Truncated APC impairs innate immune response by targeting MAVS on mitochondria in colorectal cancer.

Introduction
Colorectal cancer (CRC) is the second leading cause of cancer-related mortality worldwide [1] and is characterized by the gradual accumulation of genetic and epigenetic alterations [2]. One of the most critical tumor suppressor genes involved in CRC is Adenomatous Polyposis Coli (APC). Germline mutations in APC gene lead to familial adenomatous polyposis (FAP), an autosomal dominant inherited syndrome that invariably progresses to CRC if not treated promptly [3]. Importantly, somatic mutations in APC are found in approximately 80% of sporadic CRC cases, where they act as early driving events in tumorigenesis [4]. These mutations often result in truncated APC proteins (Trunc-APC), primarily due to premature termination codons within the Mutation Cluster Region (MCR) [5]. Emerging evidence suggests that Trunc-APC proteins may acquire gain-of-function properties, contributing to enhanced cell proliferation, migration, chromosomal instability [6].
Approximately 90% of CRC cases are classified as microsatellite stable (MSS) or proficient mismatch repair (pMMR) subtypes, which are largely resistant to immune checkpoint inhibitor (ICI) therapy [7]. Conventional chemotherapy remains the standard treatment for most advanced or metastatic CRC patients, although its efficacy is limited by significant toxicity and the development of drug resistance [8, 9]. Therefore, there is an urgent need to develop novel therapeutic strategies to improve treatment outcomes for CRC patients.
Innate immunity is the first line of defense against pathogen infection, recognizing conserved microbial structures such as viral nucleic acids and proteins. Upon viral infection, cytosolic pattern recognition receptors (PRRs, e.g., RIG-I, MDA5, and cGAS) and endosomal PRRs (e.g., TLR3 and TLR4) initiate a cascade that recruits and activates transcription factors (e.g., IRF3 and NF-κB), leading to the production of type I interferons(IFNs) and pro-inflammatory cytokines [10, 11]. These effectors suppress viral replication, induce apoptosis in infected cells, and promote adaptive immune activation, leading to an antiviral response [12]. Viral mimicry has emerged as a promising therapeutic approach that activates antiviral responses in tumor cells via endogenous stimuli, such as double-stranded RNA (dsRNA), rather than direct viral infection [13]. In this process, transcripts from endogenous retroviruses (ERVs), long interspersed nuclear elements (LINEs), and short interspersed nuclear elements (SINEs) accumulate in the cytosol as dsRNA, activating the innate immune response [14]. Growing evidence indicates that epigenetic therapies, including DNA methyltransferase inhibitors (DNMTis) and histone deacetylase inhibitors (HDACi), play a pivotal role in initiating viral mimicry [15, 16].
Recent studies have highlighted the immunological implications of APC. APC mutations have been identified as potential predictors of immunotherapy outcomes in CRC, with evidence linking these mutations to poorer survival, lower immune scores, and reduced CD8+ T cell infiltration. However, the relationship between APC mutations, tumor mutation burden (TMB), and immune checkpoint molecule expression, such as PD-L1, remains inconsistent [17–20]. Notably, research from the Andrés Alcover group has elucidated the roles of APC and its mutations in T lymphocyte function. In APCMin/+ mice, regulatory T cells (Tregs) exhibit intrinsic defects in differentiation and anti-inflammatory IL-10 production, which are critical for preventing tumor-associated inflammation and suppressing tumor initiation [21]. Furthermore, cytotoxic T lymphocytes (CTLs) from APCMin/+ mice or Apc-silenced human CD8+ T cells display impaired cytotoxic efficiency due to destabilized immunological synapses and reduced delivery of cytotoxic granules, compromising their tumor-inhibitory function [22]. Abnormal T cell migration has also been reported in FAP patients, which is associated with impaired immune surveillance and contributes to tumor progression [23]. Additionally, reduced engagement between CTLs from APCMin/+ OT-I mice and tumor cells in 3D tumor spheroid models further underscores the immune dysfunction associated with APC mutations [24]. These findings suggest that Trunc-APC may exert vital immunosuppressive effects on T lymphocytes. However, the role of Trunc-APC in immune regulation within tumor cells and the underlying molecular mechanisms remain unclear. Identifying novel binding partners of Trunc-APC and elucidating their molecular interactions in immune regulation are critical for gaining a comprehensive understanding of the signaling network in CRC, as well as for identifying novel therapeutic targets for CRC treatment.
In this study, we investigate the role of Trunc-APC in modulating innate immune responses in CRC cells. We show that Trunc-APC partially localizes to the outer mitochondrial membrane, where it interacts with MAVS, thereby inhibiting the type I interferon signaling pathway by disrupting the interaction between MAVS and RIG-I. Targeting Trunc-APC enhances the innate immune response elicited by 5-azacytidine (AZA) and trichostatin A (TSA) treatment, resulting in increased tumor cell apoptosis and significantly suppressing tumor growth in vivo.

Materials & methods

Cell culture
SW480, HT29, HCT116, and HEK-293T cell lines were obtained from the American Type Culture Collection (ATCC). SW480 and HCT116 cells were maintained in RPMI 1640 medium (GIBCO, USA, C11875500BT) supplemented with 10% fetal bovine serum (FBS, Gibco, USA, 10099141). HT29 cells were cultured in McCoy’s 5 A medium (Procell, China, PM150710) with 10% FBS, and HEK293T cells were cultured in DMEM/High Glucose (GIBCO, USA, C11995500BT) supplemented with 10% FBS. All media were routinely supplemented with Penicillin-Streptomycin (Biosharp, China, BL505A). Cell cultures were maintained at 37 °C in a humidified incubator with 5% CO2.

Antibodies and chemicals
The antibodies used in this study were as follows: APC (abcam, ab58; IB, 1:1000), APC (abcam, ab40778; IP, 1:50), APC (abcam, ab16794; IF, 1:100), MAVS (CST, 24930, IB, 1:1000; IF, 1:200; IP, 1:50), RIG-I (CST, 3743, IB, 1:1000; IP, 1:50), RIG-I (proteintech, 67556-1-Ig, IF, 1:200), Phospho-IRF3 (CST, 37829; IB, 1:1000), Phospho-p65 (CST, 3033; IB, 1:1000), IRF3 (CST, 11904; IB, 1:1000; IF, 1:200), p65 (CST, 8242; IB, 1:1000; IF, 1:200), Ubiquitin (CST, 3933; IB, 1:1000), Histone-H3 (proteintech, 17168-1-Ig, IB, 1:1000), Tom20 (proteintech, 11802-1-Ig, IB, 1:1000; IF, 1:200), Tim23 (proteintech, 11802-1-Ig, IB, 1:1000), Flag (CST, 14793; IB, 1:1000; IP, 1:50), Myc (CST, 2276; IB, 1:1000; IP, 1:50), HA (CST, 3724; IB, 1:1000; IP, 1:50), GAPDH (proteintech, 60004-1-Ig, IB, 1:1000), α-Tubulin (proteintech, 66031-1-Ig, IB, 1:1000). The chemicals used in this study included 5-AZA (A3656) from Merck (Germany), MG-132 (13259), polyinosinic-polycytidylic acid (poly(I: C)) (107202), and Trichostatin A (15144) from MedChemexpress (USA).

Generation of knockout cell lines with CRISPR-Cas9
SW480, HT29 and HCT116 cells were transiently transfected with the lentiCRISPR v2 plasmid (Addgene, USA, 52961), which was designed to contain APC-specific single-guide RNAs (sgRNAs). The sgRNA sequences targeting APC are listed in Supplementary Table S2. Transfection was performed using Lipofectamine 2000 (Invitrogen, USA, 11668019), according to the manufacturer’s protocol. After transfection, cells were selected with 3 µg/mL puromycin (InvivoGen, France, ant-pr-1) for two weeks. Single cells were then seeded into 96-well plates, and independent clones were cultured for three additional weeks. Knockout efficiency was evaluated by western blot analysis and DNA sequencing (Tsingke, China).

Transfection of shRNA, SiRNA and plasmids
The short hairpin RNAs (shRNAs) and small interfering RNAs (siRNAs) used in this study were synthesized by RiboBio (China), and detailed sequences are provided in Supplementary Table S2. Cells were seeded at approximately 30% confluence in 6-well plates and transfected with shRNA or siRNA using jetPRIME® transfection reagent (Polyplus, France, 101000046) for 48 h, according to the manufacturer’s instructions. Truncated mutants of APC (1-1337, 1-800, 251–800, and 1-250) were subcloned into the pcDNA3.1 vector. Wild-type (WT) and truncated mutants of MAVS and RIG-I were also subcloned into the pcDNA3.1 vector. WT and Ub mutant plasmids were inherited from previous work in our laboratory. The IFNβ luciferase reporter plasmid was obtained from Miaoling Biology (China). For plasmid transfection, cells were seeded at approximately 60–70% confluence in 6-well plates and transfected using jetPRIME® for 48 h, following the manufacturer’s instructions.

Cell viability assay
Cell viability in drug susceptibility assays was assessed using the Cell Counting Kit-8 (CCK-8) (MCE, China, HY-K0301). CRC cells (1 × 104) were seeded in 96-well plates and incubated for 12 h, followed by treatment with various pharmacological regimens for 72 h. At the endpoint, culture medium without FBS containing 10% CCK-8 reagent was added to each well. After a 2 h incubation, absorbance at 450 nm was measured using a VersaMax microplate reader (MD, USA).

Apoptosis detection assay
Cultured cells were divided into two groups: untreated and treated with 1 µM azacytidine for 48 h, followed by 500 nM trichostatin A for 24 h. Cells were then stained with Annexin V-APC (BioLegend, USA, 640920) for 15 min at 37 °C in Annexin V Binding Buffer (BioLegend, USA, 422201). DAPI was added 5 min before analysis. The samples were analyzed using a flow cytometer (Beckman Coulter, USA) to determine the proportion of apoptotic cells.

Dual-luciferase reporter assay
To evaluate the effect of Trunc-APC or full-length APC knockout on the activation of the IFNβ promoter, SW480 or HCT116 cells were transfected in 24-well plates with 250 ng of pGL3-IFNβ reporter plasmid and 25 ng of Renilla plasmid. After 48 h, the relative luciferase activity was determined by calculating the ratio of Firefly to Renilla luciferase activity using a dual-luciferase reporter assay kit (Yeasen, China, 11402ES60).

RNA extraction and real-time PCR
Total RNA was extracted from cultured cells using TRIzol reagent (Invitrogen, USA, 15596026) according to the manufacturer’s instructions. For qPCR, first-strand cDNA was synthesized from 1000 ng of total RNA using random primers and a Reverse Transcription Kit (Takara, Japan, RR037A). The qPCR reaction mixture was prepared by combining SYBR Green qPCR Master Mix (EZBioscience, USA, A0012-R2), cDNA template, and specific primers. The mixture was then added to a Bio-Rad PCR plate (USA), which was subsequently loaded into the LightCycler 480II system (Roche, Switzerland) for amplification. The primers used for RT-PCR are listed in Supplementary Table S3.

RNA-seq analysis
RNA-seq analysis was conducted by BGI Genomics (Shenzhen, China) using the BGISEQ platform. The raw data were processed and filtered using SOAPnuke (v1.5.6) and subsequently aligned to the hg19 genome with HISAT (v2.1.0). For data analysis, RSEM (v1.3.1), pheatmap (v1.0.8), and DESeq2 (v1.4.5) were employed following the manufacturer’s instructions.

Subcellular fractionation and proteinase K digestion assay
The cytoplasmic and nuclear fractions were extracted from cultured cells using the Nuclear and Cytoplasmic Protein Extraction Kit (TransGen, China, DE201) according to the manufacturer’s instructions. Mitochondrial and cytoplasmic fractions were isolated using the Mitochondria Isolation Kit (Thermo Fisher Scientific, USA, 89874), also following the manufacturer’s protocol. For subsequent proteinase K digestion of intact mitochondria, the mitochondrial pellet was resuspended in buffer and treated with proteinase K (100 ng/mL), either in the absence or presence of 1% Triton X-100, on ice for 30 min. The reaction products were then analyzed by western blotting with the specified antibodies.

Co-immunoprecipitation (co-IP)
Cells were lysed in IP lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.5% NP-40, 10% glycerol) with protease inhibitors (CWBIO, China, CW2200S), then incubated on ice for 30 min or sonicated for 2 min. The lysates were centrifuged at 12,000 rpm for 20 min at 4 °C to collect the supernatants. The supernatants were incubated with Protein A/G Magnetic Beads (MCE, USA, HY-K0202) pre-coupled with specific antibodies on a rotator at 4 °C overnight. The following day, the beads were washed three times with lysis buffer, eluted in 1× SDS, and heated at 100 °C for 10 min to denature the bound proteins, which were then used for further analysis.

Western Blotting (WB)`
SDS-PAGE gels with concentrations ranging from 7.5% to 12.5% were prepared using the One-Step UV Imaging PAGE Gel Rapid Preparation Kit (Epizyme, China), following the manufacturer’s instructions. Denatured protein samples were separated by electrophoresis and then transferred onto a PVDF membrane (Roche, Switzerland, 3010040001). The membranes were blocked with 5% skim milk at room temperature for 1 h and incubated with specific primary antibodies at 4 °C overnight. The next day, membranes were incubated with the appropriate secondary antibodies (mouse or rabbit) and visualized using the ChemiDoc Touch Imaging System (Bio-Rad, USA).

Semi-Denaturing Detergent Agarose Gel Electrophoresis (SDD-AGE)
A vertical 1.5% agarose gel containing 0.1% SDS was prepared. The isolated mitochondrial fraction was resuspended in 1× SDD-AGE loading buffer (0.5× TBE, 25% glycerol, 2.5% SDS, and 0.0025% bromophenol blue) and incubated at room temperature for 5 min. Samples were loaded onto the gel and subjected to electrophoresis at a constant voltage of 100 V for 40 min at 4 °C, using electrophoresis buffer composed of 0.5× TBE and 0.1% SDS. After electrophoresis, proteins were transferred onto a PVDF membrane for subsequent immunoblotting analysis.

Silver staining and mass spectrometry analysis
After electrophoresis, the gels were washed twice with double-distilled water and stained using the Fast Silver Stain Kit (Beyotime, China, P0017S) according to the manufacturer’s instructions. Differential protein bands were carefully excised using clean instruments under a laminar flow hood. The excised protein samples were subsequently subjected to mass spectrometry for proteomic analysis at BGI Genomics, Shenzhen, China. A detaied list of the mass spectrometry results is provided in Supplementary Table S1.

Immunofluorescence (IF) and proximity ligation assay
Cells were plated at an appropriate density on confocal dishes. After allowing sufficient time for cell attachment, the dishes were washed three times with PBS and fixed with 4% paraformaldehyde for 20 min at room temperature. The fixed cells were then permeabilized with 0.3% Triton X-100 in PBS for 10 min, followed by blocking with 10% BSA for 1 h. Subsequently, cells were incubated with a specific primary antibody diluted in 2% BSA overnight at 4 °C. The next day, the confocal dishes were washed three times with PBST and incubated with a fluorescent secondary antibody (Life technologies) for 1 h at room temperature. Finally, the dishes were stained with DAPI staining solution (Beyotime, China, C1005) for 10 min at room temperature and observed under a confocal microscope. To assess the impact of Trunc-APC on the interaction between RIG-I and MAVS in poly(I: C) -stimulated cells, a proximity ligation assay was performed using the kit (Sigma-Aldrich, USA, DUO96010) according to the manufacturer’s instructions.

Xenograft models
This study adheres to all relevant ethical regulations regarding animal research and received approval from the Sun Yat-Sen University Animal Care and Use Committee (L102042023070D). BALB/c nu/nu mice (5 weeks old) were obtained from Vital River Laboratories (Beijing, China) and housed under standard conditions at the Center for Experimental Animals, Sun Yat-sen University. For the subcutaneous xenograft model, control and experimental SW480 cells (6 × 106) were suspended in 100 µl of serum-free RPMI 1640 medium and injected subcutaneously into the flanks of the nude mice (n = 5). When tumor sizes reached 3–4 mm, the mice were intraperitoneally injected with the indicated drugs over a five-day treatment cycle. The dosages and treatment regimens were as follows: DNMT inhibitor 5-Azacytidine at 2.5 mg/kg daily for three days, and HDAC inhibitor Trichostatin A at 1.25 mg/kg daily for two days, with the cycle repeating once. After 28 days, the mice were euthanized, and the tumors were harvested, fixed, and paraffin-embedded for further analysis.

Immunohistochemistry (IHC) and scoring
The paraffin-embedded tissue sections were initially baked at 56 °C for 30 min, followed by dewaxing in dimethylbenzene and rehydration with graded ethanol. Antigen retrieval was performed using a citric acid buffer with high pressure and heat. After natural cooling, the sections were incubated with 3% H2O2 for 10 min to eliminate endogenous peroxidase activity. Next, the sections were blocked with 10% BSA at 37 °C for 1 h and subsequently incubated with anti-Ki67 antibody (Abcam, USA, ab15580) at a dilution of 1:100 overnight at 4 °C. The following day, a secondary antibody was applied to bind the primary antibody at 37 °C for 1 h. DAB (Zsbio, China, ZLI9017) was then used to visualize the target protein, followed by hematoxylin staining. A semi-quantitative analysis of the tissue sections was performed to score the IHC staining using the IHC profile plugin. Specifically, at least five randomly selected fields of view were assessed for each section. Positive expression was categorized as weak expression (+, 1 point), moderate expression (++, 2 points), and strong expression (+++, 3 points), based on staining intensity. The final pathological score was calculated as “Expression Degree Score × Positive Expression Proportion (%)”.

Statistical analysis
Statistical analysis was conducted using GraphPad Prism (Version 8; USA), SPSS (Version 25.0; USA), and R software (Version 4.4.1). The significance between two groups was assessed using a two-tailed Student’s t-test, while ANOVA was employed for comparisons among three or more independent groups. Data are expressed as mean ± SD, with p-values less than 0.05 considered statistically significant.

Results

Down-regulation of Trunc-APC promotes type I IFN production in colorectal cancer cells
To investigate the potential role of Trunc-APC in CRC innate immunity, we first established stably mutant APC knockout (KO) cells using SW480 and HT29, both of which express endogenous Trunc-APC (mutated at aa 1338, 1556 and 854, respectively) (Figures S1A-S1B), and performed transcriptome sequencing. Pathway enrichment analysis of the sequence showed that KO cells are enriched in cytokine signaling and interferon signaling pathways (Fig. 1A). The heatmap further illustrates the upregulation of immune-stimulated genes and chemokines in SW480 KO cells (Fig. 1B), supporting a link between Trunc-APC and CRC innate immunity. Next, we quantified mRNA levels of essential factors of innate immunity including type I interferons (IFNα, IFNβ) and two key interferon-stimulated genes (ISGs) CCL5 and CXCL9. RT-qPCR results indicate that mutant APC knockout significantly increases ISGs and type I interferon production in both SW480 (Fig. 1C) and HT29 (Figure S1D), corroborated by dual-luciferase reporter assays (Figure S1E). TIMER2.0 analysis also reveals that CRC harboring mutated APC exhibits lower expression levels of CCL5 and CXCL9 than those with wild-type APC (Figures S1G-H). Furthermore, we examined whether mutant APC knockout activates type I IFN signaling. As illustrated in Figs. 1D-I, mutant APC knockout enhances IRF3 phosphorylation, promoting its translocation to the nucleus, while significantly increasing the phosphorylation and nuclear localization of p65. These findings suggest that loss of Trunc-APC triggers type I IFN signaling. Additionally, to test whether full-length APC can regulate type I IFN signaling, we measured ISG levels and type I interferon production after silencing APC in HCT116 cell lines, which exclusively express endogenous full-length APC (1-2843 amino acids). Our findings indicate that full-length APC has no significant effect on type I interferon production (Figure S1D, S1F).
To investigate which stimuli activates type I IFN signaling in CRC cells, we employed three synthetic pathogen-associated molecular pattern (PAMP) stimuli: polyinosinic: polycytidylic acid (poly(I: C)), interferon-stimulatory DNA (ISD), and lipopolysaccharides (LPS). These stimuli are recognized by Toll-like receptor 3 (TLR3), RIG-I-like receptors (RLRs), cyclic GMP-AMP synthase (cGAS), and Toll-like receptor 4 (TLR4), collectively known as PRRs [25, 26]. As illustrated in Fig. 2E, both poly(I: C) and LPS reduced Trunc-APC expression in SW480 cells, only poly(I: C) transfection significantly increased IRF3 and p65 phosphorylation, indicating activation of the type I IFN pathway. qPCR analysis demonstrated that poly(I: C) elevated levels of type I IFN and ISGs in HEK293T cells, which were diminished by transfection with Trunc-APC variants (1-1337 and 1-800 aa) (Fig. 2F). Moreover, Trunc-APC 1-800 aa decreased IRF3 and p65 phosphorylation induced by poly(I: C) in a dose-dependent manner (Fig. 2G). These findings indicated that Trunc-APC functions as a negative regulator of type I IFN production.

Trunc-APC targets MAVS to inhibit type I IFN production
Mitochondria serve as critical hubs for the innate immune response [27]. Mariana Brocardo et al. found that Trunc-APC, particularly forms smaller than 1400 amino acids, localizes to mitochondria [28]. Therefore, we propose that Trunc-APC affects type I interferon signaling via mitochondrial platforms. We first examined the mitochondrial localization of endogenous Trunc-APC in SW480 and HT29 cell lines. As shown in Figs. 2A-B, the 1-1337 aa and 1-853 aa forms of Trunc-APC localize to both the cytoplasm and mitochondria, while the 1-1555 aa form has no significant mitochondrial location. We also overexpressed various Flag-tagged Trunc-APC constructs (1-1337, 1-800, and 1-250 aa) in an APC knockout cell line to analyze their mitochondrial localization. Consistently, the 1-1337 aa and 1-800 aa forms exhibited significant co-localization with mitochondria, with the latter showing stronger localization. The shorter 1-250 aa form displayed minimal co-localization (Figure S2A). Additionally, proteinase K digestion assays confirmed that Trunc-APC localizes to the outer mitochondrial membrane (Fig. 2C), which was further supported by the strong co-localization of Trunc-APC with Tom20, a mitochondrial outer membrane protein, as observed in immunofluorescence (Fig. 2D).

To explore the partners of Trunc-APC in mitochondrial, we overexpressed Trunc-APC (1-800 aa) in HEK293T cells, conducted Trunc-APC immunoprecipitation (IP) to investigate candidates through mass spectrometry (Figure S3A). After overlapping with human mitochondrial localization proteins from MitoCarta3.0 and innate immunity-related genes from InnateDB, four proteins (DNM1L, PDE12, MAVS, and C1QBP) were screened out (Fig. 3A). Among these, the mitochondrial antiviral signaling protein (MAVS), a critical signaling hub in the innate immune response [29], draws our attention and were validated through exogenous and endogenous IP assays (Figs. 3B-E). IP assay using the mitochondrial fraction of CRC cells further confirmed the interaction between MAVS and Trunc-APC at the mitochondria (Fig. 3F). Additionally, immunofluorescence analysis showed substantial co-localization of Trunc-APC and MAVS (Fig. 3G). In cancer tissue samples from CRC patients with APC truncation mutants, we also observed robust co-localization of Trunc-APC and MAVS (Fig. 3H). Furthermore, MAVS depletion significantly reduced the levels of type I interferon and ISGs induced by mutant APC knockout (Figs. 3I, S3B). Collectively, these findings suggest that MAVS interacts with Trunc-APC in mitochondria, mediating its inhibitory effect on type I interferon production.

Trunc-APC 1-800 Aa interacts with MAVS
To explore the interaction domain of Trunc-APC with MAVS, we employed co-immunoprecipitation (co-IP) to evaluate which variants—1-250, 1-800, and 1-1337—associate with MAVS. All three variants contain an N-terminal oligomerization domain and Armadillo repeats (ARM domain), while only the 1-1337 variant features a 15 amino acid repeat and a partial mutation cluster region (MCR) (Fig. 4A). As illustrated in Fig. 4B, both the 1-1337 and 1-800 variants interact with MAVS, with the 1-800 variant showing a significantly strong interaction. Additionally, we observed that loss of 1-250 aa can obviously suppress the binding of MAVS and Trunc-APC, which suggests that the N-terminal oligomerization domain is vital for Trunc-APC binding with partners (Fig. 4C). But loss of ARM domain almost totally inhibits the interaction. These data suggest that both the oligomerization and ARM domain are required for Trunc-APC binding with MAVS.
Furthermore, we generated a series of truncated MAVS plasmids (Fig. 4D). Through co-IP assays, we observed that loss of the transmembrane (TM) domain abrogated the interaction with Trunc-APC 1-800, indicating that the TM domain is required for this interaction. Reports indicate that the TM domain at the C-terminus of MAVS is crucial for its localization in mitochondria [30]. Mariana Brocardo et al. identified the ARM domain as the critical region of Trunc-APC located in mitochondria [28]. In addition, the transcriptional and protein levels of Trunc-APC and MAVS do not affect each other (Figure S4 A-C). Interestingly, mitochondrial expression of MAVS significantly increased following the knockout of mutant APC, while mitochondrial localization of Trunc-APC was enhanced by the knockdown of MAVS (Figure S4 D-E). These findings suggest that Trunc-APC and MAVS inhibit each other’s mitochondrial localization, potentially due to their binding domains acting as mitochondrial targeting sequences (MTS).

Trunc-APC inhibits MAVS ubiquitination and RIG-I-MAVS association
Ubiquitination exerts a dual regulatory role in MAVS, with distinct linkages—such as K48, K63, and K27—targeted to specific sites that mediate its proteasomal degradation, aggregation, and the recruitment of downstream signaling molecules, respectively [31]. Through endogenous co-immunoprecipitation (Co-IP) assays in SW480 and HT29 cells, we observed that MAVS ubiquitination significantly increased in Trunc-APC knockout cells compared to control cells (Fig. 5A, S5A). Exogenous Co-IP assays yielded similar results (Figure S5B). Additionally, we confirmed a reduction in MAVS ubiquitination upon overexpression of Trunc-APC 1-800 in HEK293T cells (Fig. 5B).
To investigate the types of ubiquitin chains on MAVS affected by Trunc-APC, we employed plasmids encoding HA-tagged ubiquitin mutants: K63 (K63-Ub), K48 (K48-Ub), and K27 (K27-Ub). Each mutant replaced all lysine residues with arginine, except at designated positions. Overexpression of Trunc-APC 1-800 in HEK293T cells resulted in decreased levels of both K63-linked and K48-linked polyubiquitination of MAVS, with K63-linked polyubiquitination remaining at a notably higher level compared to K48 type. In contrast, K27-linked ubiquitination of MAVS remained unchanged (Fig. 5B). We validated these findings in SW480 cells, where Trunc-APC knockout or knockdown elevated K63- and K48-linked polyubiquitination levels, with K63-linked polyubiquitination consistently higher than that of K48 (Fig. 5C-D, S5C). K63-linked polyubiquitination of MAVS plays a critical role in its aggregation and subsequent signaling [32, 33]. Through semidenaturing detergent agarose gel electrophoresis (SDD-AGE) assay, we observed that loss of Trunc-APC induced an increase in MAVS aggregates in SW480 cells (Figure S5D).
It was reported that unanchored K63-linked polyubiquitin chains, which accumulate on MAVS, are recognized by RIG-I to initiate the RIG-I-MAVS signaling cascade [34]. Through both endogenous and exogenous co-immunoprecipitation, we confirmed the interaction between Trunc-APC and RIG-I, which remained unaffected by poly(I: C) stimulation or MAVS knockdown (Fig. 5E, S5E-F). Notably, only the Trunc-APC 1-800 but not 1-1337 and 1-250 variants can engage with RIG-I (Figure S5H). We also generated myc-tagged RIG-I truncated mutants that preserved the N-terminal 2CARD domain or had the 2CARD domain deleted (△2CARD) (Figure S5G). As illustrated in Figure S5I, both the full-length RIG-I and the 2CARD variant associated with Trunc-APC 1-800, suggesting that RIG-I interacts with Trunc-APC through the 2CARD domain, which also serves as the binding site for MAVS. Therefore, we hypothesized that Trunc-APC interferes with the association between MAVS and RIG-I. As shown in Fig. 5F, MAVS interacted with both Trunc-APC and RIG-I upon poly(I: C) stimulation. Notably, overexpression of Trunc-APC 1-800 significantly reduced the binding between RIG-I and MAVS. Similarly, proximity ligation assay and immunofluorescence revealed clear red PLA puncta in HEK293T cells, indicating the interaction between MAVS and RIG-I in poly(I: C) stimulated cells. However, the puncta count significantly diminished with the overexpression of Trunc-APC 1-800, correlating with reduced co-localization of PLA puncta and mitotracker (Fig. 5G). Together, these findings suggest that Trunc-APC suppresses MAVS ubiquitination and disrupts the association between RIG-I and MAVS.

Combination therapy of 5-AZA and TSA enhances the type I IFN production in Trunc-APC deficient CRC cells
DNMTi (e.g., 5-Azacytidine and decitabine) were the first agents proposed to induce viral mimicry by activating innate immune responses in cancer cells [15]. Subsequent studies have revealed several non-DNMT targets that can induce viral mimicry, including histone-modifying enzymes and other proteins not involved in covalent modifications [14]. Moreover, to maintain sustained dsRNA expression and counteract the compensatory histone modifications that occur as DNA methylation decreases (a process known as the “epigenetic switch”), histone-targeting agents are frequently combined with DNMTis [35]. This combination ensures prolonged activation of immune signaling pathways, leading to a more effective anti-tumor response [36–38].
To explore potential effective therapy of innate immunity induced cell death through synergistically with targeting Trunc-APC in colorectal cancer (CRC) cells, we conducted drug treatment across five groups: 5-azacytidine (AZA) alone, the histone deacetylase inhibitor Trichostatin A (TSA) alone, the EZH2 inhibitor GSK343 alone, a combination of AZA and TSA, and a combination of AZA and GSK343. Figure 6A illustrates the fold change in mRNA levels of CCL5, CXCL9, and IFNβ before and after treatment, calculated using statistical analysis. Our results demonstrate that only the combination of AZA and TSA significantly upregulates the transcription levels of CCL5, CXCL9, and IFNβ in Trunc-APC knockout cells compared to wild-type cells (Fig. 6A, S6A). In the earlier part of this study, we showed that Trunc-APC knockout induces type I interferon and ISGs, alleviating MAVS inhibition. However, effective activation of innate immune signaling also requires upregulation of dsRNA sources. Through assessing various ERVs, we identified increased levels of ERVL, ERVH2, ERVH4, ERVH7, MLT1C627, MER4D, and MER21C following Trunc-APC knockout (some depicted in Figure S6B), suggesting that knocking out Trunc-APC enhances dsRNA formation. However, the induction of these ERVs was not further enhanced by the combination of AZA and TSA, indicating that the combination may stimulate type I IFN production through alternative repetitive elements, rather than by upregulating these specific ERVs (Figure S6B). Furthermore, we determined optimal drug concentrations by using the CCK8 assay. The combination of 1 µM AZA for 48 h followed by 500 nM TSA for 24 h resulted in the lowest relative cell viability (Figure S6C).
We then monitored cellular death following drug treatment and used flow cytometry to identify apoptosis. Trunc-APC knockout cells were notably more sensitive to the combination of AZA and TSA, exhibiting significantly higher apoptosis compared to wild-type cells (Fig. 6B-C, S6D). The combination also stimulated type I interferon and ISGs in HEK293T cells, which could be abrogated by the ectopic expression of Trunc-APC 1-1337 aa, 1-800 aa, and the ARM domain (Fig. 6D).
For in vivo studies, we implemented a dosing schedule for nude mice, starting ten days after the subcutaneous injection of either Trunc-APC knockdown or control CRC cells. The treatment included two cycles: AZA at 2.5 µg/kg daily for three days, followed by TSA at 1.25 µg/kg daily for two days, with a repeat cycle afterward (Fig. 6E). After 25 days, tumor analysis showed that Trunc-APC knockdown significantly inhibited tumor growth and enhanced sensitivity to the AZA and TSA combination, resulting in the lowest tumor weight, volume and proliferative rate in the drug-treated Trunc-APC knockdown group (Figs. 6F-J). In addition, the treatment has no significant effect on the body weight of the mice (Figure S6E).

In summary, our findings indicate that Trunc-APC can localize to the outer mitochondrial membrane, inhibits MAVS ubiquitination and the RIG-I-MAVS interaction in CRC cells. The combination of AZA and TSA enhances type I interferon production synergistically with mutant APC knockout, leading to increased cell apoptosis (Fig. 7), which provides a promising therapeutic strategy for CRC.

Discussion
Trunc-APC has been reported to play an oncogenic role in CRC pathogenesis by enhancing invasion ability of CRC cells. However, the relationship between Trunc-APC and CRC immune microenvironment remains unclear. Multiple studies have revealed that APC mutation in immune cells, such as T lymphocytes [39], macrophages [40], can impair their immune surveillance function. Nevertheless, whether and how Trunc-APC regulates the innate immunity of CRC cells has yet to be elucidated. In the present study, we revealed that the innate immune response is suppressed in CRC cells expressing Trunc-APC, as Trunc-APC interacts with MAVS and inhibits its ubiquitination, RIG-I binding, and aggregation. Furthermore, under mutant APC knockout conditions, the combination of 5-azacytidine and trichostatin A triggers robust and prolonged type I interferon signaling and induces apoptosis of CRC cells.
Full-length APC is a multifunctional protein distributed across various cellular compartments, including the plus ends of microtubules, actin-dependent membrane ruffles, the mitotic spindle, centrosomes, and nuclear-cytoplasmic shuttling [41]. Since Trunc-APC lacks the SAMP repeats necessary for axin binding, the basic region required for microtubule interaction, and the C-terminal domain for EB binding, it results in dysfunction of the β-catenin destruction complex, spindle formation, and mitotic progression [6]. Despite these deficiencies, Trunc-APC retains the ability to shuttle between the nucleus and cytoplasm [42] and to localize to centrosomes [43] and membrane contact sites [44]. Brocardo et al. reported that Trunc-APC particularly localizes to the mitochondria and interacts with Bcl-2, potentially contributing to cell survival [28]. In our study, we verified the mitochondrial localization of Trunc-APC through subcellular fractionation and immunofluorescence, and we demonstrated for the first time its presence on the outer mitochondrial membrane, where it interacts with mitochondrial antiviral signaling protein (MAVS). The difference in subcellular localization between Trunc-APC and full-length APC may explain the innate immune response repression specific to Trunc-APC. And the poor mitochondrial localization of full-length APC may be attributed to auto-inhibition between the ARM domain and residues 1362 to 1540 (APC-2,3 repeats) [45].
Moreover, the Armadillo (ARM) domain of APC, which contains seven armadillo repeats, is conserved across species and retained in most Trunc-APC [46]. The APC ARM sequence has been shown to bind various partners, including Asef, IQGAP1, KAP3, and PP2A B56α, thereby influencing APC localization to different subcellular regions [41, 47]. It is postulated that the ARM domain serves as a mitochondrial targeting sequence for Trunc-APC [28]. MAVS is primarily localized on the mitochondrial outer membrane, with additional presence on peroxisomes and mitochondrial-associated endoplasmic reticulum membranes [31]. Upon detecting PAMPs, such as viral nucleic acids and proteins (e.g., RNA and DNA), PRRs like RIG-I and MDA5 become activated and interact with MAVS, thereby initiating the activation of transcription factors including IRF3/7 and NF-κB. This signaling cascade promotes the expression of various proinflammatory factors and antiviral genes, such as IFN and ISGs [48]. Our study found that APC1-800 and APC251-800 but not APC1-250 interact with MAVS. Furthermore, the interaction involving APC1-800 was stronger than that with APC251-800, which may be attributed to differences in mitochondrial localization or the requirement of APC’s N-terminal oligomerization domain. In addition, we observed that APC1-800 cannot interact with MAVS lacking the transmembrane (TM) domain, which is crucial for MAVS localization on the mitochondrial outer membrane [30]. Similarly, it has been reported that metabolites such as lactate, as well as proteins, interact with the MAVS TM domain to suppress the innate immune response [49–52]. We hypothesize that the APC ARM domain interacts with the MAVS TM domain, which could explain the mutual inhibition of their mitochondrial localization. However, further studies are needed to pinpoint the specific amino acids responsible for the interaction between Trunc-APC and MAVS, which may represent a potential therapeutic target.
Roulois et al. first introduced the concept of “viral mimicry,” based on findings that low-dose 5-AZA-CdR treatment induces colorectal cancer-initiating cells to mimic virus-infected cells, activating the MDA5/MAVS/IRF7 pathway and inhibiting tumor growth [15]. Increasing studies have suggested that viral mimicry could be a promising therapeutic strategy, as it activates the IFN response, enhances tumor immunogenicity, and induces immunogenic cell death in a dose-dependent manner [14]. According to this theory, cancer-initiating mutations or drug-induced epigenetic reprogramming serve as “priming events” that increase cytosolic dsRNA from various endogenous sources [53]. Zhou et al. found that KRAS impairs DDX60-mediated dsRNA accumulation in CRC cells [54]. Our results suggest that Trunc-APC may function as a “priming event” by inhibiting the accumulation of ERVs and suppressing MAVS activation. Moreover, cancer cells often evolve mechanisms to limit cytosolic dsRNA accumulation and prevent IFN levels from exceeding tolerance thresholds. Consequently, “booster events” such as DNMT inhibition, HDAC inhibition, and spliceosome-targeted therapies are essential to trigger a lethal IFN response and promote apoptosis. In epigenetic therapy, drug combinations are generally more effective than single agents in circumventing the “epigenetic switch” [35, 38]. Our findings further support the combination therapies, as combining 5-Azacytidine and Trichostatin A, in synergy with Trunc-APC deletion, was particularly effective in inducing cell apoptosis and inhibiting tumor growth. However, their synergistic effect may arise from different mechanisms, as they did not exhibit a cumulative effect on elevating the major ERV types. We therefore speculate that deletion of Trunc-APC releases MAVS inhibition, while the drug combination induces sustained dsRNA accumulation. Despite these findings, several limitations should be acknowledged. First, most results were obtained from colorectal cancer cell lines, and further validation in animal models and larger patient cohorts will be necessary to establish clinical relevance. Second, although we demonstrated the interaction between Trunc-APC and MAVS, the precise amino acid residues and structural determinants remain undefined. Third, while the combination of 5-azacytidine and trichostatin A showed promising synergy with mutant APC knockout in vitro, its in vivo efficacy, safety, and potential immunosuppressive effects on other immune cell populations remain to be determined. Finally, although Trunc-APC represents a critical therapeutic target in CRC [55], strategies such as PROTAC-mediated degradation are still at an early stage, and their translational feasibility and safety require careful evaluation in future studies.

Conclusions
In this study, we identified a novel function of Trunc-APC involved in the innate immune response. Trunc-APC partially localizes to the outer mitochondrial membrane, where it interacts with MAVS and limits the type I interferon cascade by reducing K63-linked polyubiquitination of MAVS and inhibiting the MAVS-RIG-I interaction. Importantly, targeting Trunc-APC enhances the innate immune response triggered by 5-azacytidine and trichostatin A, leading to increased apoptosis and tumor inhibition. These results provide a potential therapeutic strategy for CRC, especially those with Trunc-APC.

Supplementary Information
Below is the link to the electronic supplementary material.