The evolution of T-cell receptor repertoire in different stages of non-small cell lung cancer


 Background The intricate relationship between the tumor and host was not well understood, and antigen-specific T cell is fundamental in understanding the interaction. TCR repertoire analysis which described TCR clonotypes and TCR numbers has shown that TCRs with high frequency was tumor-specific T cells, while others might be ‘bystander’ T cells within tumors. However, how these “expanded” tumor-specific T cells was selected during the tumor development was not clear. Methods We retrospectively analyzed TCR sequencing and mutation sequencing results from 144 non-small cell lung cancer (NSCLC) patients. Results A rich TCR repertoire comprising thousands of different TCR sequences was identified in all stages of NSCLC, with most TCR clonotypes presented at low frequency. Interestingly, Stage IV NSCLC tumors contain more expanded TCRs as compared to earlier stages, however, lymph node metastasis or tumor size had little impact on expanded TCRs. Moreover, accumulation of mutations did not significantly change the number of TCR clonotypes, however, EGFR mutant patients had significantly lower while KRAS mutant patients had significantly higher number of TCR clonotypes especially in terms of those “expanded” TCRs. Conclusions In summary, T cells in the tumor microenvironment were gradually activated with tumor development. Critical events such as distal metastases and generation of EGFR or KRAS mutations might be the major factors affecting the changing of tumor-specific T cells in the tumor microenvironment.

different stage to explore evolution of TCR repertoire and potential factors associated with the remodeling process.
TCR clonotypes and TCR numbers, which evaluated the Richness of T cell clonotypes and the density of T cell expansion, were wildly used to characterize the anti-tumor changes in the tumor microenvironment (8-10). As reported (10,11), most T cells in the tumor microenvironment were "bystander" T cells, such as memory T cells; while tumor-specific T cells were characterized by the TCRs within the top 1% of clonotypes. These expanded tumor-specific TCRs were always with the TCR frequency higher than 2/1,000 (0.002).
It was reported that tumor patients showed a significantly lower TCR clonotypes, but similar TCR numbers comparing with healthy people (8, 10). This prompted us to study the changes rules of TCR repertoire during tumor development. Tumor, node, metastasis (TNM) staging system is an internationally accepted system used to determine the disease stage. It is based upon an evaluation of the primary tumor, the regional lymph nodes and lymphatic drainage, and the presence or absence of distant metastases to determine prognosis and guide management. We employed the TNM system to show the difference of TCR in different stage, and also investigated the different effects of T, N and M on TCR evolution. We also analyzed the correlation between TCR numbers/clonotypes and genetic mutations to observe the influence of tumor development to TCR repertoire in this study.

Subject cohorts
We collected TCR sequencing results from 144 lung cancer patients, including 95 stage I patients, 14 stage II patients, 15 stage III patients and 20 stage IV patients (Additional file 1: Table S1). All the patients were treatment naïve and received previous anti-cancer therapy at the Cancer Center, Peking Union Medical College Hospital (Beijing, China). Clinical information was collected from the hospital's case management system and confirmed with relevant doctors.

NGS base somatic mutation detection
All patients have been tested for EGFR/KRAS mutation status, and 40 of them had received 1021-gene panel sequencing by next-generating sequencing (NGS). Genetic analysis was conducted as previously described (12,13) (Additional file 2: Table S2). Briefly, serial sections from formalin-fixed High-throughput sequencing of T-cell receptor β genes TCR repertoire was sequenced using surgical tissues of patients without distant metastasis and needle biopsy tissues of patients with advanced lung cancer. TCR sequencing and TCR repertoire quantification was performed as previously research (8, 14,15). CDR3 region of TCR β chain was amplified by multiplex PCR including PCR1 and PCR2. For as much amplification as possible V(D)J combinations, a set of 32 V forward and 13 J reverse primers was adopted to perform multiplex PCR1 and assays. Universal primers were performed in PCR2. TCRβ CDR3 region was sequenced using the Illumina Novaseq System, and reads of length 151-bp were obtained. Then, the CDR3 sequences were identified and assigned by the MiXCR software package (The CDR3 sequence was defined as the amino acids between the second cysteine of the V region and the conserved phenylalanine of the J region) (16). Sequences were collapsed and filtered in order to identify and quantitate the absolute abundance of each unique TCRβ CDR3 region for further analysis, as previously described (15,17,18) Statistical analysis Mann-Whitney test was used to compare differences between groups. Correlations between variables were analyzed using Spearman's rank test. All statistical analyses were performed using GraphPad Prism 8.0. In this study, all tests were two-sided and p values < 0.05 were considered statistically significant.

1.
Elderly patients had more expanded TCRs A rich TCR repertoire comprising thousands of different TCR sequences was identified in all stages of NSCLC. To investigate the correlation between clinical characteristics and TCR repertoire, we first performed different thresholds of TCR analysis to assess the association between TCR repertoire and age, gender and smoking. We found TCR numbers showed a positive correlation with ages with TCR frequency cutoff value at 0.002, which is corresponding to TCRs within the top 1% TCRs as reported ( Fig 1A)(10). Though we had more patients in stage I than other stages, the age of different stages were comparable (p=0.17, Fig 1B). We also found neither gender nor smoking status affect TCR clonotypes or numbers (19,20) (Supplementary Fig S1, Fig S2). .

2.
Stage IV NSCLC tumors contain more expanded TCRs as compared to earlier stages Most TCR clonotypes presented at low frequency which may be tumor irrelevant bystander T cells as reported (10). When focusing on the TCR sequences at high frequency within the tumor (threshold of 0.002), stage IV samples showed a significantly higher TCR numbers than stage I, II and III samples (Fig 2A). There was no difference in TCR clonotypes among different stages ( Fig 2B). As all the stage IV patients had distal metastasis, we speculated that distal metastasis might associate with the TCR clonal expansion (recruitment and proliferation). To test our hypothesis, we further studied the correlation between lymph node metastasis (N) or tumor size (T) with TCR.

3.
Lymph node metastasis or tumor size had little impact on expanded TCRs To study the correlation between lymph node metastasis (N) or tumor size (T) with TCR, we excluded patients with distal metastases and only analyzed stage II/III patients. These patients were divided into lymph node metastasis group (LM) and non-lymph node metastasis group (non-LM). The number of expanded TCRs accounted for almost 20% of the total TCRs in each group, and no difference was found in TCR numbers among all frequency ( Fig 3A). LM group showed decreased TCR clonotypes only at a threshold of 0.002 and 0.003 ( Fig 3B).
The effect of tumor size on TCR repertoire was analyzed in stage I patients. The patients were divided into three groups according to subgroup division of T1 (T: diameter of tumor; T≤1cm; T≤2cm; T>2cm). Similar to lymph node metastasis, no difference in TCR numbers was noticed and difference in TCR clonotypes was found in the some low frequency group which were threshold of 0.0005,0.001, and 0.002, and ( Fig 3C and Fig 3D). These results indicated that lymph node metastasis or tumor size may influent the TCR clonotypes with low frequency, but had little impact on expanded TCRs.

EGFR but not accumulation of mutations was associated with expanded TCRs
It was reported that numbers of expanded TCRs was correlated with the number of nonsynonymous mutations in early stage NSCLC tumors (10). We therefore tested whether the increased number of expanded TCRs was due to accumulation of nonsynonymous mutation in advanced NSCLC. Out of our surprise, neither TCR numbers nor TCR clonotypes changed significantly following the accumulation of non-synonymous mutations at most thresholds ( Fig 4A). We then focused on the driver mutations of NSCLC. Epidermal growth factor receptor (EGFR) and Kirsten rat sarcoma viral oncogene homolog (KRAS) were the most common mutated oncogenes in NSCLC (21,22). We found there were no difference between the EGFR mutant and EGFR wildtype group in TCR numbers, but the EGFR wildtype group showed significant more TCR clonotypes of expanded TCRs (Fig 4B, 4C). This indicated that EGFR may play more significant roles in TCR clonal expansion compared with accumulation of other mutations. Given that KRAS also was a driver mutation of NSCLC, we further analyzed the relationship between KRAS mutation and expanded TCRs. As depicted in Supplementary Fig. S3, KRAS-mutant patients showed more clonotypes and numbers in highly expanded TCRs than non-KRAS patients. These results indicated that driver mutations might play crucial roles in the evolution of TCR repertoire.

Discussion
In this study, we used two direct and effective indexes of TCR repertoire: TCR clonotypes and TCR numbers to study the evolution of TCR repertoire during tumor development. To the best of our knowledge, it was the first comprehensive analysis of TCR repertoire within non-small cell lung cancer patients at different stages of disease progression ( Supplementary Fig. S4). And it went beyond the reported difference of TCR repertoire between tumor patients and healthy peoples (8, 10).
Our study demonstrated that TCR numbers of expanded TCRs was positive correlated with age as others (23,24). Immunosenescence is characterized by a progressive and global remodeling of the immune functions during aging and it involves both innate and adaptive immunity. Senescence leads to an extensive decrease of naïve CD8 + T cells (25). A reduction of apoptotic molecules such as CD95, suggest that T cells were less prone to spontaneous apoptosis and may accumulate in elderly individuals (26). Moreover, senescent T cells also secrete high levels of inflammatory cytokines (27,28) and previously has described a positive correlation between age, number of CD8 + CD28-T cells and cytotoxicity capacity of freshly isolated T lymphocytes, suggesting that an activation within the cytotoxic arm of the immune system may occur during aging (29). All this results revealed that senescence leads to change of T cells clonotypes, but not means drastic decrease in density of T cells. This was consistent with recent studies showing that elderly patients can benefit from immunotherapy (30).
Our study showed that lymph node metastasis or tumor size had little impact on expanded TCRs. It was consistent with previous studies which have revealed that tumor-infiltrating lymphocytes (TILs) in resected NSCLC were not significantly different among I-III stages (31,32). Remarkably, numbers of vast majority of expanded TCRs were increased in stage IV samples compared with earlier stages.
Increased numbers of expanded TCRs may due to more diversity (increased clonotypes) or enhanced proliferation or recruitment of T cells of each clonotypes or both. Though we could not delineate the exact mechanism in our study, the significant difference of TCR clonotypes between Stage IV and earlier stages indicated that predictive markers of immunoadjuvant therapy and immunotherapy for advanced patients should be different.
As a parallel, we also analyzed whether accumulation of non-synonymous mutation could induce TCR expansion through increased neoantigen production (33). The accumulation of mutation will lead to increase in intratumoral heterogeneity, which will affect the tumor microenvironment (9,13). The spatial heterogeneity of TCRs has been reported and numbers of expanded ubiquitous TCRs (present in all regions of tumor) correlated with the number of ubiquitous non-synonymous mutations (present in all regions of tumor), but not with the number of regional non-synonymous mutations (10). In our study, numerical but not significant correlation between TCR numbers or clonotypes and number of non-synonymous mutations was found probably due to failure to distinguish ubiquitous and regional mutations(9, 10). Interestingly, we found EGFR mutant patients showed a significantly lower TCR clonotypes in expanded TCRs. On the contrary, KRAS mutant patients had higher TCR clonotypes and numbers in expanded TCRs. Several preclinical and clinical findings suggest that the tumor microenvironment of NSCLC patients with EGFR mutations may have less CD8 + TILs, but more regulatory T cells (Tregs) accumulated in the tumor environment (34,35). A recent study showed that KRAS mutant NSCLC was positive related with tumor immunity-associated features, including PD-L1 expression, CD8 + TILs and tumor mutational burden (22). In line with these, EGFR/KRAS which were known as common ubiquitous driver mutations in NSCLC, played more significant roles than other passage mutations in the evolution of TCR repertoire.
Our study has several limitations. First, as a retrospective study, we were unable to connect our findings with clinical outcomes. Secondly, in spite of the large cohort of patients, there were more stage I patients than all other stages. Finally, due to the limited availability of tissue samples, we were unable to perform other study such as T cells staining to further delineate the mechanism.

Conclusions
Despite the gradual accumulation of mutation, and continuous development of NSCLC, critical events such as distal metastases and generation of EGFR or KRAS mutations might be the major factors affecting the changing of tumor-specific T cells in the tumor microenvironment. The significant difference of TCRs between advanced and early stage warranted further study considering that cancer immunotherapy is more and more widely used in NSCLC patients of earlier stage.
Abbreviations TCR, T cell receptor NGS, Next generation sequencing CDR3, complementarity determining region 3 NSCLC, non-small cell lung cancer Declarations Ethics approval and consent to participate We have obtained written informed consent from the participants before genetic testing for further scientific analysis. This study was approved by the Ethic Committee of Peking Union Medical https College Hospital.

Consent for publication
Written informed consent was obtained from the patient. The manuscript does not contain any personally identifiable information or images.

Availability of data and materials
The datasets used and analyzed during this study are not publicly available to protect patient privacy, which will be available from the corresponding author on reasonable request.

Additional Files
Additional file 1: Table S1. Clinical characteristic of 144 patients.
Additional file 2: Table S2. List of target regions of the pan-cancer 1021-gene panel.
Additional file 3: Figure S1.          The influence of mutation accumulation and EGFR-driver mutation to TCR abundance. A. The Spearman's rank correlation coefficient and p value (shown above each point of blue and