Hepatocellular carcinoma escapes immune surveillance through deceiving thymus into recalling peripheral activated CD8 T cells.

Background
Immune checkpoint inhibitors (ICIs) have revolutionized the treatment landscape for advanced hepatocellular carcinoma (HCC) [1,2]. However, the efficacy of ICIs is limited by the immunologically "cold" tumor microenvironment, characterized by a paucity of tumor-infiltrating lymphocytes (TILs), particularly cytotoxic CD8+ T cells [[3], [4], [5]]. Elucidating the mechanism underlying the formation of cold tumor and developing strategy to convert it into an immunologically "hot" tumor are critical for enhancing the effectiveness of ICIs in HCC.
The thymus, the primary lymphoid organ orchestrating T cell development and central tolerance, has recently emerged as a key player in shaping anti-tumor immunity [6]. During thymic selection, self-reactive thymocytes are eliminated, while those tolerant to self-antigens mature and egress to the periphery tissues [7]. This process is mediated by thymic epithelial cells (TECs), which present a diverse array of self-antigens [8]. However, the role of the thymus in sculpting the immune response against tumor-specific antigens remains largely unexplored.
Activation-induced cell death (AICD) is a key mechanism for maintaining peripheral immune tolerance by eliminating over-activated or autoreactive T cells [9,10]. This process is tightly regulated by the cellular energy sensor AMP-activated protein kinase (AMPK), which has been shown to promote the survival of activated T cells [[11], [12], [13]]. However, whether the AMPK pathway is involved in regulating AICD of tumor-specific T cells within the thymus remains to be elucidated.
In this study, we found that the apoptosis of periphery homing activated CD8+ T cells increase significantly in the thymuses of HCC-bearing mice, which is driven by AICD initiated by TECs that captured and presented HCC related antigens from circulation. Notably, inhibiting the thymic homing of CD8+ T cells by thymectomy or AMPK activator and CCR7 antagonism-dependent disruption of CCR7-CCL19/CCL21 chemokine axis considerably restrained HCC growth. These insights shed new light on the complex interplay between the thymus and tumor immunity and would pave the way for innovative therapeutic approaches in HCC.

Results

Thymic CD8+ T cell death increased in HCC mice
The thymus, comprising cortical and medullary regions, plays a crucial role in central tolerance through negative selection mediated by medullary thymic epithelial cells (mTECs) [8]. To investigate the impact of hepatocellular carcinoma (HCC) on thymic negative selection, we established an in-situ HCC model in C57BL/6N mice using diethylnitrosamine (DEN) and carbon tetrachloride (CCl4) induction (Fig. 1A). TUNEL staining revealed a significantly increased cell death within the thymus of HCC-bearing mice compared to control group (Fig. 1B, C).
Immunofluorescence analysis uncovered a marked elevation of dead CD8+ T cells in the thymus of HCC mice, particularly pronounced in the medullary region (Fig. 1D, E). These findings suggest a profound alteration in thymic cellular dynamics in the presence of HCC. Given the critical role of CD8+ T cells in tumor immunosurveillance [14,15], we wondered whether thymic CD8+ T cell death was simply due to the change of T cell development or if it also involved peripheral T cell homing.

Thymic homing CD8+ T cells express activation marker
In autoimmune disease, activated T cells from draining lymph nodes can redistribute to various sites, including the thymus. These recirculating CD8+ T cells are characterized by high expression of CD73 [16], while activated CD8+ T cells express elevated levels of granzyme B[17].To elucidate the origin of increased apoptotic CD8+ T cells in the thymus of HCC mice, we focused on the phenomenon of peripheral T cell recirculation [18].
Using an in-situ transplantation model of HCC (Fig. 2A), flow cytometry analysis revealed a significant increase in CD73+ recirculating CD8+ T cells within the thymus of HCC-bearing mice (Fig. 2B, C). Importantly, immunofluorescence staining specific to CD8, CD73, Granzyme B, and TUNEL demonstrated that a subset of these recirculating CD8+ T cells exhibited both activation and apoptosis-like cell death (Fig. 2D, E). This observation highlights a potential mechanism by which HCC may influence thymus function through modulating peripheral T cell dynamics.

HCC enhances AICD of homing CD8+ T cells via TECs
To investigate the impact of HCC on thymus function, particularly in relation to the fate of recirculating activated CD8+ T cells, we performed mRNA sequencing on thymic tissue from chemically induced HCC models (Fig. 3A). As anticipated, development of HCC profoundly altered the thymic gene expression profile (Fig. 3B), with a notable enrichment in genes associated with apoptotic and necroptotic processes (Fig. 3C-E).
Activation-induced cell death (AICD) emerged as a key mechanism for the elimination of recirculating activated T cells in the thymus [19]. Utilizing multi-color immunofluorescence experiments, we identified thymic epithelial cells (TECs) as the primary mediators of AICD in recirculating activated CD8+ T cells (Fig.3F). This finding underscores the complex interplay between HCC and thymic cellular mechanisms in shaping the immune landscape.

HCC promotes antigen processing and presentation in thymus
To explore the mechanisms underlying the increased AICD of recirculating CD8+ T cells, we investigated the antigen presentation capabilities of the thymus in HCC-bearing mice. Utilizing a multi-omics approach including mRNA-seq, LC-MS (DIA), and immunofluorescence (Fig. 4A), we uncovered a significant enhancement in antigen internalization, transport, processing, and presentation within the thymus of HCC mice (Fig. 4B-F).
Strikingly, the model thymus shared a higher number of peptides with HCC tissue compared to normal thymus and liver (Fig. 4G, H), indicating a more active antigen processing and presentation machinery. Further analysis revealed that secretory proteins, such as AFP, APOC2, and Tg, were prominently upregulated at the protein level in the thymus of HCC mice (Fig. 4J-M).
Immunofluorescence staining pinpointed Epcam positive medullary TECs as the primary cells responsible for internalizing blood-borne HCC antigens, particularly AFP (Fig. 4N). These findings suggest that HCC-specific antigen presentation in the thymus may potentially contribute to immune tolerance against HCC.

HCC cells exhibit a robust ability of protein/antigen secretion
Considering the potential role of HCC cells-derived antigens in inducing thymic tolerance, we employed mRNA sequencing to investigate the secretory capabilities of HCC tissue samples from above chemically induced HCC mice (Fig. 5A). Remarkably, we observed a significant enrichment of pathway associated with protein secretion among the upregulated genes in HCC group (Fig. 5B).
Pathway enrichment and GO analyses revealed a substantial enhancement in cellular processes related to protein synthesis, assembly, and transmembrane transport in the HCC tissue (Fig. 5C, D). Corroborating these findings, analysis of TCGA data demonstrated significant upregulation of key genes involved in protein secretion in HCC samples compared to normal liver tissue (Fig. 5E). These results highlight the heightened secretory function of HCC cells, which may contribute to the dissemination of tumor-associated antigens.

Secreted proteins preserve immunogenicity
Considering that the frequent genetic alterations in HCC cells and their potential impact on protein immunogenicity, we examined the MHC-binding capacity of upregulated proteins in HCC tissues (Fig. 6A). Intriguingly, not only did the proportion of upregulated genes encoding secretory proteins were increased (Fig. 6B), but their potential immunogenicity also remained largely intact.
Analysis using the HLA Ligand Atlas [20] revealed that the number of HLA-binding peptides derived from secreted proteins did not significantly decrease (Fig. 6C). Moreover, the proportion of secreted proteins capable of binding to HLA-I and HLA-II molecules remained stable (Fig. 6D-G). These findings suggest that despite alterations in protein expression patterns, HCC-associated secreted proteins retain their immunogenic potential, which should be important for thymic selection and immune responses against HCC.

Blocking thymic homing of peripheral CD8+ T cells restrain HCC
To test the hypothesis that thymic deletion of HCC-reactive CD8+ T cells promote immune tolerance, we established a Hepa1-6 orthotopic liver transplantation model in adult immune competent mice. Thymectomy was performed one-week post tumor transplantation and mice were sacrificed at fourth week to assess tumor growth and CD8+ T cell infiltration (Fig. 7A). Remarkably, thymectomy significantly inhibited tumor growth (Fig. 7B, C) and was associated with increased CD8+ T cell infiltration in the HCC microenvironment (Fig. 7D, E). In the long term, however, thymectomy abolished T cell development and sustained supply for immune homeostasis. Therefore, thymectomy only highlight the potential therapeutic value of modulating thymic function in HCC.
In view of the critical role of the thymus in maintaining self-tolerance and immune surveillance, we explored strategies to block the thymic homing of peripheral activated CD8+ T cells as a potential therapeutic approach. Previous studies have shown that AMPK activation can effectively inhibit the chemotactic movement of immunocytes [21,22], and the CCR7-CCL19/CCL21 axis plays a crucial role in the thymus homing [23]. Notably, in our study, targeted activation of AMPK downregulates both CCL19 and CCL21 expression in the thymus of HCC mice (Fig. 8 A-B), thereby inhibiting the thymus homing of peripheral activated CD8+ T cells (Fig.8 C-D). Additionally, CCR7 antagonist obviously impeded HCC progression (Fig.8E-H), correspondingly, the distribution of CD73 positive CD8+ T cells decreased in the thymus while increased in HCC tissue (Fig. 8I-K). This discovery opens new avenues for therapeutic interventions in HCC by modulating thymus function without resorting to complete thymectomy.
In conclusion, our study unveils a previously unrecognized mechanism by which HCC may evade immune surveillance through manipulating thymus function. The augmented apoptosis-like death of recirculating activated CD8+ T cells in the thymus of HCC mice, coupled with strengthened antigen presentation capability of TECs, suggests a complex interplay between HCC cells and thymus-mediated immunity. These findings not only shed light on the intricate mechanisms of HCC immune evasion but also provided potential therapeutic targets for HCC treatment.

Discussion
The landscape of HCC treatment has evolved significantly with the introduction of combination therapies involving anti-angiogenic agents and ICIs. While these approaches have shown promising improvements in survival outcomes, as demonstrated by Finn et al. in their landmark study [1], the persistently low objective response rates (ORR) remain a critical challenge, particularly in the context of "cold tumors" characterized by minimal immune cell infiltration [24]. Our study builds upon this foundation, unveiling a novel mechanism underlying this phenomenon and proposing an innovative strategy to transform these immunologically inert tumors into immunologically active "hot tumors".
Previous research has primarily focused on local immunosuppression within the tumor microenvironment (TME) as a key factor in HCC immune evasion. Nishida and Kudo [25] eloquently described the role of immunosuppressive cells, such as regulatory T cells and myeloid-derived suppressor cells, in creating an inhospitable environment for effector T cells. Our study complements these findings by revealing a systemic immune evasion mechanism orchestrated by HCC, which operates at the level of T cell development and selection in the thymus. This discovery represents a significant expansion of our understanding of tumor-induced immune evasion.
We demonstrate that circulating HCC-derived antigens are captured and presented by thymic epithelial cells (TECs), leading to the deletion of HCC-reactive CD8+ T cells through AICD. This process effectively eliminates potential effector cells before they can engage with the tumor, offering a compelling explanation for the limited efficacy of current immunotherapies in some patients. Our findings align with and extend the work of Ringelhan et al., who previously highlighted the importance of systemic immune regulation in HCC progression [26].
Remarkably, our research reveals that HCC cells possess an enhanced capacity to secrete antigens compared to normal liver tissue. This observation is consistent with the findings of Gao et al., who reported altered protein secretion profiles in HCC cells [27]. We further demonstrate that these antigens, whether circulating as free molecules or encapsulated within extracellular vesicles, retain their immunogenicity - a crucial prerequisite for inducing AICD. This finding suggests that immune evasion in HCC is not a consequence of reduced antigen immunogenicity, but rather stems from the inappropriate presentation of these antigens in the thymus, a site pivotal for T cell selection and tolerance induction.
Building on these mechanistic insights, we demonstrate that thymectomy or pharmacological inhibition of CD8+ T cell thymus homing via AMPK-regulated CCL19/CCL21 suppression and CCR7 blockade significantly inhibits HCC. These approaches effectively prevent the thymic deletion of HCC-reactive CD8+ T cells, thereby enhancing anti-tumor immunity. While previous studies, such as that by Zhang et al., have explored various strategies to enhance T cell responses in HCC [28], our approach offers a novel perspective by focusing on preserving the pool of tumor-reactive T cells.
Our study not only broadens the understanding of tumor immunobiology but also provides a novel conceptual framework for developing more effective cancer immunotherapies. The findings have profound implications for the optimization of adoptive cell therapies, such as CAR T cell therapy, suggesting that strategies to prevent thymius homing upon reinfusion may enhance their efficacy. This builds upon the work of Maude et al., who highlighted the importance of T cell persistence in CAR T cell therapy outcomes [29].
Moreover, our work offers new insights into the paradoxical poor response to immunotherapy in some patients with high tumor mutational burden, a phenomenon previously described by Rizvi et al.[30]. We propose that assessment of thymic function or circulating tumor antigens could serve as novel biomarkers for immunotherapy responsiveness, potentially complementing existing biomarkers such as PD-L1 expression and tumor mutational burden.
In conclusion, by integrating our understanding of thymic function into the broader context of cancer immunotherapy, we pave the way for more comprehensive and effective strategies to combat HCC and potentially a wide range of other malignancies. This paradigm shift in our approach to cancer immunology holds promise for improving outcomes in patients with challenging cancer types and may fundamentally alter our approach to cancer prevention and treatment.

Materials and methods

Mice
Male C57BL/6N mice were utilized in this study, and two strategies were employed to build hepatocellular carcinoma (HCC), as described. All protocols were approved by the Institutional Animal Care and Use Committee of Xiamen University.

Carcinogen-induced spontaneous HCC model
For experimental HCC studies, a widely adopted model is (diethylnitrosamine) DEN and (carbon tetrachloride) CCL4-induced HCC model [31]. Briefly, two-week-old mice were intraperitoneally injected with DEN (25 mg/kg body weight, Sigma-Aldrich). After 6 weeks, mice were further intraperitoneally injected with CCl4 (2 ml/kg body weight, 1:4 v/v in corn oil, Sigma-Aldrich) or corn oil twice a week for 12 weeks. Tissues were harvested 2 weeks after the final injection. The mice in this experiment were euthanized by cervical dislocation in accordance with the AVMA guidelines for animal euthanasia (2020 Edition).

Liver orthotopic syngeneic graft model
Small-volume suspensions containing Hepa1-6 cells were injected into the subcapsular region of the median lobe parenchyma [32]. Single-cell suspension was prepared at a concentration of 2.5 × 107 cells/ml in a serum-free dulbecco's modified eagle medium (DMEM, Gibco). The cell suspension was injected with 1ml insulin disposable plastic syringe with a 28.5-gauge needle while limiting the injected volume to 40 μl to avoid tumor cell leakage and local spread leading to‘seeding’ metastasis in the peritoneal cavity. A steady and slow injection should be performed to prevent cell leakage and to minimize liver damage. The liver surface at the site of injection should be compressed for 30 seconds with a cotton swab to prevent bleeding and leakage when removing the needle. Tissues were harvested 3-4 weeks after transplantation. The mice in this experiment were anesthetized with chloral hydrate to establish the animal model. Prior to tissue collection, euthanasia of the mice was performed by cervical dislocation following the AVMA guidelines for animal euthanasia (2020 Edition).

mRNA-seq
To eliminate contamination of blood cells, mouse was flushed by perfusion with cold PBS through the inferior vena cava in situ before collection of liver, thymus and spleen. Total RNA of these samples was extracted using TRIzol (Invitrogen) according to the manufacturer's instruction. Further mRNA sequencing was performed by Aksomics Company (Shanghai, China). Briefly, mRNA was purified, fragmented, and converted to cDNA, after which adapters/barcodes were ligated. Libraries that passed quality control through the Agilent Bioanalyzer 2100 system were sequenced by the Illumina NovaSeq 6000 platform. GO and pathway enrichment analysis was performed based on the differentially expressed mRNAs, while Gene Set Enrichment Analysis (GSEA) was performed based on whole expressed mRNAs.

Mass spectrometry
Blood cells were removed from all samples according to the above method. Samples were subjected to in-solution trypsin digestion and dried. Samples were analyzed by an EASY-nLC 1200 (Thermo Fisher Scientific) coupled to an Orbitrap Fusion Lumos (Thermo Fisher Scientific) equipped with an EASY-IC ion source. Peptides were dissolved in 10 μl 0.1% formic acid and were auto-sampled directly onto a homemade C18 column (35 cm × 75 μm i.d., 2.5 μm 100Å). Samples were then eluted for 120 minutes with linear gradients of 3–35% acetonitrile in 0.1% formic acid at a flow rate of 300 nl/min. The raw files were analyzed by Proteome Discoverer 2.2 software against uniprot database.

TUNEL staining
Apoptosis-like cell death was evaluated with TUNEL assay [33]. TUNEL staining on liver sections was performed utilizing the TUNEL Apoptosis Detection Kit (YEASEN, Shanghai, China) according to the manufacturer’s instructions.

Immunofluorescence staining
Blood cells were removed from all samples according to the above method. Samples were fixed in 4% PFA for 24 hours and embedded with paraffin. Sections (3 μm) were prepared from paraffin embedded tissues and stained with primary antibodies listed in Supplemental Table 1. Fluorescent-labeled secondary antibodies (Thermo Fisher Scientific) were used to visualize the primary antibodies. The images were obtained using an LSM 900 confocal microscope (Zeiss).

Multiplex immunofluorescence staining
Sequential multiplex immunofluorescence staining with more than 3 markers on formalin-fixed, paraffin-embedded tissue sections was performed as described [34]. Briefly, sections (3 μm) were successively stained with primary antibody and fluorescent-labeled secondary antibody (Alexa Fluor™ 647, Thermo Fisher Scientific), and the images were obtained using an Evident IXplore SpinSR microscope (Olympus). Then, antibody elution was performed with the 2-mercaptoethanol/SDS stripping buffer, followed by counterstaining with the other primary antibodies and secondary fluorescent-labeled antibodies and imaging by Evident IXplore SpinSR microscope, so the cycle.

Image preprocessing and alignment
Acquired images were processed and analyzed using ZEISS software ZEN 3.1 (blue edition), Olympus cellSens Dimension, Image J FIJI and the FIJI plugin CellProfiler. Images alignment was performed using Adobe Photoshop.

Thymectomy
The thymus was resected one week after orthotopic hepa1-6 implantation [35]. Briefly, adult (8 weeks) male C57BL/6N mice were orthotopically transplanted with hepa1-6, and one week later the mice were anesthetized and restrained on a surgical board. To expose the thymus, a 2 mm incision was made on the base of the manubrium. A toothed forceps was used to grasp and exteriorize the thymic lobes, which were then excised with ophthalmic scissors. Similarly, sham surgery was performed on another group of mice. After 3 weeks, tissues were harvested. Thymectomized mice were examined at the end of the experiment to confirm there were no thymic remnants.

Flow cytometry analysis
Single-cell suspension of thymus was washed in PBS containing 1% bovine serum albumin. The cells were stained with fluorochrome-conjugated antibodies as previously published methods [36,37]. Samples were acquired on a Fortessa X20 flow cytometer (BD), and data were analyzed using FlowJo software, version 10.8.1.

Quantitative real-time PCR (RT-qPCR)
RNA samples were quantified using a Nanodrop 1000 spectrophotometer (Thermo Scientific). For reverse transcription, 1 μg of RNA was processed using random primers and MultiScribe Reverse Transcriptase (Applied Biosystems). Gene expression analysis was performed via RT-qPCR. Briefly, 50 ng of complementary DNA was amplified using SYBR Green Master Mix (BioRad) and gene-specific primers on a CFX96 Real Time PCR Detection System. The thermal cycling conditions were as follows: 40 cycles of 95 °C for 30 s, 58 °C for 30 s, and 72 °C for 30 s. Expression levels were normalized to 18S RNA for all samples. Expression level of mRNA was measured by using the following specific primers:
Ccl19 forward GGGGTGCTAATGATGCGGAA, reverse CCTTAGTGTGGTGAACACAACA; Ccl21a forward GTGATGGAGGGGGTCAGGA, reverse GGGATGGGACAGCCTAAACT; Actb Forward GGCTGTATTCCCCTCCATCG, reverse CCAGTTGGTAACAATGCCATGT.

Statistical analysis
The results were presented as the mean ± SD. Statistical analyses were performed using GraphPad Prism and Microsoft Excel 2023. Student's t-test was used to determine significant differences in the data between two experimental groups, and one-way ANOVA was used to analyze multiple group comparisons. All statistical analyses were two-sided, and p ≤ 0.05 were considered significant.

Funding
We are very grateful to the institutions that have supported our research. This study was supported by grants from the 10.13039/501100012462Innovation of Science and Technology, Fujian province (2024Y9603, 2023Y9448, 2023Y9447, 2023Y9412, 2021Y9218, 2021Y9227 and 2021Y9232), the 10.13039/501100017686Fujian provincial health technology project (2022ZD01005 and 2022ZQNZD009, 2024GGA046), 10.13039/501100001809National Natural Science Foundation of China (32000550, 82472845), 10.13039/501100003392Natural Science Foundation of Fujian Province (2023J05236, 2023J06038, 2023J05239, 2023J011296 and 2023J011254), Talent research project funded by Fujian Cancer Hospital (2022YNY12, 2022YNY13, 2022YNG01 and 2023YN05), the Special Research Funds for Local Science and Technology Development Guided by Central Government (2023L3020), 10.13039/501100021171Guangdong Basic and Applied Basic Research Foundation (2021A1515110083), Fujian Province high-level talents and young talents training funding project (GBXX2024073, GBXX2024075), The Fujian “Young Eagle Program” Youth Top Talent Program (Xuefeng Wang) and the XMU-Fujian Cancer Hospital cooperation grant for the Research Center of Metabolism and Tumor.

Data and materials availability
All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data available from authors upon request. The accession number of RNA sequencing data is GSE275715. Proteomics data are deposited in iProX (URL: https://www.iprox.cn/page/PSV023.html;?url=1729434123527Hnjq; Password: 2LpB).

CRediT authorship contribution statement
Qiaoting Hu: Investigation, Writing – original draft. Xuefeng Wang: Funding acquisition, Investigation, Writing – review & editing. Yundan You: Investigation. Jun Liu: Investigation. Bin Lan: Investigation. Fangfang Chen: Investigation. Hong Wen: Investigation. Haili Cheng: Investigation. Weibin Zhuo: Investigation. Ting Xu: Investigation. Jingxian Zheng: Investigation. Yuchuan Jiang: Investigation. Xiaojie Wang: Investigation. Jing Lin: Investigation. Zengqing Guo: Investigation. Sha Huang: Investigation. Gang Chen: Methodology. Yu Chen: Funding acquisition. Jingfeng Liu: Funding acquisition.

Declaration of competing interest
The application of Cmp2105 (HY-133073) in reducing HCC-educated thymus dependent immune evasion described in this paper have been filed for a patent application (CN 202510074921.0). Authors W.X.F., Q.T.H., B.L., Y.D.Y., Y.C., and J.F.L are listed as inventors on the patent (CN 202510074921.0). All other authors declare no competing interests.