Background
The 5-year overall survival rates across gastrointestinal (GI) cancer types vary, with oesophageal adenocarcinoma (OAC) low at 18–19%, whereas colonic and rectal adenocarcinoma rates stand at 58–59% according to the National Cancer Registry Ireland [
1]. OAC rates have increased by almost half in recent years in Western countries, mirroring the increase in obesity [
2]. Colorectal cancer (CRC) is the third most common cancer worldwide with rectal cancer accounting for approximately 35% of CRC cases. Typically, epidemiological and scientific studies group colon and rectal cancer together, despite their different standard treatment regimens [
3,
4]. Standard treatment for oesophageal and rectal adenocarcinoma involves neoadjuvant chemoradiotherapy (CRT) to shrink the tumour prior to surgical resection, whereas for colonic adenocarcinoma the standard treatment involves surgical resection followed by adjuvant targeted therapies [
5‐
7]. Response to CRT is highly variable with just 10–30% of patients achieving a complete pathological response, which is linked with higher 5-year survival rates, for both oesophageal and rectal cancer [
8‐
13]. Tumours have different levels of radiosensitivity and a number of cellular processes and immune mechanisms have been implicated in radioresponse phenotypes [
14‐
16]. Understanding the key components of the immune system which are modulated by the tumour microenvironment (TME) may offer insights into ways to improve the clinical outcome for patients with GI cancers by identifying either prognostic biomarkers or novel therapeutic strategies.
Dendritic cells (DCs) are professional antigen-presentation cells responsible for activation of T cells and thus orchestration of the adaptive immune response [
17]. Immature DCs recognise and capture antigens and are characterised by low expression of maturation and co-stimulatory markers such as CD83, CD54, CD80, CD86; HLA-DR for antigen presentation and other immunoinhibitory markers such as PD-L1 [
18,
19]. DC maturation, a crucial factor for efficient T cell activation, is triggered in response to various inflammatory mediators and TLR-dependent activation, such as bacterial LPS via TLR4, leading to the increased expression of several cell surface markers by DCs, migration to lymph nodes and presentation of antigens via MHC class I and II molecules to activate CD4
+ and CD8
+ T cells [
20]. Factors such as IL-10 and VEGF in the TME influence DC function and these can inhibit IL-12p70 and TNF-α production from DCs [
21,
22]. DCs which secrete high levels of IL-12p70 induce anti-tumour immunity, as they have increased capacity to enhance natural killer cell activity, skew T cell responses to T helper (Th)-1 type and prime tumour antigen specific T cells [
23,
24]. Decreased IL-12p70 expression is associated with suppressed endocytic activity and antigen-presentation machinery, and also decreased motility of anti-tumour immune cells to the tumour site [
25]. TNF-α released by immunostimulatory DCs can also act to enhance T cell stimulatory capacity, while increasing IL-12 production from DCs and decreasing production of the immunosuppressive cytokine IL-10 [
26,
27].
Known risk factors for the development of GI cancers include inflammatory disorders, specifically Barrett’s oesophagus for OAC and inflammatory bowel disease for CRC [
28‐
30]. Not only is inflammation a hallmark of cancer, it plays a pivotal role in modulating radiation responsiveness of tumours [
31]. Radiation can elicit the systemic release of the TLR ligands, damage-associated molecular patterns (DAMPs), after oesophageal irradiation or locally after targeted tumour irradiation, such as treatment for CRC [
32‐
37]. TLR-dependent activation of DCs after irradiation supports the use of low dose hypofractionated radiotherapy as an adjuvant to immunotherapy to enhance its effect, however either very low or high levels may be immunosuppressive [
32]. Direct immunomodulatory effects of irradiation on immune cells have been reported, such as altered IL-12 production from DCs [
33,
34]. It is important to understand the immunosuppressive nature of the TME for tumour-infiltrating DCs, which may limit the success of different treatments, e.g. immunotherapies, including DC vaccines [
38].
We have previously described the immunosuppressive effect of the colonic TME which inhibits LPS-induced DC maturation [
18,
20,
39‐
41]. Using a similar experimental outline as we described previously, in this study we investigated the effects on DC maturational capacity across different human GI cancers, using conditioned media from cell lines (in vitro conditioned media) and treatment-naïve tumour biopsies (ex vivo TCM) (Supplementary Fig.
1 A). While in vitro conditioned media represents the secretome from cancer epithelial cells, ex vivo conditioned media is a more complex model containing the soluble contributions from many different cells within the tumour microenvironment [
18,
20,
39‐
41]. Due to our interest in understanding the tumour microenvironment in both upper and lower GI tract cancers; oesophageal, rectal and colonic adenocarcinoma were investigated. This study describes for the first time that there were unexpected differences induced by the TCM on maturation of monocyte-derived DCs. Here, oesophageal cancer induced the highest level of DC maturation markers, rectal cancer induced moderate levels of DC maturation markers and colonic cancer significantly inhibited DC maturation markers. Interestingly for all GI cancer types examined here, in vitro and ex vivo TCM significantly inhibited TNF-α secretion from DCs. In addition, we modelled radiotherapy of oesophageal and rectal biopsies and found that TCM from 2Gy-irradiated tumours inhibited LPS-induced DC markers. Differential levels of specific inflammatory and angiogenic mediators were detectable in ex vivo TCM of GI cancers that correlated with DC maturation.
Discussion
The aim of this study was to examine if DC maturation was influenced by three distinct cancers of the GI tract - oesophageal, rectal and colonic adenocarcinoma. As radiotherapy induces unclear effects in terms of immunomodulation, we also investigated the effect of radiotherapy in this setting. The influence of the gastrointestinal TME on infiltrating DCs was modelled by conditioning immature monocyte-derived DCs with TCM, followed by subsequent LPS maturation, to investigate the effect of the TME on the maturational capacity of DCs as previously described [
18]. The effect of TCM on LPS-induced levels of DC markers CD54, CD80, HLA-DR, CD86, CD83 and PD-L1, and two secreted cytokines IL-12p70 and TNF-α in DC supernatants, as indicators for DC maturation were examined (Supplementary Fig.
1 A). We describe the levels of DC maturational capacity induced by conditioned media from oesophageal and colorectal cell lines. While TCM from OE33 oesophageal cell lines inhibited levels of HLA-DR only, TCM from CRC lines inhibited five DC markers - CD54, CD80, HLA-DR, CD86 and CD83. These markers are upregulated on the surface of DCs in order to ensure DCs can stimulate an effective T cell response [
18,
19]. CD83 is the most prominent surface marker for fully matured human DCs and enhances DCs’ T cell stimulatory capacity [
45]. CD80 and CD86 are also co-stimulatory and engage T cells, CD54 promotes DC-T cell binding and HLA-DR, otherwise known as MHC class II, allows for antigen presentation to CD4+ T cells [
45‐
47]. Altered HLA class II cell surface expression, a mechanism by which tumour cells escape from T cell responses, has been reported in many types of cancer [
48]. Normally antigen presentation cells, including DCs, constitutively express HLA class II molecules on the cell membrane, while only minor percentage of tumours and tumour cells express HLA-DR. In this study, we found opposing results for HLA-DR for oesophageal cancer, where the in vitro model inhibited HLA-DR on DCs and the ex vivo model enhanced HLA-DR on DCs. The reason for the modulation of HLA-DR in oesophageal cancer is unclear. This finding could potentially indicate that the tumour epithelial cells may contribute to the inhibition of HLA-DR, rather than the more complex oesophageal tumour microenvironment which contains both epithelial and non-epithelial cells. Whereas we found that both in vitro and ex vivo models of colonic cancer inhibited HLA-DR. These are contradictory findings and the reasons are unclear, in particular because we have previously shown that HLA-DR expression in tumour epithelium is an independent prognostic indicator in oesophageal adenocarcinoma patients, and speculate that for patients with enhanced survival, tumour epithelial cells may be compensating for the loss of HLA on antigen presentation cells [
49,
50].
The SW480 and SW620 cell lines are routinely categorised as colorectal adenocarcinoma [
51]. There were minor differences in the effects of SW480 and SW620 cell lines, which are models of primary versus metastatic lesions respectively. While both inhibited three DC markers, SW480 inhibited CD80 and SW620 inhibited CD54. This may suggest that, in addition to the type and localization of the primary GI tumour, but also the nature of the malignant lesion could potentially have a distinct effect on DC function in the TME. Using ex vivo TCM, generated using tumour biopsy explants from oesophageal, rectal and colonic adenocarcinoma, we assessed their differential effects on LPS-induced DC maturation. For DC surface markers in the LPS-stimulated setting, oesophageal cancer enhanced five DC markers (CD54, CD80, HLA-DR, CD86 and CD83), rectal cancer enhanced the levels of three DC markers (CD80, CD86 and CD83), whereas colonic cancer inhibited the levels of five DC markers (CD54, HLA-DR, CD86, CD83 and PD-L1) compared to LPS-induced levels in respective background media. Saying that, there was inter-individual variability apparent in the DC maturational capacity induced by ex vivo TCMs from the same GI tract tumour type. In this study, we confirmed the previously reported inhibition of DC markers by colonic adenocarcinoma, and we identified that LPS-induced PD-L1, which has a key role in the resolution of inflammation as the ligand for PD-1, is also inhibited [
18‐
20,
52]. While the reason is unclear, we speculate that this highlights the poor potential of DCs to respond to maturational stimuli in any capacity in the colonic cancer setting. Immunophenotyping tumour-infiltrating DCs across GI cancers would be important to confirm if this effect occurs in vivo, although quantification of maturation markers, including CD83, can be difficult [
53]. In our future studies, additional DC membrane markers will be included such as TLRs and other innate receptors in order to better understand the DC priming phenotypes induced by TCM. While functionally mature DCs are required in order to enable antigen presentation and T cell clonal expansion, co-culturing TME-conditioned DCs with T cells would further elucidate the overall functional significance in future studies [
20]. Different types of immunosuppressive dendritic cells have been found in cancer patients and animals [
54]. Lutz et al., 2002 proposed that tolerance occurs with either partial- or semi-maturation of DCs, whereas only full DC maturation is immunogenic and the decisive signal is the release of proinflammatory cytokines from DCs [
55].
Interestingly, in vitro conditioned media from both OAC and CRC lines inhibited LPS-induced levels of TNF-α in DC secretions, but had no significant effect on IL-12p70. Similarly, ex vivo TCM from all three GI cancer types significantly inhibited LPS-induced levels of TNF-α in DC secretions. There were differential effects on LPS-induced levels of IL-12p70 in DC supernatants based on adenocarcinoma type with oesophageal having no effect, rectal significantly enhancing, and colonic significantly reducing levels compared to LPS in corresponding background media alone. Thus, we confirmed the known IL-12p70 inhibition and identified that DC TNF-α is also significantly inhibited by the colonic TME [
18,
20]. As TNF-α from DCs is considered immunostimulatory, this then supports an extensive immunosuppressive phenotype induced by the colonic TME, in particular as shown by DC IL-12p70 inhibition, a requirement for optimal anti-tumour immunity [
24,
26]. There is a self-regulatory feedback on DCs of IL-12 and TNF-α, which may occur in order to limit the effects of DCs [
26]. The notable limitations of this study are the use of a single PBMC donor to derive DCs for each experiment, which was performed to confine inter-individual variability to cancer donors, and the limited number of treatment-naïve human tumour biopsies (
n = 8–14). While these need to be addressed to confirm reproducibility of our observations, the finding of TNF-α inhibition occurred across all cancer types and using both in vitro and ex vivo cancer samples. As the finding was common to both in vitro and ex vivo models, this indicates that soluble factors from the tumour epithelial cells may underlie this inhibition, rather than non-epithelial cells of the tumour microenvironment. Our finding of common TNF-α inhibition by GI cancers may have implications for TNF-α blockade, as has been proposed to overcome resistance to anti-PD-1 treatment [
56]. Bertrand et al., 2017 proposed that TNF deficiency may favour DC accumulation in tumours, while reducing the expression of PD-1 ligands [
56]. In line with this, our findings of reduced TNF-α could potentially indicate accumulation in the GI tract tumour setting of DCs with altered maturational capacities. We have previously shown that the TME of both early- and late-stage colonic cancer is equally suppressive for maturational capacity in DCs [
41]. Interestingly, Scarlett et al., 2012 demonstrated that tumour-resident DCs are transformed from immunostimulatory to immunosuppressive during tumour progression in a mouse model of ovarian cancer [
57]. Thus if DCs have a dynamic, even immunosuppressive, function in cancer as suggested by our study, then one could speculate from our findings that activation of suboptimally matured DCs in oesophageal adenocarcinoma could potentially result in a poorer outcome than in a setting where DC maturation is more completely suppressed, such as in colonic adenocarcinoma. When considering the implications for the findings here, it should be noted that this study is based on a model reflecting only a part of a complex system, specifically it is a model of the humoral components of the TME. These findings support conducting a larger study to determine if a negative correlation exists with outcomes, such as radioresponse and 5-year survival, across GI cancers. A larger study is also required to address some important experimental limitations of this study – in particular that DCs from multiple healthy donors should be employed, additional treatment-naïve tumour samples and non-cancerous matching GI tissues and /or cells as controls.
In this study, 2Gy-irradiation of cell lines and tumour explants was performed to correspond with the physiological effects of radiotherapy treatment at the tissue level as is performed clinically for oesophageal and rectal adenocarcinoma [
42]. We found an inhibitory effect of irradiation on DC maturation through conditioned media from both in vitro and ex vivo models. Whether this finding, that TCM from 2Gy-irradiated TME inhibited DC maturation, has relevance to tumour response to radiotherapy warrants further investigation as mentioned above. However, no significant effect was observed with the clinical outcome of tumour regression grade on this small oesophageal and rectal adenocarcinoma patient cohort in this study (
n = 14 and 10 respectively, data not shown). We propose that irradiation of the TME alters release of unknown soluble factors that are discernible to DCs, which fits with a mechanism of the radiation-induced bystander effect via altered levels of inflammatory cytokines produced by the TME [
58‐
62]. However, the irradiated TCMs had no effect on DC secretion levels of TNF-α or IL-12p70, therefore the functional significance of the effect on DC markers is difficult to decipher in terms of the ability of DCs to potentiate any immunomodulatory message to other bystander cells. Inflammation and angiogenesis are closely related and may underlie immune inhibition and radioresponse [
20]. Unfortunately, profiling the ex vivo TCM for inflammatory and angiogenic factors did not identify differences in levels of any of the mediators between 0Gy- and 2Gy-irradiated patient-matched biopsies. Radiotherapy has been described as both immunostimulatory and immunosuppressive with radiation dose proposed to be a key influencer in this, where low dose radiation, such as 2Gy, may be immunosuppressive [
32,
37,
63]. This data could suggest that radiotherapy of the GI TME may further reduce DC maturation, which we speculate could be beneficial in improving outcome in a setting where DCs may only have the capacity to sub-optimally mature, such as oesophageal adenocarcinoma.
As tumour biopsies contain tumour epithelial cells in addition to other cell types, ex vivo TCM contains many different tumour associated soluble factors, and therefore closely mimics the inflammatory milieu of the tumour in situ. Several cytokines and chemokines have been described to be present at high levels in the colonic TME compared to normal tissues, such as CXCL1 and CXCL5 (which function to attract and activate neutrophils) and CCL2 (a chemoattractant for monocytes, memory T cells and DCs) [
20]. We do not yet know the mechanistic pathways or secreted factors that induce the DC phenotypes we observed. Although NF-κB would be a candidate pathway given its reported roles in both LPS-induced DC maturation that ultimately results in the activation of NF-κB and the production of proinflammatory cytokines and in radiation-triggered TNF-α - NF-κB cross-signalling [
62,
64,
65]. In this study, levels of specific inflammatory and angiogenic mediators in ex vivo TCM of oesophageal, rectal and colonic adenocarcinoma were correlated with DC maturation marker - CD54. Thus it is possible that low levels of the cytokine IL-2 and high levels of angiogenic mediators - Ang-2 and bFGF, in TCM of tumour biopsies may confer a more DC inhibitory environment and this fits with some expected roles for these mediators. As CD54 (also known as intercellular adhesion molecule 1, ICAM-1) promotes DC-T cell binding, this indicates possible negative effects on the capacity of DCs to activate T cells in order to induce an adaptive immune response in GI tumours with lower levels of IL-2 and higher levels of Ang-2 and bFGF. IL-2 has key functions in the immune system, tolerance and immunity, primarily via direct effects on T cells, both effector and regulatory type. Interestingly, a role has been proposed for DC maturation as a mediator of systemic IL-2 effects [
66]. DCs can differentiate into endothelial-like cells when cultured in the presence of angiogenic growth factors – bFGF, VEGF and IGF-1, and these altered DCs have a reduced functional potency [
67]. We have previously shown that levels of tumour vasculature maturity or DC inhibition negatively correlate with survival of colonic adenocarcinoma patients on anti-angiogenic treatment [
40,
68]. We have previously identified that multiple mediators influence DC inhibition in colonic TCM - increasing levels of CXCL1, CXCL5, CCL2 and VEGF in the TCM correlated with inhibition of IL-12p70 secretion from DCs, however these isolated factors were not sufficient to induce all aspects of the extensive DC inhibition as observed for colonic TCM [
20]. The data from this study further supports the concept that it may be the cumulative effect of many mediators in the TME that may influence DCs. In particular, the levels of TGF-beta may be relevant due to its pivotal role in inducing immunological tolerance in DCs in colonic tissue [
69,
70]. While the effects of distinct ex vivo TCMs on DCs are described relative to their corresponding background media, profiling the levels of mediators may not be representative of the in vivo setting due to differences in culturing conditions. In future studies, healthy tissue controls and cells will be examined in conjunction with cancerous samples. GI cancers are well known to be molecularly heterogeneous with, for example, a substantial proportion of tumours (15% of CRC) displaying microsatellite instability phenotype and these tumours are commonly highly immunogenic, densely infiltrated with activated T cells and respond well to immune checkpoint blockade therapy [
71]. Therefore it would be interesting to examine if this subset of CRC tumours are not able to inhibit DC maturation to the same extent as observed in this study for which we do not know the MSI phenotype of the tumour biopsies.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.