PD-L1/PD-1 crosstalk in colorectal cancer: are we targeting the right cells?

Colorectal cancer (CRC) is the third-most common cancer worldwide and the second in Europe [1‐3]. Current treatments include tumour surgery in the early stages, followed by chemo- and radiotherapies for patients in advanced stages [4]. Although the 5-year life expectancy is close to 90% with early detection, up to 40% of patients experience recurrence, mainly in the form of regional or distant metastases [5], which has driven oncologists to search for further preventive treatments such as immunotherapies [6].

Immunotherapy is a rapidly expanding field, and significant effort is being made to improve the antitumour immune response by shifting immune checkpoint (IC) molecules [7, 8]. IC molecules are key regulators of T cell activation and self-tolerance [9], offering a new avenue of potential therapeutic targets in immune response diseases. ICs comprise a heterogeneous superfamily of molecules that either costimulate or inhibit T cell responses to mediate immune tolerance and mitigate collateral tissue damage when the immune system is responding to pathogenic infection [10]. Pathological conditions affecting both the adaptive and the innate systems, such as cancer progression, have been observed to evolve by altering the expression of these proteins. The surface ICs might act as ligands of lymphocyte receptors, modulating the duration and range of the adaptive immune response, as both stimulatory and adaptive response inhibitors. Within the inhibitory ICs, members of the B7 superfamily of molecules, and especially the ligand of programmed cell death proteins (PD-L1), emerge as promising molecules in various clinical contexts [11‐13].

Unfortunately, there are still no effective immunotherapies for CRC [14]. The continuous interaction of the gastrointestinal tract with pathogens, and hence the continuous action of the immune system on this tissue, represents a problem for the use of immune-based therapies [15, 16]. Nevertheless, initial studies have shown the importance of the immune system in prognoses, highlighting the crucial role of tumour-infiltrating cells such as lymphocytes, natural killer cells and macrophages [17]. In addition, analyses of the type, density and location of tumour-infiltrating immune cells within CRC tumour samples have revealed that, in addition to genetic mutations and tumour/node/metastasis staging, immunological data are an independent predictor of patient survival [18, 19]. Along these lines, there is broad consensus in that the subset of patients with a clinical response to PD-1 therapy consist of those harbouring a tumour microsatellite instability-high (MSI-H) phenotype, also called deficient DNA mismatch repair (dMMR) CRC, in contrast to proficient mismatch repair (pMMR) CRC [20], with MLH1, MSH2, MSH6 and PMS2 as the main MMR gene products.

Herein, we have analysed the presence and primary features of monocytes and T-lymphocytes in both tumour and peritumour tissues of patients with CRC, to elucidate which are the main tumour cells involved with ICs effects. Our data reinforce the importance of innate immune cells in the tumour microenvironment context, emphasising that crosstalk between tumour cells and immune components is significantly driven through the interaction of PD-L1/PD-1 immune checkpoint shifting, despite pMMR status.

Although cells isolated from CRC patients’ tumours exhibited proper stem cell marker expression, which differs from those with peritumour origin, we did not detect a significantly associated expression of PD-L1. Hence, these data suggest that tumour cells, per se, are not the primary source of PD-L1 in tumour samples. This suggestion is in line with other authors’ reports showing that methylation of the PD-L1 gene in CRC cells can result in low transcription and translation of PD-L1 in these cells [26].

Due to the increased relevance of immune system components in the tumour microenvironment [27], we first studied the presence and phenotype of infiltrated CD14⁺ monocytes/macrophages, and CD4⁺ or CD8⁺ T-lymphocytes, noting that the tumour zones were significantly enriched in CD14⁺ cells. Our results suggest that not only lymphocytes but also monocytes/macrophages play an important role in the evolution of CRC. These cells serve as the first line of host defence and are equipped to recognise and respond to tumour cells by triggering inflammation. However, we confirmed that tumour-infiltrating monocytes/macrophages exhibited an alternative, M2-like activation [28] in the samples. Thus, their protective influence might be suppressed in this context, given that tumours counteract the cytotoxic and proinflammatory activities (M1 phenotype) of tumour-associated macrophages (TAMs) in their local environments by neutralising them via multiple mechanisms, such as the production of certain tumour cytokines and prostaglandins [29]. In fact, findings on the importance of this polarisation process have not been only reported for CRC progression [30] but also for other several cancers, such as glioblastoma [31], as a consequence of M2-macrophage production of important factors that augment tumour growth (e.g., IL-10).

Furthermore, these infiltrated CD14⁺ cells exhibited high PD-L1 expression, which allowed them to interact with CD3⁺PD-1⁺ T-cells, and might provoke the phenomenon known as ‘T-cell exhaustion’ thus impairing the T-cell response to tumour expansion. PD-L1 expression in CRC cells appears to be paradoxically associated with a high number of CD8⁺ cells [32], and it correlates with early tumour stages. Nevertheless, we found no significant PD-L1 levels on cancer cells or CD3⁺CD8⁺ cells in our tumour samples. In fact, our data showed a significant number of infiltrated CD4⁺ T-cells expressing PD-1, which strongly suggests an interaction with the CD14⁺PD-L1⁺ resident population. These findings are in line with those of Llosa et al. [33], who postulated that CD4⁺ T-cells infiltrated populations, in both pMMR and dMMR CRC tumours, might play an important role in PD-L1/PD-1 axis function through the T-cell exhaustion phenomenon. In addition, we have recently reported a similar behaviour in some other clinical contexts, such as sepsis [34] and obstructive sleep apnoea [35], leading to T-cell exhaustion and the progression of illness due to the inability of T-cells to work steadily.

It is noteworthy that when an α-PD-1 was added to cocultures of naïve immune lineages and pMMR tumour cells, an increase in CD4⁺ cell proliferation was observed, suggesting that the T-cell exhaustion mechanism due to PD-L1/PD-1 crosstalk had been abolished [36, 37]. These data are in agreement with the observed positive outcome of patients with CRC who were treated with new α-PD-1 drugs such as pembrolizumab, nivolumab and other related antagonists [38], and match with the observed predisposition of tumour-infiltrating lymphocytes (TILs) and TAMs in dMMR tumours to respond to immune checkpoint blockade therapy (ICBT) [33, 39]. The benefits of α-PD-1 antibody therapy for pMMR and dMMR tumours remain controversial [40]. Whilst a majority of authors concur on defining dMMR as the sole tumour type able to respond to this therapy [41], there are also various studies stating its relevance against pMMR tumour progression [32, 42]. In this sense, the most common argument for dMMR performance is based precisely on immune-infiltrated population activity in highly mutated MMR-deficiency tumour microenvironment (TME) [43]. However, it could apply also to pMMR (although to a lesser extent of specific TME cases due to their associated gene stability) in which up-regulators of PD-L1 expression (e.g., IFNγ) are present in higher proportions [41]. In addition, the responsiveness to α-PD-1 observed in these pMMR tumours might be linked to alternative mechanisms rather than to a deficiency of DNA repairing enzymes. For all these cases, other features should also be taken into consideration when mechanisms for PD-L1/PD-1 crosstalk are dug into. These features include the generation of neo-antigens that then activate T-cell response to tumours, changes in signalling pathways of chemokines and cytokines expression, and the loss of MHC class I molecules in tumour cells [44]; all indicating the solid implication of TILs and TAMs. Hence, our results reinforce the suitability of either active or passive therapies [43] to restore patient immunity, e.g., M1-macrophage polarisation, together with ICBT.

Finally, some recent studies have found significant differences in PD-L1/PD-1 axis regulation among CRC primary and metastatic lesions. In fact, epigenetic mechanisms, PD-L1/PD-1 regulatory factors in the TME and TILs/TAMs composition, might be quite different in both tissues [41], making regimens for these metastatic patients clinically challenging with a single approach. Even though these differences need further investigation, combinatory treatments for fighting the complex relationships between immune surveillance and the evolving properties in tumour cells appear to be the most advantageous for treatment of the various CRC subtypes. Accordingly, our findings open the possibility of reconsidering combination therapies for pMMR tumours, including both ICBT and driving local infiltrated immune populations into a “naïve-like” stage to overcome tumour resistance.

Altogether, our results support previous findings indicating that the tumour microenvironment induces the expression of PD-L1 in CD14⁺-TAMs, and that their interaction with CD4⁺PD-1⁺ cells triggers T-cell exhaustion, thus allowing tumour propagation. In this sense, inhibition of the crosstalk at this immune checkpoint, PD-1/PD-L1, would abrogate this effect even in pMMR tumours under specific conditions. Although currently there are still no effective immune therapies for CRC, our results indicate that this line of research should not be abandoned, given both ICs and “naïve-like” stage monocytes/macrophages appear to be crucial for the development of CRC. Therefore, new strategies to avoid such interference must be addressed in this clinical context.