Antiangiogenic therapies that target the VEGF pathway have been used clinically to combat cancer for over a decade. Beyond having a direct impact on blood vessel development and tumor perfusion, accumulating evidence indicates that these agents also affect antitumor immune responses. Numerous clinical trials combining antiangiogenic drugs with immunotherapies for the treatment of cancer are ongoing, but a mechanistic understanding of how disruption of tumor angiogenesis may impact immunity is not fully discerned. Here, we reveal that blockade of VEGF-A with a mAb to VEGF augments activation of CD8+ T cells within tumors and potentiates their capacity to produce cytokines. We demonstrate that this phenomenon relies on the disruption of VEGFR2 signaling in the tumor microenvironment but does not affect CD8+ T cells directly. Instead, the augmented functional capacity of CD8+ T cells stems from increased tumor hypoxia that initiates a hypoxia-inducible factor-1α program within CD8+ T cells that directly enhances cytokine production. Finally, combinatorial administration of anti-VEGF with an immunotherapeutic antibody, anti-OX40, improved antitumor activity over single-agent treatments. Our findings illustrate that anti-VEGF treatment enhances CD8+ T-cell effector function and provides a mechanistic rationale for combining antiangiogenic and immunotherapeutic drugs for cancer treatment.

Antiangiogenic drugs disrupting the VEGF pathway are utilized in clinic for the treatment of several types of cancers. VEGF-A promotes tumor angiogenesis and exerts its biological activity by binding specific VEGFRs, namely VEGFR1 (FLT-1) and VEGFR2 (FLK-1; ref. 1). VEGF-A stimulates endothelial mitogenesis and promotes endothelial cell survival and migration through VEGFR2 signaling (1). Consequently, disruption of VEGFR2 signaling by pharmacologic inhibition of its kinase activity or by blocking interactions with its ligand, VEGF-A, with mAbs inhibits new blood vessel growth, thereby starving tumors of necessary nutrients and oxygen leading to increased hypoxia and tumor cell death (2).

In addition to its antiangiogenic activity, VEGF-A blockade improves antitumor immune responses (3). Several clinical trials are evaluating the combination of bevacizumab (a VEGF-A mAb) with immunotherapies including mAbs that block the T-cell inhibitory receptor pathways, cytotoxic T-lymphocyte–associated protein-4 (anti–CTLA-4), and programmed cell death protein 1 (anti–PD-1/-PD-L1; ref. 4). Although bevacizumab treatment in combination with immune-modulating agents has clinical activity, the mechanism by which VEGF blockade enhances immunotherapeutic efficacy is unclear. This combinatorial benefit may stem from the compound effect of eliciting independent mechanisms, but bevacizumab also may elicit immunomodulatory responses that support immunotherapeutic activity. In favor of this latter premise, pathologic amounts of VEGF-A can mediate immunosuppression by several mechanisms, including: (i) inhibiting T-cell functionality (5–8), (ii) blunting dendritic cell maturation (9), (iii) recruiting immunosuppressive cells to tumor (10–14), and (iv) disrupting leukocyte adhesion to endothelium (15–17).

The effect of VEGF blockade on tumor vasculature can result in changes that impact oxygen tension in tumors leading to hypoxia (18), to which T cells must adapt. In human and murine T cells, hypoxia can lead to a heightened effector state characterized by increased cytolytic activity and production of IFNγ (19–27) with changes in IFNγ being inconsistent across studies. This heightened effector state is attributed to hypoxia-induced stabilization of HIF1α, a transcription factor that regulates multiple cellular processes including glycolysis, angiogenesis, cell survival, and others (22, 24, 25, 28).

Here, we demonstrate using murine tumors models that treatment with anti-VEGF is associated with improved CD8+ T-cell functionality as a result of increasing hypoxia within the tumor microenvironment. Aggravated hypoxia mediated by disruption of VEGFR2 signaling within the tumor enhanced HIF1α activity within CD8+ T cells, augmenting their effector response. Combining anti-VEGF therapy with an agonist antibody against the costimulatory receptor OX40 improved antitumor activity compared with single-agent administration. Collectively, our findings demonstrate that CD8+ T cells can adapt and augment their functionality within increasingly hypoxic conditions in response to inhibition of angiogenesis. Our results provide an explanation for the outcome of ongoing clinical trials and a mechanistic rationale for antiangiogenic and immunotherapeutic combinations for cancer treatment.

Mice and heterotopic murine tumor models

All studies were conducted in accordance with the American Association of Laboratory Animal Care guidelines. Experimental studies were conducted with approval of the Institutional Animal Care and Use Committee at Genentech. Female BALB/c mice (6–8 weeks old, IMSR catalog no. CRL:028, RRID:IMSR_CRL:028), C57Bl/6 (IMSR catalog no. CRL:027, RRID:IMSR_CRL:027), and DBA-2 (IMSR catalog no. CRL:026, RRID:IMSR_CRL:026) were purchased from Charles River Laboratory and housed in pathogen-free conditions at Genentech. The following tumor cell lines were used for in vivo studies: colon carcinoma (CT26, ATCC catalog no. CRL-2638, RRID:CVCL_7256), skin melanoma (Cloudman, ATCC catalog no. CCL-53.1, RRID:CVCL_4512 and B16.F10, ATCC catalog no. CRL-6475, RRID:CVCL_0159), mammary adenocarcinoma (JC, ATCC catalog no. CRL-2116, RRID:CVCL_3530), and mammary carcinoma (EMT6-luciferase, Genentech, Inc.; originally obtained from ATCC, ATCC catalog no. CRL-2755, RRID:CVCL_1923). Cell lines were grown and expanded at Genentech and were screened using MycoAlert (Lonza, catalog no. LT07-318) the passage before inoculation into animals. Cell lines utilized were expanded for no more than 25 passages. All cell lines were authenticated at Genentech by next-generation sequencing. For survival analysis, mortality was based on mouse removal from study once tumor volumes achieved ≥1,500 mm3. Mice that were withdrawn from study due to ulcerations at the tumor site and had tumor volumes <1,500 mm3 were not included in the analysis. Percent change in tumor volume was calculated on the basis of the size of the tumor at the time of caliper measurement divided by tumor size when mice were enrolled on study multiplied by 100.

Tumor inoculations and treatments

Tumor cells were mixed in a solution of 50% BD Matrigel (BD Biosciences) and 50% Hank's Balanced Salt Solution at a density of 1 × 106 cells/mL. One-hundred microliters (1 × 105 tumor cells) were inoculated subcutaneously in the right flank of mice for CT26, B16.F10, and Cloudman tumors. For EMT6 and JC tumors, 100 μL (1 × 105 tumor cells) was inoculated in the left fifth mammary fat pad. Unless specified elsewhere, mice were randomly grouped out into cohorts of equivalent mean tumor volumes of approximately 150 mm3. Mice were dosed intraperitoneally with 5.0 mg/kg of anti-mouse VEGFA (B20.4.1.1, Genentech, Inc.) or a corresponding isotype control antibody specific for GP120 (Genentech, Inc.), a human immunodeficiency virus envelope glycoprotein, intraperitoneally twice a week, for 2 weeks. For blockade of VEGFRs, CT26 tumor–bearing mice were treated with a DANG-mutated anti-VEGFR1 (30 mg/kg twice a week, i.p., clone 9E8, Genentech, Inc.) or rat IgG1 anti-VEGFR2 (30 mg/kg twice a week, i.p., DC101, BioXCell). Corresponding isotype, species-specific, and mutant antibodies against GP120 were used as controls. Tumors were harvested between days 13 and 15 after treatment initiation unless specified elsewhere.

For combinatorial studies with anti-OX40, CT26, or B16.F10 tumor-bearing mice were grouped once tumor volumes ranged between 150 and 300 mm3. CT26 tumor–bearing mice were treated with anti-VEGF or anti-GP120 as described above, and anti-OX40 (clone OX-86, Genentech, Inc.) or corresponding isotype control anti-GP120 (0.1 mg/kg day 1 only, i.v.). B16.F10 tumor–bearing mice were treated with anti-VEGF or anti-GP120 as stated above and anti-OX40 (clone OX-86, Genentech, Inc.) or corresponding isotype control anti-GP120 (10 mg/kg three times a week, first dose i.v. and all subsequent doses intraperitoneal, Genentech, Inc.). For CD8+ T-cell depletion, 40 mg/kg of anti-CD8 (clone 53-6.7, BioXCell) or corresponding isotype control was administered intraperitoneally. For untreated mice, antibodies were administered on days 1, 3, 7, and 14. For anti-VEGF–treated mice, antibodies were administered on days 7, 9, 14, and 21 post-anti-VEGF treatment initiation. Control and anti-VEGF–treated mice were treated with 1 mg/kg of chetomin (Tocris) once a day for 3 days prior to takedown starting at 10 days post-initial antibody treatment. Chetomin was solubilized in DMSO for a stock concentration then diluted 1:10 in PBS prior to administration of 200 μL i.p. For all in vivo studies, tumor volumes were measured at least twice per week unless specified elsewhere. Tumor volumes were calculated as length × width2 × 0.5 using calipers.

RNA isolation and real-time PCR

Total RNA was purified from tumor samples pulverized into a homogeneous powder under frozen conditions using RNeasy Mini Kit (Qiagen). RNA (1 μg) was reverse transcribed using a high capacity cDNA Transcription Kit (Applied Biosystems). β-actin was the internal control gene for normalization and fold change of gene expression between control- and anti-VEGF–treated groups was calculated with the ΔΔCt method. TaqMan probes purchased from Life Technologies (Supplementary Table S1) were used for real-time PCR. CD8+ T cells were sorted directly into RLT buffer and RNA was extracted (kit manufacturer). Fluidigm analysis on sorted CD8+ T cells was performed using Biomark HD using 96.96 chips.

Flow cytometry

Tumor, lungs, spleen, and tumor-draining lymph nodes (tdLN) were harvested from mice for flow cytometric analysis. Tumors were processed as described previously (29). Lung single-cell suspension was generated using the MACS Tissue Dissociator and Lung Dissociation Protocol (Miltenyi Biotec) and processed similar to tumors. Spleen and tdLNs were disrupted through a 100-μm strainer and red blood cells were lysed. Cells were stained with Fixable Viability Dye eFluor 780 (eBioscience) and Fcγ receptors were blocked with anti-CD16 and anti-CD32 cocktail (Clone: 93; eBioscience). A cocktail of fluorochrome-conjugated antibodies (Supplementary Table S2) was added to the cells for 30 minutes at 4°C. Cells were then washed twice with FACS buffer and fixed using 4% PFA for 10 minutes at room temperature. For intracellular staining, cells were fixed and permeabilized using the FoxP3 Transcription Factor Fixation/Permeabilization Buffer (eBioscience) prior to intracellular staining according to the manufacturer's instructions (Supplementary Table S2). Cells were analyzed using LSRII and Fortessa Flow Cytometers (BD Biosciences) and data analyzed using FlowJo Software (Tree Star).

Histology, image acquisition, and analysis

Tumors were bisected at midline. Paraffin-embedded sections were fixed overnight with 4% paraformaldehyde and processed using the Sakura Tissue-TEK VIP machine, embedded in Paraplast Embedding Medium (McCormick Scientific), and sectioned at 7 μm using a Leica Microtome. Fresh frozen tumors were embedded in Optimal Cutting Temperature Compound OCT (Tissue Tek, 4583), and sectioned at 20 μm using a Cryostat (Leica Model CM3050S). Tissues were permeabilized in a PBS solution supplemented with 0.5% Triton X-100, 0.05% Tween-20, and 10% goat serum. Primary antibodies MECA-32 (Becton Dickinson, 553849) and CD3 (Clone SP6, Thermo Fisher Scientific) were incubated at room temperature for 1 hour. Tumor sections were mounted with Fluoromount-G (SouthernBiotech 0100-01). For assessment of hypoxia, pimonidazole hydrochloride (0.1 mg/mouse, Hypoxyprobe Inc) dissolved in 0.9% sterile saline solution was injected intravenously 1 hour prior to euthanasia. Following euthanasia, mice were perfused with PBS solution followed by 4% paraformaldehyde before explantation of tumors.

Images (magnification, 20×) of the entire tumor section were acquired using the automated Tissue Gnostics slide scanner system with a Zeiss Imager.Z1 microscope (http://www.tissuegnostics.com) or Panoramic Scan 150 digital slide scanner (http://www.3dhistech.com/pannoramic_scan). Both slide scanners utilized a Carl Zeiss Plan-Apochromat 20×/N.A. 0.8 objective. Image data were analyzed using custom algorithms developed in an object-oriented image analysis software (Definiens Developer XD, RRID:SCR_014283). For all image analysis, quantification was performed in viable tumor regions of the tissue section. Necrotic areas were excluded on the basis of nuclear morphology and defined as areas with small-punctate nuclei and/or areas of low cell density. Within viable tumor regions, individual cell objects were separated using a watershed segmentation routine applied to the DAPI channel. For analysis of MECA-32, FITC background signal was equalized using a top-hat filter. Total MECA-32+ vascular areas were segmented and normalized to viable tumor area. For T-cell quantification within hypoxic areas, AF488 fluorescent intensity was measured within the cell object boundaries to classify CD3+ cells. Hypoxic regions within viable areas were segmented by positive DyLight 594 fluorescent intensity. Four concentric zones were defined around the edge of the hypoxic regions. The depth of each of zones 1, 2, and 3 was set to approximately 52 μm (equivalent to ∼5 cell bodies). Zone 1 regions were immediately adjacent to hypoxic regions, whereas zone 4 regions represented distances greater than 156 μm from the hypoxic edge. CD3+ cells were counted within each zone.

T-cell stimulation and analysis of cytokine production

A total of 3 × 106 cells from tumors were stimulated with phorbol 12-myristate 13-acetate (PMA) and ionomycin (eBioscience) for 5 hours. For analysis of antigen-specific CD8+ T-cell stimulation, CT26 tumor single-cell suspensions were cultured in the presence of the CT26 H-2Ld-restricted immunodominant peptide AH1 (SPSYVYHQF; ref. 30) or the negative control H-2Ld–restricted NP118 peptide from the nucleoprotein of lymphocytic choriomeningitis virus (ref. 31; control, RPQASGVYM) at 2 μg/mL for 5 hours. Stimulations were performed in the presence of brefeldin A. Cells were stained intracellularly as described above (Supplementary Table S2).

For bone marrow–derived dendritic cell (BMDC) coculture assays, bone marrow cells obtained from mouse tibiae and femurs were resuspended in culture medium consisting of RPMI1640 supplemented with 10% FBS and 10% conditioned medium from GM-CSF–transduced B16 cells. Bone marrow cells (7 × 106) were seeded in 100 × 20 mm nontreated cell culture plates in 10 mL of culture medium and incubated for 9 days at 37°C at 5% CO2. Five milliliters of fresh culture medium was added to the plates on days 5 and 7. BMDCs were pulsed with AH1-peptide for 1 hour at 37°C. AH1-peptide–pulsed BMDCs were plated in a 96-well flat-bottom plate at a concentration of 1 × 105 cells per well. A total of 3 × 104 CD8+ T cells sorted from individual CT26 tumors were added to wells containing AH1-peptide–pulsed BMDCs. Cocultures were incubated for 37°C for 5 hours in 5% CO2. Following incubation, samples were processed for flow cytometry.

For antigen-presenting cell ex vivo stimulation assays, tumors and tdLNs were harvested from control- and anti-VEGF–treated mice on day 13 posttreatment. For myeloid cell stimulations, CD11b+/CD45+/Thy1.2/viability dye cells were purified from tumor single-cell suspensions by FACS. 50K myeloid cells were cocultured with 50K CD8+ T cells isolated from pooled tdLNs of the same mice by CD8-negative selection using Magnetic Bead Separation (Miltenyi Biotec). For dendritic cell (DC) stimulations, CD11c+/MHC class II+/CD45+/Thy1.2/F480/viability DCs and CD8+/Thy1.2+/CD45+/CD11b cells (CD8+ T cells) were isolated from tumor single-cell suspensions. 30K purified DCs were then cocultured with 100K sorted CD8+ T cells. Cocultures were plated in the presence of 2 μg/mL of AH1 peptide and GolgiStop (BD Biosciences) for 5 hours at 37°C at 5% CO2. Cells were then processed for surface and intracellular staining as described previously.

For experiments under physioxic (5% O2) and hypoxic conditions, a hypoxia chamber was utilized according to the manufacturer's instructions (Hypoxia Incubator Chamber, Stemcell Technologies). All media and reagents used in these experiments were incubated in a hypoxia chamber (1% O2) for 48 hours prior to the experiment. In brief, stimulation cultures were set up according to methods outlined above. In some experimental conditions, 10 μmol/L BAY 87-2243 or DMSO alone was added to tumor single-cell suspensions. Chambers were assembled according to the manufacturer's instructions and 1% or 5% O2 gas mixture was circulated through the chambers. O2 concentration was monitored within the chambers with O2 Monitors (PAC7000 O2 Monitor, Drager). Chambers were sealed and incubated for 6 hours at 37°C.

Western blot analysis

Whole-cell lysates from MS1 cells, lung endothelial cells, and CD8+ T cells from tumors and spleens were separated by NuPAGE 4%–12% Bis-Tris Gel (Invitrogen) and transferred to a polyvinylidene difluoride membrane using iBlot Dry Blotting System according to the manufacturer's instructions (Invitrogen). The membrane was washed three times with 5% nonfat milk in TBST (10.0 mmol/L Tris, pH 8.0, 150.0 mmol/L NaCl, 0.1% Tween 20) and incubated with anti-VEGFR2 (Cell Signaling Technology, Clone 55B11) at 1:1,000 for 1 hour. The membrane was washed three times for 10′ with TBST and incubated with ECL HRP-conjugated rabbit IgG (Amersham) at 1:10,000 for 1 hour. The membrane was washed three times for 10′ with TBST and developed with ECL Plus Western Blotting Detection Reagents (Amersham). Membrane was stripped with Restore Plus Western Blot Stripping Buffer (Thermo Fisher Scientific) and probed using a β-Actin Antibody (Novus) at 1:5,000 for a sample loading control using the aforementioned conditions.

Statistical analysis

For all figures, statistical analyses were performed using GraphPad Prism version 7.0e (http://www.graphpad.com/). Unless specified elsewhere, 5–15 mice were used for experiments to collect biologically meaningful data for statistical purposes based on previous studies using these tumor models. Mouse number variation between studies was due to exclusion of mice prior to study initiation based on tumor sizes (<100 or >250 mm3) to provide consistency between experiments. Parametric assumptions of normality and equal variances were checked using D'Agostino and Pearson Omnibus test and by determining SD equivalency between experimental groups, respectively. Unpaired Student t test was used to compare groups when data met parametric assumptions and Mann–Whitney test was used to compare groups when parametric assumptions were not met. For unpaired t test, Welch correction was used if SDs were not equivalent (>10% variation) between groups. For datasets with sample size too small for D'Agostino and Pearson Omnibus normality test, normal distribution was assumed on the basis of the appearance (Gaussian distribution) of the data. All statistical tests used above were two-tailed. Measurements for tumor growth studies were conducted under partially blinded conditions. Statistical differences for mouse survival were calculated with the Gehan–Breslow–Wilcoxon test. For all graphs: *, P = 0.05; **, P = 0.005; ***, P = 0.0005.

Anti-VEGF–induced gene expression inconsistent with lymphocyte infiltration

Observations of human tumors suggest that upregulation of adaptive immune gene signatures and high densities of tumor-resident lymphocytes correlate with improved immunotherapeutic clinical outcome (32). Cloudman, but not CT26, tumors from anti-VEGF–treated mice had significantly increased IFNγ (IFNG), TNFα (TNF), granzyme-B (GZMB), and perforin (PRF1) gene expression (Fig. 1A and B) approximately 2 weeks after initiation of treatment compared with control-treated mice. Increased proinflammatory gene expression observed with anti-VEGF treatment was not associated with tumor lymphocyte accumulation. Natural killer (NK) cell, CD4+ (FoxP3) T-cell, and CD8+ T-cell numbers were not significantly different in Cloudman tumors after anti-VEGF treatment (Fig. 1C; Supplementary Fig. S1A), whereas CT26 tumors had a significant reduction in accumulation of immune cell populations compared with control-treated mice (Fig. 1D). Three additional tumor models responsive to anti-VEGF treatment, B16.F10, EMT6, and JC, had inconsistent lymphocyte infiltration across models (Supplementary Fig. S1B–S1D), irrespective of consistent anti-VEGF–mediated tumor growth inhibition (Fig. 1E). The extent of CD8+ T-cell content did not significantly correlate with tumor size in either of the treatment groups among the five tumor models evaluated (Supplementary Fig. S1E). Thus, the effects of anti-VEGF treatment on lymphocyte accumulation were tumor-dependent and increased proinflammatory gene expression was not a product of increased lymphocyte content, but likely a result of enhanced lymphocyte function.

Figure 1.

Lack of concordance between effector gene signature and lymphocyte content. Tumors were harvested approximately 2 weeks posttreatment. qPCR for IFNG, TNF, GZMB, and PRF1 was performed on Cloudman (A) and CT26 (B) total tumor mRNA. n ≥ 4 mice/group. Flow cytometric analysis of NK cell, CD4+ (FoxP3) T-cell, and CD8+ T-cell numbers/g tumor for Cloudman (C) and CT26 (D) tumors. E, Tumor mass of five different tumor models. Circles depict individual mice, and horizontal lines represent the mean. Data representative of two to five independent experiments with 5 to 12 mice/group. If data were deemed normal, the unpaired t test or the unpaired t test with Welch correction was used if SD was similar or different between groups, respectively. If data did not pass normality, the Mann–Whitney test was used (*, P < 0.05; **, P < 0.005; ***, P < 0.0005).

Figure 1.

Lack of concordance between effector gene signature and lymphocyte content. Tumors were harvested approximately 2 weeks posttreatment. qPCR for IFNG, TNF, GZMB, and PRF1 was performed on Cloudman (A) and CT26 (B) total tumor mRNA. n ≥ 4 mice/group. Flow cytometric analysis of NK cell, CD4+ (FoxP3) T-cell, and CD8+ T-cell numbers/g tumor for Cloudman (C) and CT26 (D) tumors. E, Tumor mass of five different tumor models. Circles depict individual mice, and horizontal lines represent the mean. Data representative of two to five independent experiments with 5 to 12 mice/group. If data were deemed normal, the unpaired t test or the unpaired t test with Welch correction was used if SD was similar or different between groups, respectively. If data did not pass normality, the Mann–Whitney test was used (*, P < 0.05; **, P < 0.005; ***, P < 0.0005).

Close modal

VEGF blockade augments CD8+ T-cell effector function

CD8+ T cells contribute to antitumor immune responses and immunotherapeutic efficacy. Anti-VEGF treatment significantly reduced expression of the inhibitory receptors PD-1 and T-cell immunoglobulin and mucin domain-containing protein 3 (TIM3), and upregulated the costimulatory receptor OX40 on total CD8+ T cells from CT26 tumors (Fig. 2A) and on CT26 tumor antigen-specific CD8+ T cells identified by MHC class I–restricted (H-2Ld) tetramer staining for the immunodominant AH1-peptide (MuLVgp70423-431; Supplementary Fig. S2A). Similar to CT26 tumors, anti-VEGF treatment increased expression of OX40 on CD8+ T cells from B16.F10 and Cloudman tumors, with CD8+ T cells from the former also showing reduced expression of inhibitory receptors (Supplementary Fig. S2B and S2C). In addition, OX40 (TNFRSF4) expression from total tumor mRNA was elevated in tumors from anti-VEGF–treated mice consistent with increased cell-surface expression (Supplementary Fig. S2D). These phenotypic changes were consistent with improved T-cell functionality and prompted us to assess their effector function by means of cytokine production fitness.

Figure 2.

CD8+ T cells exhibit augmented effector function. Tumor single-cell suspensions approximately 2 weeks posttreatment were generated. A, Expression of PD-1, TIM3, and OX40 on CD8+ T cells from CT26 tumors. Data displayed above histograms represent average MFI ± SEM for total CD8+ T cells. CT26 tumor single-cell suspensions were stimulated with PMA and ionomycin (iono). B, Percentages of cytokine-producing CD8+ T cells. C, Representative dot plots illustrating IFNγ and TNFα expression within total CD8+ T cells. Numbers within the plots denote the average frequency ± SEM for given quadrants. D, Frequency of IFNγ+/TNFα+ CD8+ T cells from four additional tumor models. E, IFNγ MFI from IFNγ+ CD8+ T cells from CT26 tumors. CT26 tumor single-cell suspensions were stimulated ex vivo with AH1-peptide. F, Proportion of CD8+ T cells producing IFNγ and TNFα. G, Dot plots of total CD8+ T cells producing IFNγ and TNFα. Values within dot plots represent the mean frequency ± SEM for given quadrants. H, Frequency of AH1-tetramer+ (AH1-Tet+) CD8+ T cells expressing IFNγ or TNFα. Circles represent individual mice, and horizontal lines depict the mean. Data representative of two to five independent experiments with 5 to 12 mice/group. I, CT26 tumor volume percent change following CD8 depletion starting 1 week after anti-VEGF treatment. Time points reflect the average ± SEM. If data were deemed normal, the unpaired t test or the unpaired t test with Welch correction was used if SD was similar or different between groups, respectively. If data did not pass normality, the Mann–Whitney test was used (*, P < 0.05; **, P < 0.005; ***, P < 0.0005).

Figure 2.

CD8+ T cells exhibit augmented effector function. Tumor single-cell suspensions approximately 2 weeks posttreatment were generated. A, Expression of PD-1, TIM3, and OX40 on CD8+ T cells from CT26 tumors. Data displayed above histograms represent average MFI ± SEM for total CD8+ T cells. CT26 tumor single-cell suspensions were stimulated with PMA and ionomycin (iono). B, Percentages of cytokine-producing CD8+ T cells. C, Representative dot plots illustrating IFNγ and TNFα expression within total CD8+ T cells. Numbers within the plots denote the average frequency ± SEM for given quadrants. D, Frequency of IFNγ+/TNFα+ CD8+ T cells from four additional tumor models. E, IFNγ MFI from IFNγ+ CD8+ T cells from CT26 tumors. CT26 tumor single-cell suspensions were stimulated ex vivo with AH1-peptide. F, Proportion of CD8+ T cells producing IFNγ and TNFα. G, Dot plots of total CD8+ T cells producing IFNγ and TNFα. Values within dot plots represent the mean frequency ± SEM for given quadrants. H, Frequency of AH1-tetramer+ (AH1-Tet+) CD8+ T cells expressing IFNγ or TNFα. Circles represent individual mice, and horizontal lines depict the mean. Data representative of two to five independent experiments with 5 to 12 mice/group. I, CT26 tumor volume percent change following CD8 depletion starting 1 week after anti-VEGF treatment. Time points reflect the average ± SEM. If data were deemed normal, the unpaired t test or the unpaired t test with Welch correction was used if SD was similar or different between groups, respectively. If data did not pass normality, the Mann–Whitney test was used (*, P < 0.05; **, P < 0.005; ***, P < 0.0005).

Close modal

Anti-VEGF treatment resulted in a significantly greater frequency of CD8+ T cells producing IFNγ, TNFα (Fig. 2B), or both cytokines (Fig. 2C) compared with the control group in response to PMA and ionomycin stimulation. Similar findings were also observed in Cloudman, B16.F10, EMT6, and JC tumors (Fig. 2D). This effect was not due to tumor size, because the percentage of IFNγ+/TNFα+ CD8+ T cells was not significantly correlated with tumor mass in all tumor models (Supplementary Fig. S2E). In CT26 tumors, both the frequency of IFNγ-expressing CD8+ T cells and the abundance of IFNγ expression on a per-cell basis were augmented, as measured by mean fluorescence intensity (MFI; Fig. 2E). The extent of effector cytokine production by CD8+ T cells, particularly IFNγ, correlates with cytolytic activity (33). We observed that intratumoral CD8+ T cells from anti-VEGF–treated mice produced more granzyme-B compared with corresponding cells from control-treated mice (Supplementary Fig. S2F). Also, sorted CD8+ T cells from anti-VEGF–treated CT26 tumors exhibited a significant upregulation of PRF1, a gene that encodes perforin, which is essential for CTL activity (Supplementary Fig. S2G). Cytokine expression and elevated expression of CTL-associated molecules show that anti-VEGF treatment augments the functional phenotype of intratumoral CD8+ T cells.

VEGF blockade induced functional changes in an antigen-specific manner in CD8+ T cells from CT26 tumors exposed to AH1-peptide ex vivo, with a higher frequency of cells producing IFNγ, TNFα (Fig. 2F), or both cytokines simultaneously (Fig. 2G). The prevalence of AH1-peptide–specific CD8+ T cells in anti-VEGF–treated tumors was not significantly different from the control group (Supplementary Fig. S2H). The same was true in B16.F10 tumors as evaluated by GP100 tetramer staining (Supplementary Fig. S2I). A significantly higher frequency of IFNγ+ and TNFα+ cells was detected within the AH1-tetramer+ CD8+ T-cell population from anti-VEGF–treated mice in response to AH1-peptide exposure (Fig. 2H), demonstrating increased cytokine production by CT26 tumor–specific CD8+ T cells. To confirm that the enhanced activation observed was due to inherent in vivo changes to CD8+ T-cell biology and not due to enhanced intratumoral antigen-presenting cell activation within the stimulation cultures, sorted CD8+ T cells from treated CT26 tumors were cultured with AH1-peptide–pulsed BMDCs. A higher frequency of CD8+ T cells from tumors of anti-VEGF–treated mice produced IFNγ compared with control treatment, demonstrating that the enhanced functional phenotype is intrinsic to CD8+ T cells ex vivo (Supplementary Fig. S2J).

CD8+ T cells were depleted with antibody to assess their contribution to antitumor activity after anti-VEGF therapy. This is of particular interest within the CT26 tumor model because CD8+ T cells are activated by anti-VEGF treatment, although the number of tumor-infiltrating T cells was reduced (Fig. 1D). CD8+ T-cell depletion in untreated mice accelerated tumor growth, confirming that such cells elicit antitumor activity in the absence of anti-VEGF administration (Supplementary Fig. S2K). CD8+ T cells were depleted in tumor-bearing mice 1 week after the initiation of anti-VEGF treatment, which accelerated tumor growth (Fig. 2I; Supplementary Fig. S2L). Thus, CD8+ T cells contribute to CT26 tumor growth inhibition mediated by anti-VEGF treatment.

Treatment with anti-VEGF did not alter regulatory T cell (Treg) numbers in four of five tumor models (Supplementary Figs. S1A and S2M). Among the tumor models investigated, CT26 was the only model exhibiting an alteration in the CD8+ T cell to Treg ratio, illustrated by a significant reduction in response to anti-VEGF treatment (Supplementary Fig. S2N). Phenotypic markers of anti-VEGF–treated CT26-infiltrating Tregs associated with suppressive functions (CD25, CD39, CD73, CD103, GARP, and Helios) were unchanged, with the exception of a significant increase in the frequencies of CD73+ and GARP+ Tregs (Supplementary Fig. S2O–S2Q). The enhanced activation and functional status of CD8+ T-cell activation promoted by anti-VEGF treatment was thus unlikely to be due to alterations in the functional phenotype or prevalence of Tregs within tumors.

Enhanced CD8+ T-cell effector function is not due to myeloid cells or DCs

VEGF signaling induces the accumulation of immunosuppressive myeloid cells within tumors that can confer refractoriness to VEGF blockade (3, 13, 34–38). Three distinct populations of CD11b+ cells were identified within five tumor models that phenotypically resembled granulocytic myeloid-derived suppressor cells (gMDSC), monocytic MDSCs (mMDSC), and tumor-associated macrophages (TAM; Supplementary Fig. S3A). VEGF blockade resulted in a decrease in TAMs and an increase in gMDSCs and/or mMDSCs in three of five tumor models (Supplementary Fig. S3B). Tumor model–inconsistent alterations in the frequencies of phenotypic and functional markers (Supplementary Fig. S3C; Supplementary Table S3) within tumor-resident myeloid cell populations were noted following anti-VEGF treatment, suggesting changes in functional activity in some models. AH1-peptide–pulsed CT26 tumor-resident CD11b+ cells, from control or anti-VEGF–treated mice, stimulated AH1-tetramer+ CD8+ T cells from the dLNs of CT26 tumor–bearing mice similarly (Supplementary Fig. S3D and S3E). Although VEGF blockade altered the phenotypic characteristics of myeloid cells within some tumor models, these changes did not influence their CD8+ T-cell stimulatory capacity. Augmented tumor-resident CD8+ T cell functionality is likely independent of myeloid cells within tdLNs because there were few of these cells and they exhibited no change or inconsistent alterations in their frequency, numbers, and functional marker expression across all tumor models between control and anti-VEGF treatment (Supplementary Fig. S3F–S3H; Supplementary Table S4). Also, CD4+ and CD8+ T-cell content in the tdLNs from five separate tumor models were similar or reduced (Supplementary Fig. S3I) and numbers of tumor-specific cells within the tdLNs from CT26 and B16.F10 tumor–bearing mice were similar (Supplementary Fig. S3J) between control and anti-VEGF treatment. In addition, similar percentages of CD69+ and Ki67+ T cells were detected within tdLNs of control- and anti-VEGF–treated mice from three separate tumor models (Supplementary Fig. S3K and S3L). The frequency of IFNγ+ CD8+ T cells within the tdLNs of CT26 tumor–bearing mice were equivalent regardless of treatment following AH1-peptide challenge (Supplementary Fig. S3M). We conclude that augmented CD8+ T-cell function within the tumor in response to anti-VEGF treatment is not a result of differential activation of such cells within the tdLN.

Inhibition of the VEGF pathway can augment the stimulatory capacity of DCs, which may influence CD8+ T-cell activation within tumors (9, 39, 40). Two predominant populations of DCs, characterized as F480/CD11c+/MHC class II+ were detected within the five tumor models evaluated, CD11b+ and CD11bint (Supplementary Fig. S4A). These populations expressed little to no B220, PDCA-1, and CD8α, whereas the CD11bint population expressed CD103 (Supplementary Fig. S4B). VEGF blockade diminished the prevalence of CD11bint DCs in three of five tumor models, whereas the CD11b+ DC population was reduced in two of five tumor models (Supplementary Fig. S4C). CT26 tumors had similar percentages of PD-L1+ and CD86+ DCs, but an increased proportion of OX40L+ cells and greater MHC class II surface expression (Supplementary Fig. S4D and S4E), suggesting augmented activation following anti-VEGF treatment. Sorted CD8+ T cells from both treatment cohorts had a greater proportion of IFNγ+ cells when cocultured with sorted DCs from anti-VEGF–treated mice in the presence of AH1-peptide (Supplementary Fig. S4F and S4G). DCs from JC tumors exhibited similar increases in OX40L positivity and MHC class II expression (Supplementary Fig. S4H), however, this was not observed with tumor-resident DCs from B16.F10 and EMT6 tumors (Supplementary Fig. S4I and S4J). Considering enhanced CD8+ T-cell activation is consistent among all five tumor models, whereas DC functional marker expression is not, the effects of VEGF blockade on DC activation is likely not responsible for the consistent augmentation in CD8+ T-cell functionality observed, also supported in the CT26 model by a lack of correlation between ratios of DCs to CD8+ T cells or DC MHC class II expression with CD8+ T-cell IFNγ positivity (Supplementary Fig. S4K and S4L). In addition, anti-VEGF treatment did not alter DC content or MHC class II expression on DCs within the tdLN from four separate tumor models (Supplementary Fig. S4M–S4O). We conclude that the effects of anti-VEGF treatment on CD8+ T-cell activity occurs within the tumor and not is due to differential activation within the tdLN or modulation of the myeloid and DC compartments.

Disrupted VEGF/VEGFR2 signaling heightens effector function

VEGF-A binds to VEGFR1 with higher affinity, but reduced kinase activity, compared with VEGFR2, suggesting that it may act as a decoy receptor in some contexts (1). T cells can express both VEGFR1 and VEGFR2 on their surfaces (41–43) and therefore, may be affected by VEGF blockade. Mice bearing CT26 tumors were treated with anti-VEGFR1– or anti-VEGFR2–blocking antibodies to identify the receptor responsible for augmenting CD8+ T-cell function. Anti-VEGFR2, but not neutralization of VEGFR1 signaling, mimicked anti-VEGF treatment in reducing tumor growth (Supplementary Fig. S5A), enhancing the frequency of polyfunctional IFNγ+/TNFα+-expressing CD8+ T cells in response to PMA and ionomycin stimulation (Fig. 3A), and significantly increasing the frequency of IFNγ-expressing CD8+ T cells in response to AH1-peptide challenge (Fig. 3B). Thus, disruption of VEGF/VEGFR2 signaling is the primary mechanism driving enhanced CD8+ T-cell effector function mediated by anti-VEGF therapy.

Figure 3.

Disrupting VEGFR2 signaling enhanced CD8+ T-cell functionality. CT26 tumor–bearing mice were treated with anti-VEGF, anti-VEGFR1, or anti-VEGFR2 and tumors were harvested approximately 2 weeks after treatment. Tumor single-cell suspensions were stimulated with PMA and ionomycin. A, Representative dot plots demonstrating the frequency of IFNγ- and TNFα-producing CD8+ T cells. Average frequency ± SEM for IFNγ+/TNFα+ CD8+ T cells denoted within dot plots. B, Frequency of IFNγ+ CD8+ T cells following stimulation with AH1-peptide. Circles depict individual mice, and horizontal lines identify the mean. If data were deemed normal, the unpaired t test or the unpaired t test with Welch correction was used if SD was similar or different between groups, respectively. If data did not pass normality, the Mann–Whitney test was used. ***, P < 0.0005; NS, not significant. Data representative of two independent experiments with 5 to 10 mice/condition. C, VEGFR2 expression on CD8+ T cells (CT26, Cloudman, and B16.F10 tumors) and CD8+ T cells from tdLNs and spleens of CT26 tumor–bearing mice. Values represent the average ± SEM percentage of VEGFR2+ CD8+ T cells. Data representative of two to four independent experiments with 5 to 10 mice/group. D, Sorted CD8+ T cells from CT26 tumors (5 × 105 cells) or spleens (2 × 106 cells) probed for VEGFR2 by Western blot analysis. Sorted lung endothelial cells from CT26 tumor–bearing mice and the mouse endothelial cell line, MS1, were used as positive controls.

Figure 3.

Disrupting VEGFR2 signaling enhanced CD8+ T-cell functionality. CT26 tumor–bearing mice were treated with anti-VEGF, anti-VEGFR1, or anti-VEGFR2 and tumors were harvested approximately 2 weeks after treatment. Tumor single-cell suspensions were stimulated with PMA and ionomycin. A, Representative dot plots demonstrating the frequency of IFNγ- and TNFα-producing CD8+ T cells. Average frequency ± SEM for IFNγ+/TNFα+ CD8+ T cells denoted within dot plots. B, Frequency of IFNγ+ CD8+ T cells following stimulation with AH1-peptide. Circles depict individual mice, and horizontal lines identify the mean. If data were deemed normal, the unpaired t test or the unpaired t test with Welch correction was used if SD was similar or different between groups, respectively. If data did not pass normality, the Mann–Whitney test was used. ***, P < 0.0005; NS, not significant. Data representative of two independent experiments with 5 to 10 mice/condition. C, VEGFR2 expression on CD8+ T cells (CT26, Cloudman, and B16.F10 tumors) and CD8+ T cells from tdLNs and spleens of CT26 tumor–bearing mice. Values represent the average ± SEM percentage of VEGFR2+ CD8+ T cells. Data representative of two to four independent experiments with 5 to 10 mice/group. D, Sorted CD8+ T cells from CT26 tumors (5 × 105 cells) or spleens (2 × 106 cells) probed for VEGFR2 by Western blot analysis. Sorted lung endothelial cells from CT26 tumor–bearing mice and the mouse endothelial cell line, MS1, were used as positive controls.

Close modal

Despite existing studies indicating surface expression of VEGFR2 on CD8+ T cells, expression was undetectable on intratumoral CD8+ T cells from CT26, Cloudman, and B16.F10 tumors or on such cells from spleen and tdLNs from CT26 tumor–bearing mice by flow cytometry (Fig. 3C) using validated anti-VEGFR2 reagents (Supplementary Fig. S5B). Furthermore, VEGFR2 expression was undetectable by immunoblot of lysates from CD8+ T cells sorted from CT26 tumors and matching spleens of tumor-bearing mice (Fig. 3D). Various numbers of lung endothelial cells from these same mice or a mouse endothelial cell line (MS1) were used as positive controls in the Western blot analysis and were positive for VEGFR2 protein. These results indicate that VEGFR2 was undetectable on CD8+ T cells. Therefore, these cells are not direct targets of anti-VEGF treatment, demonstrating that their immunomodulation was secondary to other phenomena such as inhibition of angiogenesis.

Anti-VEGF treatment intensifies T-cell exposure to hypoxia

It has been proposed that anti-VEGF treatment normalizes dysfunctional tumor vasculature, improving tumor perfusion and oxygenation (44). Treatment with anti-VEGF significantly reduced mean vascular density (MVD) within CT26 tumors compared with tumors from control-treated mice (Fig. 4A). Reduced MVD was accompanied by a significant increase in hypoxia within viable tumor area measured by immunofluorescence detection of exogenously administered pimonidazole, a 2-nitroimidazole compound that forms covalent bonds with cellular macromolecules when oxygen is below 1.3% (Fig. 4B; ref. 45). When we assessed localization of T cells in respect to hypoxic tumor area within treatment cohorts, we found T cells in tumors from anti-VEGF–treated mice were more concentrated within hypoxic areas (Fig. 4C). In addition, T cells outside of hypoxic areas were in closer proximity to hypoxic regions within tumors from anti-VEGF–treated mice (Fig. 4D). These results infer a relationship between hypoxia and enhanced CD8+ T-cell effector function, suggesting adaptation to the anti-VEGF–induced hypoxic environment.

Figure 4.

Anti-VEGF treatment increases T-cell localization within hypoxic regions. Treated CT26 tumors were harvested approximately 2 weeks posttreatment and used for immunofluorescence imaging. A, Representative tumor section images stained with DAPI (blue) and MECA-32 (green) for identification of cell nucleus and vasculature, respectively (left). Images in grayscale for MECA-32 and DAPI (middle and right, respectively). Percent MVD of viable tumor area depicted in accompanying scatter plot (right graph). B, Pimonidazole hydrochloride (hypoxic area) staining as illustrated on representative whole-tumor composites (left). Percent hypoxia of viable tumor area (right). C, Normalized T-cell counts within pimonidazole-positive area. D, Zone classification of proximity to red fluorescent hypoxic area (left) within the representative image overlay (middle). Images in grayscale for pimonidazole and T-cell staining (right images, top and bottom, respectively). Zonal normalized T-cell counts (right). Circles depict individual mice, and horizontal lines represent the mean. Data representative of ≥2 independent experiments with 4 to 7 mice per group. Statistical analysis determined using the Mann–Whitney test (*, P < 0.05; **, P < 0.005).

Figure 4.

Anti-VEGF treatment increases T-cell localization within hypoxic regions. Treated CT26 tumors were harvested approximately 2 weeks posttreatment and used for immunofluorescence imaging. A, Representative tumor section images stained with DAPI (blue) and MECA-32 (green) for identification of cell nucleus and vasculature, respectively (left). Images in grayscale for MECA-32 and DAPI (middle and right, respectively). Percent MVD of viable tumor area depicted in accompanying scatter plot (right graph). B, Pimonidazole hydrochloride (hypoxic area) staining as illustrated on representative whole-tumor composites (left). Percent hypoxia of viable tumor area (right). C, Normalized T-cell counts within pimonidazole-positive area. D, Zone classification of proximity to red fluorescent hypoxic area (left) within the representative image overlay (middle). Images in grayscale for pimonidazole and T-cell staining (right images, top and bottom, respectively). Zonal normalized T-cell counts (right). Circles depict individual mice, and horizontal lines represent the mean. Data representative of ≥2 independent experiments with 4 to 7 mice per group. Statistical analysis determined using the Mann–Whitney test (*, P < 0.05; **, P < 0.005).

Close modal

Enhanced HIF1α activity supports CD8+ T-cell effector function

In response to hypoxia, HIF1α accumulates and dimerizes with HIF1β thereby initiating transcription required for adaptation to hypoxic conditions (46). In addition, HIF1α enhances CD8+ T-cell effector function in response to chronic antigen stimulation (22). Anti-VEGF treatment significantly increased the frequency of intratumoral HIF1α+ CD8+ T cells and HIF1α protein content per cell, indicative of protein stabilization (Fig. 5A and B). Sorted CD8+ T cells from CT26 tumors following anti-VEGF treatment showed a significant increase in HIF1α target genes that encode proteins of cellular glycolysis (Fig. 5C) and target genes not associated with glycolysis (Fig. 5D). CD8+ T cells from anti-VEGF–treated CT26 tumors contained a greater frequency of CD71+ and GLUT3+ cells, also HIF1α target genes (Fig. 5E and F). Thus, in response to anti-VEGF treatment, CD8+ T cells have increased amounts of transcriptionally active HIF1α protein.

Figure 5.

HIF1α activity enhanced effector function of CD8+ T cells. CT26 tumor single-cell suspensions approximately 2 weeks posttreatment were used. A, Frequency of HIF1α+ CD8+ T cells. B, Typical histogram for HIF1α staining within CD8+ T cells (left) with HIF1α MFI for HIF1α+ cells (right). C, Gene expression of sorted CD8+ T cells evaluated by qPCR using Fluidigm. Glycolysis-associated HIF1α target genes as log2 fold change normalized to control treatment. D, Expression of additional HIF1α target genes not associated with glycolysis. Box and whisker graphs plotted as minimum to maximum, with box representing 25th to 75th percentiles and line depicting the median. Representative histograms (left) and percentages (right) of CD8+ T cells expressing the HIF1α target genes CD71 (E) and GLUT3 (F). G, Representative dot plot demonstrating CD8+ T-cell OX40 and HIF1α expression (left) with HIF1α representative histograms in OX40+ and OX40 cells (right). Single-cell suspensions from anti-VEGF–treated mice were subjected to AH1-peptide stimulation. H, Representative topographic plot of IFNγ and HIF1α (left) with typical HIF1α histograms for IFNγ+ or IFNγ CD8+ T cells (right). Frequency of AH1-peptide–stimulated IFNγ+ CD8+ T cells under hypoxic conditions (1% O2) in the presence or absence of BAY 87-2243 (I) or stimulation under hypoxic or physioxic (5% O2) conditions (J). CT26 tumor–bearing control and anti-VEGF–treated mice received chetomin or vehicle 3 days prior to takedown. K, Percentage of CD71+ CD8+ T cells. L, Tumor single-cell suspensions were stimulated with AH1-peptide under normoxic conditions. Frequency of IFNγ+ CD8+ T cells. Data depict 5 to 6 mice per group (C and D). Circles depict individual mice, and horizontal lines illustrate the mean (A, B, E, and F and IL). Average frequency ± SEM of gated populations within dot plots and values within histograms depict the average MFI ± SEM for HIF1α (G and H). Data representative of two independent experiments with 4 to 15 mice per group. If data were deemed normal, the unpaired t test or the unpaired t test with Welch correction was used if SD was similar or different between groups, respectively. If data did not pass normality, the Mann–Whitney test was used (*, P < 0.05; **, P < 0.005; ***, P < 0.0005).

Figure 5.

HIF1α activity enhanced effector function of CD8+ T cells. CT26 tumor single-cell suspensions approximately 2 weeks posttreatment were used. A, Frequency of HIF1α+ CD8+ T cells. B, Typical histogram for HIF1α staining within CD8+ T cells (left) with HIF1α MFI for HIF1α+ cells (right). C, Gene expression of sorted CD8+ T cells evaluated by qPCR using Fluidigm. Glycolysis-associated HIF1α target genes as log2 fold change normalized to control treatment. D, Expression of additional HIF1α target genes not associated with glycolysis. Box and whisker graphs plotted as minimum to maximum, with box representing 25th to 75th percentiles and line depicting the median. Representative histograms (left) and percentages (right) of CD8+ T cells expressing the HIF1α target genes CD71 (E) and GLUT3 (F). G, Representative dot plot demonstrating CD8+ T-cell OX40 and HIF1α expression (left) with HIF1α representative histograms in OX40+ and OX40 cells (right). Single-cell suspensions from anti-VEGF–treated mice were subjected to AH1-peptide stimulation. H, Representative topographic plot of IFNγ and HIF1α (left) with typical HIF1α histograms for IFNγ+ or IFNγ CD8+ T cells (right). Frequency of AH1-peptide–stimulated IFNγ+ CD8+ T cells under hypoxic conditions (1% O2) in the presence or absence of BAY 87-2243 (I) or stimulation under hypoxic or physioxic (5% O2) conditions (J). CT26 tumor–bearing control and anti-VEGF–treated mice received chetomin or vehicle 3 days prior to takedown. K, Percentage of CD71+ CD8+ T cells. L, Tumor single-cell suspensions were stimulated with AH1-peptide under normoxic conditions. Frequency of IFNγ+ CD8+ T cells. Data depict 5 to 6 mice per group (C and D). Circles depict individual mice, and horizontal lines illustrate the mean (A, B, E, and F and IL). Average frequency ± SEM of gated populations within dot plots and values within histograms depict the average MFI ± SEM for HIF1α (G and H). Data representative of two independent experiments with 4 to 15 mice per group. If data were deemed normal, the unpaired t test or the unpaired t test with Welch correction was used if SD was similar or different between groups, respectively. If data did not pass normality, the Mann–Whitney test was used (*, P < 0.05; **, P < 0.005; ***, P < 0.0005).

Close modal

CD8+ T cells from CT26 tumors of anti-VEGF–treated mice that were positive for OX40 or IFNγ, expressed significantly more HIF1α than marker cells (Fig. 5G and H). Exposure of CD8+ T cells from tumors of anti-VEGF–treated mice to BAY 87-2243, an inhibitor of mitochondrial complex I that subsequently inhibits HIF1α activity (47), reduced HIF1α expression, confirming pharmacologic activity (Supplementary Fig. S6A). Exposure of CD8+ T cells from anti-VEGF–treated mice to BAY 87-2243 (Fig. 5I) or culturing in physioxic conditions (5% O2; Fig. 5J) significantly reduced the frequency of cells producing IFNγ in response to AH1-peptide challenge and significantly reduced HIF1α protein expression (Supplementary Fig. S6B and S6C). Reduced IFNγ production was not attributed to diminished cell viability (Supplementary Fig. S6D and S6E). The effects of BAY 87-2243 and physioxic conditions on IFNγ production were specific to CD8+ T cells from anti-VEGF–treated mice (Supplementary Fig. S6F and S6G).

To understand the in vivo relevance of HIF1α signaling in the context of anti-VEGF treatment, hypoxia-induced transcription was inhibited in CT26 tumor–bearing mice by blocking HIF1α/HIF2α interactions with coactivator p300 with chetomin (48). Short-term chetomin treatment significantly reduced CD45+ cell accumulation (Supplementary Fig. S6H), with reductions in most immune cell subtypes (Supplementary Fig. S6I), but not CD8+ T cells (Supplementary Fig. S6J). Chetomin treatment significantly reduced the percentages of CD8+ T cells expressing CD71 in both control and anti-VEGF–treated mice, demonstrating a pharmacodynamic effect on HIF1α activity (Fig. 5K). Treatment with chetomin reduced the frequency of CD8+ T cells expressing IFNγ within anti-VEGF–treated mice to numbers equivalent to control-treated mice (Fig. 5L), which was not because of reduced frequencies of AH1-specific CD8+ T cells (Supplementary Fig. S6K). CD8+ T cell/DC ratios remained unchanged (Supplementary Fig. S6L), and DCs expressed equivalent amounts of MHC class II and OX40L when compared with the vehicle control–treated counterparts (Supplementary Fig. S6M and S6N), suggesting that CD8+ T-cell activation was independent of DCs within the tumor. Markers associated with hypoxia-induced S-2-hydroxyglutarate production that can mediate enhanced T-cell functionality (49) were not increased in CD8+ T cells following anti-VEGF treatment (Supplementary Fig. S7). Thus, increased tumor hypoxia and augmented HIF1α activity within CD8+ T cells contributed to the enhanced effector function of such cells in response to anti-VEGF treatment.

Anti-VEGF treatment enhances anti-OX40 antitumor activity

CD8+ T cells exposed to hypoxic conditions may increase costimulatory receptor expression and stimulating these pathways can contribute to antitumor activity (22, 24, 50). We found an increase in OX40 expression on CD8+ T cells and elevated OX40L expression on DCs within some models suggesting this pathway may be operative. Combinatorial administrations of anti-VEGF and an agonist antibody against OX40 improved survival of CT26 tumor–bearing mice compared with treatment with either agent alone (Fig. 6A). For confirmation, combinatorial treatment was performed in the B16.F10 model, but due to the rapid tumor growth in this model, survival was not significantly affected (Supplementary Fig. S8). However, combinatorial treatment significantly inhibited tumor growth compared with single-agent administration (Fig. 6B). In summary, these results establish that VEGF blockade potentiates CD8+ T-cell effector function that contributes to the therapeutic effect of anti-VEGF treatment and can potentiate immunotherapeutic antitumor activity.

Figure 6.

Anti-VEGF treatment augments anti-OX40 therapy. CT26 tumor–bearing mice were administered anti-VEGF, anti-OX40 (OX86), and/or corresponding control antibodies. A, Survival graph (left) with individual mouse tumor growth curves (right). Statistical analysis for mouse survival were calculated with the Gehan–Breslow–Wilcoxon test. B16.F10 tumor–bearing mice were treated as stated above. B, Percent tumor volume change (left) with individual mouse tumor growth curves (right). Time points reflect average ± SEM. Data are representative of two to three independent experiments with 6 to 10 mice per treatment group. For percent tumor volume change, if data were deemed normal, the unpaired t test or the unpaired t test with Welch correction was used if SD was similar or different between groups, respectively. If data did not pass normality, the Mann–Whitney test was used (*, P < 0.05; ***, P < 0.0005).

Figure 6.

Anti-VEGF treatment augments anti-OX40 therapy. CT26 tumor–bearing mice were administered anti-VEGF, anti-OX40 (OX86), and/or corresponding control antibodies. A, Survival graph (left) with individual mouse tumor growth curves (right). Statistical analysis for mouse survival were calculated with the Gehan–Breslow–Wilcoxon test. B16.F10 tumor–bearing mice were treated as stated above. B, Percent tumor volume change (left) with individual mouse tumor growth curves (right). Time points reflect average ± SEM. Data are representative of two to three independent experiments with 6 to 10 mice per treatment group. For percent tumor volume change, if data were deemed normal, the unpaired t test or the unpaired t test with Welch correction was used if SD was similar or different between groups, respectively. If data did not pass normality, the Mann–Whitney test was used (*, P < 0.05; ***, P < 0.0005).

Close modal

VEGF contributes to tumor growth both by promoting angiogenesis and by limiting antitumor immune responses (1, 51). In downregulating expression of adhesion molecules on endothelial cells, VEGF restricts immune cell adhesion (15–17). Therefore, VEGF neutralization may enhance immune cell infiltration into tumors. VEGF blockade augments tumor infiltration of preactivated adoptively transferred gp100 TCR-transgenic CD8+ T cells in a syngeneic melanoma model (B16.F10; ref. 52). Here, we show that the effect of anti-VEGF on CD8+ T-cell accumulation within tumors varies across five tumor models. CD8+ T-cell content was enhanced only in the B16.F10 melanoma tumor model; additional immunomodulatory mechanisms are likely at play following VEGF blockade in other tumor models.

In all five tumor models, anti-VEGF treatment improved effector cytokine production by tumor-infiltrating CD8+ T cells. This effect was VEGFR2-mediated as receptor blockade mimicked the effects of anti-VEGF treatment. Similarly, disruption of VEGFR2 signaling enhanced effector function of T cells from patients with ovarian cancer (7, 8) and in mouse intratumoral T cells when treated with anti-VEGFR2 (53) or a bispecific antibody against angiopoetin-2 and VEGF-A (54). Our results challenge previous hypotheses that anti-VEGF enhances T-cell functionality through disruption of T-cell VEGFR2 signaling (5, 7, 8). We confirmed (41) a lack of VEGFR2 expression on CD8+ T cells. We conclude that the primary effect of anti-VEGF treatment on CD8+ T cells is independent of T-cell VEGFR2 signaling.

VEGF supports Treg and myeloid cell tumor accumulation and facilitates Treg immunosuppressive activities within tumors (3). VEGF blockade did not uniformly influence Treg and myeloid cell accumulation or functional phenotypic marker expression suggesting that alterations in numbers or suppressive function were not the cause for augmented CD8+ T-cell functionality. In addition, CD11b+ cells from CT26 tumors displayed similar CD8+ T-cell stimulatory capacity between treatment cohorts. VEGF inhibits DC maturation and activation resulting in diminished T-cell stimulatory capacity (3). We observed increased DC activation in response to VEGF blockade that was inconsistent between tumor models. The activated DC phenotype translated into increased T-cell stimulatory capacity suggesting DCs may contribute to augmented T-cell responses in select models. Neither DC content nor DC MHC class II amounts correlated with the frequency of IFNγ+ CD8+ T cells following antigen recall. Chetomin administration to CT26 tumor–bearing anti-VEGF–treated mice reduced IFNγ+ CD8+ T-cell frequencies even though the CD8+ T cell/DC ratio and DC activation status was sustained. We conclude that enhanced CD8+ T-cell activity following anti-VEGF administration is likely independent of Treg, myeloid, and DC compartment alterations.

Anti-VEGF therapy blunted tumor vascularization and increased hypoxia within viable tumor area as described previously (55–57). We observed in anti-VEGF–treated mice that T cells preferentially resided in close proximity to hypoxic tumor regions. Increased hypoxia within the tumor microenvironment was associated with increased HIF1α protein and HIF1α transcriptional target mRNA and protein within CD8+ T cells. We observed that HIF1α inhibition in vitro and in vivo reduced IFNγ production within CD8+ T cell from anti-VEGF–treated mice, suggesting that hypoxia drives effector function in this context. A similar link between HIF1α activity and T-cell effector function has been described in viral infection and tumor immunity (22, 24, 25). The phenotypic and functional effects on CD8+ T cells with conditional HIF1α activation in T cells (22) parallel our observations including: (i) elevated production of IFNγ, TNFα, and granzyme B; (ii) augmented expression of the costimulatory receptor OX40; and (iii) increased expression of glycolysis-associated molecules. Other studies have used chemical inhibition, hypoxic culturing conditions, or genetic modifications to demonstrate that HIF1α augments CD8+ T-cell activation (19–25). We conclude that increasing tumor hypoxia through inhibition of angiogenesis augments CD8+ T-cell effector function by enhancing HIF1α activity. This strategy may be translatable given existing antiangiogenic therapies.

The pathologic nature of the tumor microenvironment triggers metabolic adaptation that can impair immune cell activation and function (58): tumor cells compete for glucose with T cells, restricting antitumor immune responses (59–61). We speculate that hypoxia-mediated HIF1α stabilization might confer upon CD8+ T cells a metabolic advantage, but this remains unconfirmed. Several HIF1α targets upregulated within CD8+ T cells are associated with glycolysis (62). A greater proportion of GLUT3+ CD8+ T cells were detected within anti-VEGF–treated tumors, which may facilitate their survival in a resource-deprived tumor environment. The contribution of GLUT3 and HIF1α to CD8+ T-cell functionality during hypoxic stress could prove useful in engineering chimeric antigen receptor T cells to withstand the tumor microenvironment.

In responding patients, the duration of VEGF-targeted monotherapy is short, on the order of months. Combinatorial strategies may provide longer lasting antitumor activity. In preclinical tumor models, the combination of antiangiogenic agents with immunotherapy, such as anti–PD-1/PD-L1, have been effective (5, 53, 54, 63). The FDA has approved the combination of bevacizumab and atezolizumab (anti–PD-L1) with chemotherapeutic agents for the treatment of metastatic nonsquamous non–small cell lung cancer that lack ALK and EGFR genomic aberrations. Our findings indicate that combination of anti-VEGF with an agonizing antibody targeting costimulatory T-cell receptors, as demonstrated here with anti-OX40, represents a promising strategy. The anti-OX40 agonist used here depletes Tregs (64). Whether efficacy was mediated by costimulation and/or Treg depletion, this treatment regimen was efficacious, providing preclinical proof of concept for this combination strategy. On the basis of the effects of anti-VEGF therapy on the tumor microenvironment and T-cell adaptability, modifying the tumor vasculature or augmenting hypoxia prior to administration of immunotherapeutic drugs may potentiate antitumor immunity.

R. Jesudason is a scientific manager at Genentech, Inc. J.M. Kim is a scientist at Genentech, Inc. K.B. Walsh is a scientist at Genentech, Inc. No potential conflicts of interest were disclosed by the other authors.

Conception and design: P.E. de Almeida, G. Hernandez, J. Borneo, J.M. Kim, K.B. Walsh

Development of methodology: P.E. de Almeida, J. Mak, G. Hernandez, J. Borneo, K.B. Walsh

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.E. de Almeida, J. Mak, G. Hernandez, R. Jesudason, J. Borneo, K.B. Walsh

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P.E. de Almeida, J. Mak, G. Hernandez, R. Jesudason, J. Borneo, K.B. Walsh

Writing, review, and/or revision of the manuscript: P.E. de Almeida, J. Mak, G. Hernandez, A. Herault, J.M. Kim, K.B. Walsh

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Mak, V. Javinal, K.B. Walsh

Study supervision: J.M. Kim, K.B. Walsh

Other (contributed technically to some in vivo experiments): A. Herault

The authors thank Genentech's flow cytometry core laboratory personnel, J. Paw, T. Ho, L. Appling, C.K. Poon, A.P. Martinez, and J. Cupp; W. Ye for discussions, input, and article revisions; G. Hatzivassilou and R. Cubas for article revisions and input; N. Ramamoorthi and J. Long for assistance with fluidigm assays; E. McNamara for assistance with in vivo studies; and C. Bais for discussion and antibody to VEGFR1.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Apte
RS
,
Chen
DS
,
Ferrara
N
. 
VEGF in signaling and disease: beyond discovery and development
.
Cell
2019
;
176
:
1248
64
.
2.
Ye
W
. 
The complexity of translating anti-angiogenesis therapy from basic science to the clinic
.
Dev Cell
2016
;
37
:
114
25
.
3.
Yang
J
,
Yan
J
,
Liu
B
. 
Targeting VEGF/VEGFR to modulate antitumor immunity
.
Front Immunol
2018
;
9
:
978
.
4.
Fukumura
D
,
Kloepper
J
,
Amoozgar
Z
,
Duda
DG
,
Jain
RK
. 
Enhancing cancer immunotherapy using antiangiogenics: opportunities and challenges
.
Nat Rev Clin Oncol
2018
;
15
:
325
40
.
5.
Voron
T
,
Colussi
O
,
Marcheteau
E
,
Pernot
S
,
Nizard
M
,
Pointet
AL
, et al
VEGF-A modulates expression of inhibitory checkpoints on CD8+ T cells in tumors
.
J Exp Med
2015
;
212
:
139
48
.
6.
Mulligan
JK
,
Rosenzweig
SA
,
Young
MR
. 
Tumor secretion of VEGF induces endothelial cells to suppress T cell functions through the production of PGE2
.
J Immunother
2010
;
33
:
126
35
.
7.
Ziogas
AC
,
Gavalas
NG
,
Tsiatas
M
,
Tsitsilonis
O
,
Politi
E
,
Terpos
E
, et al
VEGF directly suppresses activation of T cells from ovarian cancer patients and healthy individuals via VEGF receptor type 2
.
Int J Cancer
2012
;
130
:
857
64
.
8.
Gavalas
NG
,
Tsiatas
M
,
Tsitsilonis
O
,
Politi
E
,
Ioannou
K
,
Ziogas
AC
, et al
VEGF directly suppresses activation of T cells from ascites secondary to ovarian cancer via VEGF receptor type 2
.
Br J Cancer
2012
;
107
:
1869
75
.
9.
Gabrilovich
DI
,
Chen
HL
,
Girgis
KR
,
Cunningham
HT
,
Meny
GM
,
Nadaf
S
, et al
Production of vascular endothelial growth factor by human tumors inhibits the functional maturation of dendritic cells
.
Nat Med
1996
;
2
:
1096
103
.
10.
Linde
N
,
Lederle
W
,
Depner
S
,
van Rooijen
N
,
Gutschalk
CM
,
Mueller
MM
. 
Vascular endothelial growth factor-induced skin carcinogenesis depends on recruitment and alternative activation of macrophages
.
J Pathol
2012
;
227
:
17
28
.
11.
Li
B
,
Lalani
AS
,
Harding
TC
,
Luan
B
,
Koprivnikar
K
,
Huan Tu
G
, et al
Vascular endothelial growth factor blockade reduces intratumoral regulatory T cells and enhances the efficacy of a GM-CSF-secreting cancer immunotherapy
.
Clin Cancer Res
2006
;
12
:
6808
16
.
12.
Hansen
W
,
Hutzler
M
,
Abel
S
,
Alter
C
,
Stockmann
C
,
Kliche
S
, et al
Neuropilin 1 deficiency on CD4+Foxp3+ regulatory T cells impairs mouse melanoma growth
.
J Exp Med
2012
;
209
:
2001
16
.
13.
Roland
CL
,
Lynn
KD
,
Toombs
JE
,
Dineen
SP
,
Udugamasooriya
DG
,
Brekken
RA
. 
Cytokine levels correlate with immune cell infiltration after anti-VEGF therapy in preclinical mouse models of breast cancer
.
PLoS One
2009
;
4
:
e7669
.
14.
Terme
M
,
Pernot
S
,
Marcheteau
E
,
Sandoval
F
,
Benhamouda
N
,
Colussi
O
, et al
VEGFA-VEGFR pathway blockade inhibits tumor-induced regulatory T-cell proliferation in colorectal cancer
.
Cancer Res
2013
;
73
:
539
49
.
15.
Griffioen
AW
,
Damen
CA
,
Blijham
GH
,
Groenewegen
G
. 
Tumor angiogenesis is accompanied by a decreased inflammatory response of tumor-associated endothelium
.
Blood
1996
;
88
:
667
73
.
16.
Dirkx
AE
,
Oude Egbrink
MG
,
Kuijpers
MJ
,
van der Niet
ST
,
Heijnen
VV
,
Bouma-ter Steege
JC
, et al
Tumor angiogenesis modulates leukocyte-vessel wall interactions in vivo by reducing endothelial adhesion molecule expression
.
Cancer Res
2003
;
63
:
2322
9
.
17.
Dirkx
AE
,
oude Egbrink
MG
,
Castermans
K
,
van der Schaft
DW
,
Thijssen
VL
,
Dings
RP
, et al
Anti-angiogenesis therapy can overcome endothelial cell anergy and promote leukocyte-endothelium interactions and infiltration in tumors
.
FASEB J
2006
;
20
:
621
30
.
18.
Ribatti
D
. 
Tumor refractoriness to anti-VEGF therapy
.
Oncotarget
2016
;
7
:
46668
77
.
19.
Caldwell
CC
,
Kojima
H
,
Lukashev
D
,
Armstrong
J
,
Farber
M
,
Apasov
SG
, et al
Differential effects of physiologically relevant hypoxic conditions on T lymphocyte development and effector functions
.
J Immunol
2001
;
167
:
6140
9
.
20.
Gropper
Y
,
Feferman
T
,
Shalit
T
,
Salame
TM
,
Porat
Z
,
Shakhar
G
. 
Culturing CTLs under hypoxic conditions enhances their cytolysis and improves their anti-tumor function
.
Cell Rep
2017
;
20
:
2547
55
.
21.
Nakagawa
Y
,
Negishi
Y
,
Shimizu
M
,
Takahashi
M
,
Ichikawa
M
,
Takahashi
H
. 
Effects of extracellular pH and hypoxia on the function and development of antigen-specific cytotoxic T lymphocytes
.
Immunol Lett
2015
;
167
:
72
86
.
22.
Doedens
AL
,
Phan
AT
,
Stradner
MH
,
Fujimoto
JK
,
Nguyen
JV
,
Yang
E
, et al
Hypoxia-inducible factors enhance the effector responses of CD8(+) T cells to persistent antigen
.
Nat Immunol
2013
;
14
:
1173
82
.
23.
Roman
J
,
Rangasamy
T
,
Guo
J
,
Sugunan
S
,
Meednu
N
,
Packirisamy
G
, et al
T-cell activation under hypoxic conditions enhances IFN-gamma secretion
.
Am J Respir Cell Mol Biol
2010
;
42
:
123
8
.
24.
Palazon
A
,
Tyrakis
PA
,
Macias
D
,
Velica
P
,
Rundqvist
H
,
Fitzpatrick
S
, et al
An HIF-1α/VEGF-A axis in cytotoxic T cells regulates tumor progression
.
Cancer Cell
2017
;
32
:
669
83
.
25.
Xu
Y
,
Chaudhury
A
,
Zhang
M
,
Savoldo
B
,
Metelitsa
LS
,
Rodgers
J
, et al
Glycolysis determines dichotomous regulation of T cell subsets in hypoxia
.
J Clin Invest
2016
;
126
:
2678
88
.
26.
Labiano
S
,
Palazon
A
,
Melero
I
. 
Immune response regulation in the tumor microenvironment by hypoxia
.
Semin Oncol
2015
;
42
:
378
86
.
27.
Noman
MZ
,
Hasmim
M
,
Messai
Y
,
Terry
S
,
Kieda
C
,
Janji
B
, et al
Hypoxia: a key player in antitumor immune response. A review in the theme: cellular responses to hypoxia
. Am J Physiol Cell Physiol
2015
;
309
:
C569
79
.
28.
Schito
L
,
Semenza
GL
. 
Hypoxia-inducible factors: master regulators of cancer progression
.
Trends Cancer
2016
;
2
:
758
70
.
29.
Li
J
,
Ybarra
R
,
Mak
J
,
Herault
A
,
De Almeida
P
,
Arrazate
A
, et al
IFNγ-induced chemokines are required for CXCR3-mediated T-cell recruitment and antitumor efficacy of anti-HER2/CD3 bispecific antibody
.
Clin Cancer Res
2018
;
24
:
6447
58
.
30.
Huang
AY
,
Gulden
PH
,
Woods
AS
,
Thomas
MC
,
Tong
CD
,
Wang
W
, et al
The immunodominant major histocompatibility complex class I-restricted antigen of a murine colon tumor derives from an endogenous retroviral gene product
.
Proc Natl Acad Sci U S A
1996
;
93
:
9730
5
.
31.
Whitton
JL
,
Tishon
A
,
Lewicki
H
,
Gebhard
J
,
Cook
T
,
Salvato
M
, et al
Molecular analyses of a five-amino-acid cytotoxic T-lymphocyte (CTL) epitope: an immunodominant region which induces nonreciprocal CTL cross-reactivity
.
J Virol
1989
;
63
:
4303
10
.
32.
Gajewski
TF
,
Corrales
L
,
Williams
J
,
Horton
B
,
Sivan
A
,
Spranger
S
. 
Cancer immunotherapy targets based on understanding the T cell-inflamed versus non-T cell-inflamed tumor microenvironment
.
Adv Exp Med Biol
2017
;
1036
:
19
31
.
33.
Horton
H
,
Russell
N
,
Moore
E
,
Frank
I
,
Baydo
R
,
Havenar-Daughton
C
, et al
Correlation between interferon- gamma secretion and cytotoxicity, in virus-specific memory T cells
.
J Infect Dis
2004
;
190
:
1692
6
.
34.
Jayaraman
P
,
Parikh
F
,
Lopez-Rivera
E
,
Hailemichael
Y
,
Clark
A
,
Ma
G
, et al
Tumor-expressed inducible nitric oxide synthase controls induction of functional myeloid-derived suppressor cells through modulation of vascular endothelial growth factor release
.
J Immunol
2012
;
188
:
5365
76
.
35.
Horikawa
N
,
Abiko
K
,
Matsumura
N
,
Hamanishi
J
,
Baba
T
,
Yamaguchi
K
, et al
Expression of vascular endothelial growth factor in ovarian cancer inhibits tumor immunity through the accumulation of myeloid-derived suppressor cells
.
Clin Cancer Res
2017
;
23
:
587
99
.
36.
Courau
T
,
Nehar-Belaid
D
,
Florez
L
,
Levacher
B
,
Vazquez
T
,
Brimaud
F
, et al
TGF-β and VEGF cooperatively control the immunotolerant tumor environment and the efficacy of cancer immunotherapies
.
JCI Insight
2016
;
1
:
e85974
.
37.
Shojaei
F
,
Wu
X
,
Malik
AK
,
Zhong
C
,
Baldwin
ME
,
Schanz
S
, et al
Tumor refractoriness to anti-VEGF treatment is mediated by CD11b+Gr1+ myeloid cells
.
Nat Biotechnol
2007
;
25
:
911
20
.
38.
Shojaei
F
,
Wu
X
,
Qu
X
,
Kowanetz
M
,
Yu
L
,
Tan
M
, et al
G-CSF-initiated myeloid cell mobilization and angiogenesis mediate tumor refractoriness to anti-VEGF therapy in mouse models
.
Proc Natl Acad Sci U S A
2009
;
106
:
6742
7
.
39.
Gabrilovich
DI
,
Ishida
T
,
Nadaf
S
,
Ohm
JE
,
Carbone
DP
. 
Antibodies to vascular endothelial growth factor enhance the efficacy of cancer immunotherapy by improving endogenous dendritic cell function
.
Clin Cancer Res
1999
;
5
:
2963
70
.
40.
Mimura
K
,
Kono
K
,
Takahashi
A
,
Kawaguchi
Y
,
Fujii
H
. 
Vascular endothelial growth factor inhibits the function of human mature dendritic cells mediated by VEGF receptor-2
.
Cancer Immunol Immunother
2007
;
56
:
761
70
.
41.
Shin
JY
,
Yoon
IH
,
Kim
JS
,
Kim
B
,
Park
CG
. 
Vascular endothelial growth factor-induced chemotaxis and IL-10 from T cells
.
Cell Immunol
2009
;
256
:
72
8
.
42.
Basu
A
,
Hoerning
A
,
Datta
D
,
Edelbauer
M
,
Stack
MP
,
Calzadilla
K
, et al
Cutting edge: vascular endothelial growth factor-mediated signaling in human CD45RO+ CD4+ T cells promotes Akt and ERK activation and costimulates IFN-gamma production
.
J Immunol
2010
;
184
:
545
9
.
43.
Lapeyre-Prost
A
,
Terme
M
,
Pernot
S
,
Pointet
AL
,
Voron
T
,
Tartour
E
, et al
Immunomodulatory activity of VEGF in cancer
.
Int Rev Cell Mol Biol
2017
;
330
:
295
342
.
44.
Tong
RT
,
Boucher
Y
,
Kozin
SV
,
Winkler
F
,
Hicklin
DJ
,
Jain
RK
. 
Vascular normalization by vascular endothelial growth factor receptor 2 blockade induces a pressure gradient across the vasculature and improves drug penetration in tumors
.
Cancer Res
2004
;
64
:
3731
6
.
45.
Gross
MW
,
Karbach
U
,
Groebe
K
,
Franko
AJ
,
Mueller-Klieser
W
. 
Calibration of misonidazole labeling by simultaneous measurement of oxygen tension and labeling density in multicellular spheroids
.
Int J Cancer
1995
;
61
:
567
73
.
46.
Samanta
D
,
Semenza
GL
. 
Metabolic adaptation of cancer and immune cells mediated by hypoxia-inducible factors
.
Biochim Biophys Acta Rev Cancer
2018
;
1870
:
15
22
.
47.
Ellinghaus
P
,
Heisler
I
,
Unterschemmann
K
,
Haerter
M
,
Beck
H
,
Greschat
S
, et al
BAY 87–2243, a highly potent and selective inhibitor of hypoxia-induced gene activation has antitumor activities by inhibition of mitochondrial complex I
.
Cancer Med
2013
;
2
:
611
24
.
48.
Kung
AL
,
Zabludoff
SD
,
France
DS
,
Freedman
SJ
,
Tanner
EA
,
Vieira
A
, et al
Small molecule blockade of transcriptional coactivation of the hypoxia-inducible factor pathway
.
Cancer Cell
2004
;
6
:
33
43
.
49.
Tyrakis
PA
,
Palazon
A
,
Macias
D
,
Lee
KL
,
Phan
AT
,
Velica
P
, et al
S-2-hydroxyglutarate regulates CD8(+) T-lymphocyte fate
.
Nature
2016
;
540
:
236
41
.
50.
Palazon
A
,
Martinez-Forero
I
,
Teijeira
A
,
Morales-Kastresana
A
,
Alfaro
C
,
Sanmamed
MF
, et al
The HIF-1α hypoxia response in tumor-infiltrating T lymphocytes induces functional CD137 (4–1BB) for immunotherapy
.
Cancer Discov
2012
;
2
:
608
23
.
51.
Ramjiawan
RR
,
Griffioen
AW
,
Duda
DG
. 
Anti-angiogenesis for cancer revisited: is there a role for combinations with immunotherapy?
Angiogenesis
2017
;
20
:
185
204
.
52.
Shrimali
RK
,
Yu
Z
,
Theoret
MR
,
Chinnasamy
D
,
Restifo
NP
,
Rosenberg
SA
. 
Antiangiogenic agents can increase lymphocyte infiltration into tumor and enhance the effectiveness of adoptive immunotherapy of cancer
.
Cancer Res
2010
;
70
:
6171
80
.
53.
Allen
E
,
Jabouille
A
,
Rivera
LB
,
Lodewijckx
I
,
Missiaen
R
,
Steri
V
, et al
Combined antiangiogenic and anti-PD-L1 therapy stimulates tumor immunity through HEV formation
.
Sci Transl Med
2017
;
9
.
pii: eaak9679
.
54.
Schmittnaegel
M
,
Rigamonti
N
,
Kadioglu
E
,
Cassara
A
,
Wyser Rmili
C
,
Kiialainen
A
, et al
Dual angiopoietin-2 and VEGFA inhibition elicits antitumor immunity that is enhanced by PD-1 checkpoint blockade
.
Sci Transl Med
2017
;
9
.
pii: eaak9670
.
55.
Franco
M
,
Man
S
,
Chen
L
,
Emmenegger
U
,
Shaked
Y
,
Cheung
AM
, et al
Targeted anti-vascular endothelial growth factor receptor-2 therapy leads to short-term and long-term impairment of vascular function and increase in tumor hypoxia
.
Cancer Res
2006
;
66
:
3639
48
.
56.
Shi
Y
,
Oeh
J
,
Hitz
A
,
Hedehus
M
,
Eastham-Anderson
J
,
Peale
FV
 Jr
, et al
Monitoring and targeting anti-VEGF induced hypoxia within the viable tumor by (19)F-MRI and multispectral analysis
.
Neoplasia
2017
;
19
:
950
9
.
57.
Miyazaki
S
,
Kikuchi
H
,
Iino
I
,
Uehara
T
,
Setoguchi
T
,
Fujita
T
, et al
Anti-VEGF antibody therapy induces tumor hypoxia and stanniocalcin 2 expression and potentiates growth of human colon cancer xenografts
.
Int J Cancer
2014
;
135
:
295
307
.
58.
O'Sullivan
D
,
Sanin
DE
,
Pearce
EJ
,
Pearce
EL
. 
Metabolic interventions in the immune response to cancer
.
Nat Rev Immunol
2019
;
19
:
324
35
.
59.
Chang
CH
,
Qiu
J
,
O'Sullivan
D
,
Buck
MD
,
Noguchi
T
,
Curtis
JD
, et al
Metabolic competition in the tumor microenvironment is a driver of cancer progression
.
Cell
2015
;
162
:
1229
41
.
60.
Zhao
E
,
Maj
T
,
Kryczek
I
,
Li
W
,
Wu
K
,
Zhao
L
, et al
Cancer mediates effector T cell dysfunction by targeting microRNAs and EZH2 via glycolysis restriction
.
Nat Immunol
2016
;
17
:
95
103
.
61.
Ho
PC
,
Bihuniak
JD
,
Macintyre
AN
,
Staron
M
,
Liu
X
,
Amezquita
R
, et al
Phosphoenolpyruvate is a metabolic checkpoint of anti-tumor T cell responses
.
Cell
2015
;
162
:
1217
28
.
62.
Buck
MD
,
Sowell
RT
,
Kaech
SM
,
Pearce
EL
. 
Metabolic instruction of immunity
.
Cell
2017
;
169
:
570
86
.
63.
Shigeta
K
,
Datta
M
,
Hato
T
,
Kitahara
S
,
Chen
IX
,
Matsui
A
, et al
Dual programmed death receptor-1 and vascular endothelial growth factor receptor-2 blockade promotes vascular normalization and enhances antitumor immune responses in hepatocellular carcinoma
.
Hepatology
2020
;
71
:
1247
61
.
64.
Bulliard
Y
,
Jolicoeur
R
,
Zhang
J
,
Dranoff
G
,
Wilson
NS
,
Brogdon
JL
. 
OX40 engagement depletes intratumoral Tregs via activating FcgammaRs, leading to antitumor efficacy
.
Immunol Cell Biol
2014
;
92
:
475
80
.