TIE-2 and VEGFR Kinase Activities Drive Immunosuppressive Function of TIE-2–Expressing Monocytes in Human Breast Tumors

TEMS suppress tumor-specific T-cell responses

We examined the impact of TEMs on antitumor T-cell responses in a group of untreated breast cancer patients with primary invasive breast carcinoma (Supplementary Table S1). We first examined the presence of tumor-specific T cells in the peripheral blood of the patients. To this end, DCs were generated in vitro from peripheral blood monocytes of patients, pulsed with autologous tumor lysate, and cocultured with autologous T cells labeled with the cell tracker CFSE. T-cell proliferation was monitored at day 5 by measuring CFSE dilution by flow cytometry. Peripheral CD4 T cells from patients proliferated in response to DCs pulsed with autologous tumor lysate and the specificity of these responses for the tumor was shown by the lower frequency of T cell proliferation when alone or in the presence of unpulsed DCs (Supplementary Fig. S1).

On the basis of these observations, we next measured peripheral blood T-cell response induced by tumor TEMs and DCs. Flow cytometric analysis of freshly dissociated breast tumors revealed that TEMs and CD11c⁺ DCs together represented 80% of the CD45⁺ hematopoietic infiltrate and that CD11c⁺ DCs were 2.6 (±0.3)-fold more abundant than TEMs (22% ± 2.7% of CD45⁺ cells, n = 10). Moreover, flow cytometry and confocal microscopic analysis revealed that, in breast tumors, the vast majority (>95%) of CD14⁺ cells coexpressed TIE-2 and VEGFR-1 (Supplementary Fig. S2). In order to measure tumor-specific T-cell responses, we immunomagnetically isolated TEMs and DCs from freshly dissociated breast tumors and cocultured them immediately with autologous T cells isolated from peripheral blood and labeled with CFSE. In this assay, tumor antigens are presented by CD11c⁺ DCs isolated from the tumor, and T-cell proliferation in response to tumor antigens was monitored by measuring CFSE dilution using flow cytometry. Among the patients examined (n = 6), the frequency of proliferating T cells following exposure to CD11c⁺ DC ranged between 0.8% and 12.8% and was set at 100% for normalization (Fig. 1A and B and Supplementary Table S2). However, when TEMs and CD11c⁺ DCs were cocultured with T cells, their proliferation was significantly (P < 0.01) reduced (Fig. 1A and B). The proliferation of CD4 and CD8 T cells was similarly and significantly (P < 0.05) impacted by TEMs (Fig. 1C), suggesting that TEMs suppresses DC-mediated T-cell proliferation. Following 5-day exposure of CD4 T cells to conditioned medium of TEMs, CD11c⁺ cells, or TEMS/CD11c⁺ coculture, the T-cell viability was not significantly affected, thus excluding any toxic effect of TEMs on T cells (Fig. 1D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

TEMs suppress CD11c⁺ DC–mediated tumor-specific T-cell proliferation. CD11c⁺ DCs and TEMs from breast tumors were exposed to autologous CFSE-labeled T cells and their proliferation was assessed 5 days later by flow cytometry. A, representative dot plots showing the percentage of proliferating CD4 and CD8 T cells. B and C, graphic representation of the changes of proliferating T cells normalized to T-cell proliferation in the presence of CD11c⁺ DC [(% CFSE^low T cells in the presence of DC) − (% CFSE^low T cells when alone) = 100%]. D, CD4 T cells negatively selected from peripheral blood were exposed for 5 days to the conditioned medium of TEM, DC, or TEM/DC coculture and their viability was assessed by flow cytometry following DAPI staining. Cumulated data from 6 independent experiments and 6 distinct patients are shown (B–D); *, P < 0.05; **, P < 0.01.

Immunosuppressive and proangiogenic functions of TEMs are driven by TIE-2 and VEGFR kinase activities

In breast tumor, TEMs express high levels of TIE-2 and VEGFR-1 (Supplementary Fig. S2), both of which control their high proangiogenic activity (Xenarios et al., under revision). We assessed whether combined blockade of TIE-2 and VEGFR kinase activities with specific receptor tyrosine kinase inhibitors (RTKI, i.e., TIE-2 kinase inhibitor compound 7 and VEGFR kinase inhibitor PTK787) would reduce the proangiogenic activity of TEMs. To this end, we measured in vitro the ability of tumor TEMs to induce sprouting of human umbilical vein endothelial cells (HUVEC; see Materials and Methods). Tumor TEMs treated with RTKI induced the formation of much shorter HUVEC sprouts relative to untreated TEMs (Fig. 2A), thus confirming that the proangiogenic activity of TEMs is controlled by the activities of TIE-2 and VEGFR kinase. We next examined the impact of these RTKI on the ability of TEMs to suppress CD11c⁺ DC-mediated T-cell proliferation. TEMs were treated for 6 hours with RTKI, extensively washed, and cocultured for 5 days with tumor DCs and CFSE-labeled autologous T cells. Importantly, we observed that RTKI treatment simultaneously abolished the proangiogenic activity of tumor TEMs (Fig. 2A) and increased significantly (P < 0.05) the fraction of proliferating T cells (T cells + DC = 100%, T cells + DC + TEMS = 44% ± 18, T cells + DC + TEMS/RTKI = 68% ± 27) and the number of cell divisions (Fig. 2B). We, therefore, calculated the T-cell proliferative index, defined as the average number of cell division that the responding cells underwent assessed on the basis of CFSE dilution measurements (see Materials and Methods). The T-cell proliferation index was significantly increased (P < 0.05) when TEMs were treated with RTKI. In addition, CD4 and CD8 tumor-specific T-cell responses were similarly enhanced (Fig. 2C). Furthermore, activities of TIE-2 and VEGFR kinase contribute more to the suppressive function of TEMs than the immunosuppressive cytokines VEGF and IL-10 (Fig. 2C), both of which are released in much larger amounts by TEMs than CD11c⁺ DCs (Fig. 2D). Thus, RTKI treatment revealed that the proangiogenic and suppressive activities of TEMs operate through common pathways involving activities of TIE-2 and VEGFR kinase.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Proangiogenic and suppressive functions of tumor TEMs are driven by TIE-2 and VEGFR kinase activities. A, TEMs were isolated from breast tumor and treated with RTKI or VEGF function-blocking antibodies and their proangiogenic activity was measured by sprouting of HUVEC cultured on microcarrier beads. The percentage of cumulated sprout length normalized to untreated TEM is shown. B, representative histograms showing proliferating CD4 T cells cocultured for 5 days with DC and untreated TEMs (filled histogram) or TEMs treated for 6 hours with RTKI (black line histogram). The number of cell divisions on the basis of CFSE content is indicated. C, TEMs were treated with RTKI or VEGF and IL-10 function-blocking antibodies and the T-cell proliferation index was calculated as the average number of cell division undergone by the responding cells. Changes in the proliferation index normalized to coculture with untreated TEMs are shown. Control experiments conducted in the absence of treatment or with isotype control antibodies showed the same proliferation index. D, VEGF and IL-10 release measured in a conditioned medium of TEMs and DC 30 hours postisolation from breast tumors. Cumulative data for TEMs from 5 independent experiments and 5 distinct patients are shown in A, B, and D. *, P < 0.05; **, P < 0.01.

Tumor TEMs support a weak T-cell proliferation and expand regulatory T cells

We found that tumor CD11c⁺ DCs were significantly (P < 0.01) more potent than tumor TEMs at inducing T-cell proliferation (Fig. 1A and B). Indeed, tumor TEMs induced a weak but significant CD4 T-cell proliferation (Fig. 3A, n = 8), which was abolished by neutralizing antibodies to MHC (Fig. 3A and B). These results indicate that tumor TEMs are functional APCs displaying, however, a much weaker aptitude to present antigen to T cells relative to CD11c⁺ DCs. Interestingly, tumor TEMs, but not CD11c⁺ DCs, induced a massive (>20%) conversion of CD4 T cells into CD25⁺ or FOXP3⁺ CD4 Tregs. Importantly, RTKI treatment inhibited the ability of TEMs to induce FOXP3 expression on CD4 T cells (Fig. 3C and D). These CD4⁺CD25⁺FoxP3⁺ cells endowed a phenotype and functional characteristics of Treg as they efficiently suppressed the proliferation of CD4⁺ CD45RA⁻ CD25⁻ effector cells in vitro (Fig. 3C, right). These results suggest that activities of Tie2 and VEGFR kinase control both, the ability of TEMs to support T-cell proliferation and the conversion of T cells into Treg.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Tumor TEMs support T-cell proliferation and expand regulatory CD4 T cells. TEMs from breast tumors were exposed to autologous CFSE-labeled CD4 T cells and their proliferation was assessed 5 days later by flow cytometry. A, percentage of proliferating T cells in eight distinct patients. B, changes in CD4 T cell proliferation exposed to TEMs in the presence or absence of MHC I and II neutralizing antibodies; representative dot blots are shown. C, frequency of CD25⁺FOXP3⁺ cells among CD4 T cells (left). Following 5-day coculture with TEMs, CD4⁺CD25⁺ were isolated by immunomagnetic selection and cocultured for 2.5 days with CD4⁺CD45RA⁻CD25⁻ effector cells previously labeled with DDAO and activated with CD3 and CD28 antibodies. Histograms show DDAO dilution in proliferating effector cells (right). D, frequency of CD25⁺FOXP3⁺ cells among CD4 T cells in the presence of CD80 or CD86 blocking antibodies normalized to CD4 + TEM (100%). Cumulative data from 4 independent experiments and 4 distinct patients are shown in B and C. *, P < 0.05; **, P < 0.01.

Antiangiogenic treatments alter TEMs phenotype toward myeloid DCs in vitro

We recently reported that, in TEMs from breast tumor or differentiated in vitro from cord blood CD34⁺ precursors, Tie2 and VEGFR1 pathways synergized with the TNF-α and TGF-β pathways to induce and reduce the proangiogenic activity of TEMs respectively (Xenarios et al., under revision). We compared the impact of RTKI and ANG-2/TGF-β treatments on the proangiogenic activity of TEMs using the HUVEC sprouting assay (see Materials and Methods section). ANG-2/TGF-β (Fig. 4A; Xenarios et al., under revision) or RTKI treatments resulted in a marked reduction in sprout length, confirming that TIE-2 and VEGFR pathways are critical for the proangiogenic activity of both TEMs differentiated in vitro (Fig. 4A) and tumor TEMs (Fig. 2A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Antiangiogenic treatments shift the gene expression profile and the functional phenotype of TEMs differentiated in vitro toward that of myeloid dendritic cells. A, proangiogenic activity of TEMs differentiated in vitro in response to ANG-2/TGF-β and TIE-2 and VEGFR kinase inhibitors (RTKI) measured by sprouting of HUVEC cultured on microcarrier beads. The percentage of cumulative sprout length normalized to untreated TEMs is shown. B, a total of 355 genes were significantly differentially expressed (adjusted P value ≤ 0.05) between ANG-2/TGF-β and untreated cells. Differentially expressed genes were manually annotated and classified in categories. The genes functionally related to angiogenesis, immune response, cell differentiation, and antigen processing are shown. C, monitoring by flow cytometry of markers of DC differentiation and activation in TEMs treated with ANG-2/TGF-β or TIE-2 and RTKI. Percentage of increase in the expression levels relative to untreated TEMs are shown. D, in vitro differentiated TEMs were cocultured with CFSE-labeled autologous T cells previously stimulated for 5 days with anti-CD3 and CD28 antibodies. In the absence of TEMs, the rate of proliferating T cells, as assessed by CFSE content, was 5% and 30% for stimulated (filled histogram) and unstimulated T cells respectively. TEMs were either untreated or treated with ANG-2/TGF-β or RTKI before coculture with T cells. Mean cumulative data of 5 experiments from 5 patients and 3 distinct cord blood samples are shown in A, C, and B respectively. A representative experiment out of 4 is shown in D. *, P < 0.05; **, P < 0.01.

To gain insight into the impaired proangiogenic phenotype of TEMs, we conducted gene expression profiling of TEMS differentiated in vitro treated with ANG-2/TGF-β (Xenarios et al., under revision; Supplementary Methods). We identified 355 genes that were significantly differentially expressed, with false discovery rate (FDR) 5% or less, upon ANG-2/TGF-β treatment relative to untreated TEMs. Approximately 70% of these genes were downregulated and encoded for functions related to cell cycle, metabolism, apoptosis, and motility (Xenarios et al., under revision). ANG-2/TGF-β treatment downregulated proangiogenic genes, while genes encoding for angiostatic proteins (VASH1: vasohibin 1, UCN: urocortin, the angiostatic cytokine CXCL9; ref. 19) were upregulated (Fig. 4B). This impaired proangiogenic activity was associated with the downregulation of genes whose products are involved in immune suppression (IL-10; HSD11B1, hydroxysteroid 11-β dehydrogenase 1; SLC16A7, solute carrier family 16, member 7) including S100A9 (S100 calcium-binding protein A9), which inhibits DC differentiation and promotes MDSC accumulation in mouse tumors (20). Upregulation of genes enhancing the immune response was observed with increased expression of genes mediating T-cell activation and chemotaxis or associated with the synthesis of proinflammatory cytokines (Fig. 4B). Moreover, ANG-2/TGF-β treatment induced downregulation of genes related to the maturation of monocytes into macrophages and, conversely, upregulation of genes encoding for markers of a promyelocytic stage. In addition, genes encoding for DC markers showed upregulated expression (Fig. 4B). Some of these genes have been previously reported to be upregulated upon in vitro maturation of DCs following exposure to lipopolysaccharide and interferon-γ (21, 22). Consistent with this observation, the expression of genes related to antigen processing and presentation, which are specific functions of APCs, were also upregulated (Fig. 4B). These data indicate that TEMs are plastic cells that shift their gene expression profile toward one resembling myeloid DC in response to treatments that impair their proangiogenic function.

In order to validate this phenotypic reversion, we monitored by flow cytometry the expression of markers of DC differentiation and activation in TEMs treated with ANG-2/TGF-β or RTKI. Relative to untreated TEMs, short and longer exposure to RTKI treatment resulted in significant (P < 0.05) increase in the expression of antigen-presenting molecules HLA-DR, integrin CD61, as well as costimulatory molecules CD80 and CD86. Moreover, CD11c expression was significantly (P < 0.05) induced by RTKI but not ANG-2/TGF-β treatment, whereas CD14 expression was strongly reduced by both treatments (Fig. 4C). These results confirmed the phenotypic change toward a more myeloid-DC-like phenotype that we observed in the gene expression profiling experiment. Furthermore, in order to test whether these treated TEMs could have altered their immunosuppressive function, we cocultured them with CFSE-labeled autologous T cells that had been previously stimulated with anti-CD3, anti-CD28 antibodies and IL-2 for 24 hours and rested 5 days (see Materials and Methods). In vitro differentiated TEMs treated with ANG-2/TGF-β or RTKI significantly increased the frequency of proliferating T cells relative to untreated TEMs (Fig. 4D). A stronger effect was seen with RTKI compared with ANG-2/TGF-β treatment (42% and 57% CFSE^low T cells respectively, compared with 31% CFSE^low T cells with untreated TEMs; Fig. 4D).

Thus, our data confirm that TEMs differentiated in vitro are plastic cells that can be induced to change from a proangiogenic immunosuppressive phenotype to one of immunocompetent cells resembling myeloid DCs. This phenotypic reversion was induced by modulating pathways driving the proangiogenic function of TEMs, indicating that the angiogenic and immunosuppressive functions of TEMs are probably linked.

In breast tumors, TEMS and CD11c⁺ DCs express markers of APCs

The reversion of the suppressive phenotype that we observed for TEMs differentiated in vitro in response to antiangiogenic treatments (Fig. 4) prompted us to examine whether TEMs and CD11c⁺ DCs display overlapping phenotypes in breast tumors. To this end, we characterized TEM and DC populations by confocal microscopy in sections of frozen breast tumor. Our results showed that TEMs were CD14⁺, CD11c⁻, TIE-2⁺, VEGFR-1⁺ CD31⁻, whereas CD11c⁺ DCs were CD14⁻, CD11c⁺, TIE-2⁻, VEGFR-1^low CD31⁻, and CD141⁻ (Supplementary Fig. S2). Both TEMs and DCs expressed APC markers, such as HLA-DR, CD80, CD86, and CD1a (Fig. 5A), consistent with their ability to function as APCs (Figs. 1A, 3A, and 3B). Quantification of confocal microscopy images (see Materials and Methods) indicated that TEMs expressed relatively higher levels of HLA-DR, CD86 [MFI CD86 (TEMs) = 93.3 ± 3.7; MFI CD86 (CD11c⁺ cells) = 28.4 ± 8.5; P < 0.01] and CD1a [MFI CD1a (TEMs) = 20.5 ± 12.1; MFI CD1a (CD11c⁺ cells) = 11.2 ± 6.9; P < 0.05] relative to CD11c⁺ DCs, but comparable [MFI CD80 (TEMs) = 57.2 ± 21.4, MFI CD80 (CD11c⁺ cells) = 56.6 ± 14.1] expression levels of CD80 (Fig. 5A). Thus in human breast tumor, while TEMs and CD11c⁺ DC populations show no detectable overlap of CD14 and CD11c expression, they share, to various extents, the APC markers HLA-DR, CD1a, CD80, and CD86 (Fig. 5A). Finally, in contrast to CD11c⁺ DCs, and consistent with their suppressive function (Fig. 1), we measured in confocal microscopic images that TEMs expressed a 34-fold higher expression level of arginase-1, a marker associated with MDSCs in the mouse, than did CD11c⁺ DCs (P < 0.05; ref. 3; Fig. 5A). These data suggest that, in breast tumor, CD11c⁺ DCs show an APC phenotype, whereas TEMs display rather a phenotype of proangiogenic suppressive APCs.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

In breast tumors, TEMs and CD11c⁺ DC express markers of APC, and TEMs alter the function and the maturation of CD11c⁺ DC. A, sections of frozen breast tumor tissues were examined by confocal microscopy for the expression of HLA-DR, CD80, CD86, CD1a, and arginase-1 and representative images from 6 tumors and 57 images per staining are shown (insert, high magnification). B, DCs display heterogeneous expression of CD11c, with HLA-DR expression significantly lower in CD11c^low cells (<40% of maximal CD11c normalized intensity) relative to CD11c^high cells (≥40% of maximal CD11c normalized intensity). C, CD11c expression levels were quantified (A.U. arbitrary units) in tumor zones containing increasing rates of TEM infiltration. Quantification and analysis of confocal microscopic images were carried out as described in Materials and Methods. **, P < 0.01. Scale bar, 25 μm.

TEMS impair CD11c⁺ DC maturation in breast tumors

Because tumor TEMs suppressed tumor-specific T-cell responses mediated by tumor CD11c⁺ DCs (Fig. 1), we next examined whether TEMs may directly alter the ability of DCs to present antigen by lowering their HLA-DR expression levels. By quantifying confocal microscopic images of breast tumor sections (see Materials and Methods), we observed that the CD11c⁺ DC population was heterogeneous and encompassed a subset of CD11c^low and CD11c^high cells. The expression levels of CD11c were normalized in each patient to the maximal CD11c expression detected (100%) and CD11c⁺ DC was classified into subsets of CD11c^low (<40% maximal CD11c intensity) and CD11c^high (≥ 40% maximal CD11c intensity) cells. Results show that HLA-DR expression was significantly higher (P < 0.01) in CD11c^high cells relative to CD11c^low cells (Fig. 5B). Importantly, CD11c expression levels were strongly dependent on the local rate of TEM infiltration. Using immunofluorescence images, we defined the rate of TEM infiltration relative to CD11c⁺ cells in a tumor zone as follows: [TEMs surface area/(TEMs + CD11c⁺ cells) surface area] * 100. We measured a significant reduction (P < 0.01) of CD11c expression with increasing rates of TEMs infiltration (Fig. 5C), indicating that TEMs impaired DC function and maturation by reducing their HLA-DR expression.

In breast tumors, high levels of CD86 expression are required for TEMs to engage Treg cells and to suppress tumor-specific immune responses

In order to understand further how TEMs suppress T-cell responses in tumor, we examined the distribution of TEMs and T cells in frozen sections of breast tumor tissues. By analyzing images of the sections, we could estimate that 60% of infiltrating T cells were in close proximity and potentially engaged by TEMs and CD11c⁺ cells. Surprisingly, while TEMs were 2.6 (±0.3)-fold less abundant than CD11c⁺ DCs, TEMs engaged 80% (±6.8%) of T cells compared with 27% (±12%) for CD11c⁺ DCs (Fig. 6A). Importantly, we observed that T cells were 1.6-fold less abundant in tumor zones containing TEMs, relative to TEMs-depleted zones. Furthermore, the vast majority of T cells (>71%) in TEMs-containing zones expressed the Treg marker CD25 and were preferentially (>92%) engaged by TEMs (Fig. 6B). We observed that, in breast tumor, Treg (CD3⁺CD25⁺ FOXP3⁺ cells) accounted for by 39% (±7%) of infiltrating T cells and were enriched in tumor zones containing TEMs where they massively (92% ±7%) form multiple tight conjugates with TEMs (Fig. 6B) but not with CD11c⁺ cells (7% ± 4%).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

In breast tumors, TEMs express high levels of CD86, thus ensuring engagement of regulatory T cells and their suppressive function. A, sections of frozen breast tumor tissues were examined by confocal microscopy for T-cell–TEM and T-cell–CD11c⁺ cell conjugates and the percentage of T cells engaged by TEMs or DCs was quantified from 117 images from 8 distinct tumors; scale bar, 25 μm. B, percentage of T cells, Treg (CD3+CD25+ cells) among T cells, and Treg–TEM conjugates among T-cell–TEM conjugates were examined in TEM-depleted (−) and TEM-containing (+) tumor zones. 98 images from 10 distinct tumors were examined; scale bar, 5 μm. C, CD86 to CD80 expression ratio was calculated in 126 confocal microscopic images from 7 distinct tumors (top). Tumor-specific proliferation of T cells (as measured in Fig. 4D) in response to CD86-specific blocking antibodies (bottom). D, changes in the expression levels of CD86 at the surface of tumor TEM measured by flow cytometry following treatment with RTKI. Cumulative data from 5 independent experiments and 5 distinct patients are shown in C, bottom and in D. *, P < 0.05; **, P < 0.01.

The immune response to tumor is often ineffective due to the downmodulated expression of HLA-DR and costimulatory ligands CD80 and CD86 on the surface of the APC, which renders tumor-specific T cells anergic (23). However, we measured by confocal microscopy that CD80 expression was comparable in TEMs and CD11c⁺ DCs while TEMs display a 3.4 (±1.1, P < 0.01)-fold increase in CD86 expression levels relative to CD11c⁺ cells. Consequently, the ratio of CD86 to CD80 expression was 3.3 higher in TEMs as compared with CD11c⁺ DCs (Fig. 6C, top). To test whether excessive CD86 engagement may support the suppressive function of TEMs, we treated TEMS with CD86-blocking antibodies before coculturing them with CD11c⁺ DCs and CFSE-labeled autologous T cells. We observed that blocking CD86 engagement increased the T-cell proliferation index to the same extent as RTKI treatment (compare Figs. 6C, bottom, and 2C) indicating that TEMs suppressed tumor-specific T-cell proliferation by excessive CD86 engagement. Interestingly, RTKI treatment decreased significantly the expression of CD86 at the surface of tumor TEMs (Fig. 6D). Furthermore, using specific antibodies, we observed that blocking CD86 engagement impaired, to a larger extent, TEMS-induced conversion of T cells into Treg relative to blocking CD80 engagement (Fig. 3D). These data indicate that CD86 contributes largely to this process. Taken together, our results suggest that the basal kinase activity of TIE-2 and VEGFR of tumor TEMs (Fig. 2A) supports the suppressive function of TEMs by maintaining high levels of CD86 expression at their surface (Fig. 6C, top, and 6D), resulting in massive conversion of T cells into Treg (Fig. 3D) and extensive engagement of Treg cells (Fig. 6B).