PRKN mediates the ubiquitination of SCAF8 to reduce the mRNA stability of KLF5 and its transcriptional activation of EFNA3 in colorectal cancer Authors: Ye Song, Yangsheng Yu, Bo Zhang, Xiao Zhang Published: 2026 ABSTRACT BackgroundDysregulation in glycolysis within the tumor microenvironment is a hallmark in the context of colorectal cancer (CRC). We aimed to identify critical drivers of glycolysis in CRC and decipher the underlying mechanism.MethodsThrough combined gene interference in CRC cells using lentivirus and treatment with glycolysis activator DASA-58, we measured glycolytic flux to evaluate how molecular mechanisms influence tumor cell activity via glycolysis. CRC cells were cocultured with NK and CD8+ T cells to analyze the anti-tumor immune response. A patient-derived xenograft mouse model and a CMT93 cell-derived allograft mouse model were developed for analyzing CRC growth, metastasis, and immune evasion.ResultsEFNA3 was elevated in CRC tissues and cell lines. The glycolytic activity, proliferative, migratory, invasive, and pro-angiogenic effects of the CRC cells were reduced, and the anti-tumor response of the CD8+ T and NK cells was enhanced after EFNA3 knockdown in CRC cells. Tumors formed by CMT93 cells with EFNA3 knockdown exhibited reduced metastasis and an increased proportion of anti-tumor immune cells infiltrated. Overexpression of KLF5 promoted EFNA3 transcription, and SCAF8 enhanced KLF5 mRNA stability. Downregulation of PRKN in CRC enhanced the SCAF8 protein expression through ubiquitination. PRKN upregulation blocked glycolysis and suppressed the malignant behavior of CRC cells, whereas the anticancer ability of PRKN was reversed by SCAF8/KLF5/EFNA3 overexpression.ConclusionsActivating the E3 ubiquitin ligase activity of PRKN to suppress EFNA3 expression represents a promising therapeutic strategy for the clinical treatment of CRC.Graphical abstract Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s11658-026-00974-0. Introduction Colorectal cancer (CRC), whose major manifestation is a malignancy derived from the glandular epithelium of the colon (COAD) and rectum (READ), ranks third and second in incidence and cancer-related mortality, respectively, in the world [1]. Metabolic reprogramming is a key hallmark of tumors, and there is a metabolic shift from oxidative phosphorylation to aerobic glycolysis, which facilitates tumor immune evasion and the formation of a suppressive tumor immune microenvironment [2]. The tumor microenvironment (TME) includes the surrounding blood vessels, immune cells, fibroblasts, myeloid-derived inflammatory cells, various signaling molecules, the extracellular matrix, and immune cells, mainly encompassing macrophages, neutrophils, natural killer (NK) cells, dendritic cells, T lymphocytes, and B lymphocytes, which participate in the initiation and progression of tumors [3]. The Eph/Ephrin (EFN) family is key to many developmental processes and an important mediator of cell proliferation, differentiation, adhesion, migration, and death, and EFNA3, a member of this group, was elevated in pan-cancer and negatively correlated with the infiltration of CD4+ T cells, macrophages, and neutrophils in gastric cancer [4]. Not surprisingly, EFNA3 expression was correlated with overall survival and progression-free survival in patients with lung adenocarcinoma, and it can regulate the glycolysis of lung adenocarcinoma cells [5]. However, there remains a lack of clarity on the function and underlying mechanisms of EFNA3 in CRC. Our preliminary bioinformatics prediction revealed the negative correlation between EFNA3 and the infiltration of CD8+ T cells and NK cells in CRC and the transcriptional regulation of EFNA3 by Krüppel-like factor 5 (KLF5). KLF5, whose expression and activity have been modulated by transcriptional and posttranslational modifications, can regulate the expression of a substantial set of targets and participates in diverse cellular functions, such as stemness, proliferation, apoptosis, and migration [6]. More importantly, KLF5 inactivation reduced basal-like breast tumor growth in a CD8+ T-cell-dependent manner, and KLF5 loss in tumors increased the number and function of T cells [7]. Thus, we hypothesized that EFNA3, controlled by KLF5, regulated the glycolysis and CD8+ T cells- and NK cells-mediated immune evasion of CRC cells. RNA-binding proteins (RBPs), epigenetic regulators that play critical roles in post-transcriptional events, can influence CRC pathogenesis and progression by mediating various components of the TME [8]. SR-related and CTD-associated factor 8 (SCAF8, also known as RNA-binding motif protein 16) is an RNA polymerase II elongation factor whose RNA-binding clusters are usually located throughout the transcript, preventing cleavage of early mRNA transcripts [9]. The effect of SCAF8 on cancers has not been reported, and its protein expression was elevated in CRC samples with little alteration on the transcriptional level. The highly conserved ubiquitination enabled by E3 ubiquitin ligases is another imperative mechanism to control the expression of oncogenes or tumor suppressors in CRC [10]. Parkin (PRKN) was predicted to be the E3 ubiquitin ligase with downregulated expression in CRC. Indeed, the tumor suppressive potential of PRKN is related to its capacity to regulate cell proliferation, cell migration, angiogenesis, and apoptosis, and dysfunction of PRKN contributed to the loss of ubiquitination of cell cycle-related genes, leading to tumor progression [11]. In this study, we demonstrated the roles that EFNA3 played in CRC glycolysis and CD8+ T cells- and NK cells-mediated immune evasion. Importantly, we found that KLF5 induced EFNA3 transcription to accelerate glycolysis, which stimulates tumor immune evasion and metastasis. Furthermore, the tumor suppressor PRKN catalyzed the ubiquitination of SCAF8 to cause the instability of KLF5 mRNA. In summary, our study has provided a novel mechanism for tumor evasion and metastasis in CRC. Materials and methods Patient samples and clinical data collection Primary CRC and paracancerous normal tissues were collected from 34 patients who underwent surgery at the First Affiliated Hospital of Harbin Medical University for etiological research. Among them, 23 were male (14 READ, 9 COAD) and 11 were female (7 READ, 4 COAD). Additionally, fresh tumor tissue from a female patient (T3N1M1, no history of viral hepatitis) was collected for mouse xenograft experiments. All patients were first diagnosed with CRC, and they had complete clinical data and had not received preoperative radiotherapy. All patients provided written informed consent. The research was approved by the Research Ethics Committee of the First Affiliated Hospital of Harbin Medical University (approval no. 2022185, date 11 March 2022) and adhered to the tenets of the Declaration of Helsinki. Immunohistochemical staining (IHC) After routine deparaffinizing and hydration, paraffin-embedded tumor tissue sections were subjected to antigen retrieval by citrate buffer. The sections were subsequently incubated with 3% hydrogen peroxide for 30 min at room temperature (RT). After being blocked with 5% normal goat serum for 1 h, the sections were hatched with primary antibodies to EFNA3 (1:200, 12480-1-AP, ProteinTech Group, Chicago, IL, USA), PKM2 (1:400, 15822-1-AP, ProteinTech Group), CD31 (1:4000, 28083-1-AP, ProteinTech Group), CD34 (1:4000, 14486-1-AP, ProteinTech Group), KLF5 (1:200, 21017-1-AP, ProteinTech Group), SCAF8 (1:500, 18893-1-AP, ProteinTech Group), PRKN (1:500, 14060-1-AP, ProteinTech Group), LDHA (1:200, A18574, ABclonal, Wuhan, Hubei, China), CD8A (1:50, A11856, ABclonal), and CD56 (1:200, 82715-1-RR, ProteinTech Group) overnight and with HRP-labeled goat anti-rabbit IgG antibody (1:500, GB23303, Servicebio, Wuhan, Hubei, China) at RT for 1 h. Afterward, immunostaining was assessed against 3,3′-diaminobenzidine and hematoxylin. Mouse tumor samples were quantified by ImageJ to analyze the rate of positive cells stained. IHC staining of clinical samples was analyzed semi-quantitatively by observers who were unaware of the grouping based on the intensity of staining (negative: 0, weak: 1, moderate: 2, strong: 3) and the proportion of positive cells (< 25%: 1, 25–50%: 2, 50–75%: 3, > 75%: 4). The final score was the product of the two scores (0–12 points). Cell culture and treatment Human colonic mucosal epithelial cells (HCMECs, CP-H040) and human rectal mucosal epithelial cells (HRMECs, CP-H041) were procured from Procell (Wuhan, Hubei, China). Human COAD cell lines DLD-1 (Serial: TCHu134, RRID: CVCL_0248) and LoVo (Serial: TCHu 82, RRID: CVCL_0399), READ cell line SW1463 (Serial: SCSP-5230, RRID: CVCL_1718), and mouse rectum carcinoma CMT93 (Serial: TCM43, RRID: CVCL_1986) were acquired from the National Collection of Authenticated Cell Cultures (Shanghai, China). Another READ cell line, RCM-1 (Serial: CBP60041, RRID: CVCL_1648), was purchased from Cobioer (Nanjing, Jiangsu, China). All cells have been authenticated by STR before use and were free from bacterial, fungal, or mycoplasma contamination. Cells at passage 2–3 were used for the experiment. The short hairpin RNA (shRNA) lentivirus (based on pHBLV-U6 vector) for the knockdown of EFNA3, KLF5, and SCAF8 and the overexpression lentivirus (based on pHBLV-CMV vector) for PRKN, SCAF8, KLF5, and EFNA3 were obtained from Hanbio (Shanghai, China), and the viral titer was 108 TU/mL. CRC cells were infected at a multiplicity of infection (MOI) of 5 and screened for stable cell lines by puromycin or G418. The sequence information is provided in Table 1. In CRC cells stably knocked down for EFNA3 via a lentiviral vector, cells were treated with 5 μM glycolysis activator DASA-58 (T6816, TargetMol, Shanghai, China) for 48 h [12], with cells treated with the solvent DMSO serving as controls.Table 1shRNA sequenceFor humanSequencesh-EFNA3 no. 1CGTGAACGACTATCTGGATATCTCGAGATATCCAGATAGTCGTTCACGsh-EFNA3 no. 2CCTTGTGGCTCTGGTAATGTTCTCGAGAACATTACCAGAGCCACAAGGsh-KLF5 no. 1CCCTGAGTTCACCAGTATATTCTCGAGAATATACTGGTGAACTCAGGGsh-KLF5 no. 2CCTATAATTCCAGAGCATAAACTCGAGTTTATGCTCTGGAATTATAGGsh-SCAF8 no. 1GCCAGCAGCTTCAACAATTAACTCGAGTTAATTGTTGAAGCTGCTGGCsh-SCAF8 no. 2TATTCCCTGAATGACTATAAACTCGAGTTTATAGTCATTCAGGGAATAScramble-shRNACCTAAGGTTAAGTCGCCCTCGCTCGAGCGAGGGCGACTTAACCTTAGG Table 1shRNA sequenceFor humanSequencesh-EFNA3 no. 1CGTGAACGACTATCTGGATATCTCGAGATATCCAGATAGTCGTTCACGsh-EFNA3 no. 2CCTTGTGGCTCTGGTAATGTTCTCGAGAACATTACCAGAGCCACAAGGsh-KLF5 no. 1CCCTGAGTTCACCAGTATATTCTCGAGAATATACTGGTGAACTCAGGGsh-KLF5 no. 2CCTATAATTCCAGAGCATAAACTCGAGTTTATGCTCTGGAATTATAGGsh-SCAF8 no. 1GCCAGCAGCTTCAACAATTAACTCGAGTTAATTGTTGAAGCTGCTGGCsh-SCAF8 no. 2TATTCCCTGAATGACTATAAACTCGAGTTTATAGTCATTCAGGGAATAScramble-shRNACCTAAGGTTAAGTCGCCCTCGCTCGAGCGAGGGCGACTTAACCTTAGG shRNA sequence RNA extraction and RT‑qPCR analysis Trizol reagent (Thermo Fisher Scientific Inc., Waltham, MA, USA) was used to extract total RNA from cell samples. Extracted total RNA was analyzed by reverse transcription and fluorescence quantitative PCR using SweScript One-Step RT-PCR Kit (G3335-50, Servicebio). The reverse-transcription quantitative PCR (RT–qPCR) data were analyzed using the 2−△△Ct method, with β-actin serving as an internal reference. The primer sequences for amplification were as follows: forward 5'- AGTTCTCGGAGAAGTTCCAGCG-3′ and reverse 5′-CAGCAGACGAACACCTTCATCC-3′ for EFNA3; forward: 5′-GGAGAAACGACGCATCCACTAC-3′, reverse: 5′-GAACCTCCAGTCGCAGCCTTC-3′ for KLF5; forward: 5′-AGGAGACAGTGGTCACAACCCA-3′, reverse: 5′-TCGACACAGGAAATGCTGGTGG-3′ for SCAF8, forward: 5′- CCAGAGGAAAGTCACCTGCGAA-3′, reverse: 5′-CTGAGGCTTCAAATACGGCACTG-3′ for PRKN, and forward: 5′-CCACCATGTACCCTGGCATT-3′, reverse: 5′- GTCCTCGGCCACATTGTGAA-3′ for β-actin. Total protein extraction and Western blot Cell or tissue samples on ice were lysed with ice-cold radioimmunoprecipitation assay (RIPA) buffer (Beyotime, Shanghai, China) containing protease and phosphatase inhibitors. After quantification by the BCA method, 30 μg of proteins were separated on 6–10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and transferred to polyvinylidene fluoride (PVDF) membranes (Millipore Corp, Billerica, MA, USA). The membranes were sealed in 5% bovine serum albumin (BSA) for 1 h and incubated overnight at 4 °C with the corresponding primary antibodies to EFNA3 (1:2000, 12480-1-AP, ProteinTech Group), PCNA (1:10000, 10205-2-AP, ProteinTech Group), CD31 (1:4000, 28083-–1-AP, ProteinTech Group), KLF5 (1:5000, 21017-1-AP, ProteinTech Group), PRKN (1:1000, 14060-1-AP, ProteinTech Group), LDHA (1:1000, A0861, ABclonal), PKM2 (1:2000, 15822-1-AP, ProteinTech Group), SCAF8 (1:5000, 18893-1-AP, ProteinTech Group), β-actin (1:3000, GB15003-100, Servicebio), His (1:5000, ab213204, Abcam, Cambridge, UK), Ubiquitin (Ub) (1:1000, ab134953, Abcam), and then with secondary antibody HRP-labeled goat anti-rabbit IgG (1:10,000, GB23303, Servicebio) at RT for 2 h. Finally, the protein signal was examined with the assistance of a chemiluminescence reagent kit. Metabolic assays The extracellular acidification rate (ECAR) was evaluated on a Seahorse XFe24 analyzer (Agilent Technologies, Santa Clara, CA, USA) using the Seahorse XF Glycolytic Stress Test Kit (103020-100, Agilent Technologies). Glucose, oligomycin, and 2-deoxyglucose (2-DG) were added to cells (10,000 cells/well) seeded in 24-well plates at the indicated time points. Oxygen consumption rate (OCR) in the cells was assessed on a Seahorse XFe24 analyzer using the Seahorse XF Cellular Mitochondrial Stress Test Kit (103015-100, Agilent Technologies). Oligomycin, FCCP, Antimycin A, and Rotenone were added to cells (10,000 cells/well) seeded in 24-well plates at the indicated time points. Glucose uptake levels The glucose uptake fluorometric assay kit (E-BC-F041, Elabscience Biotechnology Co., Ltd., Wuhan, Hubei, China) was used to assess the ability of CRC cells to consume glucose. Briefly, CRC cells were seeded in 96-well cell culture plates (3000 cells/well) and starved in a serum-free cell culture medium overnight. The 2-DG (10 mmol/L) was added to the cell culture plate and incubated for 30 min at 37 °C. After a 30-min incubation of the working solution at 37 °C to convert the probe into fluorescent material, the fluorescence values were read in a microplate reader (excitation wavelength 530 nm, emission wavelength 590 nm), and the glucose uptake was calculated from the standard curve of the standards. Analysis of lactic acid content For CRC cells, 5 × 104 cells were seeded into a 96-well plate and cultured for 24 h. Subsequently, cell supernatants were collected for assay. For tumor tissues, after routine homogenization using saline (0.9% NaCl solution), centrifugation was performed at 4 °C, 10,000 × g for 10 min, and the supernatant was placed on ice for measurement. Lactic acid content in tumor tissues or cells was assessed using the l-Lactic Acid Colorimetric Assay Kit (E-BC-K044-M, Elabscience) by reading the OD at 530 nm with the microplate reader, and all operations were performed according to the manufacturer’s protocol. Colony formation assay Stably infected CRC cells were cultured for 2 weeks after inoculation into six-well plates at 500 cells/well. After 2 weeks, the cells were fixed by adding 1 mL of 4% paraformaldehyde (PFA) for 30 min and stained with crystal violet for 30 min. The number of colonies formed was counted after photographing. Cell counting kit-8 (CCK-8) Stable infected CRC cells were seeded at a density of 2000 cells/well into 96-well plates and routinely incubated until designated time points (0 h, 24 h, 48 h, and 72 h). Subsequently, 10 μL of CCK-8 reagent (C0038, Beyotime) was added to each well, followed by continued incubation for 2 h. Cell proliferation was assessed by measuring absorbance at 450 nm. EdU incorporation assay Cell-Light EdU Apollo643 In Vitro Kit (C10310-2, Guangzhou RiboBio Co., Ltd., Guangzhou, Guangdong, China) was used to assess DNA synthesis activity in CRC cells. Briefly, stable CRC cells were seeded into 96-well plates (1 × 105 cells/well), and cells were labeled for 2 h using a 50 μM EdU solution. After fixation in 4% PFA for 30 min, the cells were permeabilized with 0.5% Triton X-100 for 10 min and stained with Apollo 643 staining solution for 30 min at RT in the dark. The nuclei were stained with Hoechst 33,342. The proportion of EdU-positive cells was calculated after photographing using a fluorescence microscope (Olympus, Japan). Chicken chorioallantoic membrane (CAM) assay To detect the angiogenic effects induced by CRC cells, a culture supernatant of CRC cells routinely cultured for 48 h was collected as a conditioned medium (CM). CAM assay was performed as previously described [13]. Briefly, a window of approximately 1.0 cm in diameter was made in the eggshell on day 6 of fertilization to expose the CAM, and 100 μL of CM from CRC cells was added. Eggs were placed in an incubator at 37 °C and 80–90% relative humidity for 2 days. Subsequently, CAMs were fixed with fixative (methanol:acetone = 1:1) for 15 min and photographed, and the number of blood vessels was counted. Transwell assays BeyoGold Transwell Permeable Supports (8 μm pore size, FTW076, Beyotime) were used to analyze the migratory and invasive potential of CRC cells, and Matrix-Gel (C0376, Beyotime) was used to encapsulate the Transwell chambers during the invasion assay. A culture medium (500 µL) containing 10% FBS was added to the 24-well plate, and the Transwell was placed inside the 24-well plate with forceps. CRC cells resuspended in serum-free medium (1 × 106 cells/mL, 200 µL) were seeded in the upper chamber of transwell inserts. After a 48-h incubation, the Transwell chamber was fixed with 4% PFA for 30 min. After being stained with crystal violet for 10 min, migrated and invaded cells were observed and counted under a microscope. Detection of the killing effects of cytotoxic lymphocytes Cytotoxic lymphocytes, including CD8+ T cells and NK cells, were isolated from healthy PBMCs using the CD8+ T Cell Isolation Kit (130-096-495, Miltenyi biotec, Bergisch Gladbach, Germany) and NK Cell Isolation Kit (130-092-657, Miltenyi Biotec), respectively. CRC cells were seeded in 96-well plates at 1 × 105 cells/well, and NK cells and CD8+ T cells were added to CRC cells at an effector cell to target cells (E:T) of 1:1 [14] and 5:1 [15] for 24-h coculture. Harvested supernatants were used to measure the levels of cytokines released by immune cells using Granzyme B (GzmB) ELISA Kit (E-EL-H1617, Elabscience) and interferon gamma (IFN-γ) ELISA Kit (E-EL-H0108, Elabscience). Tumor cells were stained using the LIVE–DEAD Cell Imaging Kit (R37601, Invitrogen Inc., Carlsbad, CA, USA), where Calcein-AM stained the live cells in green, and BOBO-3 iodide stained the dead cells in red. The cells were observed under a fluorescence microscope, and the survival rate (%) = green cells/(green cells + red cells) was calculated. Animal studies All animal studies were approved by the Animal Welfare Ethics Committee of the First Affiliated Hospital of Harbin Medical University (approval no. 2023139, date 6 February 2023) and in accordance with the Basel Declaration. The Animal Welfare Ethics Committee of the First Affiliated Hospital of Harbin Medical University strictly adheres to the principles and guidelines established by the International Council for Laboratory Animal Science (ICLAS) to ensure that our animal research meets international ethical standards. Female C57BL/6 mice were purchased from HFK Bioscience (Beijing, China). The 16-week-old female huHSC-B-NDG hIL15 mice expressing the human immune system were purchased from Biocytogen (Beijing, China). All the animals involved were housed in a specific pathogen-free environment at appropriate temperatures. Fresh in situ tumor tissue from patients with CRC was minced into microtissues smaller than 150 μm in diameter and mixed with Matrigel before being implanted subcutaneously into huHSC-B-NDG hIL15 mice [16]. After injection, tumor volume was measured every week: volume = π × length × width × height/6. When tumor volume reached approximately 150 mm3, 15 mice were randomly divided into three groups and received intratumoral injections of either EFNA3-knockdown lentivirus (sh-EFNA3 no. 1, no. 2) or control lentivirus (Scramble-shRNA). Tumor volume was measured weekly thereafter. After 4 weeks, mice were euthanized via intraperitoneal injection of an overdose of sodium pentobarbital (150 mg/kg), and tumor tissue was collected. For tumor growth experiments, 1 × 106 stably infected CMT93 cells [17] were subcutaneously injected into the right ventral side of C57BL/6 mice (n = 5/each group). At 4 weeks after tumor inoculation, mice were euthanized by intraperitoneal injection of excess sodium pentobarbital (150 mg/kg), and tumor tissues were collected. A CRC experimental liver metastasis model was constructed by splenic injection [18, 19]. After the C57BL/6 mice were anesthetized by inhalation through 2% isoflurane, a longitudinal incision was made in the left axillary line just below the posterior border of the rib cage to locate the spleen from the abdomen. CMT93 cells (1.5 × 106) suspended in 100 μL of sterile PBS were injected into the spleen. The spleen was excised 5 min after tumor cell injection to prevent splenic tumor formation. Mice were euthanized 21 days after tumor injection, and liver tissue was collected. Liver metastases were detected by hematoxylin–eosin (HE) staining using ImageJ software, and the area occupied by metastatic foci was divided by the total surface area. Flow cytometry Tumor tissues were minced and digested in PBS containing 0.5 mg/mL collagenase IV and 0.25 mg/mL DNase I for 30 min at 37 °C. The cell suspension was filtered through a 70-μm cell filter and resuspended in 1% BSA. FITC-coupled anti-mouse CD49b Antibody (11-5971-82, Invitrogen) and APC-coupled anti-mouse NKG2D Antibody (17-5882-82, Invitrogen) were added for 3-h incubation at RT in the dark to detect the content of murine activated NK cells. Human NK cells were detected using FITC-conjugated anti-human CD56 antibody (11-0566-42, Invitrogen) and APC-conjugated anti-human NKG2D antibody (17-5878-42, Invitrogen). To detect infiltrating activated CD8+ T cells in tumors, cells were stained with FITC-coupled CD8 (11-0032-82 or 11-0088-42, Invitrogen) for surface antigen staining. Following fixation and permeabilization, cells were stained with APC-conjugated GzmB antibodies (17-8898-82 or GRB05, Invitrogen). To detect the infiltration of active M1 macrophages, cells were fixed and permeabilized following surface antigen staining with an anti-human CD86 antibody conjugated to an APC (17-0869-42, Invitrogen) and stained with an anti-human CD68 antibody conjugated to FITC (11-0689-42, Invitrogen). To detect the infiltration of CD4+ T cells and regulatory T cells (Tregs) within them, cells were stained for surface antigens using FITC-conjugated anti-human CD3 antibody (11-0037-42, Invitrogen) and APC-conjugated anti-human CD4 antibody (MHCD0405, Invitrogen). The cells were fixed and permeabilized, followed by staining with PE-conjugated anti-human FOXP3 antibody (12-4777-42, Invitrogen). Staining was immediately detected by CytExpert software. Immunofluorescence staining The cells were simultaneously fixed and permeabilized by ice-cold 90% methanol and blocked by 5% goat serum. The cells were incubated with anti-LDHA (1:1000, A18574, ABclonal) and anti-PKM2 (1:2000, 15822-1-AP, ProteinTech Group) overnight at 4 °C, and with Alexa Fluor 488-labeled goat anti-rabbit IgG (1:500, GB25303, Servicebio) at RT in the dark for 1 h. The cell nuclei were labeled by incubating with DAPI working solution (1 µg/mL) at RT in the dark for 30 min. Staining was observed under a fluorescence microscope, and the mean fluorescence intensity (MFI) was measured by ImageJ. Chromatin immunoprecipitation (ChIP) assays The Pierce Magnetic ChIP Kit (26,157, Thermo Fisher) was used to analyze the binding of KLF5 to the EFNA3 promoter fragment. Briefly, CRC cells were fixed in 1% PFA for 10 min at RT. Chromatin was subsequently fragmented to 200–500 bps by micrococcal nuclease combined with an ultrasonic processor and incubated overnight with a primary antibody to KLF5 (1:20, 21017-1-AP, ProteinTech Group) or IgG at 4 °C. Endogenous DNA protein complexes were precipitated using Protein A/G magnetic beads. Samples were then washed and de-crosslinked. DNA fragments were recovered and purified, and the KLF5-pulled-down EFNA3 promoter fragment was quantified using qPCR. Dual-luciferase assays The EFNA3 upstream promoter fragment (chr1: 155078644-155078970) was queried from the UCSC Genome Browser (https://genome.ucsc.edu/index.html). The wild-type (WT) EFNA3 promoter fragment or the mutant (MT) promoter fragment lacking the KLF5 binding site was inserted into the pGL6 firefly luciferase plasmid (D2102, Beyotime) in the polyclonal site to construct a WT/MT reporter plasmid driven by the EFNA3 promoter. The above plasmids were transfected into CRC cells by Lipofectamine 2000 (Thermo Fisher) together with the pRL-TK endogenous gene plasmid (D2760, Beyotime), which carries the Renilla luciferase. After 48 h, the assay was performed using the Dual-Luciferase Reporter Gene Assay Kit (RG027, Beyotime). RNA-immunoprecipitation (RIP) The RIP kit (Bes5101, BersinBio, Guangzhou, Guangdong, China) was used to evaluate the binding capacity of SCAF8 protein to KLF5 mRNA. Briefly, 1 × 107 cells were lysed on ice by supplementing the lysis buffer with protease inhibitor and RNase inhibitor. DNA was removed from cell lysates by incubation with DNase salt stock, DNase (10 U) at 37 °C for 10 min. The cells were incubated with anti-SCAF8 (1:20, 18893-1-AP, ProteinTech Group) at 4 °C overnight. Protein A/G beads were added and incubated at 4 °C for 1 h to capture protein–RNA complexes. RNA was eluted from the magnetic beads by polysome elution buffer, and the enriched level of KLF5 mRNA was detected by RT–qPCR after the collection of purified RNA. mRNA or protein stability assay CRC cells (3 × 105 cells/well) in six-well plates were treated with 5 μg/mL of Actinomycin D (S8964, Selleck, Houston, TX, USA), and the cells were lysed and RNA extracted at the indicated time points (0 h, 4 h, 8 h, and 12 h). KLF5 mRNA levels were detected by RT–qPCR. CRC cells (3 × 105 cells/well) in six-well plates were prepared at the indicated time points (0 h, 4 h, 8 h, and 12 h) after the addition of 10 μM cycloheximide (CHX, S7418, Selleck) to prepare protein lysates, and protein expression of SCAF8 was evaluated using western blot analysis. Co‑immunoprecipitation assay (Co-IP) SCAF8 mutant protein plasmids (Flag-SCAF8K119R) with K119 mutated to arginine (R119), and recombinant tagged protein plasmids Flag-SCAF8 and His-PRKN were obtained from Servicebio. The above plasmids were cotransfected into 293 T cells (GNHu17, National Collection of Authenticated Cell Cultures) by Lipofectamine 2000 (Thermo Fisher), and the Ubiquitin protein (Abcam) was added. After 48 h, the cells were lysed on ice with a lysis buffer containing protease inhibitors (Beyotime). Flag antibody for co-IP (ab205606, Abcam) was added to the samples, and the cell lysate/antibody mixtures were gently spun on a spinner at 4 °C overnight. The protein complexes were captured by rotary incubation at 4 °C for 2 h. The beads were rinsed with IP lysis buffer and added to the loading buffer. Finally, the lysed immune complexes were separated by SDS–PAGE and analyzed by western blot. Quantification and statistical analysis At least five independent experiments were used for data analysis. Statistical analyses and graphing were performed using GraphPad Prism 10.4.2 (GraphPad, San Diego, CA, USA). Nonlin fit was used to calculate the half-life of the mRNA or protein, and Fisher’s exact test was used for columnar table analysis. The Shapiro–Wilk test or the Kolmogorov–Smirnov test was used to analyze whether the data conforms to a normal distribution or not. For nonparametric tests, the Wilcoxon matched-pairs signed rank test was used for the analysis of paired two sets of data, and the Mann–Whitney test for the analysis of data from two unpaired groups. Spearman’s rank correlation was applied to examine the correlation between the two variables. For parametric tests, the unpaired t-tests were employed for comparisons between the two groups, and multiple group comparisons were performed by one-way or two-way analysis of variance (ANOVA), followed by Sidak’s multiple comparisons test or Tukey’s multiple comparisons test. Statistical significance was denoted by p < 0.05. Results EFNA3 is associated with glycolytic activity and immune evasion in tumors of patients with CRC We first downloaded the top 1000 differential expression genes (DEGs) in COAD and READ (ranked according to adjusted p-value) from GEPIA (http://gepia.cancer-pku.cn/index.html) and cross-screened them. A total of 772 common DEGs were found ( Additional file 1:Supplementary Fig. S1A). To search for potential therapeutic targets for glycolysis in CRC, we then downloaded glycolysis-related human gene sets from The Molecular Signatures Database (https://www.gsea-msigdb.org/gsea/msigdb/index.jsp), including WP_AEROBIC_ GLYCOLYSIS.v2023.2.Hs, KEGG_GLYCOLYSIS_GLUCONEOGENESIS.v2023.2.Hs, HALLMARK_GLYCOLYSIS.v2023.2.Hs. Glycolysis-related genes were cross-screened against DEGs in CRC, and there were 17 intersecting genes: ADH1B, MET, CENPA, KIF20A, AURKA, CHPF, HMMR, CDK1, SOX9, POLR3K, EFNA3, KDELR3, SLC2A1, NSDHL, ALG1, DEPDC1, and TGFBI (Additional file 1:Supplementary Fig. S1B). The prognostic significance of the expression of these 17 genes on disease-free survival (DFS) in CRC (COAD + READ) patients was analyzed in GEPIA with p < 0.05 as the cut-off threshold. The expression of CHPF, EFNA3, and SLC2A1 was closely associated with DFS in patients, and all three of them predicted lower DFS in patients ( Additional file 1:Supplementary Fig. S1C). The effects of CHPF [20] and SLC2A1 [21] on CRC glycolysis have been previously reported, whereas it is not clear whether EFNA3 is involved in CRC glycolysis. Analysis in GEPIA informed that EFNA3 was highly expressed in CRC ( Additional file 1:Supplementary Fig. S1D). The correlation between EFNA3 expression and immune cell infiltration in tumor tissues of patients with CRC was evaluated using the QUANTISEQ algorithm in Sangerbox 3.0 (http://vip.sangerbox.com/). EFNA3 expression was negatively correlated with the infiltration of CD8+ T cells, NK cells, Tregs, and M1 macrophages, and positively correlated with the infiltration of CD4+ T cells in CRC ( Additional file 1:Supplementary Fig. S1E). Cytotoxic lymphocytes (including NK cells and CD8+ T cells) play an important role in tumor surveillance [22]. The presence of cytotoxic lymphocytes is a relevant prognostic factor in CRC [23], and the infiltration of NK cells and CD8+ T lymphocytes in the CRC microenvironment has been shown to improve the prognosis of patients with CRC [24]. However, whether EFNA3 mediates immune evasion by inducing insufficient cytotoxic lymphocyte infiltration in CRC through the glycolytic pathway requires further analysis. We first assessed the expression of EFNA3 in clinical samples by IHC staining and observed that EFNA3 expression was significantly elevated in CRC tissues (Fig. 1A). Patients with CRC were categorized into the EFNA3High (n = 17) and EFNA3Low groups (n = 17) based on the mean value of the EFNA3 IHC score (6.82). There was an insignificant difference in the male-to-female ratio (Fig. 1B) of patients and the tumor site (Fig. 1C) in the two cohorts. However, patients with EFNA3High have more advanced TNM staging (Fig. 1D). The expression of glycolytic enzymes LDHA and PKM2 was assayed in CRC tissues, and it was observed that glycolytic activity was significantly enhanced in EFNA3High patients (Fig. 1E). In addition, EFNA3High tumors had significantly attenuated IHC scores for CD8A (CD8+ T-cell marker) and CD56 (NK-cell marker), representing an immunosuppressive TME (Fig. 1F). Enhanced immunoreactivity for CD31 and CD34 and increased angiogenesis were observed in EFNA3High CRC tumor tissues (Fig. 1G).Fig. 1Correlation of EFNA3 expression with clinical features of CRC. A IHC detection of EFNA3 expression in CRC tumor tissues and adjacent tissues (n = 34). Differences in clinical parameters, including sex (B), tumor site (C), and TNM stage (D), in patients with CRC with different EFNA3 expression. IHC analysis of differences between the expression of glycolytic markers LDHA and PKM2 (E), immune cell markers CD8A (CD8+ T-cells) and CD56 (NK-cells) (F), and vascular markers CD31 and CD34 (G) in tumor tissues of patients with CRC with different EFNA3 expression. Data were presented as mean ± standard error of the mean (SEM). Wilcoxon matched-pairs signed rank test (A) was used for the analysis of paired two sets of data, and Mann–Whitney test (E–G) for analysis of data from two unpaired groups. Fisher’s exact test B–D was used for columnar table analysis Correlation of EFNA3 expression with clinical features of CRC. A IHC detection of EFNA3 expression in CRC tumor tissues and adjacent tissues (n = 34). Differences in clinical parameters, including sex (B), tumor site (C), and TNM stage (D), in patients with CRC with different EFNA3 expression. IHC analysis of differences between the expression of glycolytic markers LDHA and PKM2 (E), immune cell markers CD8A (CD8+ T-cells) and CD56 (NK-cells) (F), and vascular markers CD31 and CD34 (G) in tumor tissues of patients with CRC with different EFNA3 expression. Data were presented as mean ± standard error of the mean (SEM). Wilcoxon matched-pairs signed rank test (A) was used for the analysis of paired two sets of data, and Mann–Whitney test (E–G) for analysis of data from two unpaired groups. Fisher’s exact test B–D was used for columnar table analysis EFNA3 affects CRC cell growth and immune evasion through the glycolytic pathway EFNA3 expression was analyzed in HCMECs, HRMECs, and CRC cell lines, and the results again showed the high expression of EFNA3 in CRC (Additional file 1: Supplementary Fig. S2A, 2B). Representative DLD-1 and RCM-1 cell lines were selected for subsequent experiments. CRC cells with stable knockdown of EFNA3 were developed using lentiviral infection of two shRNAs, and western blot assays detected that the expression of LDHA and PKM2 was also significantly reduced after the knockdown of EFNA3 (Fig. 2A, Additional file 1: Supplementary Fig. S2C). The aerobic glycolysis of the cells was assessed using ECAR, and the knockdown of EFNA3 significantly reduced glycolysis and glycolytic reserve in CRC cells (Fig. 2B, Additional file 1: Supplementary Fig. S2D). Moreover, the knockdown of EFNA3 resulted in a significant elevation of basal respiration and respiratory capacity (Fig. 2C, Additional file 1: Supplementary Fig. S2E) and inhibited glucose uptake and lactic acid release in CRC cells (Supplementary Fig. S2F).Fig. 2Knockdown of EFNA3 inhibits DLD-1 cell activity and enhances immune cell-mediated tumor killing by blocking the glycolytic pathway. A The EFNA3, LDHA, and PKM2 protein expression in DLD-1 cells infected with two shRNAs targeting EFNA3 was examined by western blot assays. B Glycolysis and glycolytic reserve in DLD-1 cells infected with two shRNAs targeting EFNA3 were examined by ECAR. C Basal respiration and spare respiratory capacity in DLD-1 cells infected with two shRNAs targeting EFNA3 were examined by OCR. D Effect of knockdown of EFNA3 on DLD-1 cell proliferation detected by colony formation assays. E Effect of knockdown of EFNA3 on DLD-1 cell proliferation detected by CCK-8 assays. F CM-induced angiogenesis in DLD-1 cells detected by CAM assay. Migration (G) and invasion (H) of DLD-1 cells infected with two shRNAs targeting EFNA3 were examined by Transwell assay. I Killing of DLD-1 cells by CD8+ T and NK cells was detected by live–dead cell staining. J Cytokines IFN-γ and GzmB in culture supernatants of DLD-1 cells and the immune cell coculture system. N = 5 biologically independent replicates. Data were presented as mean ± SEM. The one-way (D, F–H) or two-way ANOVA (A–C, E, I–J) statistical analysis, followed by Tukey’s multiple comparisons test, was used Knockdown of EFNA3 inhibits DLD-1 cell activity and enhances immune cell-mediated tumor killing by blocking the glycolytic pathway. A The EFNA3, LDHA, and PKM2 protein expression in DLD-1 cells infected with two shRNAs targeting EFNA3 was examined by western blot assays. B Glycolysis and glycolytic reserve in DLD-1 cells infected with two shRNAs targeting EFNA3 were examined by ECAR. C Basal respiration and spare respiratory capacity in DLD-1 cells infected with two shRNAs targeting EFNA3 were examined by OCR. D Effect of knockdown of EFNA3 on DLD-1 cell proliferation detected by colony formation assays. E Effect of knockdown of EFNA3 on DLD-1 cell proliferation detected by CCK-8 assays. F CM-induced angiogenesis in DLD-1 cells detected by CAM assay. Migration (G) and invasion (H) of DLD-1 cells infected with two shRNAs targeting EFNA3 were examined by Transwell assay. I Killing of DLD-1 cells by CD8+ T and NK cells was detected by live–dead cell staining. J Cytokines IFN-γ and GzmB in culture supernatants of DLD-1 cells and the immune cell coculture system. N = 5 biologically independent replicates. Data were presented as mean ± SEM. The one-way (D, F–H) or two-way ANOVA (A–C, E, I–J) statistical analysis, followed by Tukey’s multiple comparisons test, was used EFNA3 silencing repressed cell proliferation (Fig. 2D, E, Additional file 1: Supplementary Fig. S3A, 3B) and the pro-angiogenic capacity of CRC cells (Fig. 2F, Additional file 1: Supplementary Fig. S3C). In addition, the migration (Fig. 2G, Additional file 1: Supplementary Fig. S3D) and invasion (Fig. 2H, Additional file 1: Supplementary Fig. S3E) of CRC cells were also significantly attenuated after the knockdown of EFNA3. Immune cells (NK and CD8+ T) were cocultured with CRC cells, and a significant reduction in the survival of cells knocked down for EFNA3 after coculture with immune cells was observed by live–dead cell staining (Fig. 2I, Additional file 1: Supplementary Fig. S3F). The levels of IFN-γ and GzmB in the coculture supernatant were significantly elevated upon knockdown of EFNA3 (Fig. 2J, Additional file 1: Supplementary Fig. S3G). CRC cells with sh-EFNA3 #1 were treated using the glycolysis activator DASA-58, which significantly restored cellular glycolytic flux (Additional file 1: Supplementary Fig. S4A) and also enhanced cellular proliferation (Additional file 1: Supplementary Fig. S4B). Furthermore, the restored glycolytic flux in DASA-58-treated cells strengthened their survival capacity following coculture with immune cells (Additional file 1: Supplementary Fig. S4C) and reduced the levels of tumor-killing factors IFN-γ and GzmB in the coculture system (Additional file 1: Supplementary Fig. S4D). Knockdown of EFNA3 inhibits in vivo growth and immune evasion in patient-derived xenografts Fresh tumor tissue from patients with CRC was injected subcutaneously into human immune system-reconstituted mice (huHSC-B-NDG hIL15). When tumors reached a specific volume (approximately 150 mm3), mice received intratumoral injections of EFNA3-knockdown lentivirus or control lentivirus. In vivo tumor growth was significantly inhibited in the sh-EFNA3 nos. 1, 2 groups (Fig. 3A), and the weight of harvested tumors was significantly reduced (Fig. 3B). Western blot confirmed the reduction of EFNA3 expression in transplanted tumors in the sh-EFNA3 nos. 1, 2 groups, while the expression of proliferation marker PCNA, vascular marker CD31, and glycolytic enzyme LDHA was also significantly attenuated (Fig. 3C). Lactic acid content was decreased in tumor tissues with the knockdown of EFNA3 (Fig. 3D). Immune cells potentially associated with EFNA3 expression were analyzed, and flow cytometry analysis revealed that the knockdown of EFNA3 increased the proportion of activated NK cells (CD56+NKG2D+) and CD8+ T (CD8+GzmB+) cells in the tumor tissue (Fig. 3E). However, knocking down EFNA3 did not significantly affect the infiltration of active M1 macrophages (Fig. 3F), total CD4+ T cells, or Tregs (Fig. 3G) in tumor tissue.Fig. 3EFNA3 knockdown inhibits the growth and immune evasion of patient-derived xenografts. A Fresh tumor tissue from patients with CRC was subcutaneously injected into mice with a humanized immune system. After the xenograft grew to approximately 150 mm3, the mice were injected with a lentivirus carrying the EFNA3 knockdown, and the growth of the xenograft was monitored. B Weight of harvested tumors. C Expression of EFNA3, PCNA, LDHA, and CD31 in tumors detected by western blot analysis. D Effect of knockdown of EFNA3 on lactic acid content in tumor tissues. E The infiltration of activated NK cells (CD56+NKG2D+) and CD8+ T (CD8+GzmB+) cells in tumors was analyzed using flow cytometry. F The infiltration of active M1 macrophages (CD68+CD86+) in tumors was analyzed using flow cytometry. G The infiltration of CD4 T cells (CD3+CD4+) in xenografts, as well as the proportion of Tregs (CD4+FOXP3+) among CD4 T cells, was analyzed using flow cytometry. N = 5 biologically independent replicates. Data were presented as mean ± SEM. The one-way (B–G) or two-way ANOVA (A) statistical analysis, followed by Tukey’s multiple comparisons test, was used EFNA3 knockdown inhibits the growth and immune evasion of patient-derived xenografts. A Fresh tumor tissue from patients with CRC was subcutaneously injected into mice with a humanized immune system. After the xenograft grew to approximately 150 mm3, the mice were injected with a lentivirus carrying the EFNA3 knockdown, and the growth of the xenograft was monitored. B Weight of harvested tumors. C Expression of EFNA3, PCNA, LDHA, and CD31 in tumors detected by western blot analysis. D Effect of knockdown of EFNA3 on lactic acid content in tumor tissues. E The infiltration of activated NK cells (CD56+NKG2D+) and CD8+ T (CD8+GzmB+) cells in tumors was analyzed using flow cytometry. F The infiltration of active M1 macrophages (CD68+CD86+) in tumors was analyzed using flow cytometry. G The infiltration of CD4 T cells (CD3+CD4+) in xenografts, as well as the proportion of Tregs (CD4+FOXP3+) among CD4 T cells, was analyzed using flow cytometry. N = 5 biologically independent replicates. Data were presented as mean ± SEM. The one-way (B–G) or two-way ANOVA (A) statistical analysis, followed by Tukey’s multiple comparisons test, was used KLF5 promotes the EFNA3 transcription To reveal the cause of elevated transcription of EFNA3 in CRC, we used UniBind: maps of high-confidence direct TF-DNA interactions across species in UCSC Genome Browser (https://genome.ucsc.edu/index.html) to observe the presence of transcription factors with binding sites in the promoter region upstream of EFNA3 (Additional file 2). The correlation between the expression of EFNA3 and the expression of transcription factors in CRC tumor tissues was analyzed by GEPIA, and the transcription factors that were correlated with the expression of EFNA3 were screened at p < 0.05, including HIF1A, KLF5, MXI1, RELA, and SPI1 (Additional file 1: Supplementary Fig. S5A). The correlation between KLF5 and EFNA3 expression was the most prominent. Analysis in GEPIA revealed that KLF5 expression was remarkably elevated in CRC (Additional file 1: Supplementary Fig. S5B). The presence of multiple binding sites for KLF5 at the EFNA3 promoter was also predicted by JASPAR (https://jaspar.elixir.no/) (Additional file 1: Supplementary Fig. S5C). We therefore hypothesized that KLF5 mediated the transcription of EFNA3 in CRC. IHC detected that the expression of KLF5 was elevated in tumor tissues of patients with CRC (Fig. 4A), and the expression of EFNA3 was significantly and positively correlated with KLF5 in CRC tissues (Fig. 4B). RT–qPCR (Fig. 4C) and Western blot analysis (Fig. 4D) also confirmed the overexpression of KLF5 in CRC cell lines. Stable knockdown of KLF5 expression in CRC cells by shRNAs was observed, and the transcript level of EFNA3 was also significantly reduced (Fig. 4E). The ChIP–qPCR assay confirmed that KLF5 bound to the promoter fragment of EFNA3 (Fig. 4F). By inserting the WT EFNA3 promoter sequence or a mutant promoter sequence lacking the KLF5 binding sequence (BS) into a luciferase reporter construct, we confirmed through dual-luciferase assay that KLF5 knockdown inhibited the transcriptional activity of the WT EFNA3 promoter. By contrast, the deletion of BS2 prevented KLF5 from affecting the transcriptional activity of the EFNA3 promoter, indicating that this was the BS for KLF5 (Fig. 4G).Fig. 4KLF5 mediates the transcription of EFNA3 in CRC. A KLF5 expression in CRC tissues and adjacent tissues was examined using IHC staining (n = 34). B The correlation between EFNA3 expression and KLF5 expression in tumor tissues of patients with CRC (n = 34) was analyzed using Spearman’s correlation analysis. The expression of KLF5 in HCMECs, HRMECs, and CRC cell lines was examined using RT–qPCR (C) and western blot analysis (D). E KLF5 and EFNA3 transcription in CRC cells infected with shRNAs targeting KLF5 was detected by RT–qPCR. F The binding ability of KLF5 to the EFNA3 promoter in CRC cells was examined using ChIP–qPCR. G The effect of the knockdown of KLF5 on the transcriptional activity of the WT/MT EFNA3 promoter was examined using dual-luciferase reporter assays. N = 5 biologically independent replicates. Data were presented as mean ± SEM. Wilcoxon matched-pairs signed rank test (A) was used for the analysis of paired two sets of data. The one-way (C–D) or two-way ANOVA (E–G) statistical analysis, followed by Tukey’s (C, D, E, G) or Sidak’s (F) multiple comparisons test, was used KLF5 mediates the transcription of EFNA3 in CRC. A KLF5 expression in CRC tissues and adjacent tissues was examined using IHC staining (n = 34). B The correlation between EFNA3 expression and KLF5 expression in tumor tissues of patients with CRC (n = 34) was analyzed using Spearman’s correlation analysis. The expression of KLF5 in HCMECs, HRMECs, and CRC cell lines was examined using RT–qPCR (C) and western blot analysis (D). E KLF5 and EFNA3 transcription in CRC cells infected with shRNAs targeting KLF5 was detected by RT–qPCR. F The binding ability of KLF5 to the EFNA3 promoter in CRC cells was examined using ChIP–qPCR. G The effect of the knockdown of KLF5 on the transcriptional activity of the WT/MT EFNA3 promoter was examined using dual-luciferase reporter assays. N = 5 biologically independent replicates. Data were presented as mean ± SEM. Wilcoxon matched-pairs signed rank test (A) was used for the analysis of paired two sets of data. The one-way (C–D) or two-way ANOVA (E–G) statistical analysis, followed by Tukey’s (C, D, E, G) or Sidak’s (F) multiple comparisons test, was used KLF5 mediates glycolysis and immune evasion in CRC cells in an EFNA3-dependent manner The overexpressing lentivirus of EFNA3 was coinfected in CRC cells with stable knockdown of KLF5, and western blot confirmed that exogenous overexpression rescued EFNA3 expression suppressed by knockdown of KLF5 (Fig. 5A, Additional file 1: Supplementary Fig. S6A). Immunofluorescence staining detected that knockdown of KLF5 decreased the expression of LDHA and PKM2 in CRC cells, whereas overexpression of EFNA3 restored the expression of glycolytic enzymes (Fig. 5B, Additional file 1: Supplementary Fig. S6B). It was observed by ECAR assay that knockdown of KLF5 inhibited cellular glycolysis and reduced glycolytic reserve, which was dependent on deficient EFNA3 expression (Fig. 5C, Additional file 1: Supplementary Fig. S6C). EdU staining (Fig. 5D, Additional file 1: Supplementary Fig. S6D) and CCK-8 assays (Fig. 5E, Additional file 1: Supplementary Fig. S6E) revealed that KLF5 knockdown reduced DNA synthesis and inhibited cell proliferation in CRC cells, whereas the combined overexpression of EFNA3 rescued the inhibited DNA synthesis and restored cell proliferation. Likewise, knockdown of KLF5 reduced angiogenesis in CRC cells, whereas overexpression of EFNA3 rescued the suppressed angiogenesis (Fig. 5F, Additional file 1: Supplementary Fig. S6F). In addition, the knockdown of KLF5 repressed the migratory (Fig. 5G, Additional file 1: Supplementary Fig. S6G) and invasive abilities (Fig. 5H, Additional file 1: Supplementary Fig. S6H) of CRC cells in an EFNA3 inhibition-dependent manner. CRC cells with knockdown of KLF5 exhibited higher cell death and significantly fewer surviving cells after coculture with CD8+ T cells or NK cells. By contrast, the upregulation of EFNA3 enhanced the survival of CRC cells under the simulated immunosuppressive microenvironment (Fig. 5I, Additional file 1: Supplementary Fig. S6I).Fig. 5KLF5 mediates EFNA3 expression to regulate the malignant biological behavior of DLD-1 cells. A The protein expression of KLF5 and EFNA3 in DLD-1 cells infected with sh-KLF5 no. 1, sh-KLF5 no. 1 + NC-OE, or sh-KLF5 no. 1 + EFNA3-OE was examined using western blot analysis. B The expression of glycolytic enzymes LDHA and PKM2 in DLD-1 cells was detected by immunofluorescence staining. C Glycolysis and glycolytic reserve in DLD-1 cells were examined by ECAR. D Viability of DLD-1 cells was detected by EdU. E The proliferation of DLD-1 cells was detected using CCK-8 assays. F CM-induced angiogenesis in DLD-1 cells detected by CAM assay. Migration (G) and invasion (H) of DLD-1 cells infected with two shRNAs targeting EFNA3 were examined by Transwell assay. I Killing of DLD-1 cells by CD8+ T and NK cells was detected by live–dead cell staining. N = 5 biologically independent replicates. Data were presented as mean ± SEM. The one-way (D, F–H) or two-way ANOVA (A– C, E, I) statistical analysis, followed by Tukey’s multiple comparisons test, was used KLF5 mediates EFNA3 expression to regulate the malignant biological behavior of DLD-1 cells. A The protein expression of KLF5 and EFNA3 in DLD-1 cells infected with sh-KLF5 no. 1, sh-KLF5 no. 1 + NC-OE, or sh-KLF5 no. 1 + EFNA3-OE was examined using western blot analysis. B The expression of glycolytic enzymes LDHA and PKM2 in DLD-1 cells was detected by immunofluorescence staining. C Glycolysis and glycolytic reserve in DLD-1 cells were examined by ECAR. D Viability of DLD-1 cells was detected by EdU. E The proliferation of DLD-1 cells was detected using CCK-8 assays. F CM-induced angiogenesis in DLD-1 cells detected by CAM assay. Migration (G) and invasion (H) of DLD-1 cells infected with two shRNAs targeting EFNA3 were examined by Transwell assay. I Killing of DLD-1 cells by CD8+ T and NK cells was detected by live–dead cell staining. N = 5 biologically independent replicates. Data were presented as mean ± SEM. The one-way (D, F–H) or two-way ANOVA (A– C, E, I) statistical analysis, followed by Tukey’s multiple comparisons test, was used SCAF8 enhances KLF5 mRNA stability We predicted the potential RBPs of KLF5 in ENCORI (https://rnasysu.com/encori/index.php) (the RBP–mRNA interactions supported by CLIP-seq data), with binding sites of RBPs on KLF5 (clipExpNum) > 5 (Additional file 1: Supplementary Fig. S7A). The correlation between KLF5 expression and the expression of these 10 RBPs in CRC tissues was analyzed by GEPIA. There are eight RBPs with significant correlation (p < 0.05), including ELAVL1, TARDBP, DDX3X, RNPS1, RBMX, SCAF8, SCAF4, and YTHDF1 (Additional file 1: Supplementary Fig. S7B). The correlation between SCAF8 and KLF5 expression was the most outstanding, with a significant positive correlation between the two. Proteomics data from UALCAN (https://ualcan.path.uab.edu/analysis-prot.html) revealed that protein expression of SCAF8 was significantly elevated in CRC ( Additional file 1:Supplementary Fig. S7C). However, the effect of SCAF8 on cancers has not been reported, and whether it is involved in CRC progression by enhancing KLF5 mRNA expression remains to be investigated. The protein expression of SCAF8 was elevated in the collected CRC tissues as detected by IHC (Fig. 6A), and the expression of SCAF8 was significantly and positively correlated with KLF5 and EFNA3 (Fig. 6B). Western blot assay also confirmed the high expression of SCAF8 in CRC cells relative to HCMECs and HRMECs (Fig. 6C). The mRNA expression of KLF5 and EFNA3 was also significantly reduced in the CRC cells with stable knockdown of SCAF8 (Fig. 6D). RIP assays detected that more KLF5 mRNA was pulled down by anti-SCAF8 in CRC cells relative to HCMECs and HRMECs (Fig. 6E). Knockdown of SCAF8 led to diminished stability and shorter half-life of KLF5 mRNA (Fig. 6F).Fig. 6SCAF8 regulates KLF5 mRNA stability. A SCAF8 expression in CRC tissues and adjacent tissues was examined using IHC staining (n = 34). B The correlation between SCAF8 and EFNA3 or KLF5 expression in tumor tissues of patients with CRC (n = 34) was analyzed using Spearman’s correlation analysis. C The protein expression of SCAF8 in HCMECs, HRMECs, and CRC cell lines was examined using western blot analysis. D SCAF8, KLF5, and EFNA3 transcription in CRC cells infected with shRNAs targeting SCAF8 was detected by RT–qPCR. E The binding of anti-SCAF8 to KLF5 mRNA in HCMECs, HRMECs, and CRC cell lines was detected by RIP assay. F Effect of knockdown of SCAF8 on KLF5 mRNA stability detected by RT–qPCR. N = 5 biologically independent replicates. Data were presented as mean ± SEM. Wilcoxon matched-pairs signed rank test A was used for the analysis of paired two sets of data, and Nonlin fit F for analyzing half-life changes in mRNA. The one-way (C, E) or two-way ANOVA (D) statistical analysis, followed by Tukey’s multiple comparisons test, was used SCAF8 regulates KLF5 mRNA stability. A SCAF8 expression in CRC tissues and adjacent tissues was examined using IHC staining (n = 34). B The correlation between SCAF8 and EFNA3 or KLF5 expression in tumor tissues of patients with CRC (n = 34) was analyzed using Spearman’s correlation analysis. C The protein expression of SCAF8 in HCMECs, HRMECs, and CRC cell lines was examined using western blot analysis. D SCAF8, KLF5, and EFNA3 transcription in CRC cells infected with shRNAs targeting SCAF8 was detected by RT–qPCR. E The binding of anti-SCAF8 to KLF5 mRNA in HCMECs, HRMECs, and CRC cell lines was detected by RIP assay. F Effect of knockdown of SCAF8 on KLF5 mRNA stability detected by RT–qPCR. N = 5 biologically independent replicates. Data were presented as mean ± SEM. Wilcoxon matched-pairs signed rank test A was used for the analysis of paired two sets of data, and Nonlin fit F for analyzing half-life changes in mRNA. The one-way (C, E) or two-way ANOVA (D) statistical analysis, followed by Tukey’s multiple comparisons test, was used PRKN influences the stability of the SCAF8 protein through ubiquitination Although the protein level of SCAF8 was observed to be significantly elevated in CRC cells, its transcriptional level was not significantly altered ( Additional file 1:Supplementary Fig. S8A), which was consistent with the prediction results in the GEPIA database (Additional file 1: Supplementary Fig. S8B). We therefore hypothesized that the elevated protein expression of SCAF8 was caused by post-translational modifications. We predicted ubiquitin ligases (E3) ( Additional file 1:Supplementary Fig. S8C) targeting the SCAF8 protein in UbiBrowser 2.0 (http://ubibrowser.bio-it.cn/ubibrowser_v3/home/index). The expression of these E3 ubiquitin ligases in CRC tissues was analyzed by GEPIA. Among them, only PRKN expression was found to be reduced in CRC (Additional file 1: Supplementary Fig. S8D). UbiBrowser 2.0 predicted the lysine site at position 119 (K119) of the SCAF8 protein (UniProt ID: Q9UPN6) as its ubiquitination site (Additional file 1: Supplementary Fig. S8E) affected by PRKN. IHC also confirmed the attenuated expression of PRKN in CRC tissues (Fig. 7A), and a significant negative correlation between PRKN and SCAF8/KLF5/EFNA3 expression was noted (Fig. 7B). The reduction of PRKN expression in CRC cells was verified using both RT–qPCR (Fig. 7C) and western blot analyses (Fig. 7D). CRC cell lines stably overexpressing PRKN were developed, and PRKN overexpression downregulated the protein expression of SCAF8, KLF5, and EFNA3 (Fig. 7E), as well as the mRNA levels of KLF5 and EFNA3, without affecting the mRNA level of SCAF8 (Fig. 7F). This suggests that PRKN regulates the function of SCAF8 through post-translational modification. The protein stability of SCAF8 was reduced after overexpression of PRKN, and the half-life was significantly shortened (Fig. 7G). SCAF8 mutant protein plasmid (Flag-SCAF8K119R) with K119 mutated to arginine (R119) was constructed, and Co-IP experiments observed that PRKN enhanced the level of ubiquitination modification of Flag-SCAF8 protein, resulting in the weakening of SCAF8 expression. However, PRKN was unable to affect the ubiquitination and expression of Flag-SCAF8K119R level, confirming that this site is the ubiquitination modification site of SCAF8 (Fig. 7H).Fig. 7PRKN inhibits protein expression of SCAF8 by ubiquitination modification. A PRKN expression in CRC tissues and adjacent tissues was examined using IHC staining (n = 34). B The correlation between PRKN and SCAF8/EFNA3/KLF5 expression in tumor tissues of patients with CRC (n = 34) was analyzed using Spearman’s correlation analysis. The expression of PRKN in HCMECs, HRMECs, and CRC cell lines was examined using RT–qPCR C and western blot analysis D. The expression of PRKN, SCAF8, KLF5, and EFNA3 in CRC cells infected with NC-OE or PRKN-OE was detected by western blot analysis E and RT–qPCR (F). G The effect of overexpression of PRKN on the stability of SCAF8 protein was examined using western blot analysis. H The ubiquitination modification of SCAF8 by PRKN in 293 T cells was examined using Co-IP assays. N = 5 biologically independent replicates. Data were presented as mean ± SEM. Wilcoxon matched-pairs signed rank test (A) was used for the analysis of paired two sets of data, and Nonlin fit (G) for analyzing half-life changes in protein. The one-way (C–D) or two-way ANOVA (E–F) statistical analysis, followed by Tukey’s multiple comparisons test, was used PRKN inhibits protein expression of SCAF8 by ubiquitination modification. A PRKN expression in CRC tissues and adjacent tissues was examined using IHC staining (n = 34). B The correlation between PRKN and SCAF8/EFNA3/KLF5 expression in tumor tissues of patients with CRC (n = 34) was analyzed using Spearman’s correlation analysis. The expression of PRKN in HCMECs, HRMECs, and CRC cell lines was examined using RT–qPCR C and western blot analysis D. The expression of PRKN, SCAF8, KLF5, and EFNA3 in CRC cells infected with NC-OE or PRKN-OE was detected by western blot analysis E and RT–qPCR (F). G The effect of overexpression of PRKN on the stability of SCAF8 protein was examined using western blot analysis. H The ubiquitination modification of SCAF8 by PRKN in 293 T cells was examined using Co-IP assays. N = 5 biologically independent replicates. Data were presented as mean ± SEM. Wilcoxon matched-pairs signed rank test (A) was used for the analysis of paired two sets of data, and Nonlin fit (G) for analyzing half-life changes in protein. The one-way (C–D) or two-way ANOVA (E–F) statistical analysis, followed by Tukey’s multiple comparisons test, was used PRKN blocks KLF5/EFNA3 signaling-mediated malignant biological behavior in CRC cells by inhibiting protein expression of SCAF8 Whether PRKN overexpression can affect CRC progression by blocking SCAF8/KLF5/EFNA3 signaling is not yet known; we performed exogenous overexpression of SCAF8/KLF5/EFNA3 jointly in CRC cells overexpressing PRKN and confirmed the expression intervention effect by western blot analysis (Fig. 8A, Additional file 1: Supplementary Fig. S9A). Overexpression of PRKN inhibited CRC proliferation, while combined overexpression of SCAF8/KLF5/EFNA3 reversed the effect of PRKN (Fig. 8B, C, Additional file 1: Supplementary Fig. S9B, C). In addition, PRKN inhibited the migratory (Fig. 8D, Additional file 1: Supplementary Fig. S9D) and invasive (Fig. 8E, Additional file 1: Supplementary Fig. S9E) capacities of CRC cells and reduced the angiogenic effects (Fig. 8F, Additional file 1: Supplementary Fig. S9F) induced by CRC cells, whereas restoration of SCAF8/KLF5/EFNA3 signaling reversed the effects of PRKN.Fig. 8PRKN/SCAF8/KLF5/EFNA3 signaling influences the malignant biological behavior of DLD-1 cells. A The SCAF8, KLF5, and EFNA3 protein expression in DLD-1 cells infected with PRKN-OE + NC-OE, PRKN-OE + SCAF8-OE, PRKN-OE + KLF5-OE, and PRKN-OE + EFNA3-OE was examined by western blot assays. DLD-1 cell proliferation detected by colony formation B and CCK-8 assays C. Migration D and invasion E of DLD-1 cells were examined by Transwell assay. F CM-induced angiogenesis in DLD-1 cells detected by CAM assay. G Lactic acid released from DLD-1 cells. H Glycolysis and glycolytic reserve in DLD-1 cells were examined by ECAR. (I) Killing of DLD-1 cells by CD8+ T and NK cells was detected by live–dead cell staining. N = 5 biologically independent replicates. Data were presented as mean ± SEM. The one-way (B, D–G) or two-way ANOVA (A, C, H–I) statistical analysis, followed by Tukey’s multiple comparisons test, was used PRKN/SCAF8/KLF5/EFNA3 signaling influences the malignant biological behavior of DLD-1 cells. A The SCAF8, KLF5, and EFNA3 protein expression in DLD-1 cells infected with PRKN-OE + NC-OE, PRKN-OE + SCAF8-OE, PRKN-OE + KLF5-OE, and PRKN-OE + EFNA3-OE was examined by western blot assays. DLD-1 cell proliferation detected by colony formation B and CCK-8 assays C. Migration D and invasion E of DLD-1 cells were examined by Transwell assay. F CM-induced angiogenesis in DLD-1 cells detected by CAM assay. G Lactic acid released from DLD-1 cells. H Glycolysis and glycolytic reserve in DLD-1 cells were examined by ECAR. (I) Killing of DLD-1 cells by CD8+ T and NK cells was detected by live–dead cell staining. N = 5 biologically independent replicates. Data were presented as mean ± SEM. The one-way (B, D–G) or two-way ANOVA (A, C, H–I) statistical analysis, followed by Tukey’s multiple comparisons test, was used Lactic acid release from tumor cells was significantly reduced after overexpression of PRKN, while SCAF8/KLF5/EFNA3 promoted lactic acid secretion from CRC cells (Fig. 8G, Additional file 1: Supplementary Fig. S9G). PRKN overexpression inhibited glycolytic phenotype and reduced glycolytic reserve in CRC cells, which was dependent on its inhibitory effect on the SCAF8/KLF5/EFNA3 axis (Fig. 8H, Additional file 1: Supplementary Fig. S9H). PRKN expression enhanced immune cell killing of CRC cells with a significant reduction in cell survival, whereas overexpression of SCAF8/KLF5/EFNA3 resulted in resistance of CRC cells to immune cell killing (Fig. 8I, Additional file 1: Supplementary Fig. S9I) . SCAF8/KLF5/EFNA3 overexpression overturns the anti-tumor effects of PRKN in vivo PRKN overexpression significantly impaired tumorigenicity in subcutaneous mouse allografts formed by CMT93 cells (Fig. 9A) and reduced the tumor burden (Fig. 9B). The tumorigenic capacity of CMT93 cells was restored upon overexpression of SCAF8/KLF5/EFNA3. Western blot analysis detected enhanced PRKN expression in allografts from the PRKN-OE group, while SCAF8/KLF5/EFNA3 expression was reduced. Conversely, overexpression of SCAF8/KLF5/EFNA3 rescued oncogenic signaling in the allografts (Fig. 9C). Lactic acid content was significantly reduced in tumors formed by CMT93 cells overexpressing PRKN, whereas the overexpression of SCAF8/KLF5/EFNA3 promoted lactic acid accumulation in tumor tissues (Fig. 9D). The protein expression of LDHA and CD31 in tumors was suppressed by PRKN overexpression and restored by overexpression of SCAF8/KLF5/EFNA3 (Fig. 9E). In PRKN-overexpressing allografts, infiltration of activated NK cells (CD49b+NKG2D+) and CD8 T cells (CD8+GzmB+) was significantly increased, whereas overexpression of the SCAF8/KLF5/EFNA3 axis suppressed immune cell infiltration (Fig. 9F). In addition, CMT93 cells overexpressing PRKN formed fewer liver metastases after splenic injection, whereas overexpression of SCAF8/KLF5/EFNA3 led to an increase in liver tumor infiltration (Fig. 9G).Fig. 9PRKN/SCAF8/KLF5/EFNA3 signaling influences the tumor growth and immune evasion of CRC cells. A Growth of tumors after subcutaneous injection of the mouse CRC cell line CMT93. B Weight of harvested tumors. C The protein expression of PRKN, SCAF8, KLF5, and EFNA3 in transplanted tumors was examined using western blot analysis. D Lactic acid content in tumor tissues. E Expression of LDHA and CD31 in tumors detected by IHC. F Analysis of NK cell and CD8+ T cell infiltration in tumors by flow cytometry. G Representative images of harvested liver and HE staining images of the tumor metastasis area 21 days after spleen injection of CMT93 cells. N = 5 biologically independent replicates. The one-way (B–G) or two-way ANOVA (A) statistical analysis, followed by Tukey’s multiple comparisons test, was used PRKN/SCAF8/KLF5/EFNA3 signaling influences the tumor growth and immune evasion of CRC cells. A Growth of tumors after subcutaneous injection of the mouse CRC cell line CMT93. B Weight of harvested tumors. C The protein expression of PRKN, SCAF8, KLF5, and EFNA3 in transplanted tumors was examined using western blot analysis. D Lactic acid content in tumor tissues. E Expression of LDHA and CD31 in tumors detected by IHC. F Analysis of NK cell and CD8+ T cell infiltration in tumors by flow cytometry. G Representative images of harvested liver and HE staining images of the tumor metastasis area 21 days after spleen injection of CMT93 cells. N = 5 biologically independent replicates. The one-way (B–G) or two-way ANOVA (A) statistical analysis, followed by Tukey’s multiple comparisons test, was used Discussion To the best of our knowledge, our study holds implications for understanding the multifaceted role of EFNA3 in various aspects of CRC cell behavior, which includes cancer cell proliferation, glycolysis, angiogenesis, and immune evasion. Furthermore, our findings indicate a regulation of transcription factor KLF5 on EFNA3 expression. EFNA3 was upregulated in most cancer types, and high expression of EFNA3 was associated with tumor progression and worse prognosis in patients with hepatocellular carcinoma [25]. Here, we predicted EFNA3 as a glycolysis-specific gene and observed the higher expression of LDHA and PKM2 in patients with CRC with high EFNA3 expression. The energy obtained from glycolysis is conducive to the proliferation, invasion, migration, and epithelial–mesenchymal transition of cancer cells, as well as the proliferation and function of immune cells [26]. Cancer cells can modify cytokines to inhibit the functioning of cytotoxic T cells, which leads to the suppression of CD8+ T lymphocytes, preventing them from recognizing foreign antigens [27]. Consistent with the predictive results that EFNA3 was negatively correlated with the infiltration of NK cells and CD8+ T cells, we also found the repressed IHC scores of CD8A and CD56 in patients with CRC with higher EFNA3 expression and enhanced levels of IFN-γ and GzmB in the coculture supernatant comprising CRC cells with EFNA3 knockdown. As demonstrated by Husain et al., EFNA3 was frequently upregulated in hepatocellular carcinoma, and its overexpression was linked to more aggressive tumor behaviors, including self-renewal, proliferation, and migration in hepatocellular carcinoma cells [28]. Likewise, in vivo evidence here showed that the mice had reduced tumor burden with EFNA3 knockdown. Among the transcription factors of EFNA3, KLF5 showed the most outstanding correlation coefficient with EFNA3 expression in CRC. It has been recently shown that KLF5 functions as a transcription factor for the leukemia inhibitory factor and bound to its promoter region in CRC [29]. The overexpression of KLF5 has been implicated in the progression of lung cancer and ovarian cancer [30, 31]. More importantly, tumors with KLF5 overexpression had reduced infiltration of CD8+ T cells, and NK cells, CD4+ effector, and CD8+ effector memory T cells were enriched within tumors with KLF5 knockdown [32]. KLF5 has also been reported to regulate the expression of CXCL12 and CXCL5 in CRC to improve the immunosuppressive TME, which contributed to the enhanced efficacy of immunotherapy and reduced metastasis [33, 34]. Our observation here that the repressive effects of KLF5 knockdown on angiogenesis, glycolysis, and immunosuppression on CRC cells were overturned by EFNA3 overexpression might add new insights into the downstream mechanism of KLF5 in CRC. As for its upstream modifiers, Y-box binding protein 1 has been revealed to stabilize KLF5 mRNA in an RNA 5-methylcytosine modification-dependent manner in basal-like breast cancer [35]. In this study, we verified the stabilizing effects of RBP SCAF8 on KLF5 mRNA. Subsequently, the ubiquitination site of SCAF8 was identified at K119, since the mutation of this site caused the stability of SCAF8. PRKN is a 52 kDa protein with 465 amino acids, and the PRKN gene localizes on human ch6q25‐27, a frequent loss region in cancers [36]. Under the condition of cervical cancer, overexpression of PRKN catalyzed the proteasomal degradation of IGF2BP3 [37]. Regarding its functional role, PRKN acted as a tumor suppressor by inhibiting the proliferation and migration of bladder cancer cells [38]. Intriguingly, the silencing of PRKN almost completely reversed the reduction in OCR and clonogenic capacity of CRC cells caused by USP26 silencing [39]. IHC staining of orthotopic breast tumor tissues showed that PRKN knockout increased the levels of its substrate PHGDH and staining of Ki-67 [40]. In addition, Perales-Linares et al. showed that PRKN deficiency led to aggressive tumor growth, which was linked to poor tumor antigen presentation and constrained antitumor CD8+ T-cell infiltration and activation, and the impact of PRKN inhibition on tumor growth was lost after depletion of CD8+ T cells [41]. Here, by using lentiviruses, we found that overexpression of PRKN repressed the glycolysis, angiogenesis, and immune evasion of CRC, which were all reversed by the overexpression of its direct substrate SCAF8 and the indirect targets KLF5 and EFNA3. The present study is not without its limitations. First, the findings of this study underscore the pivotal role of EFNA3 as a central regulator of glycolysis and immune modulation in CRC. However, the precise mechanisms linking EFNA3 to LDHA/PKM2 expression and immune-related changes remain to be elucidated, representing a significant direction for future investigation. Additionally, while the present data establish SCAF8 as an upstream regulator of KLF5 mRNA stability, the specific region of KLF5 mRNA bound by SCAF8 remains to be elucidated. Subsequent research endeavors will seek to ascertain the precise binding sites. Conclusions Taken together, these findings suggest that PRKN plays a significant role in regulating the SCAF8/KLF5/EFNA3 axis in CRC. Overexpression of PRKN can impair glycolysis and promote antitumor response of NK cells and CD8+ T cells by repressing SCAF8 protein expression and blocking the KLF5/EFNA3 signaling. Moreover, EFNA3 is considered a potential biomarker of glycolysis and a novel immunotherapeutic target in patients with CRC. Supplementary Information Additional file 1. Additional file 2. Additional file 1. Additional file 2.