Colorectal cancer (CRC) ranks third for incidence, after breast and lung cancers, and second for mortality, after lung cancer (1). Mortality is mainly due to metastatic disease, which affects nearly 22% of CRC patients at diagnosis, and more than 50% thereafter due to relapses (2). CRC is an age-related malignancy, since 90% of cases are diagnosed after the age of 50, and the average age at diagnosis is 70 (1, 3), therefore the overall aging of the population, will inevitably lead to an increase in the number of new cases diagnosed in the near future, which makes CRC a significant public health challenge.

Advances in patient-tailored therapy, which is based on the mutational, transcriptomic, and proteomic profiles of individual CRC (4), and on its immune cell context, can fulfill this need. Genetic and epigenetic alterations play a key role in shaping the CRC microenvironment (5). Defects in DNA mismatch repair proteins and subsequent microsatellite instability lead to the accumulation of mutation loads in cancer-related genes and the generation of neoantigens, which promote intra-tumoral immune cell influx, essential for immunotherapy. Mismatch repair deficiency (dMMR) or microsatellite instability (MSI) is significantly associated with long-term immunotherapy-related responses and better prognosis in cancers treated with immune checkpoint inhibitors (6). However, many of the patients that initially respond to treatment later become resistant or relapse due to the occurrence of further genetic and epigenetic alterations, which generate new cancer clones with different molecular traits. These molecular events re-shape the tumor microenvironment (TME), upregulating other inhibitory receptors and immunosuppressive mediators, and expanding the landscape of immune cell exhaustion. Exhaustion is a functional status attributed to T lymphocytes and characterized by the loss of effector function, including proliferation, release of cytokines, and secretion of cytolytic molecules, due to persistent antigen stimulation and inflammatory signals in pathological contexts, such as chronic infections or inflammatory diseases and cancer (7, 8). However, in the last few years it has become progressively evident that exhaustion is not exclusive of T cells, but can also affect other immune cell subsets (9–12) homing the TME.

In this study, using bioinformatics, we analyzed the transcriptional profile, and concomitant immune cell context, of the immune exhaustion in CRCs of the TCGA PanCancer collection, and its relationship with patients’ clinical-pathological characteristics and follow-ups. A validation cohort of CRC patients, which includes short- and long-term survivors, provided tumor samples to localize and quantify, by immunohistochemistry and computerized morphometric analysis, immune exhaustion gene expression at protein level.

Our findings reveal that high levels of expression of specific immune exhaustion genes occurs in MSI molecular subtype, however, independently from the molecular subtype of CRC, high expression levels of a range of immune exhaustion genes regulating different immune cell subsets, are significantly associated with worst prognosis. Biostatistical co-occurrence and morphological co-localization of the expression of immune exhaustion genes and markers of T and B lymphocytes, and of conventional (c) NK cells, strongly suggest that, regardless of the molecular subtype of tumor, CRC-associated immune exhaustion is not only a “T cell matter”. CIBERSORTx analysis corroborates this concept revealing that the expression of relevant immune exhaustion-related genes, detected in the different T lymphocyte subpopulations, also involves B and NK cell subpopulations.

Profiling of functional immune exhaustion in individual CRC can lead to the development of a successful patient tailored therapy to counteract metastasis and disease recurrence.

Despite the survival benefit of early detection through increased screening, CRC remains a leading cause of cancer related death (35), which is primarily due to disease progression or to the failure of current therapies to counteract metastasis (36). Immunotherapy, which has recently revolutionized the treatment of advanced tumors (37, 38), needs to be improved for CRC care, since: a) eligible patients are limited to a subset diagnosed with mismatch-repair-deficient mutations or microsatellite instability tumors (39) and, b) responsive patients will eventually experience resistance to treatment and relapse (40), concurrently with a weakening of their immune response, namely immune exhaustion (8). Exhaustion, i.e. the progressive loss of effector function due to chronic low-affinity antigen stimulation, is essential in maintaining immune homeostasis by regulating the duration and the magnitude of T cell responses, through a reduced proliferation, impaired effector functions, and elevated and sustained expression of multiple inhibitory receptors, known as immune checkpoint proteins (41). It is becoming progressively clear that the range of checkpoint inhibitors extends well beyond the few targets of current immunotherapy (41), and that functional exhaustion is not an exclusive matter of T lymphocytes, but heavily involves other immune cell populations (12, 42–44), that take part in mounting an effective immune response (19, 45, 46).

Through biostatistics and immunopathological investigations, this study explores the transcriptional program and immune cell context, that characterize the immune exhaustion landscape of CRC. While confirming the previously reported observation of high levels of expression of PD-L1, LAG3 and T-bet in the context of MSI tumors (39, 47, 49), our findings reveal that, independently of tumor stage and molecular subtype, high levels of a wide range of inhibitory receptors and exhaustion-related transcription factors, such as EOMES, T-bet, PD-L1, PD-1, LAG3, BTLA, FCRL4, SIGLEC6 and TRAF1 are associated with reduced survival in CRC patients.

TRAF1 belongs to the Tumor necrosis factor receptor (TNFR)-associated factors (TRAF) protein family, which plays important roles in the immune system as key intracellular signaling molecules in TNFR, Toll-like receptor (TLR), cytokine and antigen receptor signaling pathways (19). TRAF1 expression in T and B lymphocytes can be induced by different stimuli that activate the transcription factor NF-κB (49–51), and may act as both positive and negative regulator of immune signaling. Our immunopathological analyses show TRAF1 expression in CRC infiltrating immune cells and lymphoid aggregates, mostly in CRC samples from short-term survivors, coherently with its negative prognostic value. Besides being produced by immune cells, TRAF1 can be overexpressed by neoplastic cells in different B cell malignancies, and in solid tumors including non-small cell lung cancer for which TRAF1 has been proposed as a biomarker of tumor progression and worst clinical outcome (52–54).

Both PD-1, which binds to PD-L1 and PD-L2, and LAG3, which binds to MHC class II molecules, can be found on T and B lymphocytes, macrophages, dendritic cells (DCs), and NK cells (55–59). In CRC their expressions are strongly correlated with each other and, along with TIGIT and BTLA, are tightly associated with markers of T lymphocytes and cNK cells, in which they can inhibit receptor dependent activation (60), thereby dampening anti-tumor effector functions.

BTLA is primarily expressed on B and T cells, and on monocytes and DCs to a lesser extent, and shares structural similarity with PD-1, but its only known ligand is the Herpes virus entry mediator (HVEM), a member of the tumor necrosis factor receptor superfamily (TNFR-SF) (61). The inhibition of T cells by BTLA is stronger than the positive stimulatory effect of HVEM, a co-stimulatory tumor-necrosis factor receptor on T cells, and prevents the excessive activation of T cells (62), thereby maintaining T cell tolerance (63). It is prominently expressed on human T cells in the TME and can inhibit tumor specific CD8⁺ T cells (64). BTLA blockers have been found to enhance human T cell responses when used alone or in combination with antibodies against PD-1 (65–67). Although it is well established that BTLA mainly functions as a negative regulator of lymphocytes, recent studies indicate that the role of BTLA in tumor-resident T cells is complex, as engagement by its ligand, HVEM, inhibits proliferation and cytokine production, but promotes survival of tumor-infiltrating lymphocytes (68). Furthermore, BTLA associates with the BCR and, upon binding to HVEM, recruits the tyrosine phosphatase Src homology 2 domain-containing phosphatase 1, reduces activation of signaling molecules downstream of the BCR and inhibits B cell activation (69).

The role of TGFβ1 in the development of T regulatory cells and immunosuppression is well established (70, 71). In CRC its expression has been found to be increased when compared to adenoma and normal colon tissue (72). Elevated TGFβ1 levels were observed in the primary tumor and in plasma from CRC patients and were correlated with metastasis and poor prognosis (73, 74). TGFβ1 has also been implicated in T cell exhaustion. Notably, inhibition of TGFβ1 can prevent CRC metastasis by unleashing a cytotoxic T cell response against cancer cells, implying that TGFβ1 signaling suppresses cancer recognition by the immune system (75, 76). In vivo, inhibition of TGFβ1 signaling in CD8⁺ T cells, through expression of a dominant-negative receptor, improved the function of exhausted T cells (77). Enhanced and sustained TGFβ1/Smad signaling is a distinctive feature of virus-specific CD8⁺ T cells during chronic viral infections in vivo and leads to the up-regulation of the pro-apoptotic protein Bim and to T cell apoptosis. By contrast diminished TGFβ receptor signaling reduces T cell death and indirectly inhibits PD-1 and IL10 expression, and enables the acquisition of effector CD8⁺ T cell functions (77).

While PD-L2 expression is mainly restricted to professional APCs, such as DCs and macrophages, PD-L1 is constitutively expressed by T and B cells, macrophages and DCs and is up-regulated by inflammatory mediators (78). However, it can also be expressed by cancer cells in different tumors, including CRC, and inhibit the adaptive immune response preventing tumor cell apoptosis, which could explain the association of its high levels of expression with a worse clinical outcome (79–81).

EOMES and T-bet are T-box transcription factors that drive the differentiation and function of cytotoxic CD8⁺ T lymphocytes and NK cells (82) and that promote type 1 innate and adaptive cell-mediated immunity (83). In HCV and HIV infected patients, EOMES^high T cells display high expression of inhibitory receptors and severe functional defects, that lead to the persistent stage of chronic infection (84, 85). In acute myeloid leukemia, it has been reported that EOMES binds to the promoter of TIGIT and up-regulates the expression of this inhibitory receptor on patient-derived T cells. Moreover, in these patients a high frequency of EOMES⁺ T-bet^low CD8⁺ T cells was associated with poor response to chemotherapy and shorter OS (86, 87). Co-expression of high levels of EOMES and T-bet, and its correlation with the expression of PD-L1, PD-1, LAG3, and TIGIT has been determined, trough bioinformatic analyses, in CRC from patients with a shorter OS, and confirmed by immunopathological investigations.

TIGIT is a checkpoint receptor which mediates T and NK cell exhaustion in tumor-bearing mice and in patients with colon cancer (88). Immunostainings of CRC samples from short-term survivors show a strong expression of TIGIT in intra-tumoral immune cells, scattered within the cancer and lymphoid structures, but also in fibroblast-like stromal cells, resulting in a robust barrier against immune attack. Blockade of TIGIT, in colon cancer-bearing mice, prevents NK cell exhaustion and promotes a potent tumor-specific T cell immunity in an NK cell-dependent manner (88). Conventional NK cells belong to group 1 ILCs and are essential in killing infected or transformed cells that have null or low expression levels of MHC-I molecules (89, 90). NK cells are involved in the control of CRC progression and metastasis, but they are also susceptible to tumor-directed dysregulation (60, 91, 92). Similar to T cells, NK cells undergo exhaustion during tumor progression or chronic infections and reveal impaired cytolytic activity and production of effector cytokines, downregulation of activating receptors and overexpression of inhibitory receptors (93). Analysis of RNA-seq data from the TCGA PanCancer cohort and morphological investigations of CRC samples from the validation cohort, reveal that expression of PD-L1, PD-1, LAG3, EOMES, T-bet and TIGIT is associated and co-localizes not only with CD8⁺ and CD4⁺CD3⁺ T cells, but as well as with the expression of PD-L2 and IDO1, it co-occurs with typical markers of cNK cells, which are identified as NKp46⁺EOMES⁺T-bet⁺CD94⁺, a subset of group 1 ILCs (34). Group 1 ILCs were found to be increased in CRC, when compared with normal colonic mucosa (94), and to exhibit high level of inhibitory receptor in the late stage of the disease (92). The CD94/NKG2-A functions as a natural killer cell inhibitory receptor for different HLA class I alleles (95) including HLA-E, its predominant ligand, which confers resistance to NK-cell-mediated lysis (96). PD-1 overexpression in NK cells results in decreased degranulation (97), indicating that PD-1 signaling is suppressive not only in T cells but also in NK cells, and identifies a subset of fully mature NK cells with low proliferative response and impaired anti-tumor activity (98). A role as inhibitor of NK effector functions has been recently described also for IDO1, a rate-limiting metabolic enzyme that converts tryptophan into downstream catabolites kynurenines (KYN) (99) and is involved in the establishment and maintenance of peripheral tolerance and tumor immune escape (33). It has been demonstrated that KYN suppresses the function of CD4⁺ T cells and DCs (100) and promotes NK cell apoptosis (101), and that tryptophan-derived L-KYN impairs NK cell cytotoxicity by decreasing NKp46 and NKG2D/NKG2DLs activating receptors (102–104).

A scenario which mimics exhaustion in T lymphocytes, can affect B cells during chronic infection, inflammatory diseases (9, 25, 105) and cancer (106). B cell exhaustion manifests through the progressive loss of effector functions, of antibody and cytokine production, and high levels of inhibitory receptors. BTLA, FCRL4 and SIGLEC6, which have been shown to inhibit BCR signaling (26), have been found to be highly expressed in CRC from patients with shorter OS, by Kaplan Meyer curves of the PanCancer cohort, and in CRC of short-term survivors from the validation cohort, as detected by immunopathology. The tight association of high levels of BTLA, FCRL4, SIGLEC2 and SIGLEC6 (107) with the B lymphocyte markers, CD19, CD20 and CD79a, along with their negative prognostic value, strongly suggests a role for exhausted B lymphocytes in CRC immune escape and patient outcome.

The CIBERSORTx tool, which identifies immune cell subpopulations, based on their molecular signature, provides additional information on the immune exhaustion landscape of CRC, and strengthens the concept that, in the TME, exhaustion affects cells other than just T lymphocytes. Expression of FOXP1 and SIRT1, which can suppress antitumor T cells (108) and promote chemoresistance (109, 110), has been found in CRC infiltrating Tregs (FOXP1), CD8⁺ and memory resting CD4⁺ cells, but also in naïve and memory B lymphocytes (both FOXP1 and SIRT1). Expression of BATF involves CD8⁺ cells (111), but also resting NK cells, while expression of NR4A1 and TOX (112, 113) has been found in memory resting CD4⁺ cells, but also in naïve and memory B cells, respectively.

The integration of bioinformatics with immunopathology, used in the present study, provides new insight into the CRC microenvironment and its clinical relevance. Results suggest that CRC-associated immune exhaustion is not limited to the MSI subtype and is not exclusive of T lymphocytes, but it also involves B cells and cNK cells, and includes a wide range of checkpoint molecules and immune-related transcription factors, with significant prognostic and therapeutic implications. Taken as a whole, our findings emphasize the need to extend genome sequencing and comprehensive immunopathological analyses of individual CRC into the clinical practice, for the development of a multi-targeted patient tailored immunotherapy to effectively counteract disease progression.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethical Committee of the “G. d’Annunzio” University and Local Health Authority of Chieti. The patients/participants provided their written informed consent to participate in this study.

EDC conceived and supervised the study. CS performed bioinformatic analyses. LD’A, CF, and SLC performed the experiments. EDC, CS, and LD’A interpreted the data. EDC wrote the manuscript. All authors contributed to the article and approved the submitted version.

The research leading to these results has received funding from AIRC under IG 2019 - ID. 23264 project – P.I. EDC and from the Italian Ministry of University and Research (PRIN - 2017M8YMR8 – Unit 3 P.I. EDC).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.