Targeting CD73 increases therapeutic response to immunogenic chemotherapy by promoting dendritic cell maturation

Abstract

The CD39-CD73–adenosinergic pathway converts adenosine triphosphate (ATP) to adenosine for inhibiting anti-tumor immune responses. Therefore, targeting CD73 to reinvigorate anti-tumor immunity is considered the novel cancer immunotherapy to eradicate tumor cells. To fully understand the critical role of CD39/CD73 in colon adenocarcinoma (COAD), this study aims to comprehensive investigate the prognostic significance of CD39 and CD73 in stage I–IV COAD. Our data demonstrated that CD73 staining strongly marked malignant epithelial cells and CD39 was highly expressed in stromal cells. Attractively, tumor CD73 expression was significantly associated with tumor stage and the risk of distant metastasis, which suggested CD73 was as an independent factor for colon adenocarcinoma patients in univariate COX analysis [HR = 1.465, 95%CI = 1.084–1.978, p = 0.013]; however, high stromal CD39 in COAD patients was more likely to have favorable survival outcome [HR = 1.458, p = 1.103–1.927, p = 0.008]. Notably, high CD73 expression in COAD patients showed poor response to adjuvant chemotherapy and high risk of distant metastasis. High CD73 expression was inversely associated with less infiltration of CD45⁺ and CD8⁺ immune cells. However, administration with anti-CD73 antibodies significantly increased the response to oxaliplatin (OXP). Blockade of CD73 signaling synergistically enhanced OXP-induced ATP release, which is a marker of immunogenic cell death (ICD), promotes dendritic cell maturation and immune cell infiltration. Moreover, the risk of colorectal cancer lung metastasis was also decreased. Taken together, the present study revealed tumor CD73 expression inhibited the recruitment of immune cells and correlated with a poor prognosis in COAD patients, especially patients received adjuvant chemotherapy. Targeting CD73 to markedly increased the therapeutic response to chemotherapy and inhibited lung metastasis. Therefore, tumor CD73 may be an independent prognostic factor as well as the potential of therapeutic target for immunotherapy to benefit colon adenocarcinoma patients.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00262-023-03416-4.

Introduction

Colorectal cancer (CRC) is one of the most relevant malignancies with cancer-related mortality in the world [1, 2]. Despite considerable improvements in therapeutic strategies for CRC in the recent decades, most CRC patients remain incurable and quickly occur tumor relapse after surgery, and associate with poor 5-years overall survival (OS) [3, 4]. Therefore, alternative therapeutic strategies benefit CRC patients need to be applied for improving the survival outcome of CRC patients such as immunotherapy.

The infiltration of CD8⁺ T lymphocytes has been associated with favorable survival outcome in various types of malignances including colorectal cancer [1, 5, 6]. But the functions of cytotoxic T lymphocytes can be severely attenuated by the immunosuppressive status within tumor microenvironment (TME) such as exhaustive markers PD-1/PD-L1 [7, 8]. Moreover, one of the major immunosuppressive pathways involved in tumor progression and tumor relapse is the CD39-CD73-adenosinergic pathway [9–11]. CD39 and CD73 is an ecto-5′-nucleotidase that catalyzes ATP/AMP into adenosine to exert various immunosuppressive effects by recruitment of myeloid-derived suppressor cells (MDSCs) and regulatory T cell (Tregs) to constitute an immune escape mechanism [12]. The CD39-CD73–adenosinergic pathway is mainly driven by hypoxia and cytokines within TME to attenuate the function of cytotoxic tumor-infiltrating immune cells [13, 14]. CD73 expression on tumor cells is remarkably upregulated in several malignances such as colorectal, breast and prostate cancer [10, 15–17], which is associated with tumorigenesis, tumor progression and metastasis by inhibiting antitumor functions of CD8⁺ T cells for immune escape [16, 18]. Therefore, blockade of CD73 has been reported to effectively prevent tumor growth and metastasis [19], suggesting that CD73 may serve as a potentially therapeutic target in multiple malignancies.

In the present study, we aim to investigate the clinical relevance of CD39-CD73-adenosinergic pathway in a large cohort of resected colon adenocarcinoma (COAD). We comprehensively examined the association between CD39/CD73 expression and clinicopathological characteristics, including tumor-infiltrating immune cells density, microsatellite instability, and immune checkpoint proteins. We found that high CD73 expression is significantly associated with a poor survival outcome by long-term outcome analysis, particularly in high-risk stage II–III COAD patients. The level of tumor CD73 was positively correlated with tumor progression, distant metastasis and response to adjuvant chemotherapy. Moreover, high CD73 in tumor microenvironment (TME) significantly associated with less infiltration of cytotoxic T cells, suggesting that CD73 may create immunosuppressive TME to inhibit anti-tumor immunity in COAD. Furthermore, inhibition of CD73 significantly promotes dendritic cell maturation and enhances T cell activation by chemotherapy. Blockade of CD73 remarkably increases the therapeutic efficacy of immunogenic chemotherapy, recruits high density of dendritic cell and T cell infiltration. Moreover, the risk of lung metastasis is also synergistically decreased when administrated with anti-CD73 antibodies and oxaliplatin. Taken together, these findings indicate that CD73 may inhibit anti-tumor immunity to decrease the therapeutic efficacy of chemotherapy, which can be acted as an independent prognostic factor and a therapeutic target for COAD patients, especially high-risk stage II–III COAD patients.

Materials and methods

Tumor cells and reagents

BALB/c-derived colon carcinoma cell lines CT26, human colorectal cancer cells HCT15, HCT116, SW620, SW480, LoVo, human monocytic cell line THP1 and human T lymphocyte cell line Jurkat were purchased from ATCC. Cells were screened for Mycoplasma using the Lookout Mycoplasma Detection Kit (Sigma) prior to preparation of working stocks of frozen cells, and thaws 3 to 5 days before use in all experiments. Tumor cells were grown in RPMI1640 (Gibco) cell culture medium supplemented with 10% FBS, 2 mol/L L-glutamine, 100 U/mL penicillin and 100 mg/mL streptomycin. Rat mAb to mouse CD73 (TY/23, BE0209), rat IgG2a isotype control (2A3, BE0089) were purchased from BioXCell. Small molecule CD73 inhibitor (CD73-IN-1, Cat. HY-103695) was purchased from MedChemExpress (CA, USA). Lentiviruses carrying shRNAs against CD73 were obtained from the National Core Facility for Manipulation of Gene Function by RNAi, miRNA, miRNA sponges, and CRISPR/Genomic Research Center (Academia Sinica, Taipei, Taiwan). ATP detection kit was purchased from Biovision (#K354-100, CA, USA) and AMP detection kit was purchased from Abcam (#ab273275, CA, USA).

Patients, cohorts and tissue microarrays construction

421 stage I–IV colon adenocarcinoma patients who underwent initial surgical resection in China Medical University Hospital (CMUH) between 2006 and 2014 were randomly selected. 221 patients were men (52.5%) and 200 (47.5%) were women, with a mean age of 63.7 ± 13.9 years. High-risk stage II–III COAD patients (n = 177) were received post-operative adjuvant chemotherapy after surgery.

For TMA construction, the standard protocol was followed according to our previous studies [1, 20]. Briefly, hematoxylin/eosin (HE)-stained tissue slides were evaluated by a pathologist and its correspondent formalin-fixed, paraffin-embedded tissues for TMA construction. Correspondent adjacent normal and tumor tissues were punched from donor block, and transferred into one recipient paraffin block. Each TMA spot contained at least 50% of the tumor area.

In this cohort, colon adenocarcinoma receive preoperative chemotherapy, radiotherapy and targeted therapy were excluded. All the clinicopathological data were collected from cancer registry database in CMUH, which was the approved by Institutional Review Board (IRB). Tumor stage was defined according to the 7^th AJCC TNM staging system. The primary end point of this study was distant metastasis-free survival (DMFS) and disease-free survival (DFS), which is defined as the time from the surgical day to event day including tumor relapse and death.

Western blot analysis

Total cell lysates (20–40 μg) were separated via 6–12% SDS–PAGE and transferred onto a PVDF membrane (Millipore, MA, USA) [21]. The membranes were blocked with BlockPRO™ Blocking buffer (BioLion Biotech, Taipei, Taiwan), probed with specific antibodies overnight at 4 °C, and then incubated with HRP-conjugated secondary antibodies for 1 h. After antibody incubation, the membranes were incubated with Immobilon Western Chemiluminescent HRP Substrate (Millipore) and analyzed by an ImageQuant™ LAS 4000 biomolecular imager (GE Healthcare, Amersham, UK). The digital data were processed using Adobe Photoshop and quantified using ImageJ software (NIH, MD, USA). Each blot was stripped with Immunoblotting Stripping Buffer (BioLion Tech.) before reprobing with the other antibodies. The antibodies were listed: NT5E/CD73 (#13,160, Cell Signaling Technology, CA, USA), CD39 (#68,336, Cell Signaling Technology, CA, USA) and GAPDH (#IR3-8, iReal Biotechnology, Taipei, Taiwan).

Assessment of intracellular ATP and AMP level

The intracellular ATP and AMP level was assessed using ATP and AMP colorimetric assay Kit (#K354 and K229, Biovision, CA, USA). The concentration was evaluated by ELISA reader (OD570 nm).

Immunohistochemical analysis

The TMA recipient blocks were cut into 3-μm sections for IHC staining with standard protocol [1, 20]. TMA slides were deparaffinized, and rehydrated through graded alcohols before being exposed to the Antigen Unmasking Solutions (H3300, Vector Laboratories, Burlingame, CA). Endogenous peroxidase was blocked with 0.3% hydrogen peroxide for 15 min at room temperature. After incubation with the primary rabbit monoclonal antibody against CD73 (#13,160, 1:100, Cell Signaling Tech.) or CD39 (#14,481, 1:100, Cell Signaling Tech.) at room temperature for 2 h, the sections were stained according to the manufacture’s manual (VECTASTAIN Elite ABC Kit, Vector Laboratories) and incubated with the substrate DAB chromogen (Vector Laboratories), and then were counterstained with hematoxylin [21, 22]. Tumor CD73 and stromal CD39 expression was evaluated based on immunopositivity of cell on the membrane of tumor cells for Histo-score (H-score) according to the intensity by semiquantitative scale (0 for absent; 1 for weak; 2 for moderate; and 3 for strong membrane staining) and the percentage of membranous tumor CD73 or stromal CD39 cells. The range of H-score was from 0 to 300. For CD8/CD45/CD45RO tumor-infiltrating immune cells, the count of intra-epithelial immunopositive immune cells was evaluated under 40X microscopy. The counts of tumor-infiltrating lymphocytes were presented as No. of immune cells/mm².

The cancer genome atlas (TCGA) database

Stage I–IV CRC patients were included and their CD39 and CD73 mRNA expression data were retrieved from Human Protein Atlas (HPA, www.proteinatlas.org/pathology) [23, 24], which resourced from the RNA sequencing (RNA-seq) data together with clinical information on TCGA. The CD39 and CD73 mRNA level gene had the best log-rank p value based on the Kaplan–Meier analysis with average RNA expression levels as the cut-off (Best expression cut-off = 9.0 for CD73 and cut-off = 3.37 for CD39) by the algorithm on HPA website [24]. RNA-seq data for 368 COAD samples derived from primary tumors were obtained from the TCGA-COAD Dataset [25]. Patients were split into low and high CD73 according to the mean CD73 expression of total patients. Gene signatures for CD8A⁺ tumor-infiltrating immune cells were calculated based.

Mouse tumor challenge and treatment

Five-week-old female BALB/c mice were administered according to the guidelines approved by the Institutional Animal Care and Use Committee of China Medical University (Protocol No. CMUIACUC-2021–359). BALB/c mice were subcutaneous injected with 2 × 10⁵ CT26 cells for 5 days in the right flank (primary tumor) and intravenous injected with 1 × 10⁵ CT26 cells five days following primary tumor inoculation. Mice (n = 5/group) were randomized to treatment groups when primary tumors reached an average tumor volume of 70–100 mm³. Oxaliplatin (OXP, 5 mg/kg, i.p.) was administrated on Day 7, 10, 13, 16 and 19. Anti-CD73 mAb (clone TY/23, 100 μg/mouse) and IgG control mAb were intraperitoneally administrated on Day 8, 11, 14 and 17. Tumors were measured by caliper every 2–3 days and volumes calculated using the formula: length × width² /2. This experiment is repeated twice. The mice were sacrificed when the longest diameter reached 20 mm, and the survival of the tumor-bearing mice was observed and recorded every 3 days.

Flow cytometry analysis

After treatment, tumors were isolated and weighed from the mice and then placed in petri dishes containing blank RPMI media at room temperature to keep them in media to prevent dehydration. Tumors were minced into small pieces (1–2 mm) by a beaver blade, filtered through a 70 μm strainer, spun down, and then resuspended in blank RPMI media. Thereafter, the cell suspensions were layered over Ficoll-Paque media and centrifuged at 1,025 xg for 20 min. The layer of mononuclear cells was transferred into a conical tube, and 20 ml of complete RPMI media was added and then gently mixed and centrifuged at 650 xg for 10 min twice. Finally, the supernatant was removed, and the TILs were resuspended in complete RPMI media [26, 27].

Then, TILs were resuspended in 500 μL of staining buffer (2% BSA, 0.1% NaN₃ in PBS). The cells were stained with the following surface marker panels: (1) DC maturation: CD45-APC^Fire750 (BioLegend, CA, USA), CD3/CD19-PerCP/Cy5.5 (BioLegend, CA, USA), MHC-II-PE/Cy7(BioLegend, CA, USA), CD11c-AF488 (BioLegend, CA, USA), CD86-PE (BioLegend, CA, USA), CD80-APC (BioLegend, CA, USA) and their isotypes; (2) T cell phenotype, CD3-APC^Fire750 (BioLegend, CA, USA), CD4-FITC (BioLegend, CA, USA), CD8a-PE/Cy7 (BioLegend, CA, USA), CD44-PE (BioLegend, CA, USA), CD62L-APC (BioLegend, CA, USA) and their isotypes; (3) Cytotoxic T cell panel: CD3-FITC (BioLegend, CA, USA), CD45-APC^Fire750 (BioLegend, CA, USA), GamB-AF647 (BioLegend, CA, USA), IFNγ-PE (BioLegend, CA, USA) and their isotypes. Samples were analyzed on a Guava easyCyte Flow cytometer (Luminex, CA, USA).

Immunohistochemical analysis on the resected mouse tumors

The antibodies used in this study were as follows: anti-cleaved caspase-3 (#9661, Cell Signaling Technology), FITC-conjugated anti-mouse CD11c (ab210308, Abcam) and anti-mouse granzyme B (ab255598, Abcam). Tissue slides were deparaffinized, incubated with 3% H₂O₂ in water for 10 min to quench endogenous peroxidase activity, and subjected to heat-mediated antigen retrieval with Antigen Unmasking Solutions (H3300, Vector Laboratories, Burlingame, CA). Tissue Sects. (3-µm thickness) were stained with the HRP-conjugated avidin biotin complex (ABC) from the Vectastain Elite ABC Kit (Vector Laboratories, Burlingame, CA) and DAB chromogen (Vector Laboratories) and counterstained with hematoxylin. For CD11c and granzyme B, the tissue sections were stained with FITC-conjugated secondary antibodies, and counterstained with DAPI.

Staining for immune cells was positive when detected in the tumor-infiltrating lymphocytes (TILs) and was evaluated using a microscope (OLYMPUS BX53, Tokyo, Japan). Regarding the detection of TILs, the tissue was viewed at 40 × magnification, and the area with the highest density of CD11c⁺ and GzmB⁺ TILs within the malignant cells was counted at 400 × magnification (no. of TILs/high-power field). The average number of tumor-infiltrating immune cells in five high-power fields was included in the evaluation [28].

Statistical analysis

The statistical analyses were carried out by SPSS software version 22 (SPSS, Chicago, IL, USA) with p-value < 0.05 as statistically significant cut-off. The patient survival was estimated by the Kaplan–Meier log-rank test. The correlation analysis between CD73/CD39 expression (H-score) and clinicopathological parameters were analyzed by chi-squared test or Fisher’s exact test (two-tailed). Unpaired t test was used to compare the difference between two groups, and one-way ANOVA was used to compare multiple groups. Univariate and multivariate Cox proportional hazard COX regression model was constructed with influence factors including sex, age, pT stage, pN stage, pTNM, tumor location, tumor differentiation, lymphovascular invasion, perineural invasion, tumor PD-L1, tumor PD-L2, tumor CD73 and stromal CD39 with an entry criterion of p < 0.05.

Results

High tumor CD73 expression was associated with risk of distant metastasis, poor survival outcome and less infiltration of immune cells in colon adenocarcinoma.

To understand the role of CD39-CD73–adenosinergic pathway in colon adenocarcinoma (COAD), we interrogated the expression pattern of CD39 and CD73 within tumor microenvironment (TME). As presented in Table S1 and S2, 421 COAD patients who received surgery between 2006 and 2014 were enrolled in this study, and followed up until 2021. The majority of the patients were men (52.5%). We used a Histo-scoring system based on the IHC staining intensity to evaluate CD39 and CD73 expression [1] (Fig. 1). The representative images of immunohistochemical CD39 and CD73 expression were shown according to their staining intensity (Fig. 1A). We found that CD39 was expressed on tumor cell (150/421 = 35.6%, Fig. S1A, Table S1), stromal cells (205/421 = 48.7%, Fig. S1B, Table S1 and S2), and inflammatory cells (119/421 = 28.3%, Fig. S1C, Table S2). But CD73 was specific expressed on tumor cells (285/421 = 67.7%, Table S1). The patients’ clinicopathological characteristics and CD39/CD73 expression are summarized in Table S1 and Table S2. Patients with high tumor CD73 had the tendency to associate with the presence of perineural invasion (PNI, p = 0.0673, Fig. 1B). Moreover, patients with high tumor CD73 had clinically associated with the risk of distant metastasis (DM, p = 0.0353, Fig. 1B). The expression level of tumor CD73 was associated with the tumor size (pT stage, p = 0.0008, Fig. 1D). The expression of stromal CD39 was not associated with the risk of LVI, PNI, DM and tumor size (Fig. 1C, D).

Fig. 1.

The expression of tumor CD73 was significantly associated with the risk of distant metastasis after surgical operation and poor survival outcome. A CD73 proteins were mainly expressed on the membrane of cancer cells. CD39 proteins were majorly expressed in the stromal cells and partially expressed in the inflammatory cells. B The H-score of tumor CD73 was statistically associated with the risk of distant metastasis (n = 421 and unpaired T-test p = 0.0353). C Stromal CD39 expression was not associated with the present of LVI, PNI and distant metastasis (n = 421) by unpaired T-test. D Tumor CD73 expression was significantly associated with pathological T stage (n = 421, p = 0.0008). Stromal CD39 expression was not associated with pathological T stage (n = 421 and p = 0.70). One-way ANOVA test. E Patients with high tumor CD73 expression were clinically associated with worsen disease-free survival (DFS) in CRC patients (n = 421, p = 0.0076). High stromal CD39 expression was associated with favorable DFS in CRC patients (n = 421 p = 0.0034). F Tumor CD73 expression was significantly associated with distant metastasis-free survival (DMFS) and DFS in stage II-III COAD patients who received adjuvant chemotherapy (n = 177, log-rank p = 0.0077). G Tumor CD73 expression was conversely associated with the density of CD8⁺ TILs and CD45.⁺ TILs within tumor microenvironment (TME). Unpaired t test (n = 421)

We then further examined the prognostic role of CD39 and CD73 expression in colon adenocarcinoma, Kaplan–Meier (KM) analysis was used to examine their prognostic value (Table 1). KM survival curve showed that patients with high tumor CD73 expression were clinically associated with poor DFS in stage I–IV COAD patients (48.7% vs 61.0%, p = 0.019, Fig. 1E). However, patients high stromal CD39 expression were clinically associated with favorable DFS in stage I–IV COAD patients (58.0% vs 46.3%, p = 0.007, Fig. 1E). Moreover, high tumor CD73 expression was significantly associated with worsen distant metastasis-free survival (DMFS) and DFS in high-risk stage II–III colon adenocarcinoma patients who received adjuvant chemotherapy (Fig. 1F). These results suggested that tumor CD73 expression was associated with poor survival outcomes in high-risk stage II–III COAD patients.

Table 1.

Correlation between clinicopathologic parameters, DFS and OS

Parameters	Noa	DFS (%)	p value*	OS (%)	p value*
	421	52.0%		57.0%
Sex			0.928		0.605
Female	200	52.0%		58.0%
Male	221	52.0%		56.1%
Age			0.077		0.021
< 65	210	56.7%		62.9%
≥ 65	211	47.4%		51.2%
pT stage			< 0.001		< 0.001
pT1-2	70	75.7%		78.6%
pT3-4	351	47.3%		52.7%
pN stage			< 0.001		< 0.001
Negative	219	65.3%		68.9%
Positive	202	37.6%		44.1%
pTNM stage			< 0.001		< 0.001
Stage1-2	219	65.3%		68.9%
Stage3-4	202	37.6%		44.1%
Tumor location			0.724		0.491
Distal colon	237	53.2%		58.2%
Proximal colon	179	52.0%		56.4%
Tumor differentiation			0.941		0.807
Well to moderate	411	52.3%		57.2%
Poor	6	52.0%		50.0%
Lymphovascular invasion			< 0.001		< 0.001
Absent	179	66.5%		69.8%
Present	242	41.3%		47.5%
Perineural invasion			< 0.001		< 0.001
Absent	254	64.2%		68.1%
Present	166	33.7%		40.4%
Tumor PD-L1			0.002		0.001
High	143	62.2%		68.5%
Low	277	46.6%		40.9%
Tumor PD-L2			0.096		0.014
High	143	56.6%		64.3%
Low	274	49.3%		52.9%
Tumor CD73			0.012		0.038
Low	154	60.4%		64.3%
High	267	47.2%		52.8%
Stromal CD39			0.007		0.001
High	205	58.0%		64.7%
Low	216	46.3%		50.0%

^aNumber of cases may differ due to missing data

P-values marked with bold indicate statistically significant differences between these groups

CD73 plays an immunosuppressive role within TME, therefore, we evaluated the relationship between tumor CD73 and immune signatures such as CD8⁺ T cells (cytotoxic T lymphocyte marker), CD45⁺ T cells (general immune cell marker) and CD45RO⁺ T cells (effector/memory T lymphocyte marker) by immunohistochemical analysis on intraepithelial tumor-infiltrating lymphocytes (TILs). We found a significantly negative association between tumor CD73 expression with immune contexture of CD8⁺ TILs (p = 0.01) and CD45⁺TILs (p = 0.032, Fig. 1G). These results suggest that tumor CD73 expression can be a significant prognostic factor for COAD patients by inhibiting the infiltration of immune cells, especially CD8⁺ T cell-mediated immune response.

Prognostic impacts of tumor CD73 expression in high-risk COAD patients

For univariate analysis of long-term DFS by Cox regression model, several clinicopathological characters were associated with patient survival outcome including pT stage, pN stage, LVI and PNI. The immune-related parameters tumor PD-L1, tumor CD73 and stromal CD39 were also associated with survival outcome. Patients with a high tumor membranous CD73 also had an increased risk for a poor DFS (HR = 1.465, 95% CI = 1.084–1.978, p = 0.013) compared with patients carrying a low tumor CD73 (Table 2). Conversely, patients with low stroma CD39 were associated with poor DFS (HR = 1.458, 95% CI = 1.103–1.927, p = 0.008). These results showed that tumor CD73 have the clinically prognostic relevance for COAD patients. Moreover, tumor CD73 was found to be the independent prognostic factor for long-term DFS by multivariate COX analysis after adjusted with these significant parameters identified by univariate analysis (Table 2). Patients with high tumor CD73 levels (HR = 1.711, 95% CI = 1.260–2.325, p = 0.001) presented an increased risk for poor DFS. These results show that the level of tumor CD73 could independently predict the prognosis in COAD patients (Table 2).

Table 2.

Univariate and multivariate analysis of DFS and known prognostic factors in stage I-IV colon adenocarcinoma patients

Parameters	Univariate analysis					Multivariate analysis
	No. at riska	Events	HR	95% CI	p value	HR	95% CI	p value
Sex	421				0.928			0.429
Female	200	96	1.0			1
Male	221	106	0.987	0.749–1.030		1.121	0.844–1.489
Age					0.081			0.014
< 65	210	91	1			1
≥ 65	211	111	1.28	0.97–1.689		1.432	1.075–1.908
pT stage					0.001
pT1-2	70	17	1
pT3-4	351	185	2.52	1.532–4.143
pN stage					< 0.001
Negative	219	76	1
Positive	202	126	2.346	1.762–3.123
pTNM stage					< 0.001			< 0.001
1–2	219	76	1			1
3–4	202	126	2.505	1.873–3.349		2.083	1.502–2.887
Tumor location					0.726
Proximal colon	237	111	1
Distal colon	179	86	0.951	0.718–1.26
Tumor differentiation					0.942
Well to moderate	411	196	1
Poor	6	3	0.958	0.306–2.997
Lymphovascular invasion					< 0.001			0.134
Absent	179	60	1			1
Present	242	142	2.144	1.585–2.901		1.3	0.922–1.832
Perineural invasion					< 0.001			< 0.001
Absent	254	91	1			1
Present	166	110	2.326	1.76–3.074		1.853	1.370–2.508
Tumor PD-L1					0.003			0.002
High	143	54	1			1
Low	277	148	1.612	1.18–2.203		1.667	1.205–2.307
Tumor PD-L2					0.100			0.071
High	143	62	1.0			1.0
Low	274	139	1.286	0.953–1.735		1.338	0.976–1.834
Tumor CD73					0.013			0.001
Low	154	61	1.0			1
High	267	141	1.465	1.084–1.978		1.711	1.260–2.325
Stromal CD39					0.008			0.007
High	205	86	1.0			1
Low	216	116	1.458	1.103–1.927		1.481	1.115–1.966

P-values marked with bold indicate statistically significant differences between these groups

To evaluate the prognostic value of tumor CD73 in high-risk stage II–III COAD patients, we stratified stage II–III patients into two subgroups based on the administration of adjuvant chemotherapy (Table 3). We found high tumor CD73 had an increased risk for a poor DFS in stage II–III COAD patients who received adjuvant chemotherapy followed by surgery (HR = 2.440, 95% CI = 1.367–4.353, p = 0.003, Table 3). However, there was no prognostic significance of tumor CD73 in stage II–III COAD patients who received surgical operation only. These results showed that tumor CD73 could be an independent prognostic factor for COAD patients, especially for high-risk stage II–III COAD patients.

Table 3.

Multivariate analysis of DFS and known prognostic factors in high-risk stage II-III carcinoma patients

Parameters	Multivariate analysis
	HR	95% CI	p value
Stage II-III patients (surgery only)
Sex (Male vs Female)	0.787	0.398–1.555	0.49
Age (> 65 vs < 65)	5.875	2.140–16.131	0.001
pT stage (pT3-4 vs pT1-2)	0.115	0.012–1.139	0.064
pN stage (Positive vs Negative)	1.159	0.449–2.994	0.76
Tumor location (Distal vs Proximal)	0.820	0.408–1.646	0.576
LVI (Present vs Absent)	0.915	0.439–1.908	0.812
PNI (Present vs Absent)	1.970	0.868–4.471	0.105
Tumor CD73 (High vs Low)	0.533	0.276–1.027	0.06
High-risk stage II-III patients (receiving adjuvant chemotherapy)
Sex (Male vs Female)	0.865	0.527–1.420	0.567
Age (> 65 vs < 65)	1.210	0.747–1.959	0.439
pT stage (pT3-4 vs pT1-2)	2.589	0.614–10.918	0.195
pN stage (Positive vs Negative)	1.346	0.795–2.278	0.268
Tumor location (Distal vs Proximal)	1.610	0.960–2.701	0.071
LVI (Present vs Absent)	1.428	0.776–2.629	0.252
PNI (Present vs Absent)	1.782	1.076–2.951	0.025
Tumor CD73 (High vs Low)	2.440	1.367–4.353	0.003

P-values marked with bold indicate statistically significant differences between these groups

CD73 attenuated chemotherapy-induced cancer immunogenicity to inhibit anti-cancer immunity and the response to chemotherapy

We then analyzed the CD73/NT5E and CD39/ENTPD1 mRNA transcriptome results from The Cancer Genome Atlas (TCGA) database [24]. CD73 mRNA was highly expressed in tumor tissues compared to normal tissues. But CD39 mRNA was decreased in tumor tissues (Fig. 2A). To further validate the prognostic value of CD73, the results of CD73 mRNA on stage I–IV colorectal cancer was retrieved from Human Protein Atlas (HPA, www.proteinatlas.org/pathology) [23, 24], which resourced from TCGA database with the RNA sequencing (RNA-seq) data and basic clinical information. The cut-off of CD73 mRNA level for survival analysis was based on the algorithm on HPA website [24]. CRC patients with high CD73 mRNA have poor survival outcome (HR = 1.598, 95%CI = 1.042–2.452, p = 0.0426, Fig. 2B). CD39 mRNA level was not associated with survival outcome in stage I–IV CRC patients in TCGA cohort (Fig. 2B). Moreover, the CD73 mRNA level was negatively correlated with the expression of CD8⁺ T cell signatures including CD8A, GZMB, CXCR3 and CXCL9 (Fig. 2C). Taken together, these results showed that CD73 expression might inhibit anti-tumor immunity, leading to poor survival outcome in CRC patients.

Fig. 2.

Genomic correlates of NT5E and CD8 signatures in the TCGA dataset A CD73/NT5E was highly expressed on tumor tissues compared to normal tissues. B High CD73 mRNA expression was remarkably associated with poor survival outcome in CRC patients that retrieved from TCGA database (cut-off mRNA = 9.0, n = 368, p = 0.0453). Moreover, CD39 mRNA expression was not associated with poor survival outcome in CRC patients that retrieved from TCGA database (cut-off mRNA = 3.37, n = 515, p = 0.7079). C High CD73 mRNA expression was conversely associated with less expression of T-cell mediated signatures (TCGA database, n = 368, unpaired T test). ***p < 0.001

CD73 is an ectonucleotidase that catalyzes ATP/AMP into adenosine for immunosuppressive TME by recruiting MDSCs and Tregs. Moreover, the release of ATP by dying cells dictates their immunogenicity for dendritic cell maturation and recruitment when chemotherapy elicited immunogenic cell death (ICD)[29, 30]. High expression of CD73 may convert ATP into adenosine during chemotherapy-induced ICD, such as oxaliplatin (OXP). Therefore, we evaluate whether CD73 affects the therapeutic efficacy of chemotherapy by attenuating cancer immunogenicity to inhibit anti-tumor immunity. We first evaluated the level of CD73 in five colorectal cancer cell lines and found that high expression of CD73 existed in HCT116 cells (Fig. 3A). We then generated HCT116^shCD73 stable cell line (Fig. 3B). We treated HCT116^shNC and HCT116^shCD73 cells with OXP for 6 h. We found that knockdown of CD73 retained the level of OXP-induced ATP release an AMP accumulation (Fig. 3C). Similarly, inhibition with CD73 by small molecule CD73-IN-1 also maintained the level of extracellular ATP and AMP (Fig. 3D). Previous studies showed that blockade of CD73 promotes dendritic cell maturation and infiltration to enhance the anti-tumor response of radiotherapy by sustaining the level of extracellular ATP [31]. Therefore, we then performed in vitro experiments using OXP-treated HCT116^shNC and HCT116^shCD73 cancer cells cocultured with THP1-derived immature DCs (THP-iDCs, Fig. 3E). Direct co–culture experiments also showed that high expression of DC maturation marker CD86 on THP-iDCs when cocultured with OXP-treated HCT116^shCD73, compared to HCT116^shNC (Fig. 3F). Moreover, the T cell activation was also increased when cocultured with OXP-treated HCT116^shCD73 and THP-iDCs (Fig. 3G). These cytotoxic ability of T cells that co-cultured with OXP-treated HCT116^shCD73 and THP-iDCs were also superior than OXP-treated HCT116^shNC/THP-iDCs cells (Fig. 3H). These results showed that tumor CD73 decreased chemotherapy-induced cancer immunogenicity to reduce DC maturation and T cell activation.

Fig. 3.

CD73 attenuated OXP-induced anti-tumor immunity by converting ATP into adenosine. A The expression level of CD73 in five colorectal cancer cell lines. B HCT116 cells were infected with lentivirus carrying shRNA against CD73, then selected by puromycin (2 μg/mL) for 3 days. Cell lysates were analyzed by western blot. C HCT116^shNC and HCT116^shCD73 cells were treated with OXP for 6 h. The conditioned medium was harvested for ATP and AMP assay (n = 3). *p < 0.05. D HCT116 cells were treated with OXP (50 μM) and CD73-IN-1 (10 mM) for 6 h. The conditioned medium was harvested for ATP and AMP assay (n = 3). *p < 0.05 and ***p < 0.001. E The schematic diagram of co-culture experiment to evaluate the dendritic cells (DC) maturation and T cell activity. F OXP-treated HCT116^shNC and HCT116^shCD73 cells were co-cultured with THP1-immature dendritic cells (iDC) for 24 h. The level of DC maturation was analyzed by flow cytometry (n = 3). *p < 0.05 and **p < 0.01. G After coculture, Jurkat T cells were co-cultured with HCT116^shNC/THP1-iDC and HCT116 ^shCD73/THP1-iDC for 15 h. The T cell activation was analyzed by flow cytometry (n = 3). *p < 0.05 and **p < 0.01. H Jurkat T cells were individually co-cultured with HCT116^shNC and HCT116.^shCD73 cells for 24 h. The cytotoxicity was analyzed by CCK assay (n = 3). *p < 0.05 and **p < 0.01

We next investigated whether blockade of CD73 can enhance the response to immunogenic chemotherapy in vivo and reduce the risk of metastasis (Fig. 4A). We found blockade of CD73 by antibodies (clone TY/23) conferred sensitivity to OXP treatment in immunocompetent mice, suggesting that CD73 inhibition can potentiate OXP activity against colorectal cancer (Fig. 4B). The tumor volume and tumor weight were significantly decreased when combined with OXP and anti-CD73 antibodies (Fig. 4C, D). The apoptotic marker caspase-3 cleavage was markedly observed in OXP/anti-CD73 mAb group (Fig. 4E). Moreover, we found that the number of lung metastasis nodules was remarkably decreased in OXP/anti-CD73 antibodies group (Fig. 4F). The survival period was also prolonged by OXP/anti-CD73 antibodies treatment (Fig. 4G). These experiments were repeated and similar results were observed (Fig. S2A).

Fig. 4.

Blockade of CD73 enhanced the therapeutic efficacy of immunogenic chemotherapy. A The schematic diagram of treatment regimen. BALB/c mice were subcutaneous injected with 2 × 10⁵ CT26 cells for 5 days and intravenous injected with 1 × 10.⁵ CT26 cells on Day 5 (n = 5). Oxaliplatin (OXP, 5 mg/kg, i.p.) was administrated on Day 7, 10, 13, 16 and 19. Anti-CD73 mAb (clone TY/23, 100 μg/mouse) and IgG control mAb were administrated on Day 8, 11, 14 and 17. B The tumor volume was measured every 2–3 days (n = 5). **p < 0.01 and ***p < 0.001. C The tumor volume on day 30 was shown (n = 5). *p < 0.05 and ***p < 0.001. D The resected tumors were weighed (n = 4). *p < 0.05 and ***p < 0.001. E The representative images of primary tumor and apoptotic marker cleaved caspase-3 (n = 4). *p < 0.05 and ***p < 0.001. F The representative images of lung were shown (n = 4). The number of lung metastatic nodules was evaluated under microscope. *p < 0.05, *p < 0.01 and ***p < 0.001. G The mouse survival was assessed with Kaplan–Meier survival curves (n = 6)

Moreover, high density of tumor-infiltrating CD11c⁺DCs and GzmB⁺ T cells were observed in OXP/anti-CD73 mAb group (Fig. 5A–C). To evaluate the immune cell profile within tumors, we extracted tumor-infiltrating immune cells for flow cytometry. The gating strategies were shown in Fig. S2B–D. As shown in Fig. 5D, the density of CD11c⁺MHCII⁺ DCs was remarkably increased in OXP/anti-CD73 mAb group (Fig. 5D). The infiltrated mature DCs (CD86⁺CD80⁺ CD11c⁺MHCII⁺ DCs) also increased (Fig. 5E). In addition, the infiltrating number of CD4⁺ T cells, CD8⁺ cells and effector/memory CD8⁺ T cells (CD44⁺CD62L⁻CD8⁺ T_EM) was higher in OXP/anti-CD73 mAb group, compared to OXP group (Fig. 5F). The number of cytotoxic GzmB⁺ CD8⁺ T cells and IFNγ⁺ CD8⁺ T cells within resected tumors was also increased (Fig. 5G, H). Taken together, these results showed that tumor CD73 inhibited immunogenicity of dying cancer cells by converting ATP into adenosine, leading to less DC maturation and T cell-mediated anti-tumor immunity.

Fig. 5.

Blockade of CD73 enhanced DC infiltration for anti-tumor immunity when combined with OXP treatment. A The resected tumors were stained with CD11 (DC marker) and GzmB (cytotoxic T cell marker), and evaluated under immunofluorescent microscope. B The quantification of CD11c⁺ DCs was shown (n = 4). *p < 0.05 and ***p < 0.001. C The quantification of GzmB⁺ T cells was shown (n = 4). *p < 0.05 and ***p < 0.001. D The tumor-infiltrating immune cells were extracted from resected tumors for flow cytometry. The absolute number of CD11c⁺MHC-II⁺ DCs was shown (n = 3–4). *p < 0.05 and ***p < 0.001 (n = 3). E The absolute number of tumor-infiltrating CD86⁺CD80⁺CD11c⁺MHC-II⁺ DCs was shown (n = 3–4). *p < 0.05 and ***p < 0.001. F The absolute number of tumor-infiltrating CD4⁺T, CD8⁺T and CD44⁺CD62L⁻CD8⁺T_EM cells was shown (n = 3–4). *p < 0.05 and ***p < 0.001. G The absolute number of tumor-infiltrating cytotoxic GzmB⁺CD8⁺T cells was shown (n = 3–4). *p < 0.05 and ***p < 0.001. H The absolute number of tumor-infiltrating cytotoxic IFNγ⁺CD8.⁺T cells was shown (n = 3–4). *p < 0.05 and ***p < 0.001

Discussion

In this study, we found 67.7% COAD patients harbored high CD73 expression within TME, suggesting that CD73 is a tumor-specific expression in COAD patients. High tumor CD73 in the TME is associated with shorter survival outcomes, tumor relapse and resistance to adjuvant chemotherapy. Moreover, the level of tumor CD73 is conversely correlated with the infiltration of immune cells within TME, indicating that CD73 may inhibit infiltration of immune cells for immune escape. By targeting CD73, we found that dendritic cell maturation and cytotoxic T cell activity was significantly increased after treatment with immunogenic chemotherapy OXP in vitro and in vivo. Moreover, the lung metastasis was also decreased when administrated with anti-CD73 antibodies and OXP. Taken together, these findings suggest that CD73 can be considered an independent prognostic factor for high-risk COAD patients, especially for patients who receive adjuvant chemotherapy. Moreover, CD73 blockade may benefit patients from chemotherapy by increasing DC maturation for anti-tumor immunity.

Recent studies have demonstrated that many solid tumors impair proinflammatory signaling as an immune escape mechanism by converting extracellular ATP to adenosine via CD39/CD73 signaling pathway. Accumulating immunosuppressive adenosine constitutes a negative feedback mechanism to prevent immunosurveillance for anti-tumor immune responses [32, 33]. This immune escape leads to tumor growth, metastasis and resistant to chemotherapy. In line with these studies, we found that tumor CD73 expression was correlated with perineural invasion (PNI), distant metastasis and tumor size. Moreover, CD73 in tumor tissues is highly expressed, compared to normal peritumoral tissues, which correlated with the tumor progression. Supporting to our studies, high tumor CD73 expression has been reported in multiple malignancies and associated with worse clinical outcomes [34]. Moreover, high CD73 expression was also reported to associate with lymph node metastasis and distant metastasis in several malignancies including gastric carcinoma, gallbladder cancer, renal cancer and head and neck squamous cell carcinoma (HNSCC) [33–35]. These finding highlighted that high CD73 expression increased risk of lymph node metastases.

Furthermore, Messaoudi et al. found that high expression of tumor CD73 in metastatic site was resistant to peri-operative chemotherapy in patients with colorectal liver metastasis (CRLM) [33] and resistant to chemotherapy in breast cancer [36]. The poor response to chemotherapy, tumor progression and large tumor sizes are considered critical characters with chronic hypoxia [33]. In these hypoxic area, stressed cancer cells release ATP that can be directly converted into adenosine by CD73 for tumor growth and survival. Moreover, the rich of CD73 in TME also attenuated the therapeutic efficacy of immunogenic chemotherapeutic agents such as oxaliplatin and doxorubicin by converting the ATP, which is a key mediator of immunogenic cell death (ICD), into adenosine [29, 30, 37]. Consistently, we found that inhibition of CD73 increased the immunogenicity of cancer cells by oxaliplatin treatment, triggering DC maturation and T cell activation. Blockade of CD73 conferred to sensitivity to chemotherapy, resulting in tumor regression, infiltration of DCs and CD8⁺ T cells. Taken together, high CD73 in TME not only promoted cancer cell survival but resulted in immunosuppressive adenosine to prevent effective immunosurvalliance and immunoscavenging of residual tumor cells.

Several studies suggested that immunosuppressive TME by CD73-derived adenosine mainly dampened anti-tumor effect of different immune cells for immune escape including CD8⁺ immune cells and NK cells. CD73 also participated in the regulation of MDSC expansion, pro-tumoral M2 macrophages polarization and Treg inhibitory activity. Furthermore, several studies showed that tumor CD73 expression significantly weakened the immune response to immune checkpoint inhibitors (ICIs) [38]. Targeting CD73 showed favorable anti-tumor effects in several preclinical studies [39]. Stagg et al. reported that blockade of CD73 could reactivate adaptive anti-tumor immunity to delay tumor growth and metastasis [40]. One potential valuable antibody drug targeting CD73 is MEDI9447, enhancing the activity of anti-PD-1 inhibitor by increasing CD8⁺ T cells, prolonging IFNγ and granzyme B expression and reducing MDSC and Tregs within TME [38, 41, 42]. Therefore, combined with anti-CD73 monoclonal antibodies dramatically enhanced the therapeutic effect of anti-CTLA-4 and anti-PD-1 inhibitors, suggesting that CD73 was a potential biomarker for response to ICIs treatment. These findings showed that targeting CD73 may immunologically reinvigorate “cold” tumor into “hot” tumor, especially in mismatch repair proficient (MMR) colorectal cancer.

But the underlying mechanism to regulate CD73 expression within microenvironment was complicated. Recent studies found that epigenetic control on CD73 promoter was significantly associated with tumor CD73 expression in HNSCC and pancreatic cancer (PC) [43, 44]. Chen et al. showed that CD73 expression was conversely associated with the status of CD73 promoter methylation, and hypomethylation of CpG site on cg23172664 was remarkably correlated with poorer OS. Moreover, they found that PC patients with high CD73 expression contained a lower proportion of CD8⁺ T cells and γδ⁺ T cells by TCGA database, which is consistent with our observation in colon adenocarcinoma patients [44]. Our recent studies found that inhibition of DNMTs by FDA-approved 5-azacytidine (5-AC) and decitabine (DAC) increased the clinical benefit of anti-PD-L1 blockade combined with chemotherapy or radiotherapy in preclinical colorectal cancer model [30, 45]. Therefore, epigenetic control of CD73 by demethylating agents may potentially increase the therapeutic efficacy to anti-CD73 blockade. However, it is still need to investigate in future.

In addition to shape immunosuppressive TME, CD73 may reduce the therapeutic efficacy of radiotherapy and immunogenic chemotherapeutic agents by attenuating the effect of immunogenic cell death (ICD) [29, 46]. Several damage-associated molecular patterns (DAMPs) were released by radiotherapy and immunogenic chemotherapeutic agents to promote antigen-presenting cells (APCs) maturation and activation for anti-tumor immune response [47]. ATP, which is one of DAMPs, can engage P2RX7 receptor to dendritic cells (DCs) for T cell activation and anti-tumor immunity. Therefore, blockade of CD73 may increase the therapeutic efficacy of adjuvant chemotherapy and radiotherapy by increasing APCs maturation for T cell activation. Recent studies showed that CD73 blockade with radiotherapy increased the tumor-infiltrating DCs to enhance the therapeutic response [31, 48]. Moreover, CD73 blockade enhanced the radiotherapy-induced abscopal effects, suggesting that CD73 may be a radiation-induced immune checkpoint [31, 48]. More attractively, we found that the upstream CD39 is independent prognostic factor for favorable outcome [49]. Previous studies showed that rectal cancer patients with high tumor CD39 had favorable clinical parameters such as early TNM and absent of lymph node metastasis. They also found that high tumor CD39 was associated with favorable overall survival but they indicated that tumor CD39 cannot be used as an independent factor for predicting patients’ prognosis. Therefore, the prognostic value and its underlying mechanism of CD39 is needed to elucidate further.

Taken together, our findings showed that high tumor CD73 was associated with the risk of distant metastasis and less infiltration of immune cells, indicating that CD73 may promote immunosuppressive TME for immune escape and tumor relapse. Moreover, our studies showed that tumor CD73 may be an independent prognostic factor for colon adenocarcinoma, especially in chemotherapy-refractory COAD patients, suggesting that tumor CD73 may have prognostic value as well as the potential of therapeutic target for immunotherapy in colon adenocarcinoma.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Data curation, YL and TC; Formal analysis, SC, WTC, KCH, JL and AS; Funding acquisition, KSCC and KCH; Investigation, WTC, TK and PY; Animal experiments and flow cytometry analysis, KCH, CC and WH; Project administration, KCH; Resources, WTC, TK and KSCC; Supervision, KSCC; Writing—original draft, KCH; Writing—review & editing, KSCC.

Declarations

Competing interests

The authors declare no competing interests.

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical approval

This study was reviewed and approved by the Internal Review Board (IRB) of China Medical University Hospital [Protocol number: CMUH105-REC2-073]. The method was carried out in accordance with the committee’s approved guidelines.

Informed consent

Informed consents were obtained from all participants in the study.