Article Text
PDF
Abstract
Background Tumour-infiltrating T cells can mediate both antitumour immunity and promote tumour progression by creating an immunosuppressive environment. This dual role is especially relevant in hepatocellular carcinoma (HCC), characterised by a unique microenvironment and limited success with current immunotherapy.
Objective We evaluated T cell responses in patients with advanced HCC by analysing tumours, liver flushes and liver-draining lymph nodes, to understand whether reactive T cell populations could be identified despite the immunosuppressive environment.
Design T cells isolated from clinical samples were tested for reactivity against predicted neoantigens. Single-cell RNA sequencing was employed to evaluate the transcriptomic and proteomic profiles of antigen-experienced T cells. Neoantigen-reactive T cells expressing 4-1BB were isolated and characterised through T-cell receptor (TCR)-sequencing.
Results Bioinformatic analysis identified 542 candidate neoantigens from seven patients. Of these, 78 neoantigens, along with 11 hotspot targets from HCC driver oncogenes, were selected for ex vivo T cell stimulation. Reactivity was confirmed in co-culture assays for 14 targets, with most reactive T cells derived from liver flushes and lymph nodes. Liver flush-derived T cells exhibited central memory and effector memory CD4+ with cytotoxic effector profiles. In contrast, tissue-resident memory CD4+ and CD8+ T cells with an exhausted profile were primarily identified in the draining lymph nodes.
Conclusion These findings offer valuable insights into the functional profiles of neoantigen-reactive T cells within and surrounding the HCC microenvironment. T cells isolated from liver flushes and tumour-draining lymph nodes may serve as a promising source of reactive T cells and TCRs for further use in immunotherapy for HCC.
- HEPATOCELLULAR CARCINOMA
- ANTIGENS
- T LYMPHOCYTES
- IMMUNOTHERAPY
- T-CELL RECEPTOR
Data availability statement
Data are available upon reasonable request. Data are available on reasonable request. All data relevant to the study are included in the article or uploaded as online supplemental material.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Hepatocellular carcinoma (HCC) is a leading cause of cancer mortality, with no cure for advanced stages.
Neoantigen T cell therapy is promising in other solid tumours, however, its role in HCC is unclear.
HCC studies lack single-cell T cell/T-cell receptor (TCR) analysis with functional screenings involving extra-tumoural T cell immunological sites.
WHAT THIS STUDY ADDS
Single-cell RNA-seq was used to analyse antigen-experienced CD4+ and CD8+ T cells in HCC tumours, liver flushes and lymph nodes.
Functional screenings isolated neoantigen-reactive T cells and their TCRs.
Transcriptomic data reveal unique phenotypic and functional profiles of frequent neoantigen-reactive T cell clones.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
This study maps the single-cell functional landscape of T cells in HCC, covering both intra- and extra-tumoural origins.
The approach may support new immunotherapy options for HCC by targeting neoantigen-reactive T cells.
Introduction
Hepatocellular carcinoma (HCC), the most common primary liver cancer, is the third leading cause of cancer-related deaths worldwide.1 Early-stage HCC can be treated effectively with surgery or ablation, but advanced stages require palliative systemic therapy.2 3 Immune checkpoint blockade targeting PD-1, PD L-1 and CTLA-4 shows promise in HCC but benefits only a minority of patients.4–8 Tumour-infiltrating T cells, another potential therapy, particularly CD8+ cytotoxic T cells, play a dual role in tumour progression by promoting antitumour immunity through direct killing of tumour cells, but can become dysfunctional due to suppressive signals from regulatory T cells (Tregs) and myeloid-derived suppressor cells.9 10 This dysfunction may also lead T cells to support tumour growth by creating an immunosuppressive microenvironment, highlighting the need for therapeutic strategies that both enhance T cell function and overcome immunosuppressive barriers.
Adoptive cell therapy (ACT), where T cells are expanded and/or sensitised ex vivo before re-infusion,11 leverages T cells’ specificity and avidity for tumour-specific antigens (TSAs) or tumour-associated antigens (TAAs). Neoantigens—TSAs arising from mutations—are particularly attractive due to their high specificity and minimal toxicity.12 13 Genome-wide sequencing reveals multiple immunogenic neoepitopes in tumours, suggesting potential targets for ACT using tumour-specific TCR. Mutation-specific immune repertoires, such as those involving KRAS and TP53, show promise in solid cancers.14–18
In HCC, mutations in CTNNB1, TP53, AFP and AXIN1 are well-characterised.19–21 But investigation of T cell responses to these antigens are limited by tumour-infiltrating T cells (TILs) dysfunction, inadequate immune responses to existing neoantigens, and the poor expansion and persistence of transferred T cells.12 22 Key sources for identifying functional or antigen-reactive T cells include peripheral blood mononuclear cells, TILs and lymph node-derived T cells.23–25 The varied composition and functionality of T cell receptor (TCR) clones across these sources suggest that a multi-site approach may enhance immune response harnessing.23
In this study, we collected T cells from tumour-infiltrating regions, tumour-draining lymph nodes and peripheral blood T cells recirculating in the liver obtained from liver flushes. We aimed to identify tumour-reactive TCRs by screening several neoantigens through co-cultures with peptide-loaded antigen-presenting cells (APCs) and isolating 4-1BB+ cells for TCR sequencing. We additionally evaluated antigen-experienced T cells using single-cell transcriptomics and proteomics due to their high relevance for personalised treatments.26 This combined approach could accelerate the identification of antigen-enriched T cells and/or their TCRs, facilitating T cell-based ACT in HCC.
Materials and methods
Additional detailed materials and methods are included in online supplemental materials.
Supplemental material
Results
Phenotypic and functional spectrum of antigen-experienced T cells
We focus on memory-enriched CD4+ and CD8+ T cells, which can efficiently target neoantigens in cancer, offering potential for personalised T cell therapy.26 We explored antigen-experienced T cells in four HCC patients (HCC01, HCC05, HCC14 and HCC16) using single-cell RNA sequencing, focusing on T cells from tumours, draining lymph nodes and liver flushed, selected based on the limited availability of T cells and the sufficient quality of the tumour samples (figure 1A, table 1 and online supplemental figure 1A).
Phenotypic and functional spectrum of antigen-experienced T cells revealed by single-cell RNA-seq. (A) Schematic overview of the steps involved in the screening and isolation of neoantigen-reactive TCRs. Seven patients' tumours were analysed using whole-exome sequencing (WES) and RNA sequencing (RNA-seq) for neoantigen screening. (B) UMAP and clustering of integrated T cells from all patients and tissues. (C) The distribution of cells in the 2-dimensional UMAP space is shown, with cells colored according to tissue distribution. (D) UMAP and clustering of CD4+ T cells from figure 1B, Cl0, Cl4 and Cl5. (E) Stacked bar plots representing the proportions of each CD4+ cluster 0-5 for each patient and per tissue sample, UMAP visualisation of tissue-specific distribution of CD4+ T cells is presented in online supplemental figure 3A. (F) UMAP and clustering of CD8+ T cells from figure 1B, Cl1, Cl2 and Cl3. (G) Stacked bar plots representing the proportions of each CD8+ cluster 0-6 for each patient and per tissue sample, UMAP visualisation of tissue-specific distribution of CD8+ T cells is presented in online supplemental figure 4A. Single cell sequence data analysis of B–G are from 4 patients (HCC01, HCC05, HCC14, HCC16). Flush, blood flushed from explanted liver; HSP, heat shock proteins; LN, adjacent draining lymph node; TCM, T central memory; TD, terminally differentiated; TE, T effector; Treg, T regulatory; TRM, T resident memory.
Patient cohort, including clinical information, the total number of patient-specific candidate neoantigens (validated in silico), the total number of in vitro tested candidate neoantigens (patient-specific+shared+viral antigens) and detected reactivity
We first visualised the cellular distribution across sample types based on signature markers, enabling the identification of distinct T cell subsets: memory T cells, characterised by co-expression of CD45RO, CD62L, CXCR6 and CD127 (IL-7R), indicating survival and recirculation potential; exhausted T cells, defined by the co-expression of PD-1, CD39, and reduced CD127 expression, reflective of chronic activation and functional decline; exhausted memory T cells, a transitional subset expressing both memory markers (CD45RO, CXCR6) and exhaustion markers (PD-1, CD39), suggesting functional overlap between memory and exhaustion states; central memory T cells (TCM), identified by CD45RO+, CD62L+, CD127+, and the absence of exhaustion markers (PD-1, CD39), consistent with a circulating memory phenotype; effector T cells (TE), characterised by the lack of memory-associated markers (CD127, CD69), CD45RO+ and CD62L+ status, indicative of a less differentiated, short-lived effector phenotype; tissue-resident memory T cells (TRM), defined by CD103+, CD69+, CD127+ and CXCR6+, indicative of tissue retention and survival, with PD-1 expression suggesting basal activation; Tregs, identified by the co-expression of CD39, CD103 and PD-1, hallmark features of suppressive immune function; terminally differentiated T cells (TD), characterised by CD45RO+ (memory), CD62L+ (recirculation potential), CD69+ (activation), CXCR6+ and CD127+, indicative of survival and tissue-homing. The absence of PD-1 and CD39 distinguishes this subset from exhaustion states. Distinct clonal patterns were observed among these subsets, as illustrated in figure 1B,C, and online supplemental figure 1B,C. Below is the detailed phenotype of each cluster: Cl0/CD4+ TCM: CD4+ CD8– CD103– CD39– PD-1– CD672L+ CD45RO+ CD69+ CXCR6+ CD127+; Cl1/CD8+ TE: CD4– CD8+ CD103– CD39– PD-1– CD62L+, CD45RO+ CD69– CXCR6– CD127–; Cl2/CD8+ TRM: CD4– CD8+ CD103+ CD39+ PD-1+ CD62 L– CD45RO+ CD69+ CXCR6+ CD127+; Cl3/CD8+ exhausted: CD4– CD8+ CD103+ CD39+ PD-1+ CD62L+ CD45RO+ CD69+ CXCR6+ CD127–; Cl4/CD4+ Treg: CD4+ CD8– CD103+ CD39+ PD-1+ CD62L+ CD45RO+ CD69+ CXCR6+ CD127+; Cl5/CD4+ TD: CD4+ CD8– CD103+ CD39– PD-1– CD62L+ CD45RO+ CD69+ CXCR6+ CD127+; Cl6/MAIT: CD4+ CD8+ CD103+ CD39+ PD-1+ CD62 L– CD45RO+ CD69+ CXCR6+ CD127+; Cl7/γδ T: CD4– CD8+ CD103+ CD39– PD-1+ CD62 L– CD45RO– CD69+ CXCR6+ CD127–; Cl8/HSP: CD4+ CD8+ CD103+ CD39+ PD-1+ CD62L+CD45RO+ CD69+ CXCR6+ CD127+.
Among these, Cl6 T cells express high levels of MAIT-specific TCR genes, TRAV1-2 and TRBV6-4, along with a distinct activated phenotype characterised by the expression of S100A4 and S100A6 (stress response genes), CEBPD (an inflammatory transcription factor) and RORA (a regulator of cellular metabolism and tissue residency). This suggests that Cl6 T cells are likely responding to environmental stressors such as infection, inflammation or the tumour microenvironment. Cl7 T cells are characterised by high expression of γδ T cell-related TCR, TRDV1, a defining chain of the Vδ1 subset, as well as IKZF2 (Helios), a functional marker of unconventional T cells. This cluster also expresses high levels of cytotoxic and effector molecules, including NKG7, CST7, CTSW, CCL5, FCGR3A (CD16) and PLAAT4, as well as NK-like receptors such as KLRC2 (NKG2C), KLRC3 (NKG2E) and KLRD1 (CD94). These features suggest that Cl7 T cells represent unconventional, tissue-resident, cytotoxic γδ T cells, likely involved in stress responses, immune surveillance and tissue homeostasis. Cl8 T cells exhibit high expression of heat shock protein (HSP) genes, including HSPA1A, HSPA1B, HSPA6 and HSP90AA1, as well as stress response-related genes such as BAG3, UBB, ZFAND2A and CACYBP. This indicates that Cl8 T cells may either be experiencing stress or are specialised for responding to stressed environments, possibly recognising stress-induced molecules rather than antigens through classical TCR pathways. Additionally, the presence of double-positive CD4+ CD8+ cells, which are rare in peripheral blood but observed in thymus-derived or specialised subsets, supports their unique and potentially stress-adapted function (figure 1B,C and online supplemental 2A,B).
After excluding these unconventional T cells and clusters with stressed cells and limited TCRs from Cl6, Cl7 and Cl8 (online supplemental figure 2A,B), we identified 6 CD4+ T cells clusters (figure 1D) with a unique transcriptomic and phenotypic profile (online supplemental figure 3A–C): Cl0/CD4+ TEM T cells with the phenotype CD103– CD39– PD-1– CD62L– CD45RO+ CD69+ CXCR6+ CD127+; Cl1/CD4+ TCM T cells with the phenotype CD103+ CD39+ PD-1– CD62L+ CD45RO+ CD69– CXCR6+ CD127+; Cl2/CD4+ TRM T cells with the phenotype CD103– CD39– PD-1– CD62L– CD45RO– CD69+ CXCR6– CD127–; Cl3/CD4+ (long-term survival) T cells with the phenotype CD103+ CD39+ PD-1+ CD62L+ CD45RO+ CD69+ CXCR6+ CD127+; Cl4/CD4+ TRM (Tregs) T cells with the phenotype CD103+ CD39+ PD-1+ CD62L+ CD45RO+ CD69+ CXCR6+ CD127–; Cl5/CD4+ TRM (activated) T cells with the phenotype CD103+ CD39+ PD-1+ CD62 L– CD45RO+ CD69+ CXCR6+ CD127–. All tissues showed clusters associated with effector and central memory (clusters 0 and 1), but flush uniquely contained CD4+ TRM like cells with exhaustion features (PD-1+, CD39+, CD127–) (cluster 5) (figure 1E and online supplemental figure 3A,B), yet maintained local immune surveillance capabilities as indicated by activation markers (CD69) and chemotactic factors (CCL4) (figure 1D and online supplemental figure 3A,B). Moreover, cluster 3, prevalent in the lymph nodes; expressed PASK, LDHB, LEF1, SESN3 and IFITM1 (online supplemental figure 3B) indicating resilience to oxidative stress. The CD39+ and PD-1+ expression (online supplemental figure 3C) suggests a regulatory role in managing local immune responses and preventing tissue damage, particularly in tumour microenvironments. Additionally, high expression of metabolic genes (EEF1A1, EEF1B2, NACA, FAU and RACK1) (online supplemental figure 3B) suggests these cells are metabolically active, supporting sustained response to chronic antigenic exposure. Furthermore, cluster 4, likely T regulator cells, was present in the tumour and lymph node but scarce in liver flushes (figure 1E and online supplemental figure 3A–C).
We identified seven clusters of CD8+ T cells with distinct markers (figure 1F and online supplemental figure 4A–C): Cl0/CD8+ TRM T cells with the phenotype CD103+ CD39– PD1+ CD62L+ CD45RO+ CD69+ CXCR6+ CD127+; Cl1/CD8+ TEM (terminally differentiated) T cells with the phenotype CD103– CD39– PD1– CD62L+ CD45RO+ CD69– CXCR6– CD127+; Cl2/CD8+ TEM (partially exhausted) T cells with the phenotype CD103– CD39– PD1+ CD62L+ CD45RO+ CD69+ CXCR6+ CD127+; Cl3/CD8+ TE (highly cytotoxic) T cells with the phenotype CD103– CD39– PD1– CD62L+ CD45RO+ CD69– CXCR6– CD127+; Cl4/CD8+ TRM (exhausted) T cells with the phenotype CD103+ CD39+ PD1+ CD62L+ CD45RO+ CD69+ CXCR6+ CD127+; Cl5/CD8+ TE (resting) T cells with the phenotype CD103– CD39– PD1– CD62L+ CD45RO+ CD69– CXCR6– CD127+; Cl6/CD8+ TRM (advanced exhausted) T cells with the phenotype CD103+ CD39+ PD1+ CD62L+ CD45RO+ CD69+ CXCR6+ CD127+. Cluster 3 was consistently prevalent in the flush samples (figure 1G and online supplemental figure 4A), expressing high levels of GZMB, GZMH, GZMA and PRF1, indicating strong cytotoxic effector function. The presence of GNLY and NKG7 supports this cytotoxic phenotype (online supplemental figure 4B). Their CD69– status (online supplemental figure 4C) suggests a non-resident state primed for strong cytotoxic responses on antigen encounter. Additionally, a high presence of cluster 4 (TRM exhausted) cells was observed in lymph nodes and the tumour (figure 1G). This cluster, characterised by CD103, CD39, PD-1 and CXCR6, consists of exhausted tissue-resident memory T cells with cytotoxic potential (GZMK), stress response elements (HMOX1, MIR155HG) and early differentiation markers (CD27, CD28). These cells likely reside in peripheral tissues, exhibiting reduced effector functions due to chronic antigen exposure in cancer or chronic infections.
Immune signatures of TCRs derived from clonal T cell populations
In cancer immunology, high TCR expansion often correlates with the recognition of TAAs or neoantigens, which may indicate an effective antitumour response.27 To investigate this further, we performed transcriptomic and proteomic analyses of clonal TCR populations in memory-enriched CD4+ and CD8+ T cells, hypothesising that clonal expansion might vary among subsets and tissues. No significant clonal expansion was observed in CD4+ T cell subsets (figure 2A and online supplemental figure 5A), despite the predominance of specific clusters, such as effector memory T cells in cluster 0 (Cl0, as shown in figure 1D). Notably, the most expanded clones, for example, clone 1 (less than 92 clones in each patient) to clone 5, accounted for less than 1% of the total population. These clones suggest significant proliferation of T cells expressing particular TCRs in response to neoantigens, potentially reflecting high immunogenicity (figure 2A and online supplemental figure 5A). In contrast, the widespread presence of certain TCR clones across multiple samples, individuals or clusters may indicate unrelated contexts. These clones likely represent basal T-cell expansion without active immunological relevance, or they may belong to ‘public TCRs’—TCRs shared across individuals due to recognition of conserved or ubiquitous antigens. The distribution of the top five clonotypes shows enrichment of TCRs primarily in tumours, followed by liver flushes, and fewer in lymph nodes (figure 2B and online supplemental figure 5A). This pattern suggests that tumours and liver flushes are enriched sources of neoantigen-experienced T cells. The data imply that tumour-specific T cells proliferate extensively in the tumour and liver environments, but are less abundant in lymph nodes, potentially due to immune suppression within the tumour microenvironment or a lack of priming within the lymphatic compartment.
Immune signatures of TCRs derived from clonal T cell populations. (A) Stacked bar plots showing the distribution of clones in CD4+ cluster (resolution 0.5) in number of cells. UMAP analysis and clustering of CD4+ T cells, originally presented in figure 1D, depict their tissue-specific distribution as shown in online supplemental figure 3A and online supplemental 5A. (B) UMAP plots showing the localisation of the top 5 CD4+ clones for each patient. The top clones are calculated from the total number of cells, meaning that the distribution across clusters will also reflect differences in distribution of clusters between samples. UMAP and clustering of CD4+ T cells are as shown in figure 1D, and distribution of clusters in tumours, flush and LN are as shown in online supplemental figure 3A. (C) Stacked bar plots showing the distribution of clones in CD8+ clusters (resolution 0.75) in number of cells. UMAP analysis and clustering of CD8+ T cells, originally presented in figure 1F, depict their tissue-specific distribution as shown in online supplemental figure 4A and online supplemental 5A. (D) UMAP plots showing the localisation of the top 5 CD8+ clones for each patient. The top clones are calculated from the total number of cells, meaning that the distribution across clusters will also reflect differences in distribution of clusters between samples. UMAP and clustering of CD8+ T cells are as shown in figure 1F, and distribution of clusters in tumours, flush and LN are as shown in online supplemental figure 4A. Data analysis of A–D are from 4 patients (HCC01, HCC05, HCC14, HCC16).
In contrast, CD8+ T cells demonstrated evidence of dominant clonotypic expansion in certain clusters, with the top five clonotypes increasing significantly in number: from 342 to 801 in HCC01, 132 to 767 in HCC05, 97 to 287 in HCC14 and 68 to 191 in HCC16. These clonotypes collectively accounted for more than 5%–10% of the total CD8+ T cell population (figure 2C). The most expanded clonotype was primarily localised to effector and cytotoxic T cells in clusters 2 and 3 (Cl2 and Cl3) for HCC01, tumour-resident memory T cells (TRM) in cluster 0 for HCC05, terminally differentiated T cells in cluster 1 for HCC14, and both terminally differentiated and resting effector cells in clusters 1 and 5 for HCC16 (figure 2C). This expansion across different clusters in multiple patients suggests that these T cells may recognise a shared antigen or group of antigens, likely tumour-associated or neoantigens. However, their distribution across diverse functional clusters (effector, TRM, terminally differentiated) underscores the heterogeneous and context-dependent behaviour of tumour-reactive T cells. Some clonotypes appear actively engaged in antitumour activity, while others show signs of exhaustion or differentiation into less active states, reflecting adaptation to the tumour microenvironment. Visualising the top five CD8+ clonotypes for each patient (figure 2D and online supplemental 5A) revealed pronounced segregation in certain cases, such as HCC05 (cluster 0) or HCC14 and HCC16 (cluster 1), but overall, a spread across several clusters. These clonotypes were predominantly enriched in liver flushes, followed by tumours, and were less frequent in lymph nodes (figure 2D and online supplemental figure 5A). This phenotype highlights the dynamic and heterogeneous nature of tumour-reactive CD8+ T cell responses. The enrichment of TCRs in liver flushes and tumours suggests active immune responses to tumour antigens in these compartments.
Although the number of patients evaluated in this study is limited, these findings highlight the significant diversity and heterogeneity of T cell populations in HCC. This emphasises the need for personalised approaches in immunotherapy. Furthermore, the TCR clonality observed primarily in liver flushes and tumours suggests the presence and activity of tumour-reactive T cells in these sites. This underscores the importance of selecting the appropriate T cell sources—such as TILs and those from liver flushes—for further evaluation of neoantigen-specific T cell responses and their potential for targeted therapeutic strategies.
Identification of T cell reactivity to selected neoantigens in HCC
To investigate the phenotypic and functional profiles of antigen-experienced T cells, we further explored the potential neoantigens and their associated reactive T cells. Through whole exome sequencing and RNA sequencing (RNA-seq) of tumour samples from seven HCC patients, we identified a total of 15 neoantigens in HCC01, 11 in HCC02, 17 in HCC04, 84 in HCC05, 194 in HCC06, 153 in HCC14 and 68 in HCC16 (table 1). To prioritise the most immunogenic neoantigens, we applied the PIOR ranking method, and the top candidates were selectively tested, as detailed in table 1. To assess T cell reactivity against these HCC-specific neoantigens, we isolated APCs from liver flushes and T cells from HCC tumours, draining lymph nodes, and liver flushes for subsequent co-culture screenings, as outlined in figure 3A. This approach allowed us to examine the interaction between neoantigens and reactive T cells, providing insights into potential targets for immunotherapy in HCC. We manually curated coding variants from missense mutations in sequenced tumours (online supplemental table 1). Most mutations derived from single-nucleotide variants (SNVs), revealing a low to intermediate mutational burden in HCC tumours (online supplemental figure 5B).28 Frequently identified genes among our HCC samples are shown in online supplemental figure 5C. We prioritised SNV-derived neoantigens for T cell immunogenicity assays considering: (1) the biological relevance (driver genes or commonly mutated genes),29 (2) whether they are shared mutations and (3) allelic fraction frequency as an indication of clonality. Out of the seven enrolled HCC patients, T cell reactivity was detected in four patients (figure 3B and table 1). T cell reactivity, indicated by the activation marker 4-1BB on live CD4+ or CD8+ T cells, was primarily directed at patient-specific mutations, with some reactivity to shared (hotspot) mutations in BCORL1, BRCA2, ADAT2 and PIK3CA (table 1 and online supplemental table 1). No reactivity was observed for viral epitopes included in screening peptide pools for patients with a medical history of infection (figure 3B and table 1). All peptides tested are reported in online supplemental table 2. Among evaluated T cell sources, only those from lymph nodes or the liver flushes were reactive (figure 3B and table 1). In addition, the main reactivities were observed in CD4+ T cells, likely due to the use of long peptides (25 m) less efficiently presented through MHC class I (figure 3B and table 1 and online supplemental figure 5D), although the possibility that only CD4+ T cells are tumour-reactive in these patients cannot be excluded.
Identification of (prospective) T cell reactivity to selected neoantigens in HCC. (A) Schematic overview for the isolation of neoantigen-reactive T cells and/or their TCRs. Following in vitro stimulation with neoantigen-encoding peptides and FACS sorting of 4-1BB+ T cells, single-cell RNA-seq was performed on 4-1BB enriched T cells following a second rapid expansion after FACS sorting of 4-1BB+ samples. APCs and T cells were obtained from samples of 7 patients. The neoantigen screening results and peptide lists are presented in table 1 and online supplemental tables 1 and 2. Note: only T cells with double validation of mutation-specific immunogenicity after expansion were selected for further TCR single-cell sequencing, as demonstrated in the experiment shown in panel C. (B) The representative graph illustrates the positive reactivity of T cells derived from draining lymph nodes, or liver flushes from HCC patients. Reactivity is based on the percentage of 4-1BB expression in both CD4+ and CD8+ T cells following stimulation with various neoantigens. T cell reactivity screen for other HCC patients is detailed in table 1. OKT3 served as the positive control, while the absence of peptide was used as the negative control. (C) The graph shows the double validation of mutation-specific immunogenicity, with the percentage of 4-1BB expression for samples expanded after FACS sorting to enrich for 4-1BB+ cells. (D) Reactivity based on 4-1BB expression after stimulation with different peptide concentrations for SBNO2 mutant antigen, SBNO2 wild-type peptide and unstimulated (no-peptide added) sample. (E) Plot shows the CDR3β counts in the SBNO2 non-reactive (NR) and reactive (R) populations. Three CDR3β clones named TRBV12-3, TRBV6-1 and TRBV12-3 that were solely expanded in the SBNO2-R sample are labelled on the graph. (F) Venn diagram shows the CDR3β clonotypes (n=8) that are shared between SBNO2 non-reactive (NR)/4-1BB negative and reactive (R)/4-1BB positive samples. (G) Table shows the 3 reconstructed TCR pairs for these 3 clones which were screened for immune reactivity. The data analysis presented in panels C–F was derived from expanded 4-1BB+ T cells, which were exclusively available from the HCC05 patient lymph node. CDR3, complementarity-determining region; Mut, mutant; SBNO2-R (reactive) SBNO2-NR (non-reactive); ;TCR, T cell receptor; WT, wild-type.
Integration of functional screenings and single-cell RNA-seq potentiates the isolation of previously unidentified neoantigen-reactive TCRs
Following neoantigen screening, we performed FACS sorting and expanded reactive samples to enrich for 4-1BB+ clones, enabling TCR RNA sequencing. Due to the limited availability of T cells from other patients after neoantigen screening and FACS sorting, we validated the mutation-specific immunogenicity of neoantigen-reactive T cells exclusively in HCC05 LN-T cells for SBNO2, FANCA, SNTG2, SLCO2B1, REEP6 and CTNNB1 (figure 3C) prior to TCR single-cell sequencing. Among these, only SBNO2-reactive, sorted and expanded CD4+ T cells demonstrated reactivity to mutation-specific peptides (figure 3C). This finding suggests that during in vitro rapid enrichment, stochastic expansion of non-tumour-reactive clones may outcompete neoantigen-specific clones, highlighting a limitation of this enrichment method in reliably isolating neoantigen-reactive T cell clones. For the SBNO2-specific sample, reactive T cells were confirmed to be tumour-specific, with wild-type reactivity comparable to the unstimulated control (figure 3D). A comparison of TCR clonality before (using bulk TCR sequencing data from CD Genomics; data not shown) and after 4-1BB enrichment revealed differences in expanded TCR clonotypes (figure 3E). This supports the observation that non-tumour-reactive clones can outgrow neoantigen-specific ones during in vitro expansion. Notably, three TCR clones were specifically expanded in the SBNO2-reactive sample, further confirming their antigen-specific reactivity (figure 3E–G).
Evaluation of specificity and functionality of reconstructed neoantigen-reactive TCRs
Reconstructed TCRs for SBNO2 from the 4-1BB+ population were genetically transferred to autologous T cells for testing against mutated and wild-type antigens (figure 4). Autologous T cells were electroporated with RNA encoding mTCR1, mTCR2 and mTCR3 pairs (figure 3G) and co-culturing with autologous B cells loaded with mutant and wild-type peptides to assessed 4-1BB expression. The mTCR1 and mTCR3 showed strong reactivity to their cognate antigens (figure 4A and online supplemental figure 6A), with significantly lower responses to wild-type peptides, confirming their antigen-specific reactivity. TCR2 did not respond to the mutated antigen (online supplemental figure 6B). Both reactive TCRs functioned independently of co-receptors, with CD4 and CD8 T cells being 4-1BB+ (figure 4A). Cytokine production (IL-2, IFNγ, TNFα) and CD107α expression were measured after 24 hours of stimulation with peptide-loaded APCs. Both mTCR1 and mTCR3 transduced T cells exhibited cytotoxicty against specific antigens but not with wild-type peptides (figure 4B,C). Additionally, avidity differences indicated varied sensitivities to antigen concentrations (figure 4D). Overall, these findings show that while both TCRs demonstrate specificity and functional profiles, their differing avidity may influence immune responses effectiveness.
Evaluation of reconstructed TCRs for neoantigen-reactivity and functionality. (A) FACS plots show the 4-1BB expression on total CD4+ and CD8+ T cells following 24h stimulation of mTCR1- and mTCR3-engineered T cells with mutant and wild-type SBNO2 peptides. Reactivity on both CD4+ and CD8+ T cells indicate a co-receptor independent stimulation of the TCRs. Production of IFNg, TNFa, IL-2 and CD107a expression on (B) CD4+ and (C) CD8+ cells for mTCR1- and mTCR3-modified T cells after stimulation with mutant and wild-type peptides. Mock (untransduced) T cells were used as negative control. (D) Avidity of mTCR1 and mTCR3 was assessed based on percentage of 4-1BB expression on mTCRb+CD3+ T cells following peptide titration of mutant and wild-type peptides. All experiments presented in the panels were conducted exclusively using samples from the HCC05 lymph node. Mut, mutant; mTCR, murinized T cell receptor; TCR, T cell receptor; WT, wild-type.
Single-cell RNA-seq reveals unique signatures of neoantigen-reactive T cells
We finally investigated phenotypic and functional transcriptomic markers in 4-1BB positive (SBNO2-p) and negative (SBNO2-n) sorted T cells after antigen stimulation. Three main T cell populations were identified (figure 5A,B): (1) a non-antigen reactive/non-exhausted subset (cluster 0) exclusive to SBNO2-n, characterised by high IL7R/CD127 expression; (2) an activated/exhausted-like population (cluster 1), present in both samples but expanded in SBNO2-p; and (3) a neoantigen-reactive/exhausted but functional subset (cluster 2), uniquely expanded in SBNO2-p (figure 5A,B and online supplemental figure 7A,B). The TCR clonality indicated neoantigen reactivity with TCR clone A/mTCR1 co-localised in cluster 2 of SBNO2-p, while clones B/mTCR3 and C/mTCR2 were dispersed across clusters 0 and 1 (figure 5C and online supplemental figure 7C). Activation/exhaustion markers (eg, CD45RO protein, CD69 protein, TIGIT, TNF, IL2RA/RB) were identified in both exhausted clusters. Still, cluster 1 mainly expressed PD-1 protein, CTLA4, TNFRSF4, ADGRG1 and TNFRSF18, while cluster 2 showed high levels of CD39 protein, LAG3, TNFRSF9 (4-1BB), HOPX, IFNG, NKG7 and GZMB (figure 5D,E and online supplemental figure 7B,D). This indicates a unique signature for neoantigen-reactive CD4+ T cells with effector and cytotoxic properties.
Transcriptomic and proteomic characterization of neoantigen-reactive TCRs. (A) UMAP visualisation and distribution of T cell clusters 0-2 between SBNO2-p and SBNO2-n samples. (B) Stack bar plot showing the proportion of each cluster 0-2 between reactive and non-reactive SBNO2 samples. (C) UMAP localisation of TCR clones A (mTCR1), B (mTCR3) and C (mTCR2) that were previously assessed for neoantigen reactivity. (D) Ridge plots with protein expression of markers measured by antibody-derived tags for T cell clusters 0 to 2. (E) Violin plots showing the expression levels for representative gene markers for clusters 0 to 2. All experiments presented in the panels were conducted exclusively using samples from the HCC05 lymph node.
Discussion
The diverse T cell repertoire enables recognition of various mutations and antigens, playing a critical role in controlling cancers, including HCC.14 30–32 This study characterised T cells and their TCRs from multiple immunological sites in HCC patients, focusing on memory CD4+ and CD8+ precursors for personalised treatments.26 We sought to determine if selected neoantigens from HCC tumours could activate reactive T cells from both intra- and extra-tumoural sources, facilitating the isolation of neoantigen-driven TCRs.
Unlike previous studies concentrated on TILs, our research highlights T cells from liver flushes and draining lymph nodes as valuable sources of neoantigen reactivity. This approach may help overcome the challenges associated with the poor proliferation of TILs observed during the study, particularly in tumours with significant necrosis. Our findings indicated that T cell reactivities predominantly originated from these alternative sources, specifically targeting patient-specific mutations and a hotspot mutation in the PIK3CA gene.33 Notably, T cell responses to mutated neoantigens were observed in four out of seven HCC patients. Whether these responses effectively controlled tumour growth remains to be determined.
The absence of antigen responses in tumour-infiltrated T cells could stem from insufficient enrichment of neoantigen-specific TCR clones during expansions or the use of bulk T cells rather than enriched antigen-experienced cells.34 Importantly, we successfully isolated neoantigen-reactive TCR clones from the previously overlooked liver flushes and draining lymph nodes.
Following genetically transferring TCRs into patients’ autologous T cells, we identified tumour-specific reactivity against the neoantigen SBNO2, which is involved in the IL-10 pathway.35 Although the identified TCRs were clonal and significant for personalised adoptive cell therapy (ACT), shared neoantigens may broaden the applicability of TCRs to larger patient populations.15 30 33 36
Both neoantigen-reactive TCRs exhibited co-receptor-independent recognition, activating CD4+ and CD8+ T cells and eliciting polyfunctional responses. Differences in avidity were observed through 4-1BB expression during peptide titration. The neoantigen-reactive CD4+ T cells showed high levels of CD39 and expressed genes associated with effector function and cytotoxicity, indicating a Th1 polarisation essential for effective anti-cancer responses. Notably, HOPX gene expression emerged as a signature marker for these T cells, correlating with T cell persistence.37 38
Our findings underscore the distinct transcriptomic and proteomic profiles of clonally expanded neoantigen-specific T cells, suggesting strategies to differentiate them from exhausted or non-reactive cells. In conclusion, we assessed T cell responses in advanced HCC patients by analysing tumour infiltrates, liver flushes and lymph nodes, emphasising the need for individualised approaches. Our integrated pipeline successfully isolated neoantigen-reactive T cells and their TCRs, paving the way for new immunotherapy strategies in HCC.
Limitation
The limitations of this study include the relatively low mutational burden in HCC, which results in fewer neoantigens and limits the number of potential targets for TCR-based therapies. This makes it more difficult to identify specific tumour antigens. Additionally, the use of long peptides in our approach may not fully capture the diversity of T cell reactivity, as it could miss responses to shorter peptides or post-translational modifications, potentially restricting the range of immune responses detected. Finally, by focusing primarily on SNVs, we may overlook other important genetic alterations, such as insertions, deletions (indels) and structural variants, which can also contribute to tumour immunogenicity and lead to an incomplete understanding of the tumour-specific T cell landscape. To overcome these limitations, future studies should expand the scope of antigen targets by incorporating shorter peptides, post-translational modifications and other genetic variations such as indels and structural variants. Additionally, focusing on TAAs and advancing TCR engineering could enhance the effectiveness of TCR-based therapies, even in tumours with low mutational burdens.
Supplemental material
Data availability statement
Data are available upon reasonable request. Data are available on reasonable request. All data relevant to the study are included in the article or uploaded as online supplemental material.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by Etikprövningsmyndigheten Dnr 2024-01616-02. Participants gave informed consent to participate in the study before taking part.
Acknowledgments
We thank all the patients enrolled in the present study. The Transplantation Unit at Karolinska University Hospital for providing clinical samples. The Flow Cytometry Core Facility (MedH) for assisted sorting. Malin Larsson and the Bioinformatic PhD Advisory Programme for their contributions to data processing. We acknowledge support from the National Genomics Infrastructure in Stockholm funded by Science for Life Laboratory, the Knut and Alice Wallenberg Foundation and the Swedish Research Council, and SNIC/Uppsala Multidisciplinary Center for Advanced Computational Science for assistance with massively parallel sequencing and access to the UPPMAX computational infrastructure. The authors also acknowledge Agneta Andersson assistance with the Maxcyte electroporator.
References
Footnotes
HY and CC contributed equally.
Correction notice This article has been corrected since it published Online First. The affiliations have been updated for author Ola Nilsson.
Contributors Study conceptualisation: PM, MS and AP. Experimental design: PM, CC, MB and AP. Clinical samples: CJ. Samples processing: PM and AP. Sequencing support: OBN, PM and JB. Experiments: PM, HY, DNS, FG and GR. Data analysis: PM, HY, JF, CC, Y-CWL and AP. Technical/scientific support: MC, MS, PS, HY and CJ. Manuscript writing: PM, HY and AP. Manuscript reviewed and approved by all authors. AP is the guarantor.
Funding This work was supported by grants from Region Stockholm CIMED (2023-2025 AP); The Sjöberg Foundation (2023-2025 AP), The Swedish Cancer Society (20 0769 Pj AP; 2024-2026 MS; 201355PjF PS), The Swedish Research Council (2023-2025 AP), Vinnova CAMP and Vinnova LaScarna grant (MS).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.