducing the polymer into short‐chain
   oligomers of lactic acid and glycolic acid. Continued hydrolysis leads
   to the formation of monomeric units, which are metabolized into carbon
   dioxide and water, ultimately achieving complete biodegradation and
   mineralization.

Mouse Xenograft Experiments

   The mouse xenograft experiment obtained female thymus free BALB/c nude
   mice (5–6 weeks old, weight 18–22 g) from the Charles River (Beijing,
   China) group.

   KLE cells (WT or NC or SF3A3 cells) or T24 cells were collected and
   resuspended in serum‐free medium at a concentration of 10^6 cells mL^−1
   and subcutaneously injected into one side. After 14 days, the tumor
   volume reached≈100mm,^[ [234]^3 ^] and the animals were injected
   intraperitoneally with indicated doses of PEITC (16 mg/kg, every two
   days), PEITC(Hydrogel) (20 µL, 20 mg mL^−1, once a week), si‐SF3A3‐1
   (30 µL 5 nm, every two days) or vehicle (PBS or hydrogel). Every two
   days, all groups were used a vernier caliper to measure the length and
   width of the tumor and were used a balance to measure the weight of the
   mice. After three weeks of administration, euthanize the mice and
   collect tumor samples. Measure the tumor weight of the mice and divide
   them into two groups. One group is fixed in formalin for subsequent HE
   staining and immunofluorescence staining. HE staining is conducted by
   the Wuhan Servicebio Technology Co., Ltd (Wuhan, China). Another group
   is used for immunoblotting experiments; Tumor volume calculation
   formula: 1/2 * length * width * width; All animal experiments were
   conducted using double‐blind methods. All animal experiments were
   approved by the Ethical Review Committee of Laboratory Animal Welfare,
   Zhejiang University (ZJU20230106).

Immunofluorescence (IF) Staining

   Tumors were excised from the mice within 4 h of the final treatment
   dose and then fixed in 10% formalin for 24 h and stored in 70% ethanol
   in PBS at 4 °C. Tissues were sliced into 8 µm sections and stained with
   primary antibodies against SF3A3 (Cell Signaling Technology, #8057S) or
   Ki67 (Abclonal #A2094) followed by anti‐rabbit IgG secondary antibodies
   (Cell Signaling Technology, Beverly, MA, USA). Tissues were imaged
   using an inverted fluorescence microscope at 20 × magnification.

Culture of PTCs from Solid Tumors

   Studies on PTC were conducted by GeneX Precision Medicine Co., Ltd.
   (Zhejiang China). These EC specimens were collected at the first
   Hospital of Lanzhou University. Collected samples were conditioned in
   ice‐cold PBS with 10 mm HEPES and 100 U/mL penicillin‐streptomycin
   (Thermo Fisher Scientific). Tissues were washed with PBS at least five
   times. Necrotic areas and adipose tissue were removed as possible.
   Tissues were minced into small pieces and digested in 5 mL PBS/EDTA
   1 mm containing collagenase l, ll, and IV (Thermo Fisher Scientific)
   200U mL^−1, each, at 37 °C for 1 h, the digestion was pipetted every 15
   min to facilitate cell release. 40 µm filters were used to collect
   dissociated cells. After 10 min’ centrifugation (300 x g, 4 °C) cell
   pellets were resuspended in PTCs growth medium and seeded in
   low‐attachment‐surface dish at the concentration of 10^5 cells cm^−2.
   Cells were cultured in an incubator at 37 °C, 5% CO[2]. PTCs growth
   medium was refreshed every 2–3 days, as necessary.

Cell Viability Assay for PTCs

   For PTCs/spheres in a 96‐well plate, the medium was half removed and
   replaced with 80 µL Celltiter‐Gloreagent from a cell viability assay
   (Promega). Cells were incubated at room temperature for 1 h, and
   viability readings were obtained using a Synergy plate reader (BioTek).

Genomic Consistency of PTCs with the Original Tumors

   Genomic consistency analysis was carried out by Majorbio BioPharm
   Technology Co., Ltd. (Shanghai, China). Genomic DNA was extracted using
   the CTAB method, and its quality was assessed via 1% agarose gel
   electrophoresis to detect degradation or contamination. DNA
   concentration and purity were measured with a NanoDrop ND‐2000
   spectrophotometer. Only high‐quality DNA samples (OD[260/280] between
   1.8 and 2.0; OD[260/230] ≥ 2.0) were used for library preparation.
   Genomic DNA was randomly fragmented to 180–280 bp (Covaris,
   Massachusetts, USA). Fragment ends were blunted and A‐tailed before
   ligation with adapter oligonucleotides. Ligated fragments were enriched
   by PCR, indexed, and hybridized with biotin‐labeled probes. Exons were
   captured using streptavidin‐coated magnetic beads, further enriched by
   PCR, purified (Beckman Coulter, USA), quantified (Agilent Bioanalyzer
   2100), and sequenced on the DNBSeq T7 platform.

   Low‐quality reads (phred score < 20), adapter‐contaminated reads, and
   those with >10 ambiguous bases were filtered using Fastp. High‐quality
   reads were aligned to the reference genome using BWA‐MEM with default
   parameters. BAM files were sorted with Samtools, and PCR duplicates
   were marked using MarkDuplicates. Base quality recalibration was
   performed before variant calling. Germline SNPs and indels were
   identified using Sentieon's Haplotyper and GVCFTyper modules, while
   somatic variants were detected via Mutect2. Standard hard filtering
   criteria (as recommended by GATK Best Practices) were applied.
   Structural variants (SVs) were detected using Manta, which utilizes
   split reads, soft‐clipped alignments, insert sizes, and
   inter‐chromosomal mapping events. CNVs were identified using CNVkit's
   core‐window algorithm. All variants were annotated with the Variant
   Effect Predictor (VEP) and categorized by genomic location (e.g.,
   exonic, intronic, splicing, intergenic, and UTR regions).

Pharmacokinetic Analysis using UPLC‐MS/MS

   Pharmacokinetic analysis was conducted using ICR mice (equal numbers of
   males and females), which were randomly assigned to two groups. Group I
   received a single oral dose of PEITC dissolved in 1% DMSO (16 mg kg^−1
   body weight), while Group II was administered 20 µL of PEITC‐loaded
   hydrogel via subcutaneous injection, allowing in situ gel formation at
   the injection site. The blood samples were withdrawn at defined time
   points (0.5, 1, 1.5, 2, 2.5, 3, 4, 6, 8, 12, 16, and 24 h). Take 50 µL
   of serum and add 200 µL of acetonitrile to precipitate the protein.
   After vortex mixing, place the sample on an ice bath and let it stand
   for 10–15 min. Centrifuge the sample at 15 000 rpm for 10 min and
   collect the supernatant. It was determined by the UPLC‐MS/MS method and
   separated on the Waters ACQUITY UPLC HSS T3 chromatographic column
   (1.8 µm, 2.1 mm × 100 mm). The parent m/z was 164.0100, and the
   daughter m/z was 130.1900. The liquid chromatographic conditions were
   as Table [235]S8 (Supporting Information).

In Vivo Biodistribution Studies

   A biodistribution study was conducted using ICR mice (balanced by sex),
   randomly assigned to two groups. Group 1 received a single oral
   administration of PEITC (16 mg kg^−1) dissolved in 1% DMSO. Group 2 was
   subcutaneously injected with 20 µL of PEITC‐loaded hydrogel
   (20 mg mL^−1), which formed an in situ gel at the injection site. At 2
   and 4 h post‐administration, animals were euthanized via CO₂
   asphyxiation, and major organs—including the liver, kidneys, heart,
   lungs, and spleen—were collected. Issues were rinsed with ice‐cold PBS
   and preserved at −80 °C. Before analysis, samples were thawed, washed
   thoroughly with cold saline, and blotted dry. A 100 µg portion of each
   tissue was homogenized in 900 µL of acetonitrile to precipitate
   proteins. After vortexing, samples were placed in an ice bath for 10–15
   min, then centrifuged at 15,000 rpm for 10 min. The supernatant was
   collected for quantification. Tissue homogenates were further clarified
   by centrifugation at 15700 × g for 15 min at 4 °C. PEITC concentrations
   in the supernatants were measured via UPLC‐MS/MS, as detailed in the
   Pharmacokinetic Analysis section.

Biochemical Test to Assess Liver and Kidney Function

   Hepatotoxicity and nephrotoxicity across treatment groups were
   evaluated after a 21‐day study period by measuring serum biochemical
   markers. Levels of alanine aminotransferase (ALT), aspartate
   aminotransferase (AST), blood urea nitrogen (BUN), and creatinine (CRE)
   were quantified using commercially available assay kits, following the
   manufacturer's protocols.

Histopathological Assessment

   At the end of the study, mice were euthanized, and liver, kidney, and
   subcutaneous tissues were collected from each group. The samples were
   fixed in 10% neutral‐buffered formalin and processed for hematoxylin
   and eosin (H&E) staining, which was performed by Wuhan Servicebio
   Technology Co., Ltd. (Wuhan, China) following standard procedures.
   Pharmacokinetic, toxicological studies and animal care have been
   approved by the Ethics Committee for Experimental Animal Research,
   Zhejiang University of Technology. Animals receive appropriate
   treatment and are used in a scientifically effective and ethical manner
   (MGS20250423044).

Database Analysis

   The GEPIA and TGCA database was used to analyze tumor prognosis and the
   expression of SF3A3 in EC ([236]http://gepia.cancer;
   [237]https://portal.gdc.cancer.gov/). The UCSC Xena database was used
   to analyze prognosis ([238]https://xenabrowser.net/)). The drugbank
   database was used to predict SF3A3 inhibitors
   ([239]https://go.drugbank.com/).

Statistical Analysis

   Statistical analysis were performed using GraphPad Prism (version
   10.1.2). The relevant data were typeset and plotted by Adobe
   Illustrator. Results are presented as mean ± standard deviation (SD).
   Animal experiments were conducted independently at least 5 nude mice
   each group, and cell experiments were repeated at least three times.
   Differences between two groups were analyzed with independent t‐tests.
   For more than two groups, one‐way ANOVA followed by Tukey's post‐hoc
   test was used. Statistical significance was set at p < 0.05.

Conflict of Interest

   The authors declare no conflict of interest.

Supporting information

   Supporting Information
   [240]ADVS-12-e04184-s001.docx^ (10MB, docx)

Acknowledgements