Abstract
Background/Aim: Bone marrow-derived cells regulate the antitumor functions of tumor infiltrating lymphocytes (TILs) through arginase 1 (ARG1)-dependent metabolism. This study examines which ARG1-producing lineage is responsible for the inhibitory function of TILs. Materials and Methods: Multiplexed immunohistochemistry was performed for CD11b, CD163, CD68, and CD15, together with ARG1 expression and CD3+ TIL infiltration estimation in human colorectal cancer specimens. Results: Stratified survival analyses demonstrated that a large number of CD3+ TILs is a favorable prognostic factor in subgroups with a high level of ARG1+ infiltration and in the subgroup with a low level of ARG1– CD15+ infiltration. Calculation of the ARG1+/ARG1– ratio demonstrated that CD3+ TIL infiltration was prognostic in the subgroup with a low ARG1+/ARG1– ratio for CD15+ cells, contrary to other lineages. Conclusion: Tumor infiltrating CD15+ cells, the majority of which show polymorphonuclear features, are responsible for the ARG1-dependent T-cell dysfunction in human colorectal cancer.
Colorectal cancer is a common type of malignancy worldwide. Early-stage colorectal cancers are curatively resectable and have a favorable prognosis, whereas advanced cancers require multidisciplinary therapies to achieve favorable results (1-3). Various chemotherapeutic modalities in combination with molecular targeted therapies, such as those targeting epidermal growth factor receptor (4) or vascular endothelial growth factor (5) have been attempted, and more recently, immunotherapy (6) has also been attempted, but a substantial number of patients are refractory to such treatment modalities.
It is now widely recognized that characteristics of the tumor microenvironment (TME) of individual cancers affect their biological behavior and response to treatment (7). In colorectal cancer, it has been recognized from the 1980s that marked immune/inflammatory cell infiltration is associated with a favorable prognosis (8). Among the various types of immune/inflammatory cells in the TME, T-cell infiltration has been clarified to be associated with the prognosis of cancer patients (9-11). The recent success of immune checkpoint inhibitor therapy for various cancers, including colorectal cancer with mismatch repair deficiency (6, 12, 13), provides another line of evidence that the local immune response defines the biological behavior of cancers. Furthermore, several attempts, such as classification using “immunoscore” (14) have been made to predict patient prognosis and the therapeutic effects of treatments more accurately than conventional staging methods, by considering the level of infiltration of tumor-infiltrating lymphocytes (TILs).
A variety of suppressive or regulatory immune cells, such as regulatory T cells (15), inhibitory subtypes of bone marrow-derived cells, including M2-like macrophages (16), bone marrow-derived suppressor cells (MDSCs) (17), and granulocytes (18) simultaneously infiltrate cancer tissue together with effector lymphocytes. It is assumed that these suppressive cells are responsible for the lack of response to checkpoint inhibitor therapies in some cancer patients (19).
The suppression of cytotoxic lymphocyte activity by bone marrow-derived cells is mediated by L-arginine metabolism in the TME. Stimulation by Th2 cytokines, such as interleukin (IL)-4, IL-13, and transforming growth factor beta induces the production of arginase 1 (ARG1), and suppresses inducible nitric oxide synthase (iNOS), which is associated with cytotoxic function (20, 21). The consumption of L-arginine by ARG1 in the TME results in T-cell inactivation by the inhibition of CD3ζ chain expression on T cells (22, 23).
Many types of myeloid-derived cells express functional ARG1 (24), although the specific ARG1-producing cell lineages responsible for the inhibitory effects against cytotoxic T-cell reactions in the TME have not been clarified to date. In this study, we performed fluorescent multiplex immunohistochemistry on surgically resected human colorectal cancer specimens to investigate the effects of representative ARG1-producing cell lineages, including CD11b+, CD163+, CD68+, and CD15+ cells, on the association between CD3+ T-lymphocyte infiltration and patient prognosis.
Materials and Methods
Patients. Under approval of the institutional ethics committee, a series of surgically resected colorectal carcinoma specimens and clinical records were collected retrospectively from patients who consulted the Department of Gastrointestinal and Pediatric Surgery, Tokyo Medical University, between 2005 and 2014. Among the consecutive cases, patients with early stage cancers (Stage 0 and Stage I) and advanced Stage IV cancers were excluded, and 106 Stage II and Stage III patients were enrolled in this study. A summary of the characteristics of the patients are shown in Table I.
Clinical characteristics of the patients.
Multiplex immunohistochemistry. Formalin-fixed and paraffin-embedded specimens containing the largest cut surface of the tumor were used. Multiplex fluorescent immunohistochemistry was performed by the Tyramide Signal Amplification method using Opal fluorophore reagents (Opal 520, Opal 540, Opal 570, Opal 620, Opal 650, and Opal 690, Akoya Biosciences, Marlboro, MA, USA) according to the manufacturer’s instructions. The primary antibodies used were anti-human CD3 (rabbit monoclonal, clone SP7, Abcam, Cambridge, UK), anti-human CD11b (rabbit monoclonal, clone EP1345Y, Abcam), anti-human CD15 (rabbit monoclonal, clone SP192, Abcam), anti-human CD68 (mouse monoclonal, clone PG-M1, Abcam), anti-human CD163 (mouse monoclonal, clone 10D6, Thermo Fisher Scientific, Waltham, MA, USA), and anti-human arginase I (mouse monoclonal, clone C-2, Santa Cruz Biotechnology, Dallas, TX, USA). A horseradish peroxidase-labeled secondary detection system (EnVision plus, DAKO, Glostrup, Denmark) was used as a catalyst for fluorophore-conjugated tyramide. High pH Target Retrieval Solution (DAKO) was used for primary antigen heat retrieval, and Immunoactive Retrieval Buffer (pH6, Matsunami Glass, Osaka, Japan) was used for antibody stripping.
Quantitative imaging analyses. Fluorescent tissue images were captured from 3 randomly selected high-power fields (669×500 μm) containing cancer cell nests of each specimen, using an automated imaging system (Vectra ver. 3.0, Akoya Biosciences). An image analyzing software program (InForm ver. 2.4.8, Akoya Biosciences) was used to divide images into tissue frameworks (tumor nest and stromal region) and to segment tissue into each cell. Using an analytic software program (Spotfire, TIBCO, Palo Alto, CA), the cutoff value of each fluorescent signal was determined, and the number of immune cells with a specific phenotype was counted.
Statistical analyses. Considering the rightward skewed distribution of the immune cell count, the lower quartile value was used as the cutoff point for grading. Statistical analyses were performed using SPSS Statistics ver. 22 (IBM, Armonk, NY). A p-value of less than 0.05 was considered to indicate a statistically significant difference between groups.
Results
Fluorescent signal detection. CD3 and the various bone marrow-derived cell markers were all detectable by fluorescent multiplex immunohistochemistry. Representative images are shown in Figure 1. The majority of CD11b+ cells were mononuclear, but some multinucleated CD11b+ cells were also observed. CD68+ cells and CD163+ cells showed a mononuclear morphology. Most of the CD15+ cells demonstrated polymorphonuclear features. Some of the CD11b+ cells showing mononuclear features, co-expressed CD15.
Representative multiplexed labeling images. (A) CD3+ lymphocyte infiltration (green) was clearly detectable in cancer tissues. The inset is a bright-field converted image. (B) CD11b+, (C) CD68+, (D) CD163+, and (E) CD15+ cell infiltration (green) was also evaluated together with simultaneous ARG1 expression (red). Insets are bright-field converted images. Note the mononuclear morphology of CD68+ and CD163+ cells and the polynuclear characteristics of CD15+ cells. CD11b+ cells are predominantly mononuclear, with some polynuclear cells. (F) Multinuclear CD11b+ cells often coexpress CD15 (CD11b in green and CD15 in red). Bar=25 μm. The color tones of the fluorescent signals were schematically adjusted by a software program.
Prognostic significance of CD3+ intratumoral lymphocytes. A larger number of infiltrating CD3+ T cells was associated with a more favorable overall survival of patients (Figure 2A). To analyze the spatial features of the cancer tissue, CD3+ lymphocyte infiltration in the cancer cell nest (intratumoral) and in the stromal area was then separately evaluated. Marked intratumoral CD3+ lymphocyte infiltration was associated with a favorable overall prognosis (Figure 2B), whereas the amount of stromal CD3+ lymphocytes did not significantly correlate with the overall survival of patients (Figure 2C). Similar results were observed for disease-free survival (data not shown).
Prognostic significance of CD3+ TILs. (A) A higher number of CD3+ TILs was significantly associated with a more favorable overall prognosis. Analysis considering the spatial distribution of TILs demonstrated that (B) a higher amount of intratumoral CD3+ TILs correlated with a more favorable prognosis, although (C) stromal CD3+ TILs were not significantly associated with overall survival. TIL: Tumor infiltrating lymphocytes.
Immunomodulatory features of arginase 1-positive bone marrow-derived cells. We next aimed to evaluate the effects of stromal bone marrow-derived cell infiltration on the prognostic effects of intratumoral CD3+ T-cell infiltration. In the tumor groups with a high number of infiltrating CD11b+, CD163+, and CD68+ bone marrow-derived cells, intratumoral CD3+ T-cell infiltration was associated with a favorable prognosis; however, intratumoral CD3+ T-cell infiltration was not significantly associated with patient prognosis in the groups with a low number of these bone marrow-derived cells (Figure 3). In CD15+ cells, although the difference was not statistically significant, a similar trend was observed (Figure 3). For disease-free survival, similar results were also obtained and a significant difference was also confirmed for CD15+ cells (data not shown). Subsequently, ARG1 expression by these stromal bone marrow-derived cells, and its effects on the association between intraepithelial CD3+ T-cell infiltration and prognosis were analyzed. First, all stromal ARG1+ cells together, regardless of their cell lineage, were evaluated. The amount of intraepithelial CD3+ T-cell infiltration significantly correlated with prognosis in the tumor subgroup with a high level of stromal ARG1+ cell infiltration, whereas such an association was not observed in the tumor subgroup with a small number of stromal ARG1+ cells (data not shown). In each tumor subgroup with a high level of ARG1+CD11b+, ARG1+CD163+, ARG1+CD68+, and ARG1+CD15+ cell infiltration, intratumoral CD3+ cell infiltration was associated with favorable overall survival, whereas such an association between survival and CD3+ lymphocyte infiltration was not confirmed in the tumor subgroup with a low number of ARG1+ bone marrow-derived cells (Figure 4). Similar results were also obtained for disease-free survival (data not shown). Next, the ARG1– counterparts of these cell lineages were analyzed in a similar manner. Regarding the ARG1– counterparts of CD11b+ and CD163+ in the subgroup with a high level of ARG1– cell infiltration, prognosis correlated with the number of infiltrating CD3+ intratumoral T-cells (Figure 5). Regarding ARG– CD68+ cells, its amount did not affect the correlation between CD3+ TIL and overall survival (Figure 5). In each of the tumor subgroups with a low number of ARG1–CD11b+, ARG1–CD163+, and ARG1–CD68+ infiltrating cells, no significant correlation between CD3+ TILs and overall survival were observed (Figure 5). In contrast, in the tumor subgroup with a high number of stromal ARG1–CD15+ cells, no significant correlation between CD3 intratumoral lymphocyte infiltration and overall survival was detected, whereas CD3 intratumoral lymphocyte infiltration was found to associate with a favorable overall survival in the tumor subgroup with a lower number of stromal ARG1–CD15+ infiltrating cells (Figure 5). Similar results were also obtained for disease-free survival (data not shown). These results suggested that ARG1 expression in CD15+ cells may have a different effect on CD3+ TILs than ARG1 expression in CD11b+, CD68+, and CD163+ cells. Then, to confirm the effect of ARG1 expression on CD15+ cells, we calculated the ratio of ARG1+/ARG1– cells among CD15+ cells (Figure 6). As 10% of the samples had no ARG1–CD15+ cell infiltration in the stromal area, CD15+ cell infiltration in the entire tumor tissue was evaluated to calculate the ratio. In the tumor subgroup with a high ARG1+/ARG1– ratio among the CD15+ cells, the number of infiltrating CD3+ lymphocytes did not correlate with prognosis. In contrast, in the tumor subgroup comprising CD15+ cells that were predominantly ARG–, a higher number of CD3+ T-cell infiltration was associated with favorable overall survival. Similar analyses regarding ARG1 expression in CD163+, CD68+, and CD11b+ cells demonstrated contradictory results to CD15+ cells (Figure 6). In the tumor subgroups of CD163+, CD68+, and CD11b+ cells that were predominantly ARG1+, marked CD3+ T-cell infiltration was associated with favorable survival, whereas such a significant association with prognosis was not observed in the tumor subgroups of CD163+, CD68+, and CD11b+ cells that were predominantly ARG1–. Identical results were obtained regarding disease-free survival (data not shown).
Effects of bone marrow-derived cell lineages on the prognostic effects of intratumoral CD3+ TILs. In the tumor subgroups with a high number of infiltrating (A) CD11b+, (C) CD163+, and (E) CD68+ cells, a higher number of CD3+ TILs was significantly associated with a more favorable overall prognosis. On the other hand, in the tumor subgroups with a low number of infiltrating (B) CD11b+, (D) CD163+, and (F) CD68+ cells, the number of infiltrating CD3+ TILs was not a prognostic factor. A similar trend was observed in the tumor subgroups with a (G) high and (H) low number of infiltrating CD15+ cells, although the difference was not statistically significant. TIL: Tumor infiltrating lymphocytes.
Prognostic analysis of CD3+ intratumoral TILs stratified by ARG1+ bone marrow-derived cell lineages. In the tumor subgroups with a large number of infiltrating ARG1+ bone marrow-derived cells, for any lineage, including (A) CD11b+, (C) CD163+, (E) CD68+, and (G) CD15+, the infiltration of intratumoral CD3+ TILs was favorably prognostic. On the other hand, the infiltration of intratumoral CD3+ TILs was not significantly prognostic in the tumor subgroups with a small number of infiltrating ARG1+ cells among these cell lineages (B, D, F, H). TIL: Tumor infiltrating lymphocytes.
Prognostic analysis of intratumoral CD3+ TILs stratified by ARG1(–) bone marrow-derived cell lineages. Intratumoral infiltration of CD3+ TILs was a favorable prognostic factor for the overall survival of tumor subgroups with a high number of infiltrating (A) ARG1–CD11b+ or (C) ARG1–CD163+ cells. In the tumor subgroups with a low number of infiltrating (B) ARG1–CD11b+ or (D) ARG1–CD163+ cells, intratumoral infiltration of CD3+ TILs did not significantly correlate with overall survival. Inversely, for CD15+ cells, intratumoral infiltration of CD3+ TILs was prognostic in the tumor subgroup with (H) a small number of ARG1–CD15+ cells, whereas such prognostic significance was not observed in the tumor subgroup with (G) high ARG1–CD15+ cell infiltration. In the tumor subgroups stratified by the amount of ARG1– CD68+ cell infiltration, the infiltration of CD3+ TILs showed no significant correlation with overall survival time, regardless of the amount of ARG1–CD68+ cell infiltration (E, F). TIL: Tumor infiltrating lymphocytes.
Prognostic analysis of intratumoral CD3+ TILs stratified by ARG1+/ARG1– ratio in various bone marrow-derived cell lineages. Intratumoral infiltration of CD3+ TILs was a favorable prognostic factor for overall survival in tumor subgroups with a high ARG1+/ARG– ratio in (A) CD11b+, (C) CD163+, and (E) CD68+ cell lineages, whereas the infiltration of CD3+ TILs did not significantly correlated with overall survival in the tumor subgroups with a low ARG1+/ARG1– ratio in these cell lineages (B, D, F). In contrast, for the CD15+ lineage, intratumoral infiltration of CD3+ TILs was prognostic in the tumor subgroup with (H) a low ARG1+/ARG1– ratio. In the tumor subgroupwith (G) a high ARG1+/ARG1– ratio among the CD15+ cell lineages, intratumoral infiltration of CD3+ TILs were not a prognostic factor. TIL: Tumor infiltrating lymphocytes.
Discussion
This was an observational study using surgically resected human colorectal cancer specimens. Although this study was based on immunohistochemistry and statistical validation, multiplex immunohistochemistry enabled for accurate counting of target immune/inflammatory cells in strictly identical fields of view. In addition, the analyses were performed using an analytic software program, thus eliminating arbitrary interpretations by the observer. Therefore, we were able to obtain objective results, even though this was an observational immunohistochemical study.
In the present study, similar to preceding studies on colorectal cancer (9-11), favorable overall and disease-free survival times of patients were confirmed in the group with a high level of intratumoral T-cell infiltration. The association between TILs and patient prognosis differed depending on the spatial distribution of TILs. The number of TILs in the tumor nest correlated with patient prognosis, whereas the number of stromal CD3+ TILs was not significantly associated with prognosis. Similar features have been reported by immunohistochemical studies of CD8+ TILs in colorectal cancer, showing that the number of intratumoral TILs presents the most significant association with patient survival compared to stromal TILs and TILs infiltrating the invasive margin (11). This study confirms the necessity to consider the spatial distribution of TILs for their use as a prognostic or predictive biomarker. Furthermore, it is presumed that intratumoral CD3+ TILs are mostly CD8+ cytotoxic T lymphocytes, and it can be inferred that the number of CD3+ TILs correlates with patient prognosis. On the other hand, stromal CD3+ TILs may contain a mixture of a variety of helper subsets, including regulatory T cells which may explain the lack of a significant association between stromal CD3+ TIL and prognosis.
The effect of immune/inflammatory cells expressing representative myeloid-derived cell markers on the correlation between intratumoral CD3+ TIL infiltration and patient prognosis was evaluated by stratified survival analyses. Among the cell surface markers used in this study, CD11b is a representative cell-surface marker of MDSCs (25, 26). CD15 is a phenotypic marker of so-called “polymorphonuclear (PMN)” leukocytes (25, 26), which are expressed on neutrophils and PMN-MDSCs (27). CD68 is a pan-macrophage marker, and CD163 is a cell-surface marker of M2-like suppressive macrophages (28). Although the expression of each marker is not sufficient to strictly identify the above indicated cell lineages, these markers were used for general categorization.
Bone marrow-derived cells expressing these markers are assumed to include immunosuppressive cells to some extent, although a higher level of CD3+ lymphocyte infiltration was associated with a favorable prognosis in the tumor subgroup with a higher number of infiltrating CD11b+, CD163+, CD68+, and CD15+ cells. Conversely, in the tumor subgroup with a lower number of infiltrating bone marrow-derived cells, CD3+ TIL infiltration was not found to be a significant prognostic factor. These bone marrow-derived cells were then analyzed for concurrent ARG1 expression using the multiple labeling method. A high level of CD3+ TIL infiltration was favorably prognostic in the tumor subgroup with a large number of infiltrating ARG1+ bone marrow-derived cells of any cell lineage, whereas such prognostic significance of CD3+ TILs was not observed in the tumor subgroup with a low number of infiltrating ARG1+ cells.
The same analyses were then performed for the ARG1– counterparts. The number of CD3+ TILs was identified as a significantly prognostic factor in the tumor subgroup with a large number of cells of the ARG1–CD11b+ and ARG1–CD163+ lineages, whereas the opposite was observed for the ARG1–CD15+ lineage. CD3+ TIL infiltration was prognostic for the tumor subgroup with a low number of infiltrating ARG1–CD15+ cells. Although statistically significant results were not observed for the CD68+ lineage, our results suggests that ARG1 expression on CD15+ cells, i.e., neutrophils and PMN-MDSCs, has a different effect on CD3+ TILs than ARG1 expression on other bone marrow-derived cells.
The large number of infiltrating ARG1+ cells of each bone marrow-derived cell lineage is not synonymous with the low number of infiltrating ARG1– cells. Therefore, we performed similar analyses using the ARG1+/ARG1– ratio of each lineage as a stratification factor. Our results demonstrated that for the CD15+ lineage, CD3+ TIL infiltration was a significantly favorable factor in the tumor subgroup with a low ARG1+/ARG1– ratio, whereas the number of CD3+ TILs were not a significantly prognostic factor in the tumor subgroup with a high ARG1+/ARG1– ratio. Regarding CD11b+, CD163+, and CD63+ lineages, the results were opposite to those of the CD15+ lineage, namely, CD3+ TIL infiltration was prognostic in the tumor subgroup with a high ARG1+/ARG1– ratio, but not prognostic in the tumor subgroup with a low ARG 1+/ARG1– ratio. These results suggest that local ARG1-dependent immunosuppression in colorectal cancer is mainly regulated by CD15+, multinucleated bone marrow-derived cells.
Analyses under simple stratification by each bone marrow-derived cell lineage without using the ARG1+/ARG1– ratio demonstrated that bone marrow-derived cells are enhancing the anti-tumor effect of CD3+ TILs. Our results indicated the favorable prognostic significance of CD3+ TILs in the subgroups of tumors with a large number of bone marrow-derived cells, including CD15+ cells, regardless of ARG1 expression. This appears to contradict the immunosuppressive features of bone marrow-derived cells, although the infiltration of bone marrow-derived cells can be interpreted as part of the immunoinflammatory status of the TME, which is accompanied by the infiltration of various types of immune cells, including inhibitory immune cells. Another possibility underlying this contradiction is the expression of iNOS in bone marrow-derived cells. iNOS is involved in immunostimulatory metabolism, and can be coexpressed with ARG1 (20). The ratio of iNOS-producing cells to ARG1-producing cells in tissues, or the balance of iNOS and ARG1 in individual bone marrow-derived cells may explain this contradiction. In this study, ARG1 expression was measured as a quantitative value, although we categorized ARG1 expression into 2 groups by setting the lowest value at which a positive signal can be detected, to conduct preliminary analyses. Quantitative evaluation to analyze the iNOS/ARG1 ratio in bone marrow-derived cells using continuous quantitative data should be performed in the future.
Regarding the CD11b+ lineage, some of the cells appear to have a polymorphonuclear morphology, and co-express CD15, which may correspond to polymorphonucleated MDSCs. Mononuclear CD11b+ cells that do not express CD15 and CD15+ multinucleated CD11b+ cells can be regarded as mononucleated and PMN-MDSCs, respectively. The ARG1+CD15+ cells detected in this study were assumed to include PMN-MDSCs and neutrophils. The biological function of “polymorphonuclear leukocytes”, including PMN-MDSCs and neutrophils in cancer tissue is remarkably diverse (18). Besides ARG1-dependent suppression, other immunosuppressive mechanisms of neutrophils, such as complement-dependent T-cell suppression (29), have also been focused on. Regarding the diversity of these polymorphonulcear leukocytes, there are conflicting reports on the prognostic significance of tumor-infiltrating neutrophils in colon cancer, with some reports correlating them with a favorable prognosis (30) and others with a poor survival rate (31). Besides immunosuppressive function, polymorphonuclear leukocytes may promote cancer metastasis (32), elucidating diverse functions of polymorphonuclear leukocytes will be exceedingly helpful for the development of cancer therapeutic strategies.
Although it remains difficult to strictly identify human neutrophils and PMN-MDSCs in tissue specimens, the detailed phenotypic characteristics of ARG1+CD15+ cells in cancer tissue remain an important issue to be clarified.
This study demonstrated that CD15+ polymorphonuclear cells in tumor foci are likely to be the mainstay of ARG1-dependent immunosuppression in human colon cancer. Our results highlight the immunosuppressive function of CD15+ polymorphonuclear bone marrow-derived cells, and are essential for the development of therapeutic strategies that target this suppressive mechanism.
Conclusion
Among various lineages of bone marrow derived ARG1 producing cells, CD15+ polymorphonuclear cells are major suppressor of cytotoxic immune response in human colorectal cancer.
Footnotes
↵* These Authors contributed equally to this study.
Authors’ Contributions
Study concept and design: KM, ES and KK. Data acquisition and analyses: KM, ES, GK, and EM. Sample preparation and organization of clinical information: RU, TT, KK, JM, ME, TI. Drafting of the manuscript: KM, ES. Critical revision of the manuscript: KK, YN, and AT.
Conflicts of Interest
The Authors have no potential conflicts of interest to declare.
- Received September 15, 2021.
- Revision received November 7, 2021.
- Accepted November 8, 2021.
- Copyright © 2022 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.











