Background

Epstein-Barr virus (EBV), a type of human herpesvirus, possesses a double-stranded DNA genome with an estimated length of 172 KB. EBV predominantly infects B lymphocytes and epithelial cells during its normal life cycle. Since its discovery in a patient with Burkitt’s lymphoma in 1964, EBV has been implicated in the development of a variety of tumors, primarily lymphomas and epithelial malignancies such as nasopharyngeal and gastric cancers (GCs), which may also be associated with the cellular susceptibility to EBV infection.1–3 The carcinogenic mechanism of EBV infection varies in different types of cancer. While it can promote the malignant conversion of B lymphocytes into persistently proliferating lymphoblastoid cells,4 it does not directly induce the malignant transformation of epithelial cells. However, in nasopharyngeal carcinoma, EBV infection promotes metastasis and angiogenesis by modulating the expression of non-coding RNAs.5 6 It is common to observe a large amount of lymphocyte infiltration in nasopharyngeal carcinomas. However, these cells do not appear to fully function as tumor killers and instead seem to play a role in tumor maintenance and immune evasion.3 The exact impact of EBV infection on the formation of an immunosuppressive microenvironment in tumors is still not well defined.

T-cell exhaustion is a prevalent phenomenon in cancer. Exhausted T cells exhibit inhibited effector functions, high expression of suppressor molecules (eg, PD-1, LAG-3 and Tim-3) and metabolic and epigenetic alterations.7 Although the development of PD-1/PD-L1 immunotherapy has made it possible to reverse T-cell exhaustion to some extent, its clinical efficacy remains limited.8 T-cell exhaustion is influenced by the tumor microenvironment, with the presence of suppressor molecules on tumor cells and the infiltration of suppressor cells such as Tregs, myeloid-derived suppressor cells (MDSCs), and M2 macrophages within the tumor. Persistent infection also contributes to T-cell exhaustion,7 although there is no direct evidence establishing the relationship between EBV infection and T-cell exhaustion in malignant tumors. The existence of this intricate regulatory network may be responsible for the difficulty in reversing T-cell exhaustion. It is necessary to further investigate the mechanisms underlying T-cell exhaustion in tumors and identify novel drug targets to reverse T-cell exhaustion.

Mononuclear macrophages originate from the bone marrow or yolk sac. In the process of tumor formation, mononuclear macrophages are recruited into the tumor via blood circulation to differentiate into macrophages, the most abundant type of immune cells in the tumor microenvironment.9 10 Tumor-associated macrophages (TAMs) are plastic, with anticancer M1 and procancer M2 types coexisting in the tumor immune microenvironment, and they can switch from one state to another in response to different signals.11 Although the specific markers for distinguishing M1 and M2 macrophages are not yet fully defined, some surface markers have been identified for M1 (CD16, CD64, CD86) and M2 (CD163, CD206, CD226) macrophages.12 In progressive tumors, procancer M2 macrophages tend to predominate in the microenvironment and play a detrimental role in tumor progression by promoting tumor invasion and metastasis through the secretion of extracellular matrix-degrading enzymes such as metalloproteinases and serine proteases.13 Additionally, M2 macrophages can also inhibit the tumor-killing abilities of T cells or natural killer (NK) cells by downregulating the expression of tumor antigens.14 M2 macrophages are an important component of the immunosuppressive tumor microenvironment, but the precise mechanisms of the interactions between M2 macrophages and other immune cells, such as T cells, are still not completely understood.

Matrix metalloprotein 9 (MMP9) is a zinc-dependent matrix metalloproteinase that can remodel the extracellular matrix by degrading gelatin and collagen. This property makes it an important factor in cancer progression.15 Apart from facilitating tumor metastasis and angiogenesis by degrading the extracellular matrix,16 17 MMP9 can also have a negative impact on immune responses by cleaving the activating receptors on immune cells or antigen-presenting complex molecules on tumor cells.18 Despite the demonstrated involvement of MMP9 in promoting cancer, there are no specific inhibitors of MMP9 that have been approved in clinical trials to date.19 Therefore, it is crucial to gain a deeper understanding of the mechanisms behind MMP9 upregulation in tumors and to explore new combination therapies to address the low effectiveness of MMP9 inhibitors.

In this study, we found that EBV-positive tumors recruited large numbers of CD8+T cells and mononuclear macrophages into the tumor by secreting CCL5. Subsequently, in the presence of CSF1 and IL10, mononuclear macrophages were polarized into CD163+M2 macrophages and produced high levels of MMP9, which significantly inhibited the tumor-killing ability of T cells. Altogether, these results revealed the mechanism through which EBV infection induced T-cell exhaustion. Moreover, the combination of MMP9 inhibitors and TCR-T-cell therapy presented a promising novel approach for the treatment of EBV-positive tumors.

MethodsCell culture

The tumor cell lines HK1, HK1-EBV, AGS and AGS-EBV were obtained from Tong Xiang, Sun Yat-sen University Cancer Center and cultured in RPMI 1640 (GIBCO) supplemented with 10% fetal bovine serum (NEZERUM). C666-1-A11-LMP2A cells and compatible TCR-T cells were provided by TCRCure Biological Technology Company; the former were cultured in DMEM (GIBCO) supplemented with 10% FBS, and the latter were cultured in X-vivo medium (LONZA)+1000 IU/mL IL2 (Beijing Four Rings Bio-Pharmaceutical). CD14+mononuclear macrophages and CD8+T cells were isolated from peripheral blood mononuclear cells (PBMCs) by magnetic bead sorting. Mononuclear macrophages were cultured with RPMI 1640 supplemented with 10% Australian fetal bovine serum (ExCell Bio), and CD8+T cells were cultured with X-vivo supplemented with 1000 IU/mL IL2 similar to TCR-T cells. All cells were cultured in a constant temperature incubator containing 5% CO2 at 37 ℃.

Multiplex immunohistochemistry staining

We conducted multiplex immunohistochemical staining using a Panovue multiplex fluorescent immunohistochemistry (IHC) kit (Panovue, 0004100100) according to the manufacturer’s guidelines. Briefly, paraffin sections were first dewaxed, hydrated, fixed and subjected to antigen repair and block. Subsequently, the primary antibody for the first marker was applied, followed by a horseradish peroxidase-conjugated secondary antibody. The appropriate tyramide signal amplification was then selected for labeling. This process was then repeated for each marker, and finally, the nuclei were labeled with DAPI. To obtain multispectral images, we used Olympus vs200 and Olympus UPLXAPO 20× objective lenses for whole-slide fluorescence image scanning. The whole-slide multispectral images were analyzed with Qupath software. The following primary antibodies were used: CD8 (CST, 1:400), Granzyme B (CST, 1:200), CD163 (CST, 1:400), and Pan-CK (CST, 1:400).

PBMC isolation and magnetic absorption cell sorting

PBMCs were isolated using density gradient centrifugation with Ficoll (TBD, LTS1077-1). Briefly, anticoagulated peripheral blood from healthy donors was diluted three times with saline after plasma removal and subsequently added to centrifuge tubes containing Ficoll for density gradient centrifugation at 800×g for 25 min. Then, the intermediate layer containing PBMCs was collected and washed with PBS for subsequent experiments.

Specific immune cell populations were isolated from PBMCs using Miltenyi magnetic absorption cell sorting (MACS). PBMCs were resuspended in 80 µL auto MACS Running Buffer (Miltenyi, 130-091-221-1) at a concentration of 1×107 cells. Then, a 20 µL of magnetic beads was added and incubated for 15 min at 4ºC in the dark. The cells were then diluted with MACS buffer and added to a separation column (Miltenyi, 130-042-401) placed in a strong magnetic field. The labeled cells were adsorbed in the separation column and washed off with buffer after removing the strong magnetic field to obtain a specific immune cell population with a purity of over 90%. CD14 magnetic beads (Miltenyi, 130-050-201) were used to sort CD14+mononuclear macrophages, and CD8 magnetic beads (Miltenyi, 130-045-201) were used to sort CD8+T cells.

Chemotaxis assay

We used 3 µm Transwell chambers (Corning, 353096, JET Biofil, TCS012024) to evaluate the migratory capacity of CD14+mononuclear macrophages and CD8+T cells. For chemotaxis experiments using tumor supernatant, 1×106 CD14+mononuclear macrophages or CD8+T cells were resuspended in 200 µL serum-free medium and then added to the upper chamber. The lower chamber contained 500 µL of tumor supernatant. The plates were then incubated at 37 ℃ with 5% CO2 for 4 hours. Afterward, the upper chamber was carefully removed, and 1–3 randomly selected fields of the lower chamber were photographed with a phase contrast microscope to visualize the cells that had migrated to the lower chamber.

For chemotaxis experiments involving chemokines, mononuclear macrophages or T cells were treated with tumor supernatant overnight, centrifuged, resuspended in 200 µL of serum-free medium and added to the upper chamber. The lower chamber contained 200 ng/mL CCL5 (PeproTech, 300-06-20), 200 ng/mL CXCL10 (PeproTech, 300-12-5) or 500 µL of serum-free medium. After 4 hours of chemotaxis in a 37°C incubator with 5% CO2, cells that had migrated to the lower chamber were photographed. For chemotaxis experiments involving CCR5-blocking, cells were first treated with 200 nM of CCR5 antibody (ab65850) overnight to block CCR5, whereas controls treated without CCR5 antibody. Both types of cells were centrifuged, resuspended in serum-free medium, and then chemotaxis with 200 ng/mL of CCL5, serum-free medium, or tumor supernatant.

Immunohistochemical staining

Paraffin sections of pathological tissues from patients with nasopharyngeal or gastric cancer were first dewaxed and hydrated and then subjected to antigen repair under high temperature and pressure conditions. Afterward, endogenous peroxidase was blocked by treatment with 3% H2O2. The sections were blocked with sheep serum for 1 hour (ZS bio.co., ZLI-9056) and incubated with primary antibody for 1 hour and secondary antibody (Dako, GK500710) for half an hour at 37 ℃. After the antibody-binding site was labeled with DAB (ZS bio.co., ZLI-9017), the nuclei were stained with hematoxylin (HUAYUN, HYR800). The following primary antibodies were used: CCL5 (Abcam, ab52562, 1:1000), CXCL10 (Abcam, ab214668, 1:1000), CD163 (Abcam, ab182422, 1:200), CD3 (ZS, ZA-0503-6.0), Granzyme B (CST, 46 890S, 1:200), and MMP9 (Abcam, ab76003, 1:1000).

ELISA

We used a CCL5 ELISA kit (Proteintech, KE00093), CXCL10 ELISA kit (Neobioscience, EHC157.96), CSF1 ELISA kit (Neobioscience, EHC028.96), IL10 ELISA kit (Neobioscience, EHC157.96) and MMP9 ELISA kit (Neobioscience, EHC115.96) to measure the levels of CCL5, CXCL10, CSF1, IL10 and MMP9 in the cell supernatant according to the manufacturer’s protocol.

Flow cytometry

For surface marker staining, CD14+mononuclear macrophages, THP-1 cells or CD8+T cells were resuspended in PBS (Bio-Channel, BC-BPBS-01) at a concentration of 1×106/100 µL. Fixable Viability Stain 780 (BD, 564996) was then added at a 1:1000 dilution and incubated for 10 min. Afterward, APC-CD163 (Biolegend, 333610), PE-CD206 (Biolegend, 321106), PE-Tim3 (BD, 563422) or BV421-LAG3 (BD, 565720) was added to the cell suspension at a 1:100 dilution and incubated for 25 min at room temperature. The cells were then centrifuged, resuspended in 300 µL of PBS and analyzed by flow cytometry.

For intracellular marker staining, CD8+T cells or TCR-T cells were first activated with Cell Activation Cocktail (Biolegend, 423304) for 5 hours. The cells were then resuspended in PBS and stained with Fixable Viability Stain 780 for 10 min, followed by PE/Cyanine7-CD8 (Biolegend, 100722) or PE/Cyanine7-CD3 (Biolegend, 300316) staining at a 1:100 dilution for 20 min. After washing with PBS, the Fixation/Permeabilization Kit (BD, 554714) was then used for cell permeabilization, fixation, and intracellular molecule staining. Briefly, cells were treated with Cytofix/Cytoperm solution for 25 min to fix and permeabilize the cells, followed by washing and resuspension in Perm/Wash solution. APC-IFN γ (Biolegend, 502512), PE-TNF α (Biolegend, 502909), FITC-perforin (Biolegend, 308104) and BV421-Granzyme B (BD, 563389) were added at a 1:100 dilution for intracellular staining and incubated for 25 min. Finally, the cells were resuspended in 300 µL of PBS and analyzed by flow cytometry.

CD14+mononuclear macrophage differentiation

CD14+mononuclear macrophages isolated from PBMCs by MACS were resuspended in culture medium (CM) at a concentration of 2×106 cells/mL. Subsequently, a 500 µL of the cell suspension was added to each well of a 24-well plate. To induce differentiation of the CD14+mononuclear macrophages or THP-1 cells, EBV−tumor/EBV+tumor cell supernatant was added at a volume of 500 µL/well and incubated for 3 days. Alternatively, for differentiation induction with cytokines, 50 ng/mL CSF1 (PeproTech, 300-25-10), 80 ng/mL IL10 (PeproTech, 200-10-10) or both were added to 24-well plates and incubated for 3 days. To block differentiation, 200 nM pexidartinib (CSF1Rin) (Selleck, S7818), 5 µg/mL anti-IL10R antibody (Biolegend, 308817) or both were added simultaneously with the induction of EBV+tumor cell supernatant for 3 days.

Coculture experiments and treatments

CD14+mononuclear macrophages were treated with tumor supernatant or cytokines for 3 days to induce CD163+M2 macrophages. Afterward, the cells were centrifuged, and the medium was replaced. Subsequently, CD163+M2 macrophages or mononuclear macrophages without polarization were added to 24-well plates at a density of 1×106 cells per well. After that, 1×106/well CD8+T cells or TCR-T cells were added to the same 24-well plate and cocultured with CD163+M2 macrophages or mononuclear macrophages without polarization for another 3 days with or without 20 µM MMP9 inhibitor (MCE, HY-135232). Then, the cells were either analyzed using flow cytometry or used directly for killing assays or apoptosis assays after replacing with fresh CM. Additionally, pure TCR-T cells (>95% purity) were isolated using the MACS technique after coculture and subsequently employed in the killing experiments or apoptosis assays.

RNA sequencing and analysis

CD14+mononuclear macrophages were treated with TRIzol after induction of differentiation with tumor supernatant or cytokines. The extracted RNA samples were then sent to Gene Denovo for purification, sequencing and analysis. Differential gene expression between the two groups was analyzed using DESeq2 software, followed by Gene Ontology (GO) pathway enrichment analysis. Additionally, gene set enrichment analysis (GSEA) was performed using software such as GSEA and MSigDB.

Lactate dehydrogenase release

C666-1-A11-LMP2A cells were seeded in 96-well plates overnight at a concentration of 1×104 cells per well and allowed to adhere to the well surface. Afterward, 5×104 TCR-T cells, 5×104 mock-T cells, 5×104 mononuclear macrophages+5×104 TCR-T cells or 5×104 CD163+M2 macrophages+5×104 TCR-T cells were added to each well to kill the tumor cells for 16 hours. Subsequently, the culture supernatant was collected. Lactate dehydrogenase (LDH) is originally present in the cytoplasm and can be released into the CM on cell death.20 To assess the level of cytotoxicity, we used a CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega, G1780) according to the manufacturer’s protocol to measure the LDH level in the collected supernatant.

Apoptosis assay

C666-1-A11-LMP2A cells (2×105/well) were added to 24-well plates overnight for adherence. Subsequently, 1×106 TCR-T cells or 1×106 mononuclear macrophages/CD163+M2 macrophages+1×106 TCR-T cells were added to induce tumor cell apoptosis for 16 hours. Afterward, the cells were collected and labeled with an Annexin V-AF647/PI apoptosis kit (ESscience, AP006), and then the apoptosis rate of C666-1-A11-LMP2A cells was determined by flow cytometry.

Animal studies

Five-week-old female NCG mice were obtained from Gempharmatech Company. A total of 5×106 C666-1-A11-LMP2A cells were subcutaneously inoculated in the right flank of the mice. The injection of immune cells via the tail vein was started in the second week after subcutaneous inoculation of the tumor cells, when the tumor maximum diameter reached 5 mm. CD14+mononuclear macrophages were induced to differentiate into CD163+M2 macrophages (>90% purity) by treatment with CSF1 (50 ng/mL)+IL10 (80 ng/mL). A total of 1×107 CD163+M2 macrophages were then injected into the tail vein of mice once a week. Additionally, 1×107 TCR-T cells, prepared by TCRCure Biological Technology Company, were injected either alone or together with CD163+M2 cells via the tail vein. For mice treated with TCR-T cells, 10,000 IU IL2 was intraperitoneally injected every other day to maintain TCR-T-cell activity. Furthermore, MMP9 inhibitors (MCE, HY-135232) were intraperitoneally administered at a dose of 20 mg/kg twice a week from the time of immune cell injection. The length (L) and width (W) of the tumors were measured every 2 days with Vernier calipers, and the tumor volume was calculated based on the following formula: tumor volume=(L×W2)/2. The xenograft tumors were harvested on day 26 for tumor weight evaluation, IHC staining, and other analyses.

Statistical analysis

GraphPad Prism V.8 was used for statistical analysis. The statistical methods used are described in the figure legends. Significance was determined as *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Student’s t-test was used to make comparisons between two groups. One-way analysis of variance (ANOVA) was employed to assess differences between multiple groups while two-way ANOVA was used to compare tumor growth curves. Pearson’s correlation coefficient was used to assess the correlations between CD163+cells and Granzyme B/CD8.

ResultsA higher abundance of exhausted CD8+T cells in EBV-positive tumors than EBV-negative tumors is associated with an increased population of CD163+ M2 macrophages

EBV infection is a risk factor for the progression of several malignancies.3 T-cell exhaustion is a state manifested by a decrease in T-cell effector function, accompanied by a possible upregulation of the expression of suppressor molecules, which is an important mechanism of immune evasion in tumors.7 To investigate whether there is a potential link between EBV infection and T-cell exhaustion, we collected paraffin-embedded tumor sections from three types of EBV-associated tumor patients: nasopharyngeal cancer (NPC), gastric cancer (GC), and lymphoepithelioma-like carcinoma of the lung (LELC), who were initially diagnosed at the Sun Yat-sen University Cancer Center and underwent surgical procedures to obtain gross specimens. First, we performed EBER in situ hybridization to determine whether the tumors were infected by EBV. EBER-positive ones were defined as EBV-positive tumors, and vice versa as EBV-negative, and ultimately, three pairs of roughly matched EBV± tumor paraffin sections for each type of carcinoma were used to perform multiplex immunohistochemical staining. Interestingly, in all three tumor types, we observed that EBV-positive tumors recruited more CD8+T cells than EBV-negative tumors (figure 1A,B). However, the expression of Granzyme B, a major effector molecule of intratumoral CD8+T cells, did not show a significant upregulation in the EBV-positive tumors except in the LELC sections (figure 1A,C). We further investigated the ratio of Granzyme B/CD8 and found that this ratio was significantly lower in EBV-positive tumors (figure 1D), indicating a loss of T-cell effector function and a higher state of exhaustion. Many factors contribute to T-cell exhaustion, and an increase in suppressor immune cells is one of them. Previous studies have suggested that EBV can activate ATR and subsequently promote M2 polarization.21 Although there is a lack of specific markers that distinguish between M1 and M2 macrophages, CD163 is a widely used surface marker for M2 macrophages.12 Accordingly, we included the marker CD163 for M2 macrophages in our multiplex IHC staining and found a significant increase in the number of suppressor immune cells (CD163+M2 macrophages) in EBV-positive tumors (figure 1E,F). Furthermore, we observed a negative correlation between the number of CD163+M2 macrophages and the Granzyme B/CD8 ratio (figure 1G), suggesting a potential connection between T-cell exhaustion and the increase in CD163+M2 macrophages in EBV-positive tumors. Overall, our findings indicate an elevated numbers of exhausted T cells in EBV-positive tumors, which may be associated with the increased number of CD163+M2 cells.

Figure 1

Increased numbers of exhausted CD8+T cells and CD163+M2 macrophages within EBV-positive tumors. (A) Representative single-stained and merged images of panels (CD8, Granzyme B, DAPI) used for multiplex immunohistochemistry in EBV-positive and EBV-negative tumors. Magnification: ×400 (single-stained images and big merged images), ×200 (small merged images in the bottom right corner). (B–D) Statistical analysis of the density of CD8 (B) or Granzyme B (C) and the ratio of Granzyme B/CD8 (D) in multiplex immunohistochemistry. Mean±SD, n=3, two-tailed t-test. (E) Representative merged images of panels (CD8, Granzyme B, CD163, Pan-CK, DAPI) used in multiplex immunohistochemistry. Magnification: ×400 (small merged images in the top left corner), ×100 (big merged images). (F) Statistical analysis of the density of CD163 in EBV-positive and EBV-negative tumors by multiplex immunohistochemistry. Mean±SD, n=3, two-tailed t-test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. (G) Correlation analysis between CD163/mm2 and the Granzyme B/CD8 ratio in NPC, GC and LELC. EBV, Epstein-Barr virus; GC, gastric cancer; LELC, lymphoepithelioma-like carcinoma of the lung; NPC, nasopharyngeal cancer.

EBV infection in tumor promotes CCL5 secretion to recruit CD8+T cells and mononuclear macrophages

The numbers of both CD8+T cells and CD163+M2 macrophages were found to be increased in EBV-positive tumors. To further investigate the underlying mechanism, we conducted immune cell chemotaxis assays. We used two pairs of cell lines, one infected with EBV (HK1-EBV and AGS-EBV) and the other not infected (HK1 and AGS), as previously described.22 We collected the supernatants from both EBV-infected and uninfected tumor cell lines. Then, both EBV-infected and EBV-uninfected tumor supernatants were used to recruit CD8+T cells or CD14+mononuclear macrophages, which were isolated from healthy donor PBMCs using MACS with a purity of over 90% (online supplemental figure 1A). The results showed that the supernatant from EBV-infected tumors had a significantly stronger recruitment ability for both CD8+T cells and CD14+mononuclear macrophages than the supernatant from EBV-uninfected tumors (figure 2A,B). Then, to investigate the chemokines involved in the recruitment of CD8+T cells and CD14+monocytes, we performed RNA sequencing of EBV-infected and uninfected tumor cell lines. Pathway enrichment analysis revealed that the chemokines CCL5 and CXCL10 were upregulated in pathways related to T-cell chemotaxis and mononuclear macrophage chemotaxis (online supplemental figure 1B,C). Immunohistochemical staining and ELISA measurements confirmed that EBV-infected tumors secreted increased levels of CCL5 and CXCL10, consistent with the sequencing results (figure 2C,D). Next, we used 200 ng/mL CCL5, 200 ng/mL CXCL10 or serum-free RPMI 1640 (control) to recruit CD8+T cells and CD14+mononuclear macrophages treated with different EBV-infected tumor supernatants overnight and found that CCL5 had a significantly higher recruitment capacity than CXCL10 or serum-free RPMI 1640 for both cell types (figure 2E,F and online supplemental figure 1D,E), suggesting that EBV-infected tumors may recruit these immune cells by secreting more CCL5. CCL5 interacts with three known receptors: CCR1, CCR3, and CCR5. CCL5 has a stronger affinity for CCR5 than CCR1 and CCR3.23 To verify whether CCL5 secreted by EBV-infected tumors binds predominantly to CCR5 with the highest affinity, we used a 200 nM CCR5 antibody overnight to block CCR5 on CD8+T cells and CD14+mononuclear macrophages. The recruitment of CCL5 to immune cells treated with both EBV-infected tumor supernatant and blockade of CCR5 was markedly attenuated in comparison to cells treated with tumor supernatant only overnight (figure 2G and I and online supplemental figure 1F and H). Similarly, EBV-infected tumor supernatants had a diminished ability to recruit CD8+T cells and CD14+mononuclear macrophages when their CCR5 was blocked (figure 2H and J and online supplemental figure 1G and I). In summary, EBV-infected tumors secrete CCL5, which binds to CCR5 on immune cells and subsequently recruits CD8+T cells and CD14+monocyte macrophages.

Figure 2

EBV-infected tumors secrete CCL5 to recruit T cells and monocytes. (A, B) Representative images (left) and quantification (right) of CD8+T cells (A) or CD14+monocytes (B) recruited to the lower chamber by EBV-infected or EBV-uninfected tumor supernatants. Magnification: ×400. Mean±SD, n=3, two-tailed t-test. (C, D) Left: Representative images of EBV-uninfected or EBV-infected nasopharyngeal and gastric cancer pathology sections stained with antibodies targeting CCL5 (C) or CXCL10 (D). Magnification: ×200. Right: Measurement of CCL5 (C) and CXCL10 (D) levels in the supernatant of EBV-uninfected or EBV-infected nasopharyngeal and gastric cancer cell lines by ELISA. Mean±SD, n=4, two-tailed t-test. (E, F) Quantification of CD8+T cells (E) or CD14+ monocytes (F) recruited to the lower chamber by 200 nM CCL5, 200 nM CXCL10 or serum-free RPMI 1640. CD8+T cells and CD14+monocytes were pretreated overnight with AGS-EBV or HK1-EBV cell supernatants. Mean±SD, n=3, one-way ANOVA. (G, I) Quantification of CD8+T cells (G) and CD14+monocytes (I) recruited to the lower chamber. CD8+T cells and CD14+monocytes were pretreated overnight with AGS-EBV or HK1-EBV cell supernatant, and anti-CCR5 or CCL5+anti-CCR5 groups were simultaneously treated with 200 nM CCR5 antibody overnight to block CCR5. CD8+T cells or CD14+ monocytes were then recruited with 200 nM CCL5 or serum-free RPMI 1640. Mean±SD, n=3, one-way ANOVA. (H, J) CD8+T cells or CD14+ monocytes were treated with or without CCR5 antibody overnight to block CCR5. The above cells were recruited with EBV-infected or EBV-uninfected tumor supernatants. Quantification of CD8+T cells (H) and CD14+ monocytes (J) recruited to the lower chamber. Mean±SD, n=3, one-way ANOVA. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ANOVA, analysis of variance; EBV, Epstein-Barr virus.

EBV infection in tumors induces CD163+M2 macrophages polarization by secreting CSF1 and enhancing autocrine IL10 production in mononuclear macrophages

EBV-infected tumors recruit large numbers of mononuclear macrophages, but their direction of polarization within the tumor remains unclear, and the mechanisms underlying the increase in CD163+M2 macrophages within EBV-positive tumors still need to be further investigated. We first induced THP-1 cell or CD14+mononuclear macrophage differentiation with EBV-infected or EBV-uninfected tumor supernatant and found that EBV-infected tumor supernatant significantly promoted the polarization of both toward CD163+M2 macrophages than EBV-uninfected tumor supernatant (figure 3A), suggesting that the increase in CD163+M2 macrophages within EBV-positive tumors may be related to the polarization of CD163+M2 macrophages promoted by EBV infection. In the RNA sequencing results of the EBV-infected and EBV-uninfected tumor cell line, CSF1 enrichment scores were the highest in pathways associated with mononuclear macrophage differentiation (online supplemental figure 1C). Further analysis using ELISA revealed elevated levels of CSF1 in EBV-infected tumor supernatants compared with EBV-uninfected tumor supernatants (figure 3B). Monocytes can differentiate into mature macrophages in response to CSF1, but macrophage polarization to CD163+M2 still requires the involvement of other inhibitory molecules such as IL10 or TGF β.24 25 When treated with recombinant CSF1 alone, CD14+monocytes showed a tendency to differentiate more toward CD163+CD206+ M2 cells. However, when treated with IL10 alone or with both CSF1 and IL10, they tended to differentiate toward CD163+CD206- M2 cells, similar to the effect of treatment with EBV-infected tumor supernatant (figure 3C). We thus hypothesized that IL10 plays an important role in the EBV-induced differentiation of CD163+M2 macrophages. Although there was a slight increase in the IL10 level in the EBV+tumor supernatant, the absolute level was relatively low (figure 3D). It has been reported that macrophages themselves can secrete IL10 in addition to tumor cells.26 Therefore, we examined the level of IL10 secreted by mononuclear macrophages treated with EBV-infected or EBV-uninfected tumor supernatant and found that mononuclear macrophages treated with EBV-infected tumor supernatant exhibited higher autocrine production of IL10 (figure 3E). These results revealed that IL10 was more frequently produced by macrophages than tumor cells in EBV-infected tumor. Next, we further explored the role of CSF1 in IL10 autocrine secretion. We found that blocking the receptor of CSF1 impaired the ability of EBV-infected tumor supernatant to promote IL10 autocrine secretion while treatment of macrophages with CSF1 alone increased IL10 secretion (figure 3F), suggesting that CSF1 was involved in the promotion of EBV-induced IL10 autocrine secretion by mononuclear macrophages. Next, to further confirm these findings, we performed blocking experiments. Blocking the CSF1 receptor reduced the differentiation of CD163+M2 macrophages when CD14+mononuclear macrophages were treated with EBV-infected tumor CM (EBV+CM), while blocking IL10 receptor alone or blocking both the CSF1 and IL10 receptor resulted in minimal differentiation toward CD163+M2 macrophages (figure 3G). These findings indicate that EBV-infected tumors secrete higher levels of CSF1, which in turn promotes IL10 autocrine secretion by mononuclear macrophages recruited into the tumor. The combined effect of CSF1 and IL10 then drives the differentiation of monocytes toward CD163+M2 macrophages.

Figure 3

EBV-infected tumors promote CD163+M2 macrophages polarization. (A) Flow cytometry analysis of the CD163+M2 phenotype in THP-1 or CD14+ monocytes treated with EBV-infected or EBV-uninfected tumor supernatants. Mean±SD, n=4 (HK1), n=3 (AGS), two-tailed t-test. (B) Level of CSF1 in EBV-infected and EBV-uninfected tumor supernatants as measured by ELISA. Mean±SD, n=4, two-tailed t-test. (C) Flow cytometry analysis of the CD163+M2 phenotype of CD14+monocytes treated with 50 ng/mL CSF1, 80 ng/mL IL10, or both together. Mean±SD, n=3, one-way ANOVA. (D) Level of IL10 in EBV-infected and EBV-uninfected tumor supernatants as measured by ELISA. Mean±SD, n=4, two-tailed t-test. (E) Levels of IL10 in supernatants of mononuclear macrophages treated with EBV-infected and EBV-uninfected tumor supernatants for 3 days as measured by ELISA. Mean±SD, n=3, two-tailed t-test. (F) Levels of IL10 in mononuclear macrophage supernatants treated for 3 days with EBV-infected tumor supernatant, EBV-infected tumor supernatant and CSF1R inhibitor, 50 ng/mL CSF1, or no treatment as measured by ELISA. Mean±SD, n=4, two-tailed t-test. (G) Flow cytometry analysis of the CD163+M2 phenotype of CD14+monocytes treated with EBV-infected tumor supernatant (control), EBV-infected tumor supernatant and CSF1R inhibitor, EBV-infected tumor supernatant and anti-IL10R antibody, or EBV-infected tumor supernatant and CSF1R inhibitor and anti-IL10R antibody. Mean±SD, n=3, one-way ANOVA. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ANOVA, analysis of variance; EBV, Epstein-Barr virus.

EBV-induced CD163+M2 polarization contributes to T-cell exhaustion

To further investigate whether there is a direct correlation between increased CD163+M2 macrophage numbers and T-cell exhaustion in EBV-positive tumors, we first conducted RNA sequencing on mononuclear macrophages induced by EBV-infected or EBV-uninfected tumor supernatant. We identified a total of 94 upregulated genes and 327 downregulated genes in mononuclear macrophages treated with EBV-infected tumor supernatant compared with those treated with EBV-uninfected tumor supernatant (figure 4A). GO pathway enrichment of these differentially expressed genes revealed several pathways associated with T-cell activation in the top 20 pathways (figure 4B). Next, GSEA also suggested that mononuclear macrophages induced by EBV-infected tumor supernatants downregulated pathways associated with lymphocyte activation or T-cell activation (figure 4C,D). We examined the expression of inhibitory molecules on mononuclear macrophages induced by EBV-infected or EBV-uninfected tumor supernatant and found that EBV-infected tumor supernatant upregulated the expression of TGFB1, Galectin9, Galectin3 and LSECtin (online supplemental figure 2A) on mononuclear macrophage. Subsequently, we cocultured mononuclear macrophages induced by EBV-infected or EBV-uninfected tumor supernatant with CD8+T cells and found that mononuclear macrophages induced by EBV-infected tumor supernatant significantly inhibited T-cell function (figure 4E) and upregulated the expression of inhibitory receptors (Tim3 and LAG3) on the surface of T cells (online supplemental figure 2B), leading to the manifestation of T-cell exhaustion characteristics. These evidences suggested a direct link between macrophage polarization induced by EBV infection and T-cell exhaustion, but whether it was EBV-induced CD163+M2 macrophages that led to T-cell exhaustion still needed to be further explored. We then generated CD163+M2 macrophages (>90% purity) by treated CD14+monocytes with CSF1 and IL10 and then cocultured them with CD8+T cells, and found that these CD8+T cells similarly exhibited reduced function compared with those co-cultured with untreated monocyte macrophages (control) (figure 4F,G). In contrast, reduced CD163+M2 macrophage differentiation by blocking either CSF1 or IL10 receptors alone or together during the treatment of mononuclear macrophages with EBV-infected tumor CM (EBV+CM) improved T-cell function (figure 4H,I). These results confirmed that EBV-induced generation of CD163+M2 macrophages contributed to T-cell exhaustion. Overall, we find that there is a direct link between T-cell exhaustion and the induction of CD163+M2 macrophage polarization by EBV infection via CSF1 and IL10 within EBV-positive tumors.

Figure 4

CD163+M2 macrophages polarized by EBV suppresses T-cell function. (A) Volcano plot of differentially expressed genes between mononuclear macrophages induced with EBV-uninfected tumor supernatant and those induced with EBV-infected tumor supernatant for 3 days. (B) GO pathway enrichment analysis of differentially expressed genes between mononuclear macrophages treated with EBV-infected and EBV-uninfected tumor supernatant. (C, D) GSEA of differentially expressed genes between mononuclear macrophages treated with EBV-infected and EBV-uninfected tumor supernatant. (E) Flow cytometry analysis of T-cell functional molecules (Granzyme B, IFN γ, TNF α, Perforin) after 3 days of coculture with mononuclear macrophages polarized with EBV-uninfected or EBV-infected tumor supernatants. Mean±SD, n=3, two-tailed t-test. (F) Flow cytometry analysis of T-cell functional molecules after 3 days of coculture with mononuclear macrophages untreated (control) or polarized with 50 ng/mL CSF1+80 ng/mL IL10 (M2). (G) Statistical analysis of T-cell functional molecules in (F). Mean±SD, n=3, two-tailed t-test. (H) Flow cytometry analysis of T-cell functional molecules after 3 days of coculture with mononuclear macrophages induced by EBV-infected tumor culture medium (control), EBV-infected tumor culture medium and CSF1R inhibitor (CSF1Rin), EBV-infected tumor culture medium and anti-IL10R antibody (anti-IL10) or EBV-infected tumor culture medium and CSF1R inhibitor and anti-IL10R antibody (CSF1Rin+anti-IL10). (I) Statistical analysis of T-cell functional molecules in (H). Mean±SD, n=3, one-way ANOVA. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. ANOVA, analysis of variance; EBV, Epstein-Barr virus; GO, Gene Ontology; GSEA, gene set enrichment analysis.

CD163+M2 macrophages impair the cytotoxic potential of TCR-T cells in vitro and in vivo

To research the effect of CD163+M2 polarized by EBV-infected tumors on the cytotoxicity of T cells, we constructed C666-1-A11-LPM2A cells as well as TCR-T cells that have TCRs (AU011) capable of directly recognizing and targeting C666-1-A11-LPM2A cell antigen (A11-LMP2A) (online supplemental figure 3A). When cocultured with C666-1-A11-LMP2A cells, TCR-T cells secreted more effector molecules, such as perforin, TNF-α, IFN-γ and granzyme B, than mock-T cells (online supplemental figure 3B). Furthermore, TCR-T cells effectively killed C666-1-A11-LPM2A tumor cells, leading to a higher release of LDH than mock-T cells (online supplemental figure 3C). These results confirmed that our TCR-T cells can directly recognize, target and kill C666-1-A11-LPM2A cells. However, in the presence of CD163+M2 macrophages, the release of effector molecules by TCR-T cells was significantly reduced (figure 5A). Isolation of TCR-T cells cocultured with CD163+M2 cells using MACS revealed their decreased cytotoxic capacity against C666-1-A11-LMP2A cells by LDH release and apoptosis assay (figure 5B,C). Similar results were observed when TCR-T cells were not isolated (online supplemental figure 3D,E). We then subcutaneously implanted NCG mice with C666-1-A11-LPM2A cells and injected CD163+M2 macrophages and TCR-T cells through the tail vein on day 13 and day 21 separately or together to investigate the effect of M2 macrophages on the cytotoxicity of TCR-T cells in vivo (figure 5D). Compared with the control group (without treatment), the treatment group injected with TCR-T cells alone or M2+TCR T cells exhibited decreased tumor volume and weight (although the difference between the M2+TCR T and control groups was not statistically significant) while M2 macrophages injection alone had little effect on tumor growth. However, tumor growth was accelerated in the M2+TCR T group compared with the TCR-T group, indicating that M2 macrophages inhibited the tumor-killing capability of TCR-T cells in vivo (figure 5E and online supplemental figure 4A–D). Hematoxylin-eosin (HE) and immunohistochemical staining of the tumors revealed infiltration of injected M2 macrophages and TCR-T cells into the tumor, but coinjection of M2 with TCR-T cells resulted in a significant reduction in the functional molecule granzyme B compared with TCR-T cells alone. Interestingly, although the difference of the number of CD163+cells between M2 and M2+TCR T was not statistically significant, injection of TCR-T cells appeared to promote infiltration of CD163+M2 macrophages, which we speculated that apoptotic tumor cells resulting from TCR-T cell activity released certain chemokines that further recruited CD163+M2 macrophages (figure 5F). In conclusion, CD163+M2 macrophages inhibited the targeted killing of C666-1-A11-LMP2A cells by TCR-T cells both in vitro and in vivo.