Cancer vaccine overcomes immune evasion of nasopharyngeal carcinoma by restoring MHC-I through transcriptional regulation of NLRC5 Authors: Chai Phei Gan, Sau Yee Kok, Bernard Kok Bang Lee, Natasha Zulaziz, Sok Ching Cheong, Natalia Savelyeva, Kue Peng Lim Published: 2025 ABSTRACT BackgroundNasopharyngeal carcinoma (NPC) is considered an immune-hot tumour. However, 30–80% of cases exhibit downregulation of antigen processing and presentation machinery (APM), enabling the evasion of host immunosurveillance. While cancer vaccines aim to trigger immune responses against tumour antigens, their efficacy in APM-deficient tumours remains uncertain. This study evaluates the efficacy of cancer vaccine targeting tumour-associated antigens in APM-downregulated NPC cells in vitro and further validates the modulation of APM pathways in vivo.MethodsAPM status was evaluated via differential gene expression analysis of 42 NPC tumours and 4 non-NPC tissues using a 17-gene APM signature. Genes downregulated in NPC and associated with reduced MHC-I expression were identified. MHC-I expression was further examined by immunohistochemistry in 35 tumours and 5 non-NPC tissues. The effect of cancer vaccine on APM gene expression was examined by co-culturing peptide-trained T cells with NPC cells. Next, T cell-mediated cytotoxicity was assessed in an APM-deficient model generated by siRNA-mediated knockdown of NLRC5. Finally, cancer vaccine-induced modulation of APM genes was validated in a poorly immunogenic mouse tumour model.ResultsAmong the 17-gene APM signature, NLRC5 was the most significantly downregulated gene in NPC and strongly correlated with reduced MHC-I expression. Immunohistochemistry confirmed MHC-I downregulation in 63% of tumours. Co-culture of NPC cells with peptide-trained T cells upregulated NLRC5, and key MHC-I assembly genes (TAP1 and B2M), enhancing MHC-I expression and antigen-specific cytotoxicity in NPC cells expressing the target antigen. Notably, the knockdown of NLRC5 was reversed upon co-culture with peptide-trained T cells, resulting in T cell-mediated cytotoxicity. In vivo, cancer vaccine treatment consistently induced APM gene expression, including NLRC5, supporting its potential in restoring antigen presentation in NPC.ConclusionsThis study demonstrates that peptide-trained T cells can upregulate NLRC5 and MHC-I expressions on tumour cells, thereby restoring antigen presentation and enhancing tumour immunogenicity. These findings underscore the therapeutic potential of cancer vaccines in treating APM-downregulated NPC.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12967-025-07418-x. Background Nasopharyngeal carcinoma (NPC) is a subset of head and neck cancer, and exhibits a distinct population-specific prevalence, occurring nearly 40 times more frequently in Asians than in Caucasians [1]. Due to its non-specific early symptoms, over 80% of patients are diagnosed at a locally advanced stage or with distant metastasis [2]. While early and locally advanced NPC respond well to radiotherapy, recurrent or metastatic NPC carries a poor prognosis, with a median overall survival of 20 months [3]. Immune checkpoint inhibitors (ICIs) are increasingly used to treat patients with recurrent or metastatic disease, considering that NPC is an immune hot tumour characterised by its association with Epstein-Barr virus (EBV) infection, dense lymphocytic infiltration and elevated PD-L1 expression [4]. However, overall response rates to anti–PD-1 monotherapy remain modest at 10–43% and fail to produce a durable clinical benefit [5, 6]. One proposed reason for this limited efficacy is the deficiency in antigen processing and presentation machinery (APM), which is essential for effective T cell-mediated anti-tumour immunity [7–9]. Downregulation of the APM pathway has been reported in 30–80% of NPC patients [10–13]. Specifically, the expression of major histocompatibility complex (MHC-I) molecules, including the HLA-A and HLA-B, as well as their associated components such as TAP1 and TAP2, is significantly reduced in NPC compared to normal tissue. APM downregulation can result from “hard/irreversible” genomic alterations, such as chromosomal structural variants, single-nucleotide variants and mutations, which lead to inactivation of HLA-A and HLA-B in 13% and 9% of NPC, respectively [10, 14]. “Soft/reversible” alterations driven by epigenetic, transcriptional and posttranscriptional mechanisms may also contribute to APM downregulation, although their prevalence in NPC remains unclear [15, 16]. Additionally, EBV-encoded proteins such as BNLF2a and BILF1 can also cause APM downregulation [10, 17]. MHC-I downregulation is associated with poorer survival outcome, likely due to compromised antigen presentation, enabled cancer cells to evade immune detection [11, 14, 18]. Recent research has demonstrated the potential of targeting APM to overcome immune escape in cancer by promoting tumour antigenicity, enhancing immunogenicity, and stimulating T cell-mediated tumour killing. For example, platinum-based chemotherapies have been shown to upregulate MHC-I on both tumours and dendritic cells in preclinical studies, thereby improving the efficacy of anti-PD1 therapy [19]. In NPC, combining anti-PD1 therapy with gemcitabine and cisplatin has led to improved response rates of 70–91% [6, 20], likely due to chemotherapy-induced immune modulation. This effect may be mediated by increased MHC-I antigen presentation, potentially through activation of pathways such as IFNβ and NF-κB [21]. However, despite these immunostimulatory effects, the combination of chemotherapy and anti-PD1 is associated with substantial toxicity, with grade ≥3 adverse events reported in 57–94% of NPC patients. These limitations underscore the need for alternative strategies that can enhance antigen presentation with fewer side effects. Cancer vaccines targeting tumour-associated antigens (TAAs) offer a promising approach to provoke an immune response by priming tumour-reactive T cells against existing tumours [22, 23]. Notably, MAGED4B and FJX1 are highly immunogenic TAAs that are overexpressed in NPC [24, 25]. Vaccines targeting these TAAs significantly inhibited tumour growth when combined with anti-PD1 therapy [26]. In this study, we investigated the potential of cancer vaccines targeting MAGED4B and FJX1 in modulating APM to restore anti-tumour immunity in NPC. Specifically, we demonstrated that co-culturing peptide-trained T cells and NPC cells endogenously expressing these TAAs enhances antigen presentation through the upregulation of NOD-like receptor family CARD domain containing 5 (NLRC5) and MHC-I in the APM pathway to induce T cell-mediated cytotoxicity. Material and methods Bioinformatics analyses of APM gene expression The RNA sequencing dataset GSE68799, comprising 42 NPC and 4 non-NPC tissues, was obtained from the Gene Expression Omnibus database. This dataset was selected for its large number of tumour samples and its accessibility through a public repository. A curated APM signature consisting of 17 genes [27] was used to identify differentially expressed genes (DEGs) between NPC and non-NPC tissues using DESeq2, with thresholds set at log2-fold change < −1 or > 1 and p-value < 0.05 [28]. A heatmap illustrating gene expression patterns was generated using the Morpheus web tool (https://software.broadinstitute.org/morpheus). The APM genes analysed in this study form an integrated cascade that governs peptide digestion and MHC loading, beginning with transcriptional regulators (NLRC5, CIITA, RFX5) and proceeding through the immunoproteasome (PSMB10 and PA28/PSME subunits), peptide translocation (TAP1 and TAPBPL), endoplasmic reticulum trimming (ERAP1), chaperoning and loading (CALR, HSP90 family, B2M) and finally surface presentation to T cells [27]. APM genes that were significantly downregulated in NPC tissues were further analysed using Pearson correlation to assess their association with MHC-I genes (HLA-A, HLA-B, and HLA-C). Additionally, the KEGG antigen processing and presentation pathway was analysed in the GSE68799 dataset to assess the prevalence of HLA gene downregulation in NPC. Immunohistochemistry staining Formalin-fixed, paraffin-embedded (FFPE) tissues were used to assess target protein expression via immunohistochemistry (IHC) staining with the EnVision+ system (Dako, Germany). The samples included whole tumour tissue sections (n = 14) and a tissue microarray (NH1001a; TissueArrays.com, USA) comprising additional tumour samples (n = 21) and non-NPC tissues (n = 5). The whole tumour tissue sections were obtained from Penang General Hospital upon medical ethics approval (NMRR-09–944-4848). Non-NPC tissues refer to nasal polyps. The surface epithelium of nasal polyps is typically pseudostratified ciliated columnar, closely resembling healthy nasopharyngeal tissue. Importantly, nasal polyps lack neoplastic transformation and extensive immune infiltration, making them suitable control for IHC grading. FFPE tissue sections were deparaffinized and dehydrated in decreasing grades of ethanol. Antigen retrieval was performed by boiling the tissues in the respective antigen retrieval buffer in the microwave. The Tris/EDTA pH 9.0 buffer was used for antigen retrieval of FJX1, while sodium citrate pH 6.0 buffer was used for antigen retrieval of MHC-I, MHC-II and MAGED4B. Tissues were incubated with 1% bovine serum albumin in Tris-buffered saline to prevent unspecific binding. The primary antibodies were anti-HLA Class-I monoclonal antibody, clone EMR8-5 (1:100; Abcam, USA), anti-HLA Class-II monoclonal antibody clone 6C6 (1:100; Abcam, USA), anti-MAGED4B (1:100; Sigma-Aldrich, USA) and anti-FJX1 (1: 200; Sigma-Aldrich, USA). After peroxidase blocking, antigen expressions were visualized using the Envision+ Dual Link System HRP (DAB+) kit (Dako, Germany). The immunoreactivity of MHC-I was evaluated based on the criterion established by the Human Leukocyte Antigen and Cancer component of the 12th International Histocompatibility Workshop [29]. The staining intensity of MHC-I in the lymphocytes and vessel endothelium is regarded as internal positive controls. Tumour cells stained as similar or greater intensity than the internal positive control was graded, and area positivity score was given as 0 (no expression): less than 25%; 1 (low expression): 25 to 75% and 2 (high expression): greater than 75%. On the other hand, the immunoreactivity of MAGED4B and FJX1 was scored based on a 4-point intensity scoring system: 0 = negative expression; 1 = weak positive; 2 = moderate positive; 3 = strong positive. Score 0 and 1 were grouped as low expression, 2 and 3 were grouped as high expression, as reported previously [25]. Cell culture and cell transfection NPC-43 and NPC-53 cells, derived from recurrent tumours [30], and NPC-268 cell derived from a primary tumour [31] were cultured in RPMI-1640 medium (Gibco, USA) containing 10% fetal bovine serum (FBS; Gibco, USA), 100 IU Penicillin/Streptomycin (Gibco, USA) and 4 µM ROCK inhibitor (Y-27632; Enzo Life Sciences, USA). Cells were maintained at 37 °C in 5% CO2 incubator. NPC cell lines used in this study were authenticated by short tandem repeat DNA profiling to confirm their identity and rule out cross-contamination. Additionally, they were tested and confirmed to be mycoplasma-free. To knockdown NLRC5 expression, NLRC5 siRNA (5’ GGA CAC CUG GCA GUC UUU CAU UCA U3’; Invitrogen, USA) was transfected into NPC cells using Lipofectamine™ RNAiMAX (Invitrogen, USA) according to the manufacturer’s protocol. For control, cells were transfected with non-targeting (NT) siRNA (Invitrogen, USA). To evaluate the transfection efficiency of NLRC5 knockdown, cells were harvested at the indicated time points for real-time PCR and western blot assays. The mouse melanoma cell line B16F10 expressing the human-HLA-A2 gene (B16F10–A2) was cultured in RPMI-1640 medium (Gibco, USA) containing 10% FBS and 100 IU Penicillin/Streptomycin (Gibco, USA), at 37 °C in an incubator with 5% CO2. To express MAGED4B and FJX1 in B16F10–A2 cells, the pLENTI-6.3 plasmids carrying MAGED4B and FJX1 were simultaneously transfected into the cells using Lipofectamine according to manufacturer’s instruction. Cells were further selected using antibiotics (G418 and blasticidin) for one month. Western blotting was performed to confirm the expression of MAGED4B and FJX1 in B16F10–A2 prior to their use in the mouse experiment. Generation of MAGED4B- and FJX1-specific T cells Peripheral blood mononuclear cells (PBMCs) were harvested from blood samples collected into cellular preparation tubes (Becton Dickinson, USA) from three healthy donors (HD30, HD31 and HD32). Red blood cells were removed using ammonium chloride potassium (Gibco, USA) lysing buffer. The use of peripheral blood samples in this study was approved by the Medical Ethics Committee, Faculty of Dentistry, University of Malaya [DF OS 2101/0002/2179(L). CD8+ T cells were isolated through negative selection using the human CD8+ T cell isolation kit (Miltenyi Biotec, Germany). Both CD8+ and non-CD8+ cell fractions were collected, with the non-CD8+ fraction serving as autologous antigen-presenting cells (APC). The CD8+ T cells were cultured in RPMI medium supplemented with 10% FBS and 30 ng/ml anti-CD3 (OKT3 clone, Miltenyi Biotec, Germany). Antigen-specific T cells were generated using a modified protocol adapted from Erskine et al. 2012 [21]. CD8+ T cells were stimulated with 10 µg/mL of overlapping peptide pools derived from MAGED4B (M-OPP) and FJX1 (F-OPP; JPT Peptide Technologies, Germany), respectively. CD8+ T cells were supplemented with 50 IU/mL of recombinant human interleukin-2 (IL-2; R&D Systems, USA) from day 3 onwards. Restimulation with autologous APC cells was performed at day 7 and 14. To prepare APCs, the non-CD8+ fraction was pulsed with 10 µg/mL of M-OPP or F-OPP, followed by mitomycin C treatment to inhibit cell proliferation. CD8+ T cells that completed 3 rounds of stimulation were hereafter referred to as peptide-trained T cells and were used in subsequent functional assays. Modulation of APM gene expression by peptide-trained T cells The expression of APM genes in NPC cell lines following exposure to peptide-trained T cells was assessed in vitro. NPC cells (target) were seeded overnight at 3 × 105 in a 6-well plate, then co-cultured with peptide-trained T cells (effector) at an effector-to-target ratio of 10:1 for 24 hours. NPC cells stimulated with 100 ng/mL IFNγ (R&D Systems, USA) served as a positive control. Following incubation, total RNA was extracted from NPC cell lines using RNeasy Micro kit (Qiagen, Germany), and cDNA was synthesized from 1 µg of total RNA using BIOLINE SensiFAST reagent kit (Bioline Reagent Ltd, UK). Gene expression analysis was performed on the 7500 Real-Time PCR System (Applied Biosystems, USA) using specific PCR primers (Supplementary Table 1) and Power Sybr Green PCR Master Mix (Applied Biosystems, USA). Relative mRNA levels of NLRC5, CD274, TAP1, and B2M were obtained after normalisation to GAPDH. Flow cytometry analysis The expression of human leukocyte antigen ABC (HLA-ABC), corresponding to MHC-I was analysed by flow cytometry. Firstly, HLA-A typing was performed on PBMCs and NPC cell lines using specific antibodies and isotype controls. Briefly, cells were stained with FVS780 viability dye (BD Biosciences, USA) for 15 minutes followed by incubation with the respective primary antibodies: PE-conjugated mouse anti-human HLA-A2 (Clone BB7.2, BD Biosciences, USA), unconjugated HLA-A11/A3 IgG (One Lambda, USA), unconjugated HLA-A2 and A24 IgM (One Lambda, USA), and unconjugated HLA-A24/A11/A2403 IgM (One Lambda, USA). For isotype controls, cells were stained with mouse IgG2b-UNLB (SouthernBiotech, USA) and mouse IgM-UNLB (SouthernBiotech, USA). Cells stained with PE-conjugated mouse anti-human HLA-ABC (Clone W6/32, BD Biosciences, USA, 567582) served as the positive control. Unconjugated primary antibodies and isotype controls were further incubated with the corresponding secondary antibodies: Goat F(ab’)2 anti-mouse IgG (H+L), human ads-PE (SouthernBiotech, USA), and Goat F(ab’)2 anti-mouse IgM, human ads-PE (SouthernBiotech, USA). After staining, a minimum of 1 × 105 cells were analysed using the LSRFortessa™ X-20 Cell Analyzer (BD Biosciences, USA), with gating against unstained and isotype controls. To assess total MHC-I expression on NPC cells, cells were stained with FVS780, followed by HLA-ABC staining, and analysed as described above. The median fluorescence intensity (MFI) of HLA-ABC was quantified using FlowJo software (BD Biosciences, USA) to determine the relative abundance of HLA-ABC among the NPC cell lines. Western blot analysis NPC cell lines were lysed in RIPA buffer (Thermo Scientific, USA) supplemented with 1x Halt protease and phosphatase inhibitors cocktail (Thermo Scientific, USA). Protein concentrations were measured using Pierce BCA protein assay kit (Thermo Scientific, USA). Protein lysates were denatured by heating at 98 °C for 10 minutes. Equal amounts of protein were separated on a 10–12% SDS-polyacrylamide gel and transferred to PVDF membranes. The membranes were blocked with 5% skimmed milk and probed with antibodies against MAGED4B (1:1000, HPA003554, Sigma-Aldrich, USA), FJX1 (1:1000, HPA059220, Sigma-Aldrich, USA), HLA Class I A/B/C (1:1000, clone EMR8-5, Ab70328, Abcam, USA), and NLRC5 (1:500, clone 3H8, MABF260, Millipore, USA). Anti-GAPDH antibody (1:1000, Proteintech, USA) and anti-tubulin (1:1000, T5168, Sigma-Aldrich, USA) were used as a loading control. The Immobilon Forte Western HRP substrate (Milipore, USA) was used for chemiluminescence detection and images taken using Western blot imager (Azure Biosystem, USA). Suberoylanilide hydroxamic acid (SAHA) treatment assay NPC cells were seeded at 3x105 cells per well in a 6-well plate overnight and serum-starved for 24 hours. Cells were treated with SAHA (Cayman Chemical, USA) at a final concentration of 1 µM, 5 µM and 10 µM for 24 hours in the CO2 incubator. Subsequently, cells were harvested to assess MHC-I protein expression using HLA-ABC MFI quantification by flow cytometry and Western blotting. T cell-mediated cytotoxicity assay The cytotoxic activity of peptide-trained T cells against NPC cell lines was assessed using the xCELLigence Real-Time Cell Analyzer (RTCA; Agilent, USA). Briefly, background impedance values of 96 well E-Plate (Agilent, USA) filled with medium were obtained. NPC target cells were seeded into the E-plate (4x104 cells/well for NPC-43 and NPC-53 or 2x104 cells/well for NPC-268),and allowed to adhere for 24 hours in the RTCA. peptide-trained T cells were added to target cells at a 10:1 effector-to-target ratio. Impedance values were recorded hourly for 24 hours. Cell index was normalized to the time of effector T cells addition, and cytolysis was calculated in relative to control cells without effector T cells. Animal experiment AAD mice (transgenic for HLA-A2.1/H2-Dd allele) were bred and housed in the Animal Facility Centre of National University Malaysia. AAD mice (6–10 weeks) were inoculated with 1x105 B16F10–A2 cell line expressing MAGED4B/FJX1 and were randomised into 2 groups to receive (i) 40 µg plasmid DOM (pLDOM, control); (ii) 40 µg each of doggybone DNA vaccine encoding MAGED4B/FJX1 (DNA vaccine). Intramuscular administration of the DNA vaccine, aided by electroporation, was perform on day 5 and day 21 post cell inoculation; Tumour size was monitored 2 times/week until day 40. Tumour volumes were evaluated using the formula: volume = ½ (length X width2). Animal study was approved by Experimental Animal Ethics Committee, National University Malaysia (CRM/2020/KUE PENG/25-MAR./1093-APR.-2020-APR.-2023). RNA sequencing and bioinformatic analysis Total RNA was extracted from mouse tumour tissue using the Qiagen RNeasy mini kit (Qiagen, Germany), with purity and integrity assessed via Nanodrop (Thermo Scientific, USA) and the RNA Nano 6000 Assay Kit run on the Bioanalyzer 2100 system (Agilent, USA). RNA library was prepared from 1 μg of total RNA using the NEBNext® UltraTM RNA Library Prep Kit (NEB, USA) and sequenced on Illumina NovaSeq (Illumina, USA) to generate paired-end 150bp reads. Raw reads (FASTQ format) were processed with fastp to remove adapter sequences, poly-N regions, and low-quality reads. Clean reads were mapped to the reference genome using HISAT2. DEGs between the DNA vaccine and control groups were identified using a joint filtering method, with thresholds for up-regulated and down-regulated genes set at log2 fold change < −1 or > 1 and p-value < 0.01 [18]. To further analyse the functional significance of these DEGs, the KEGG pathway enrichment analysis was performed using Metascape (Broad Institute, USA) [22]. A heatmap of genes in the 17-gene APM signature was generated to examine the changes in gene expression between the DNA vaccine and control group. Statistical analysis Statistical analyses for bioinformatics studies are described in the respective sections. Experimental data were analysed using GraphPad Prism 10 (GraphPad Software, USA). IHC scoring of MHC-I expression in NPC tissues was evaluated using Fisher’s exact test to determine associations with clinical parameters. The expression of APM genes in NPC cells stimulated by peptide-trained T cells or IFNγ was compared to control cells using one-way ANOVA followed by Bonferroni’s post-hoc test for multiple comparisons. NLRC5 expression after siRNA knockdown, antigen-specific T cell-mediated killing, tumour growth in animal models, and histopathological assessment of immune cell infiltration were analysed using Student’s t-test. Statistical significance was set at p < 0.05 for all analyses. All data were reproduced with at least two independent experiments. Results Downregulation of NLRC5 expression is associated with decreased MHC-I expression in NPC tumours Gene expression analysis of the NPC dataset (GSE68799) revealed that genes within the APM signature, as defined by Thompson and colleagues [27], were collectively expressed at lower levels in NPC tumours compared to non-NPC tissues (Fig. 1a). Among the 17 APM genes, NLRC5, CIITA, PSMB10, PSME1 and PSMB9 were significantly downregulated in NPC. Specifically, NLRC5 showed the most pronounced reduction at 2.2-fold. Given that NLRC5 functions as a master transcriptional regulator of MHC class I genes, its expression showed a strong positive correlation with the expression of HLA-A, HLA-B, and HLA-C. (Fig. 1b). In comparison, PSMB10 and PSMB9 exhibited moderate positive correlation, while CIITA and PSME1 showed only weak correlation with the expression of HLA-A, HLA-B, and HLA-C (sFig. 1a). Tumours with low NLRC5 expression consistently exhibited low HLAs expression, indicating that NLRC5 downregulation may lead to impaired transcriptional activation of the MHC-I group of genes, potentially decreasing MHC-I expression in NPC. Furthermore, analysis of the KEGG antigen processing and presentation pathway in the GSE68799 dataset revealed that approximately 60–70% of NPC cases exhibit downregulation of HLAs (sFig. 1b). These findings suggest that reduced NLRC5 expression plays a critical role in the coordinated suppression of MHC-I APM in NPC, potentially facilitating tumour escape from immune surveillance.Fig. 1Antigen presentation machinery (APM) is downregulated in NPC. A Heatmap of APM-related genes from GSE68799, showing z-score normalised expression in NPC tumours compared to non-NPC tissues. The two columns on the right represent the average z-scores across all non-NPC and tumour samples. The analysis highlights widespread downregulation of APM-related genes in tumours. Asterisks denote statistically significant differential expression (*p < 0.05, **p < 0.01). B Correlation analysis reveals that NLRC5 downregulation is strongly and positively associated with reduced expression of MHC-I genes (HLA-A, HLA-B, and HLA-C) in NPC Antigen presentation machinery (APM) is downregulated in NPC. A Heatmap of APM-related genes from GSE68799, showing z-score normalised expression in NPC tumours compared to non-NPC tissues. The two columns on the right represent the average z-scores across all non-NPC and tumour samples. The analysis highlights widespread downregulation of APM-related genes in tumours. Asterisks denote statistically significant differential expression (*p < 0.05, **p < 0.01). B Correlation analysis reveals that NLRC5 downregulation is strongly and positively associated with reduced expression of MHC-I genes (HLA-A, HLA-B, and HLA-C) in NPC MHC-I downregulated tumours overexpress MAGED4B and FJX1 To validate the DEG findings indicating MHC-I downregulation, we assessed MHC-I and MHC-II protein levels by IHC in an independent cohort of 35 NPC and 5 non-NPC tissues. The demographic and clinical characteristics of NPC patients are shown in Table 1. In non-NPC tissues, MHC-I expression was observed in normal epithelial cells at levels that are comparable or higher than those in the surrounding stroma. By contrast, NPC tumour cells exhibited weaker MHC-I membrane staining compared to lymphocytes and endothelial cells in the surrounding stroma, indicating MHC-I downregulation in tumours (Fig. 2a). MHC-I downregulation was detected in 63% (22/35) of NPC cases, and none in the non-NPC tissues (Fig. 2b). The IHC findings are consistent with our gene expression analysis, confirming that MHC-I is significantly downregulated in about two-thirds of NPC cases. However, MHC-I downregulation was not associated with disease stage or lymph nodes involvement (Table 2). Meanwhile, MHC-II downregulation showed minimal differences between NPC and non-NPC tissues (29% vs 40%). Given the higher prevalence of MHC-I downregulation in NPC, further investigation into its expression and modulation was conducted in this study.Fig. 2Validation of MHC-I downregulation in NPC by IHC staining. A Representative micrograph of MHC-I and MHC-II staining in NPC and non-NPC tissues. Insets show 20x magnification of the selected regions. B Proportion of samples with low MHC expression. In comparison to the control, 63% of NPC exhibited low MHC-I, while 29% exhibited low MHC-IITable 1Demographic and clinical characteristics of NPC patientsVariablesPatients with NPC, N (%)NPC tissues35 (100%)Whole tissue section14 (34.1%)TMA21 (65.9%)AgeRange, years28–84Mean ± S.D.51.2 ± 14.5GenderMale32 (91.4%)Female3 (8.6%)StagingStage I-II22 (62.9)Stage III-IV11 (31.4)Lymph nodes metastasisNegative20 (57.1)Positive13 (37.1)Table 2The association between MHC-I and MHC-II with clinical variablesVariablesMHC-IMHC-IILowHighp-valueLowHighp-valueSample typeNPC tumours22 (62.9)13 (37.1)0.0289*10 (28.6)25 (71.4)0.6266Non-NPC0 (0.0)4 (100.0)2 (40.0)3 (60.0)StagingStage I-II14 (40.0)8 (22.9)0.71365 (14.3)17 (48.6)0.4376Stage III-IV6 (17.1)5 (14.3)4 (11.4)7 (20.0)Lymph nodes metastasisNegative13 (37.1)7 (20.0)0.71713 (8.6)17 (48.6)0.1067Positive7 (20.0)6 (17.1)6 (17.1)7 (20.0)The association between MHC-I and MHC-II expression with clinical variables was examined with the Fisher’s exact test. *A p-value < 0.05 is considered to be statistically significant Validation of MHC-I downregulation in NPC by IHC staining. A Representative micrograph of MHC-I and MHC-II staining in NPC and non-NPC tissues. Insets show 20x magnification of the selected regions. B Proportion of samples with low MHC expression. In comparison to the control, 63% of NPC exhibited low MHC-I, while 29% exhibited low MHC-II Table 1Demographic and clinical characteristics of NPC patientsVariablesPatients with NPC, N (%)NPC tissues35 (100%)Whole tissue section14 (34.1%)TMA21 (65.9%)AgeRange, years28–84Mean ± S.D.51.2 ± 14.5GenderMale32 (91.4%)Female3 (8.6%)StagingStage I-II22 (62.9)Stage III-IV11 (31.4)Lymph nodes metastasisNegative20 (57.1)Positive13 (37.1) Demographic and clinical characteristics of NPC patients Table 2The association between MHC-I and MHC-II with clinical variablesVariablesMHC-IMHC-IILowHighp-valueLowHighp-valueSample typeNPC tumours22 (62.9)13 (37.1)0.0289*10 (28.6)25 (71.4)0.6266Non-NPC0 (0.0)4 (100.0)2 (40.0)3 (60.0)StagingStage I-II14 (40.0)8 (22.9)0.71365 (14.3)17 (48.6)0.4376Stage III-IV6 (17.1)5 (14.3)4 (11.4)7 (20.0)Lymph nodes metastasisNegative13 (37.1)7 (20.0)0.71713 (8.6)17 (48.6)0.1067Positive7 (20.0)6 (17.1)6 (17.1)7 (20.0)The association between MHC-I and MHC-II expression with clinical variables was examined with the Fisher’s exact test. *A p-value < 0.05 is considered to be statistically significant The association between MHC-I and MHC-II with clinical variables The association between MHC-I and MHC-II expression with clinical variables was examined with the Fisher’s exact test. *A p-value < 0.05 is considered to be statistically significant To evaluate the relationship between TAAs expression and MHC-I status, the same set of tissues underwent IHC staining for MAGED4B and FJX1. Both TAAs were found to be overexpressed in 91% of NPC tissues (Table 3a; sFig. 2a), which is in line with previous study [25]. However, about 60% of NPC cases with high TAAs expression exhibited MHC-I downregulation, indicating a need for strategies to restore antigen presentation to enable effective immune surveillance and promote antigen-specific cytotoxicity (Table 3b).Table 3aThe association between target antigens expression in NPC tumours and non-NPC tissuesVariablesMAGED4BFJX1LowHighp-valueLowHighp-valueSample typeNPC tumours3 (8.6)32 (91.4) < 0.0001*3 (8.8)31 (91.2)0.0005*Non-NPC5 (100.0)0 (0.0)4 (100.0)0 (0.0)The association between MAGED4B and FJX1 expression with sample type was examined with the Fisher’s exact test. *A p-value < 0.05 is considered to be statistically significant.Table 3bTAAs expression and MHC-I status in NPC tissuesProtein expressionMAGED4B(n = 35)FJX1(n = 34)LowHighLowHighMHC-ILow1 (2.9%)21 (60%)2 (5.9%)20 (58.8%)High2 (5.7%)11 (31.4%)1 (2.9%)11 (32.4%) Table 3aThe association between target antigens expression in NPC tumours and non-NPC tissuesVariablesMAGED4BFJX1LowHighp-valueLowHighp-valueSample typeNPC tumours3 (8.6)32 (91.4) < 0.0001*3 (8.8)31 (91.2)0.0005*Non-NPC5 (100.0)0 (0.0)4 (100.0)0 (0.0)The association between MAGED4B and FJX1 expression with sample type was examined with the Fisher’s exact test. *A p-value < 0.05 is considered to be statistically significant. The association between target antigens expression in NPC tumours and non-NPC tissues The association between MAGED4B and FJX1 expression with sample type was examined with the Fisher’s exact test. *A p-value < 0.05 is considered to be statistically significant. Table 3bTAAs expression and MHC-I status in NPC tissuesProtein expressionMAGED4B(n = 35)FJX1(n = 34)LowHighLowHighMHC-ILow1 (2.9%)21 (60%)2 (5.9%)20 (58.8%)High2 (5.7%)11 (31.4%)1 (2.9%)11 (32.4%) TAAs expression and MHC-I status in NPC tissues Peptide-trained T cells enhance MHC-I expression in NPC cells To investigate the effect of peptide-trained T cells on MHC-I expression in NPC cells, we first screened a panel of NPC cell lines for baseline expression of MAGED4B, FJX1 and MHC-I. Both MAGED4B and FJX1 were detected across all NPC cell lines (sFig. 3a). Flow cytometry analysis revealed that NPC-268 exhibited the lowest surface MHC-I expression, followed by NPC-43, and NPC-53 (Fig. 3a). All cell lines showed MHC-I induction upon IFNγ stimulation.Fig. 3Modulation of MHC-I expression in NPC cells by IFNγ, peptide-trained T cells, and HDAC inhibitor. A Surface expression of HLA-ABC on NPC cell lines measured by median fluorescent intensity (MFI) of PE-tagged HLA-ABC at baseline (left), and the induction of HLA-ABC by IFNγ treatment detected by Western blotting (right). B Enhancement of HLA-ABC expression by peptide-trained T cells against MAGED4B (M-T) and FJX1 (F-T) after co-culture (left); bar chart shows relative changes of HLA-ABC compared to control cells grown in culture media (CM; right). C Western blot analysis of total HLA-ABC expression following SAHA treatment. D Fold change in surface HLA-ABC expression relative to the control cells grown in CM Modulation of MHC-I expression in NPC cells by IFNγ, peptide-trained T cells, and HDAC inhibitor. A Surface expression of HLA-ABC on NPC cell lines measured by median fluorescent intensity (MFI) of PE-tagged HLA-ABC at baseline (left), and the induction of HLA-ABC by IFNγ treatment detected by Western blotting (right). B Enhancement of HLA-ABC expression by peptide-trained T cells against MAGED4B (M-T) and FJX1 (F-T) after co-culture (left); bar chart shows relative changes of HLA-ABC compared to control cells grown in culture media (CM; right). C Western blot analysis of total HLA-ABC expression following SAHA treatment. D Fold change in surface HLA-ABC expression relative to the control cells grown in CM CD8+ T cells specific for MAGED4B (M-T cells,) and FJX1 (F-T cells) were generated from healthy donors (HD23, HD30 and HD32) carrying at least one of these HLA haplotypes: HLA-A2, HLA-A11 or HLA-A24 (Table S3). These peptide-trained T cells were co-cultured with HLA-A matched NPC cell lines (Table S2), resulting in an approximately two-fold increase in MHC-I expression in NPC cells relative to baseline levels (Fig. 3b). Epigenetic modulation such as DNA methylation and histone deacetylation (HDAC) is known to contribute to MHC-I downregulation in tumours. Based on previously published studies, DNA methylation is considered an unlikely mechanism of MHC-I suppression in the NPC cell lines examined in this study. Specifically, NPC-43 and NPC-268 are identified as hypomethylated, showing reduced global methylation levels relative to the immortalized normal nasopharyngeal epithelial cell line NP460, whereas NPC-53 exhibits a global methylation profile comparable to NP460 [31, 32]. Although histone deacetylation status of these cell lines was not fully characterised, HDAC inhibition was performed as an indirect approach to probe HDAC regulation of MHC-I in these cell lines. SAHA was chosen as the HDAC inhibitor in this study because it is currently being tested in combination with ICI for recurrent and metastatic head and neck cancer [33]. The response to SAHA treatment varied across NPC cell lines. NPC-43 exhibited a dose-dependent increase in total and surface MHC-I expression, indicating the potential of HDAC in mediating MHC-I expression in this cell. Meanwhile, NPC-53 and NPC-268 showed minimal changes in surface MHC-I, with relative fold change below 2 (Fig. 3c). In comparison to SAHA treatment, peptide-trained T cells induced MHC-I upregulation across all three NPC cell lines, suggesting that a vaccine-based approach may offer a more robust strategy for enhancing MHC-I expression in these NPC cell lines. Peptide-trained T cells upregulate MHC-I-associated genes and promote antigen-specific killing of NPC cell lines To elucidate the molecular basis of MHC-I upregulation in NPC by peptide-trained T cells, we analysed the expression of key genes involved in MHC-I transcription and assembly including NLRC5, TAP1, and B2M following co-culture. Given that IFNγ stimulation is known to concurrently induce both MHC-I and CD274 (PD-L1), which can have opposing effects on antitumor immunity [34], we also examined CD274 expression in response to peptide-trained T cells. Compared to control cells, NLRC5, TAP1 and B2M mRNA levels were significantly upregulated in NPC cells following co-culture with peptide-trained T cells, although the magnitude of induction varied across cell lines (Fig. 4a). Notably, CD274 expression was substantially lower following co-culture with peptide-trained T cells compared to direct IFNγ stimulation, indicating that peptide-trained T cells preferentially enhance MHC-I while limiting excessive PD-L1 upregulation, which may help to overcome immune suppression.Fig. 4Peptide-trained T cells enhance MHC-I-associated gene expression and promote antigen-specific cytotoxicity in NPC cells. A Peptide-trained T cells significantly upregulated NLRC5, TAP1 and B2M, with minimal induction of CD274 (PD-L1) compared to IFNγ treatment. Asterisks denote statistically significant differential expression (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns, not significant). B Co-culture of peptide-trained T cells (effector) with NPC cells (target) at a 10:1 effector-to-target ratio induced antigen-specific killing. ORL-195/MAGED4B which overexpresses MAGED4B served as positive control, while ORL-195LacZ, lacking MAGED4B expression served as a negative control. Cytolysis below 20% was considered background noise, indicating no antigen-specific killing Peptide-trained T cells enhance MHC-I-associated gene expression and promote antigen-specific cytotoxicity in NPC cells. A Peptide-trained T cells significantly upregulated NLRC5, TAP1 and B2M, with minimal induction of CD274 (PD-L1) compared to IFNγ treatment. Asterisks denote statistically significant differential expression (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns, not significant). B Co-culture of peptide-trained T cells (effector) with NPC cells (target) at a 10:1 effector-to-target ratio induced antigen-specific killing. ORL-195/MAGED4B which overexpresses MAGED4B served as positive control, while ORL-195LacZ, lacking MAGED4B expression served as a negative control. Cytolysis below 20% was considered background noise, indicating no antigen-specific killing Consistent with these transcriptional changes, antigen-specific T cells killing was observed across all three NPC cell lines following MHC-I upregulation induced by peptide-trained T cells (Fig. 4b). After 24 hours of co-culture, these peptide-trained T cells achieved cytolytic efficiencies ranging from 50% to 60%, exceeding the 20% baseline for non-specific killing, regardless of the differences in basal MHC-I expression in these cell lines. For example, NPC-268, which initially exhibited low MHC-I levels, showed enhanced MHC-I expression and demonstrated susceptibility to T cell-mediated cytotoxicity, achieving a similar cytolysis rate compared to NPC-43 and NPC-53 (Fig. 3b). The induction of MHC-I potentially suggests improved tumour cell recognition that leads to antigen-specific cytolysis. Furthermore, CD8+ T cells derived from multiple donors consistently demonstrated antigen-specific killing, confirming the reproducibility of this response across individuals (sFig. 3). Overall, these results indicate that peptide-trained T cells promote NLRC5-dependent MHC-I regulation in NPC cells, leading to improved antigen presentation and cytotoxic T cell recognition while avoiding excessive PD-L1 upregulation. NLRC5 knockdown reduces MHC-I expression in NPC cells without impairing T cell-mediated cytotoxicity Given that NLRC5 is an important regulator of MHC-I expression, we next investigated the effect of NLRC5 on MHC-I expression and T cell-mediated cytotoxic efficiency. We used siRNA to knockdown NLRC5 in NPC-43. NLRC5 knockdown in NPC-43 cells was validated at the gene level, showing a 76% reduction in NLRC5 mRNA expression, which remained low for up to 72 hours (Fig. 5a). This knockdown was further confirmed at the protein level, showing a corresponding decrease in both NLRC5 and MHC-I expression (Fig. 5b), supporting the essential role of NLRC5 in maintaining MHC-I expression. Interestingly, while NLRC5 knockdown markedly reduced MHC-I expression, it did not affect the expression of other key MHC-I associated genes such as B2M and CD274, but led to an increase in TAP1 expression (Fig. 5c).Fig. 5Impact of NLRC5 knockdown on MHC-I expression and antigen-specific killing of NPC-43. A NLRC5 gene expression after siRNA treatment. B Western blot showing reduced MHC-I protein levels after NLRC5 knockdown. C Relative expression of MHC-I-associated genes in NLRC5 knockdown cells compared to control cells. Dotted lines indicate baseline levels in control cells. D No reduction in cytolysis after NLRC5 knockdown compared to control cells, which endogenously express NLRC5. E Peptide-trained T cells restore both NLRC5 and MHC-I expressions in NLRC5 knockdown cells Impact of NLRC5 knockdown on MHC-I expression and antigen-specific killing of NPC-43. A NLRC5 gene expression after siRNA treatment. B Western blot showing reduced MHC-I protein levels after NLRC5 knockdown. C Relative expression of MHC-I-associated genes in NLRC5 knockdown cells compared to control cells. Dotted lines indicate baseline levels in control cells. D No reduction in cytolysis after NLRC5 knockdown compared to control cells, which endogenously express NLRC5. E Peptide-trained T cells restore both NLRC5 and MHC-I expressions in NLRC5 knockdown cells Despite reduced MHC-I expression following NLRC5 knockdown, peptide-trained T cells retained their cytotoxic activity against siNLRC5-treated NPC-43 cells, as no significant difference in tumour cell killing was observed between control and siNLRC5-treated cells (Fig. 5d). This is likely due to the restoration of NLRC5 and MHC-I expression upon co-culture with peptide-trained T cells (Fig. 5e). Collectively, these results demonstrate that NLRC5 is essential for sustaining MHC-I expression in NPC cells. However, peptide-trained T cells can reinduce NLRC5 and restore MHC-I levels, thereby maintaining efficient T cell-mediated cytotoxicity. This highlights the robustness of vaccine-induced immune activation in enhancing tumour immunogenicity through regulation of NLRC5 and MHC-I APM axis. Vaccine-induced upregulation of NLRC5 enhances antigen presentation and suppresses tumour growth in vivo We next sought to determine whether this vaccine-induced regulation of the APM also occurs in vivo and contributes to tumour control. Due to the lack of syngeneic NPC mouse model, we employed the B16F10–A2 mouse melanoma, which is characterised by poor immunogenicity and low MHC-I expression, to evaluate the efficacy of the DNA vaccine in an antigen presentation-deficient setting [35]. Although limited to a single model, our in vivo efficacy study demonstrated that a prime-boost regime of the DNA vaccine significantly suppressed tumour growth compared to the pLDOM control animals (Fig. 6a). Vaccine-treated tumours also demonstrated a significant increase in CD8+ T cell infiltration as compared to control animals (Fig. 6b), suggesting enhanced antitumor immunity in poorly immunogenic B16F10 tumours.Fig. 6Inhibition of tumour growth by DNA vaccine targeting MAGED4B and FJX1 via upregulation of APM genes in a syngeneic mouse model. A The subcutaneous tumour volume was significantly reduced in mice treated with the DNA vaccine compared to the control group. B Immune cell infiltration into the tumour increased after DNA vaccination. Insets show 20x magnification of the selected regions. C KEGG pathway analysis revealed that APM pathway is upregulated in DNA vaccine-treated group. D Heatmap showing gene expression in the 17-gene APM signature. The two columns on the right represent the average control and DNA vaccine samples, respectively. Asterisks denote statistically significant differential expressions (*p < 0.05) Inhibition of tumour growth by DNA vaccine targeting MAGED4B and FJX1 via upregulation of APM genes in a syngeneic mouse model. A The subcutaneous tumour volume was significantly reduced in mice treated with the DNA vaccine compared to the control group. B Immune cell infiltration into the tumour increased after DNA vaccination. Insets show 20x magnification of the selected regions. C KEGG pathway analysis revealed that APM pathway is upregulated in DNA vaccine-treated group. D Heatmap showing gene expression in the 17-gene APM signature. The two columns on the right represent the average control and DNA vaccine samples, respectively. Asterisks denote statistically significant differential expressions (*p < 0.05) To further investigate whether vaccine modulates antigen presentation, we performed RNA sequencing on the tumour samples. Our KEGG pathway enrichment analysis identified antigen processing and presentation as one of the most significantly upregulated pathways following DNA vaccination (Fig. 6c). This was supported by the overall upregulation of genes of the APM signature. In particular, Nlrc5, Tap1 and Rfx5 were the genes significantly increased in the DNA vaccine-treated group (Fig. 6d). The in vivo findings align with our in vitro data, where co-culture of NPC cells with peptide-trained T cells upregulated NLRC5, TAP1, and B2M, restored MHC-I expression, and enhanced antigen-specific cytotoxicity. Together, these results demonstrate that the vaccine can upregulate NLRC5 and promote antigen presentation, thereby strengthening anti-tumour immune responses. Discussion Cancers can evade immune surveillance by downregulating APM, which in turn impairs antigen presentation and reduces immunogenicity. In about 30% of NPC cases with APM downregulation, somatic alterations such as mutations and structural variants affecting MHC class I expression were found in NLRC5, HLA-A, and HLA-B, occurring in 2.9 to 8.5% of cases [10]. This observation suggests that only a small subset of APM-deficient tumours harbour hard alteration that lead to permanent loss of MHC-I expression and function. In contrast, the majority of cases likely involve soft alteration in which APM components are suppressed through potentially reversible mechanisms, such as transcriptional or epigenetic regulation. In our study, all three NPC cell lines tested demonstrated inducible upregulation of MHC-I expression upon IFNγ stimulation, peptide-trained T cell co-culture, or HDAC inhibitor treatment. This responsiveness suggests that these cell lines do not harbour disruptive mutations in MHC-I genes and are instead representative of soft alterations. While MHC-I sequencing was not undertaken in this study, future analyses of MHC-I and APM gene alterations will be key to identifying patients with reversible APM defects and refining treatment stratification. In this study, we identified that APM downregulation in NPC may arise from the transcriptional downregulation of NLRC5, a key transcriptional regulator of MHC-I expression [36]. Indeed, our transcriptomic analysis of NPC dataset revealed a strong positive correlation between NLRC5 and MHC-I expression in NPC. This relationship was further validated in vitro, where NLRC5 knockdown in the NPC-43 cell line led to reduced MHC-I expression. NLRC5 acts as a master transcriptional regulator of the MHC-I APM, directly activating genes such as B2M, TAP1, and PSMB9 which are essential for antigen processing and presentation to CD8+ T cells [37]. Consistent with this role, MHC-I downregulation in NPC patients exhibit reduced CD8+ T cell populations in both tumours and peripheral blood compared to healthy controls, indicating impaired CD8+ T cell-mediated antitumor immunity [11]. Furthermore, low NLRC5 expression is associated with poor patient survival, underscoring its clinical relevance [37, 38]. MAGED4B and FJX1 are shared TAAs and were found to be upregulated in 91% of NPC cases, consistent with the 95% reported in head and neck cancer [25]. These TAAs contribute to oncogenesis by promoting cancer cell proliferation and invasion [25, 39, 40]. Importantly, MAGED4B and FJX1 are reported to be immunogenic, and their high expression in NPC highlights their potential as viable targets for vaccine-based therapies. However, 60% of NPC cases overexpressing these TAAs exhibit downregulation of MHC-I. This paradox underscores a critical therapeutic opportunity: although TAAs are abundantly expressed, their recognition by cytotoxic T cells is compromised. Therefore, restoring MHC-I-mediated antigen presentation is essential to enhance immune recognition. Here, we demonstrate that a cancer vaccine approach can effectively restore antigen presentation and enhance immunogenicity in APM-downregulated NPC. Our in vitro experiments demonstrated that co-culturing peptide-trained T cells with NPC cell lines induced the expression of NLRC5 and MHC-I across all three NPC cell lines, along with key MHC-I-associated genes, TAP1 and B2M. While a potent induction of NLRC5 could be achieved through IFNγ stimulation, the clinical use of systemic IFNγ is limited due to its association with severe and potentially life-threatening toxicities [41]. Alternatively, the FDA-approved HDAC inhibitor SAHA has been shown to enhance MHC-I expression by reversing epigenetic silencing in various cancer models [42–44]. However, in our study, SAHA exhibited limited efficacy in the NPC cells. Only NPC-43 showed dose-dependent induction of MHC-I, while NPC-53 and NPC-268 displayed minimal response even at the highest concentration tested. Although HDAC status in NPC cell lines is not known, we postulated that HDAC-mediated epigenetic suppression may not be the underlying mechanism of MHC-I downregulation in these NPC cell lines tested. This upregulation of NLRC5 and MHC-I protein expression by peptide-trained T cells enhanced tumour immunogenicity and promoted antigen-specific cytotoxicity against NPC cell lines. To further investigate the role of NLRC5, we performed knockdown experiments in NPC cells. Although NLRC5 deficiency led to reduced MHC-I expression, it was reversed upon co-culture with peptide-trained T cells, resulting in T cell-mediated cytotoxicity. Consistently, our in vivo studies showed that antigen-specific DNA vaccination enhances antitumor immunity in a MHC-I downregulated antigen-expressing B16F10–A2 melanoma syngeneic mouse model. Notably, tumour suppression was accompanied by IFNγ secretion (unpublished data) and increased CD8+ T cell infiltration following vaccination. Consistent with our in vitro study, the APM pathway was among the most significantly upregulated pathways in DNA vaccine-treated tumours. Among the 17-gene APM signature, Nlrc5 was found to be significantly upregulated in mice responding to DNA vaccination. The induction and stable expression of Nlrc5 could be crucial for MHC-I upregulation, which further restores immunogenicity in B16F10 melanoma cells [33, 36, 45, 46]. Hence, we postulated that vaccination can induce NLRC5 expression, which in turn, binds to promoter regions to enhance the transcription of MHC-I, B2M, and TAP1, thereby promoting antigen presentation [47, 48]. This may occur through several mechanisms. First, vaccine components can activate pattern recognition receptors (PRRs; e.g., TLRs, RIG-I–like receptors, or the cGAS–STING pathway) in antigen-presenting cells, triggering type I interferon and inflammatory cytokine production, as well as activating NK and innate lymphoid cells that secrete IFN-γ [49, 50]; Second, vaccination elicits antigen-specific CD4+ and CD8+ T cells that release IFN-γ upon antigen encounter. Paracrine IFN-γ signalling to antigen-presenting and stromal cells further upregulates NLRC5 and amplifies MHC-I pathway gene expression [36]. Third, robust innate and adaptive immune activation can remodel chromatin at immune gene loci, enhancing accessibility of interferon response elements and sustaining NLRC5 and APM gene expression beyond the acute cytokine phase [51]. Future studies may consider to investigate the role of interferon signalling by blocking type I/II interferon pathways using neutralizing antibodies or JAK/STAT inhibitors during vaccination to determine whether NLRC5 induction is interferon-dependent. Strategies aimed at inducing IFNγ could potentially enhance APM expression. However, it is important to consider that while IFNγ is crucial for initiating immune responses, excessive IFNγ stimulation may induce CD274 in the absence of specific antigen recognition and lead to T cell dysfunction [52]. Therefore, approaches that rely on antigen-specific activation, such as cancer vaccines, may offer a more targeted and clinically feasible strategy for restoring antigen presentation in NPC. Moreover, it is also worth noting that MHC-I expression can be regulated through alternative pathways, such as NF-κB and cGAS-STING signalling [33], which were not examined in this study. Nevertheless, our findings highlight the potential of vaccination strategies to counteract transcriptional suppression of APM genes and overcome immune evasion mechanisms in NPC. NPC is characterised by abundant immune cell infiltration and high CD274, yet its response rate to ICIs remains limited compared to other “hot” tumours. One contributing factor is the downregulation of APM, including MHC-I [6, 53]. Restoring MHC-I expression has been shown to improve tumour sensitivity to ICI treatment [52, 54]. In our study, peptide-trained T cells robustly induced MHC-I-associated components in NPC cells, and only minimally upregulated CD274. This is notable because PD-L1 expression is often co-regulated with MHC-I in response to IFNγ. The modest induction of CD274 alongside strong MHC-I upregulation suggests that vaccination may selectively enhance tumour immunogenicity without triggering excessive immune checkpoint expression. Although our study focuses on the transcriptional regulation of MHC-I via NLRC5, it has several limitations that require further investigation. Our findings suggest that cancer vaccines may enhance immune recognition in tumours with transcriptionally suppressed MHC-I. However, MHC-I downregulation may also result from enhanced protein degradation and disruptions in endocytic trafficking, which are not explored in the current study and whose responsiveness to cancer vaccination remains unknown. Additionally, the relatively small NPC cohort and limited availability of clinical information constrain the clinical interpretation of our results. This reflects the inherent challenges in obtaining sufficient NPC tissue, as lesions are typically small, deeply located, and rarely resected surgically. Endoscopic biopsies yield limited material, which is primarily used for diagnostic purposes, leaving little for research. Moreover, the retrospective nature of the samples precluded systematic collection of clinical outcome data. These limitations highlight the need for a systematic effort to establish an NPC tissue bank that integrates tumour specimens with clinical metadata to support future research. Lastly, due to the lack of a syngeneic NPC mouse model, in vivo experiments were performed using the B16F10 melanoma model. Although this model may not fully recapitulate the NPC tumour microenvironment, it served as a useful platform to confirm APM modulation following vaccination. These limitations underscore the need for employing more representative NPC models and comprehensive analyses of MHC-I regulatory pathways to better understand immune evasion and optimize immunotherapeutic strategies in NPC. Conclusions Our study demonstrates that peptide-trained T cells can upregulate MHC-I expression in NPC cells via NLRC5 transcriptional regulation, enhancing immune recognition and cytotoxic killing of tumour cells both in vitro and in vivo. These findings highlight the potential of cancer vaccines in APM-downregulated tumours, presenting a potential strategy to enhance immunotherapy efficacy, particularly for patients with MHC-I downregulation and resistance to current treatments. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3 Supplementary Material 4 Supplementary Material 4 Supplementary Material 5 Supplementary Material 5 Supplementary Material 6 Supplementary Material 6 Supplementary Material 7 Supplementary Material 7