Introduction

Nasopharyngeal carcinoma (NPC), a prototypical neoplastic tumor, is characterized by the infiltration of various immune cells into and around the tumor, mainly lymphocytes (T and B cells), dendritic cells, mononuclear macrophages, granulocytes, and mast cells. T and B cells constitute approximately 90% of these infiltrating cells.1 Although radiotherapy is the primary treatment modality for NPC, its efficacy is limited in patients with advanced or metastatic cancers.2 Immunotherapy has emerged as a promising therapeutic approach for various cancers, including NPC, because of its potential to target unique tumor microenvironments. However, this approach has limitations owing to the complexity of the immunosuppressive microenvironment of NPC.3 Therefore, it is imperative to further investigate the mechanisms of immunosuppression in the tumor microenvironment of patients with NPC.

RNA modifications discovered in 1956 have been identified in over 150 species.4 Messenger RNA (mRNA) plays crucial roles in various post-transcriptional modifications impacting gene expression and tumor suppression processes, such as tumor immunosuppression,5 cell metabolism,6 cell proliferation,5 and invasion and metastasis.7 Major mRNA modifications include N6-methyladenosine (m6A), N1-methyladenosine, 5-methylcytosine, N7-methylguanosine (m7G), pseudouridine (Ψ), 2’-O-methylation, and N4-acetylcytidine (ac4C).8 Among these, mRNA methylation has been extensively studied in tumors. For instance, m6A modifications regulate the translation efficiency of dendritic cell histone proteases.9 In NPC, the m6A methyltransferase METTL3 induces epithelial-mesenchymal transition by modifying Snail mRNA, thereby promoting NPC progression.10 MRTTL1, an m7G methyltransferase, promotes angiogenesis by modifying the VEGFA mRNA.11 Despite being the only acetylation event in mRNA modification, ac4C is as abundant as m7G modification12; however, it has been relatively understudied in cancer.

N-acetyltransferase 10 (NAT10) is the only protein identified with both acetylase and RNA-binding structural domains, and has been identified as the sole enzyme responsible for ac4C modification.13 NAT10, a member of the GCN5 (general control non-repressible 5)-related N-acetyltransferase family, is an ATP-dependent RNA acetyltransferase.14 NAT10 acts together with THUMP structural domain 1 (THUMPD1) to ensure RNA stability. As a specific ac4C adapter, THUMPD1 assists NAT10 in catalyzing RNA acetylation to produce ac4C.15 NAT10 is highly expressed in various cancers, including gastric cancer16 and hepatocellular carcinoma.17 Recently, Liao et al showed that NAT10 facilitates ac4C modification of NOTCH3 mRNA, contributing to tumor invasion and metastasis.18 Furthermore, ac4C enhances hypoxic tolerance in gastric cancer cells by promoting excessive glycolysis.19 Yang et al demonstrated the association between NAT10 and poor prognosis in patients with head and neck squamous cell carcinoma (HNSCC).20 Based on these studies, we hypothesized that NAT10-mediated ac4C modifications may play a pivotal role in NPC progression.

Immunosuppression is a key mechanism of tumor immune evasion.21 Single-cell transcriptomics of nasopharyngeal tissues have shown significant infiltration of T cells into the NPC tumor microenvironment, where T-cell dysfunction predominates during NPC progression.1 22 23 Furthermore, a previous study demonstrated that the dual “epithelial-immune” NPC tumor cell subpopulation exhibits strong immunosuppressive and regulatory effects on T cells within the immune microenvironment. This subpopulation suppresses interferon (IFN)-γ secretion by CD8+ T cells, leading to depletion of tumor-infiltrating T cells.24 Zhong et al showed that PD-1+CXCR5–CD4+ T cells promote the formation and maturation of tertiary lymphoid structures in the NPC tumor microenvironment by secreting chemokine CXC chemokine ligand (CXCL)13.25 However, the effects of ac4C modification on the immunosuppressive microenvironment of NPC remain uncertain; in particular, the specific immune cells and underlying molecular mechanisms have not been thoroughly investigated.

In this study, we aimed to investigate the role of NAT10-driven immune reprogramming in NPC and elucidate the mechanism by which the NAT10/dead box helicase 5 (DDX5)/high mobility group box 1 (online supplemental file 1HMGB1) axis inhibits T cells within the tumor microenvironment, ultimately contributing to NPC immunosuppression. In addition, we found that NAT10 knockdown enhanced anti-programmed cell death protein 1 (PD-1) therapeutic sensitivity as a combination therapy for NPC, highlighting the potential of targeting NAT10-mediated ac4C modifications as a promising immunotherapy strategy.

ResultsNAT10-mediated ac4C modification promotes NPC progression

To investigate the effects of ac4C modification on the NPC immune microenvironment, we first assessed the expression of NAT10 in 150 NPC tissue microarrays using immunohistochemistry (IHC). The staining intensity and the positive staining rate were both categorized into four grades (scored from 1 to 4 points), the product of which formed a total score. The final staining intensity was quantified on a scale of 1–8 for low expression and 9–16 for high expression using the X-tile software. Analysis of the correlation between IHC scores and clinical outcomes of NPC showed increased expression of NAT10 in patients with NPC who were diagnosed with clinical stages III and IV compared with that in those diagnosed with clinical stages I and II. These findings indicate a potential association between dysregulated NAT10 expression and terminal-stage NPC (figure 1A,B). Additionally, NAT10 upregulation showed a significant correlation with NPC recurrence and metastasis (figure 1C,D).

Figure 1

Upregulation of NAT10 is correlated with advanced stage, recurrence, metastasis, and poor clinical outcomes in NPC. (A) Representative NAT10 IHC staining of NPC tissue microarrays (n=150, scale bar, 200 µm). (B–D) NAT10 expression at different clinical stages (one-way analysis of variance), recurrence, and metastases (Student’s t-test). (E) NAT10 staining scores divided into low expression (scores of 1–8) or high expression (scores of 9–16). Kaplan-Meier curves representing overall survival. (F) ac4C expression assessed using dot blots in normal and NPC tissues. (G) Representative anti-ac4C dot blotting was conducted on total RNA, with methylene blue staining used as a loading control. (H–I) The volume and weight of xenograft tumors derived from NPC in various experimental groups were analyzed. (J) IHC was performed to assess the expression levels of NAT10 and Ki67. (K–L) Quantification of IHC staining for NAT10 and Ki67 expression was conducted using Student’s t-test. (M) Correlation between NAT10 and dead box helicase 5 levels in the tissues. (N–O) Lung metastasis was visualized using bioluminescence imaging, with colored scale bars used to represent the normalized photon flux. (P) Representative images of H&E staining were obtained for the visualization of cancer lesions in lung tissues from two groups of nude mice. The data are presented as the mean±SD. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ac4C, N4-acetylcytidine; IHC, immunohistochemistry; NAT10, N-acetyltransferase 10; NPC, nasopharyngeal carcinoma; WT, wild-type.

Survival rate analysis indicated that patients with high NAT10 expression had poorer clinical prognosis than those with low NAT10 expression (figure 1E). These findings were confirmed using a cohort from the Gene Expression Omnibus and The Cancer Genome Atlas (TGCA) databases (online supplemental figure S1A–C). Moreover, we examined NAT10 expression in different human cancers using data from the Human Protein Atlas database, which revealed high NAT10 expression in head and neck cancers and ranked it among the top five tumors (online supplemental figure S1D,E). We also confirmed the overexpression of NAT10 in NPC cells (online supplemental figure S1F–H).

Subsequent assessment of the level of ac4C modification in NPC and normal tissues revealed a dramatic increase in ac4C expression in NPC tissues compared with that in normal tissues (figure 1F). These findings indicate that NAT10-mediated ac4C modification is a valuable survival biomarker and is associated with NPC progression. To explore the oncogenic role of NAT10 and ac4C modifications in NPC, we transfected NAT10 lentivirus with a GFP tag or CRISPR-Cas9 in CNE2 cells to overexpress or knock out NAT10 and validate its efficiency (online supplemental figure S2A–C). The ac4C modification level was decreased in Cas9 cells, as indicated by dot blotting (figure 1G).

Next, we investigated the function of NAT10 in promoting growth and metastasis in vivo using mouse tumor xenograft and metastasis models implanted with NAT10 knockout, NAT10 overexpressing, and control NPC cells. Tumor xenogeneic models showed that NAT10 promoted tumor volume and mass growth (figure 1H,I), and IHC results showed a positive correlation between NAT10 and Ki67 (figure 1J–M). The tumor metastasis model showed that NAT10 promoted NPC lung metastasis via bioluminescence (figure 1N,O) and H&E staining of lung tissue showed a positive association between the tumor lesions and NAT10 expression (figure 1P). We also showed that NAT10 promoted NPC cell proliferation and metastasis in vitro (online supplemental figure S2D–N). Together, these findings suggest that NAT10-mediated ac4C modification promotes NPC proliferation and metastasis, and is associated with NPC prognosis.

NAT10 deficiency enhances CD4+ and CD8+ T cells activity of NPC

The degree of malignancy, metastasis, and tumor recurrence is often closely related to the immunosuppressive microenvironment, especially the insufficient function of T cells.26 27 The human nasopharynx exhibits diverse immune infiltrates, with previous single-cell sequencing studies indicating that normal nasopharyngeal tissue is primarily populated with T and B lymphocytes. In cases of inflammation, B cells become more abundant in nasopharyngeal tissue, whereas T cells are prevalent in the tumor microenvironment of NPC.28 As NPC progresses, there is notable functional suppression of T cells within the tumor microenvironment. To investigate the effect of ac4C modification on the immune microenvironment of NPC, we analyzed HNSCC in the immune database TIMER2.0 from the TCGA database. The analysis revealed that NAT10 expression was significantly negatively correlated with T cells, dendritic cell (DC), natural killer cell (NK cell), B cells, neutrophil, and macrophage (figure 2A). Furthermore, to confirm the immunosuppressive effect of NAT10 in NPC, we developed NAT10 knockout C57BL/6 mice using CRISPR/Cas9, named C57BL/6 NAT10em1Smoc. Peripheral blood flow cytometry panel analysis in untreated C57BL/6 and C57BL/6 NAT10em1Smoc mice revealed that knockout of NAT10 significantly increased the percentage of CD4+ T cells, CD8+ T cells and DCs (figure 2B–D). Further, t-Distributed Stochastic Neighbor Embedding (t-SNE) subsets of 30,000 cells in the peripheral blood of C57BL/6 and C57BL/6 NAT10em1Smoc mice were analyzed, and we found that CD4+T cells and CD8+T cells were significantly increased in the NAT10 knockout group, while the number of DCs was small and had no significant change (figure 2E,F). Moreover, specific T-cell analysis of peripheral blood of C57BL/6 and C57BL/6 NAT10em1Smoc showed the same results (online supplemental figure S2J–N). Subsequently, T cell-killing experiments showed that NAT10-Cas9 NPC cells were more likely to be killed by T cells (figure 2G,H). Furthermore, human T cells were injected into the tail vein of nude mice, followed by subcutaneous implantation of NAT10 knocked out or overexpressing or control NPC cells to study the interaction between human T cells and different types of NPC cells (figure 2I). The results showed that NPC cells were sensitive to T-cell killing and that knockout of NAT10 significantly promoted T-cell immunity in NPC compared with the control group (figure 2J,K). Moreover, using CNE2 and C666-1 cell lines, we showed that NAT10 deficiency enhanced the viability of T cells and promoted their recruitment of T cells to the tumor center (figure 2L,M).

Figure 2

NAT10 depletion increases TME CD4+ and CD8+ T cells. (A) Correlation of NAT10 expression with immune infiltration levels in TIMER2.0. (B) Mouse peripheral blood immune cells panel, including CD4+ T cells, CD8+ T cells, B cells, dendritic cell (DC), natural killer cell (NK cell), neutrophil (Neu), and macrophage (Mac), was detected in C57BL/6 NAT10em1Smoc and C57BL/6 wild-type (WT) mice. B cells accounted for lymphoid cell subsets, CD4+T cells, CD8+T cells and NK cells accounted for non-B-cell lymphoid subsets, and DC, Neu and Mac accounted for myeloid cell subsets. (C–D) Flow cytometry assay of CD4+ and CD8+ T-cell levels in the peripheral blood of C57BL/6 NAT10em1Smoc and C57BL/6 WT mice. (E–F) t-SNE subsets of 30,000 cells in peripheral blood of C57BL/6 and C57BL/6 NAT10em1Smoc mice. (G–H) Detection of apoptosis in NPC cells (CNE2 and C666-1) with NAT10 Cas9 or control in T-cell co-culture. (I) Macroscopic images of NPC xenograft tumors in different groups. (J–K) Statistical analysis of tumor volume and weight on day 21. (L–M) On the basis of in situ modeling of NPC with CNE2 and C666-1 cells, exogenously injected T cells visualized using bioluminescence imaging with DiD fluorescence. (N) Chemokine levels in conditioned medium (CM) from NPC cells (CNE2 and C666-1) or Cas9 cells were quantified using a protein array. (O) ELISA analysis of CXC chemokine ligand 16 levels in CM of indicated cells (n=6 per group). (P) Detection of apoptosis of T cells co-cultured with CNE2 or CNE2 NAT10 Cas9. Mean±SEM. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. NAT10, N-acetyltransferase 10; NPC, nasopharyngeal carcinoma; t-SNE, t-Distributed Stochastic Neighbor Embedding; TME, tumor microenvironment.

Tumor cells secrete cytokines required for T-cell infiltration, such as CXCL9, CXCL10, and CXCL16, which effectively expand CD8+ T cells.29 Chemokines are small molecular weight (8–14 kDa) multifunctional cytokines that play a crucial role in directing the migration of immune cells to specific tissues during inflammation.30 Subsequently, we investigated whether additional secreted proteins were regulated by NAT10, which affects the development of the tumor immunosuppressive microenvironment. A panel of chemokines was quantified in the conditioned medium obtained from CNE2 cells and C666-1 cells with or without NAT10 Cas9 expression using protein arrays. Combined with the intersection of CNE2 and C666-1, we found that CXCL16 was the most stable, significant and unique chemokine affected by NAT10 (figure 2N). Furthermore, ELISA revealed a significant increase in CXCL16 expression following NAT10 deficiency in the cell supernatant (figure 2O). CXCL16 plays a crucial role in promoting T-cell survival and coordinating CD8+ T-cell expansion.31 32 The involvement of CXC- CXCL9 and CXCL10 as ligands of CXC-chemokine receptor 3 is important for the recruitment of TH1, CD8+ T, and NK cells to the tumor microenvironment.33 Then, co-culture experiments with T cells and CNE2 cells revealed that NPC cells knocked out NAT10 reduced the apoptosis of surrounding T cells (figure 2P). Together, these findings indicate that ac4C modification not only inhibits T-cell recruitment and function but also affects the secretion of cytokines in the tumor microenvironment, which may be responsible for T-cell recruitment and be a manifestation of immunosuppression. In summary, NAT10 knockout promotes T cell and chemokine activity, thereby inhibiting tumor progression.

NAT10-mediated ac4C modification promotes mRNA stability and translation of CCAAT enhancer binding protein γ, dead box helicase 5 and helicase-like transcription factors

We performed a combined analysis of acetylated RNA immunoprecipitation-sequencing (acRIP-seq) and RNA-sequencing (RNA-seq) using control and NAT10 Cas9 CNE2 cells to investigate the potential ac4C-modified mRNA targets associated with T-cell recruitment in NPC (figure 3A). Correlation analysis of gene expression profiles revealed 181 differentially expressed genes with at least a twofold change. Among these, 61 genes were significantly downregulated (figure 3B). Subsequent assessment of ac4C modifications showed a significant decrease in the enrichment of ac4C motifs, particularly the “CXXCXXCXX” motif, in Cas9 cells (online supplemental figure S3A). Additionally, we observed clustering of ac4C sites near the translation start site, with most sites being located within the coding sequence (online supplemental figure S3B). The wild-type (WT) group exhibited 3,448 unique ac4C-modified peaks and 2,896 unique ac4C genes, whereas the Cas9 group exhibited 2,850 unique ac4C-modified peaks and 2,458 unique ac4C genes (online supplemental figure S3C,D). Given the role of NAT10 as an acetyltransferase, we speculated that the emerging genes in the WT group might encompass authentic targets of NAT10 (online supplemental figure S3C,D). We cross-linked acRIP-seq and RNA-seq to analyze 181 differentially expressed genes (fold change ≥2, FDR<0.05) and identified 87 downregulated and 94 upregulated genes. The analysis identified 28 genes associated with ac4C hypoacetylation of mRNA levels within the Cas9 group. Among these genes, 10 were identified as transcription factors or transcriptional cofactors, representing over one-third of the total genes analyzed (figure 3C).

Figure 3

NAT10 promotes ac4C acetylation of HLTF, DDX5, and CEBPG mRNAs. (A) Heatmap illustrating the differential expression of ac4C-related RNA between CNE2 and NAT10 Cas9 CNE2 cells by combining acetylated RNA immunoprecipitation-sequencing (acRIP-seq) and RNA-sequencing (RNA-seq). (B) The ac4C four-quadrant distribution chart focusing on the deacetylation of downregulated genes. (C) Bioinformatics analysis flow chart of downstream target of NAT10. (D) Evaluation of the regulatory role of target gene mRNA levels by real-time quantitative PCR. (E) Western blot analysis of HLTF, DDX5, CEBPG, and NAT10 levels (glyceraldehyde-3-phosphate dehydrogenase was used as control). (F) Integrative Genomics Viewer shows the distribution of ac4C peaks from acRIP. (G) The acRIP-quantitative PCR analysis confirmed that NAT10 knockout disrupted ac4C modification of HLTF, DDX5, and CEBPG mRNAs. (H) The RIP analysis between NAT10 and HLTF, DDX5, CEBPG mRNAs. Mean±SEM. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ac4C, N4-acetylcytidine; CEBPG, CCAAT enhancer binding protein γ; DDX5, dead box helicase 5; HLTF, helicase-like transcription factors; mRNA, messenger RNA; NAT10, N-acetyltransferase 10; WT, wild-type; FDR, False Discovery Rate.

Gene Ontology analysis revealed that genes exhibiting reduced ac4C peaks were associated with various biological processes, including mRNA transcription and cellular growth (online supplemental figure S3E). Combined analysis of the mRNA and protein expression levels identified three genes that were significantly regulated by ac4C, including transcription factor CCAAT enhancer binding protein γ (CEBPG), transcriptional cofactor dead box helicase 5 (DDX5), and helicase-like transcription factors (HLTF), all of which could significantly downregulate mRNA and protein levels when inhibiting NAT10 (figure 3D,E, online supplemental figure S3F). The acRIP-quantitative PCR (qPCR) assay demonstrated a decrease in the ac4C modification abundance of CEBPG, DDX5, and HLTF following NAT10 knockout, as confirmed by the Integrative Genomics Viewer schematic illustrating reduced ac4C modification peaks (figure 3F,G). Further, RIP experiments confirmed that knocking out NAT10 inhibited its interaction with CEBPG, DDX5, and HLTF Rnas (figure 3H). Thus, NAT10 promotes ac4C acetylation of CEBPG, DDX5, and HLTF.

Next, we explored the potential mechanisms underlying the ac4c-mediated regulation of CEBPG, DDX5, and HLTF. Our findings indicated no significant differences in the promoter activities of CEBPG, DDX5, and HLTF between the WT and NAT10 Cas9 CNE2 cells (online supplemental figure S4A). Subsequently, WT and NAT10 Cas9 cells were treated with actinomycin-D to inhibit transcription. The results showed that deletion of NAT10 significantly affected the mRNA stability of CEBPG, DDX5, and HLTF in CNE2 cells, indicating that NAT10 plays a crucial role in regulating the transcription of these genes (online supplemental figure S4B–D). Furthermore, we investigated whether ac4C modulates the expression of CEBPG, DDX5, and HLTF beyond mRNA stability. To this end, WT cells transfected with the vector or NAT10 overexpressing constructs were treated with L-leucinamide (MG132) to inhibit proteasome activity, or cycloheximide (CHX) to block protein translation. The data indicated that the presence of MG132, but not CHX, attenuated NAT10-induced expression of CEBPG, DDX5, and HLTF (online supplemental figure S4E–H). This suggests that NAT10 regulates protein translation rather than protein stability or post-translational modifications of CEBPG, DDX5, and HLTF. No significant differences were observed in the half-lives of CEBPG, DDX5, and HLTF between WT and NAT10 Cas9 CNE2 cells under blocking protein translation (online supplemental figure S4I), further supporting this finding. In conclusion, ac4C acetylation promotes the mRNA stability and translation of CEBPG, DDX5, and HLTF.

The positive feedback loop of HLTF-NAT10

Most transcription factors are essential for controlling cellular functions and initiating the transcription of specific genes. Therefore, we explored whether these three transcription factors regulate NAT10 expression. The results showed that NAT10 levels decreased with siHLTF treatment, whereas no significant differences were observed with shDDX5 and siCEBPG treatments (figure 4A). However, in vitro experiments confirmed that CEBPG, DDX5, and HLTF promoted the proliferation and migration of CNE2 cells, thereby affecting the malignant progression of NPC (online supplemental figure S5). Then, chromatin immunoprecipitation (CHIP) experiments revealed that HLTF was specifically localized to the promoter region of NAT10 and played a role in promoting NAT10 transcription (figure 4B). Furthermore, luciferase reporter further confirmed that when the NAT10 promoter region to which HLTF binds is mutated, the transcriptional activity of HLTF was significantly inhibited (figure 4C). Considering that transcription factors need to be in the nucleus to play transcriptional regulatory roles, we explored whether NAT10 could affect HLTF entry and exit from the nucleus. The result showed that overexpression of NAT10 facilitated the nuclear localization of HLTF, whereas NAT10 knockout had the opposite effect of promoting HLTF exit from the nucleus (figure 4D–F). In summary, HLTF is regulated by NAT10 as a downstream target gene and functions as an upstream transcription factor that regulates the expression of NAT10, creating a positive feedback loop (figure 4G).

Figure 4

HLTF transcriptionally regulates the expression of NAT10, and NAT10 facilitates its nuclear entry. (A) WB analysis of NAT10 expression after transfection with siHLTF, siCEBPG, or shDDX5. (B) Chromatin immunoprecipitation assays were conducted to investigate the interactions of HLTF, DDX5, and CEBPG with the NAT10 promoter region in NPC cells. (C) HLTF transcription factors were confirmed to bind to the NAT10 promoter in NPC cells using a dual-luciferase reporter assay. (D) Localization of HLTF in cellular immunofluorescence images captured using confocal fluorescence microscopy. (E–F) WB and real-time PCR were used to assess the mRNA expression levels of HLTF in the nucleus and cytoplasm of wild-type (WT), NAT10 overexpressed, and Cas9 cells. GAPDH were used as endogenous control for cytoplasmic RNA, while 18s rRNA was selected as endogenous control for nuclear RNA. β-actin were used as endogenous control for cytoplasmic protein, while histone-H3 was selected as endogenous control for nuclear protein. (G) Schematic diagram of HLTF forming a positive feedback loop with NAT10. Mean±SEM. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. CEBPG, CCAAT enhancer binding protein γ; DDX5, dead box helicase 5; HLTF, helicase-like transcription factors; mRNA, messenger RNA; NAT10, N-acetyltransferase 10; NPC, nasopharyngeal carcinoma; WB, western blot.

HMGB1 is the direct target of DDX5

To further explore how CEBPG, DDX5, and HLTF affect T cell-mediated immunosuppression, we used CHIPBase and hTF target databases to predict their downstream genes. This analysis revealed that CEBPG regulated 1,300 downstream genes, DDX5 regulated 7,031 downstream genes, and HLTF did not yield any predictions. NAT10 is transcriptionally regulated by HLTF; therefore, we compared the genes with consistent alterations in NAT10 acetylation and expression levels (fold change ≥2, FDR<0.05) from previous sequencing data with the anticipated downstream targets of CEBPG and DDX5 and identified HMGB1 as the sole target gene (figure 5A). CEBPG, DDX5, and HLTF were found to be positively correlated with HMGB1 in HNSCC based on TCGA database (online supplemental figure S6A–C). Transfection with DDX5-specific lentiviral short hairpin RNA (shRNA) to knock down DDX5 led to a subsequent decrease in HMGB1 expression (figure 5B). As expected, silencing of CEBPG and HLTF led to a decrease in HMGB1 expression (figure 5C, online supplemental figure S6D). Using a CHIP assay, we confirmed the presence of DDX5 and CEBPG in the promoter region of HMGB1 instead of HLTF (figure 5D, online supplemental figure S6E,F). Studies have extensively explored the relationship between CEBP family members and HMGB134; therefore, we focused on DDX5. Western blot analysis indicated higher levels of DDX5 expression in NPC cells than that in NP69 cells (online supplemental figure S6G). IHC analysis of DDX5 and HMGB1 tissue microarrays indicated that DDX5 and HMGB1 upregulation was significantly correlated with clinical stage, recurrence, metastasis, and outcome of NPC (figure 5E–M). Combined with immunohistochemical scores, we confirmed that DDX5 and HMGB1 were positively correlated (figure 5N).

Figure 5

Correlation between N-acetyltransferase 10, DDX5, and HMGB1 expression levels. (A) Venn diagram of gene prediction results of downstream transcriptional regulation of CEBPG, DDX5, HLTF by CHIPBase and hTFtarget databases and acetylated RNA immunoprecipitation-sequencing. (B) WB analysis of HMGB1 expression after transfection with the DDX5 short hairpin RNA. (C) WB analysis of HMGB1 expression after transfection with siCEBPG. (D) CHIP assays were conducted to investigate the interaction of DDX5 with the HMGB1 promoter region in NPC cells. (E) Representative immunohistochemistry staining of DDX5 and HMGB1 in the NPC tissue microarrays (n=150). (F–M) Statistical comparison of DDX5 and HMGB1 expression levels across metastasis, recurrence, cancer clinical stages, and survival. (N) Correlation between DDX5 and HMGB1 levels in NPC tissues. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. CEBPG, CCAAT enhancer binding protein γ; CHIP, chromatin immunoprecipitation; DDX5, dead box helicase 5; HLTF, helicase-like transcription factors; HMGB1, high mobility group box 1; NPC, nasopharyngeal carcinoma; WB, western blot.

To further support the assertion that the NAT10-DDX5-HMGB1 axis contributes to enhanced malignancy, CNE2 cells were subjected to NAT10 Cas9 treatment and subsequently transduced with overexpressed lentiviral NAT10 or DDX5 with or without HMGB1-shRNA. Inhibition of NAT10 resulted in decreased metastasis and growth of NPC cells, although these effects were attenuated by DDX5 overexpression. Similarly, the suppressive effect of DDX5 on NPC cell proliferation and metastasis was counteracted by HMGB1 (online supplemental figure S7A–E). IHC analysis revealed elevated expression of NAT10, DDX5, and HMGB1 in the NPC xenograft group, indicating an accelerated malignant phenotype (online supplemental figure S7F–I). Taken together, these results suggest that the presence of DDX5 and HMGB1 is essential for the effect of NAT10 on the malignant characteristics of NPC.

Inhibition of NAT10 enhances tumor response to anti-PD-1 therapy

Next, we examined whether NAT10 mediates NPC immunosuppression through the secretion of HMGB1 by establishing a nude mouse tumor xenograft model through intravenous injection of exogenous human T cells. These findings revealed that NAT10 knockout suppressed the proliferative potential of NPC cells. However, the proliferative capacity was restored on transfection with lentivirus HMGB1 (figure 6A–C). We hypothesized that NAT10 might evade the killing of immune cells through the secretion of HMGB1. NAT10 has been shown to influence T-cell quantity by modulating the secretion of proteins, including HMGB1, and a series of chemokines, which in turn affects immune cell infiltration within the tumor microenvironment, leading to the establishment of an immunosuppressive microenvironment that enables evasion of immune cell attacks. Although immune checkpoint blockade therapy has demonstrated significant benefits for numerous patients with cancer, additional questions remain to be addressed, including patient selection criteria and the development of combination therapies to enhance resistance or sensitivity to monotherapy. Therefore, we determined whether NAT10 influences the response of cancer cells to PD-1 immune checkpoint blockade therapy. Additionally, the proliferative capacity of the Lewis lung carcinoma cell line (LLC) injected with luciferase lentivirus was significantly reduced in the subcutaneous xenograft models of C57BL/6 NAT10em1Smoc mice compared with that in control C57BL/6 mice (figure 6D–G). Subsequent evaluation of the effectiveness of PD-1 blockade therapy revealed that tumors in NAT10-deficient mice exhibited increased sensitivity to treatment (figure 6D–G). However, HMGB1 overexpression allowed tumors to be substantially desensitized to anti-PD-1 therapy, as evidenced by the tumor volume (figure 6D–G). In addition, we analyzed the proportion of CD4+ and CD8+ T cells in the peripheral blood of mice, demonstrating that NAT10 deficiency enhanced T cells (figure 6H,I). These results were further confirmed in the subcutaneous xenograft models of C57BL/6 NAT10em1Smoc and C57BL/6 mice with CNE2 cells (figure 6J). Further, CNE2 oeHMGB1 or CNE2 NC co-culture with T cells showed that tumor-derived HMGB1 promoted T-cell apoptosis (figure 6K). Collectively, these results suggest that NAT10 plays a role in regulating tumor response to anti-PD-1 therapy in cancer cells.

Figure 6

NAT10 inhibition renders NPC cells resistant to anti-PD-1 immunotherapy. (A) CNE2 cells with NAT10 overexpressed or silencing or controls were subcutaneously transplanted into nude mice via tail vein injection of exogenous T cells. (B–C) Subcutaneous tumor volumes and weights were measured on day 21 (Student’s t-test). (D) Macroscopic images of NPC xenograft tumors in C57BL/6 NAT10em1Smoc and C57BL/6 wild-type (WT) mice with or without anti-PD-1 therapy. (E) The tumor volume was visualized using bioluminescence imaging, with colored scale bars used to represent normalized photon flux. (F) Macroscopic images of NPC xenograft tumors. (G) Volume analysis of xenograft tumors derived from NPC in various experimental groups. (H–I) Flow assay of CD4+and CD8+T cell levels in the peripheral blood of different groups. (J) Flow assay of CD4+T cell levels in the peripheral blood of CNE2 cell xenograft tumors. (K) Flow analysis of T-cell apoptosis in co-culture of CNE2 cells and T cells. Mean±SEM. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. HMGB1, high mobility group box 1; LLC, Lewis lung carcinoma; NAT10, N-acetyltransferase 10; NPC, nasopharyngeal carcinoma; PD-1, programmed cell death protein 1.

Discussion

RNA modification is a potential new strategy to counteract tumor immunosuppression. Recently, clinical studies have explored small-molecule inhibitors targeting FTO and ALKBH5.35–37 These inhibitors can suppress m6A modifications, thereby reducing the expression of immune checkpoint genes and aiding in overcoming immune evasion within the tumor microenvironment. However, small-molecule inhibitors of other modified targets are currently lacking.

The ac4C acetylation modification holds significant research potential, with NAT10 playing a crucial role as the confirmed acetyltransferase responsible for this modification. In this study, NAT10 was highly expressed in NPC and is associated with a poor prognosis. RNA transcriptome sequencing and acRIP-seq identified CEBPG, DDX5, and HLTF as downstream genes targeted by NAT10. HLTF also acts as an upstream transcription factor of NAT10 that positively regulates NAT10, thereby forming a positive feedback loop. The stability of mRNA is primarily influenced by its nucleotide sequence, which affects its secondary and tertiary structure, as well as its interaction with various RNA-binding proteins. Recent advancements in high-throughput RNA-seq have highlighted the critical role of mRNA modifications in regulating mRNA stability.38 The balance between mRNA stability and translation efficiency is reciprocal; decreased mRNA stability results in reduced translation, which further impacts mRNA stability.39 Previous studies have showed enrichment of ac4C is enriched at the 5’ end of the coding sequence region, and is related to the stability of the substrate mRNA, whereas m6A is biased at the 3’ end, and is related to mRNA instability. The ac4C modification of mRNA thus enhances stability and improves translation efficiency.40

HMGB1 was identified as a common downstream gene of CEBPG, DDX5, and HLTF, and DDX5 transcriptionally regulated HMGB1, forming the NAT10/DDX5/HMGB1 axis. HMGB1, a non-histone protein, primarily localized in the cell nucleus also be passively released into the extracellular matrix as a “danger signal” in response to cellular stress, injury, or death, initiating multiple immune responses. However, in the tumor microenvironment, HMGB1 released by tumor cells often acts as a “tumor pro