Multiple tumor-derived factors are responsible for the accumulation and expansion of immune-suppressing myeloid-derived suppressor cells (MDSC) and M2-like tumor-associated macrophages (TAM) in tumors. Here, we show that treatment of tumor-bearing mice with docetaxel in combination with the phosphatidylserine-targeting antibody 2aG4 potently suppressed the growth and progression of prostate tumors, depleted M2-like TAMs, and MDSCs, and increased the presence of M1-like TAMs and mature dendritic cells in the tumors. In addition, the antibody markedly altered the cytokine balance in the tumor microenvironment from immunosuppressive to immunostimulatory. In vitro studies confirmed that 2aG4 repolarized TAMs from an M2- to an M1-like phenotype and drove the differentiation of MDSCs into M1-like TAMs and functional dendritic cells. These data suggest that phosphatidylserine is responsible for the expansion of MDSCs and M2-like TAMs in tumors, and that bavituximab, a phosphatidylserine-targeting antibody currently in clinical trials for cancer, could reverse this process and reactivate antitumor immunity. Cancer Immunol Res; 1(4); 256–68. ©2013 AACR.
Tumors have long been recognized as having immunosuppressive microenvironments that thwart the host's ability to control tumor growth (1, 2). Myeloid progenitors from the bone marrow populate tumors and differentiate into a heterogeneous population of myeloid-derived suppressor cells (MDSC) that are immunosuppressive (2). MDSCs secrete immunosuppressive cytokines, including interleukin (IL)-10 and TGF-β, that induce the development of regulatory T cells (Treg; refs. 3, 4), suppress immune responses mediated by CD4+ and CD8+ T cells (5–7), and the cytotoxic activities of natural killer (NK) and NKT cells (8). MDSCs of monocytic subtype (CD11b+, Gr-1+, and Ly-6CHi) differentiate into tumor-associated macrophages (TAM) and dendritic cells with impaired functionality (9–11). Dendritic cells remain immature and lack the costimulatory molecules needed to function as antigen-presenting cells (APC; ref. 12). TAMs become predominantly polarized into the alternatively activated M2-like phenotype that secrete proangiogenic factors (13) and immunosuppressive cytokines that further limit T-helper 1 immune responses (14). In contrast, classically activated M1-like TAMs secrete immunostimulatory cytokines that have direct tumoricidal activity. M1-like TAMs, however, are sparse and confined to the less hypoxic regions of tumors (15). The presence of M2-like TAMs in tumors correlates with poor prognosis (16, 17), whereas the presence of M1-like TAMs correlates with longer survival for patients (18).
Phosphatidylserine is a phospholipid that contributes to the immunosuppressed tumor microenvironment by preventing immune and inflammatory reactions (19–21). Phosphatidylserine is confined to the inner leaflet of the plasma membrane in most normal mammalian cells but becomes exposed on the outer surface of apoptotic cells, where it subverts unwanted immune reactions against dying cells (22). Antitumor responses are similarly suppressed in the tumor microenvironment because phosphatidylserine is exposed on endothelial cells in the tumor vasculature (23, 24) and on tumor-derived microvesicles (25); phosphatidylserine is expressed constitutively on some tumor cells (26). Moreover, the exposure of phosphatidylserine is increased significantly on tumor cells undergoing apoptosis in response to chemo- and radiotherapy, where it further enhances immunosuppression (27, 28). Exposed phosphatidylserine is recognized by macrophages and dendritic cells, which have receptors that recognize phosphatidylserine directly through TIM 3 and TIM 4, brain-specific angiogenesis inhibitor 1 (BAI1), stabilin-2, or receptor for advanced glycation end-products (RAGE; refs. 29–32), or indirectly through a variety of bridging proteins (33, 34). Binding to macrophage phosphatidylserine receptors triggers IL-10- and TGF-β–dependent immunosuppressive signals that stimulate them to engulf the phosphatidylserine-expressing cells without secreting inflammatory cytokines (19–21). Moreover, while intratumoral dendritic cells bind and ingest phosphatidylserine-expressing cells, they maintain an immature phenotype, lacking the costimulatory molecules required for APC activity (35, 36). These data emphasize that exposed phosphatidylserine is a major factor in maintaining the immunosuppressed state in tumors, and further suggest that chemotherapy, radiotherapy, and androgen-deprivation therapy are undermined by the phosphatidylserine in the tumor microenvironment (37).
To explore the possibility of reversing the immunosuppressive effects of exposed phosphatidylserine, we generated a family of phosphatidylserine-targeting antibodies that bind with high affinity to complexes of the phosphatidylserine-binding plasma protein, β2-glycoprotein I (β2GP1) and anionic phospholipids. The antibodies bind to phosphatidylserine-expressing membranes by cross-linking two molecules of β2GP1 bound to phosphatidylserine on the membrane. The antibody–β2GP1–PS complex is only stably formed on phosphatidylserine-containing surfaces. 2aG4, a mouse immunoglobulin G2a (IgG2a) version of the human chimeric antibody bavituximab, localizes to phosphatidylserine-expressing tumor vascular endothelium and elicits strong antitumor effects when combined with chemo- or radiotherapy in mouse tumor models (24, 27, 28). Bavituximab is currently being tested in multiple clinical trials (38–40). Despite the progress that has been made, the mechanism of action of anti-phosphatidylserine antibodies is not fully understood.
In this study, we examined whether 2aG4 can reverse the immunosuppressive effects of exposed phosphatidylserine in mouse models of human prostate cancer. We show that 2aG4 reactivates antitumor immunity on multiple levels: (i) the switching of TAMs to a tumoricidal M1-like phenotype; (ii) the reduction of MDSCs in tumors; and (iii) the maturation of dendritic cells into functional APCs. Our data show that the antitumor activity of bavituximab is due in large part to the suppression of immune tolerance and a concomitant reactivation of antitumor immunity, generating M1-TAMs that destroy phosphatidylserine-expressing tumor vasculature.
Materials and Methods
Human prostate cancer cell lines LNCaP and PC3, and the C44 hybridoma (CRL-1943) were obtained from the American Type Culture Collection. PC3 cells were stably transfected with firefly luciferase (Jer-Tsong Hsieh, University of Texas Southwestern Medical Center, Dallas, TX). These cell lines were maintained in RPMI-1640 supplemented with 10% v/v heat-inactivated FBS (Hyclone) without antibiotics. All cell lines were regularly tested for Mycoplasma.
2aG4 (mouse IgG2a), C44 (control mouse IgG2a), and bavituximab (mouse 2aG4 VH and Vκ, human IgG1 κ constant domains) were produced by Peregrine Pharmaceuticals, Inc. Human β2GP1 was purified as previously described (41). All were endotoxin-free. Rituximab was from Genentech. Rat anti-mouse CD31, rat anti-mouse CD11b (M1/70, Mac-1), rat anti-mouse F4/80, rat anti-mouse Ly-6G (Gr-1), hamster anti-mouse CD11c (integrin αM-chain), hamster anti-mouse CD80, rat anti-mouse FcγIII/II receptor (CD16/CD32) monoclonal antibodies, and rabbit anti-mouse inducible nitric oxide (NO) synthase (iNOS) polyclonal antibody were from BD Pharmingen. Rat anti-mouse CD49b and rat anti-mouse NKG2A/C/E monoclonal antibodies were from Serotec Inc. Goat anti-mouse arginase-I (Arg-1) polyclonal antibody was from Santa Cruz Biotechnology, Inc. Hamster anti-mouse CD31 was from Pierce Biotechnology. All secondary antibodies for immunohistochemistry were from Jackson ImmunoResearch Labs. Fluorescein isothiocyanate (FITC)- or phycoerythrin (PE)-conjugated anti-CD11b (clone M1/70), FITC- or PE-conjugated anti-F4/80, FITC-conjugated anti-Ly6C (clone ER-MP20), allophycocyanin-conjugated anti-Gr-1 (clone RB6-8C5), and FITC- or PE-conjugated normal hamster IgG and rat IgG were from eBiosciences.
Male severe combined immunodeficient mice (SCID; NCI-ncr) mice 5 to 6 weeks old were purchased from the National Cancer Institute-Frederick Cancer Research Facility (Frederick, MD). All experimental procedures were approved by the University of Texas Southwestern's Animal Care and Use Committee.
Prostate cancer PC3 and LNCaP tumors were selected for the study because almost all TAMs are of the M2-like phenotype, and have few transitioning cells, which makes them good tumor models in which to show M2 to M1 polarization.
Tumor cells (1 × 106 cells/0.1 mL/mouse) suspended in 50% Matrigel in PBS were injected s.c. into the right flank of male SCID mice. Tumor volumes were calculated as V = (a2 × b)/2, where a and b were the minimal and maximal diameters of the tumor, respectively.
Male mice were anesthetized and PC3 cells (5 × 105 cells in 50 μL PBS) were injected into the dorsolateral prostate lobes after surgical exposure of the prostate. Bioluminescence imaging (BLI) was carried out while mice were under anesthesia.
When LNCaP tumors reached 0.8 to 1.0 cm diameter or when BLI signals for orthotopic PC3 tumors reached 1 × 106 photons/s (equivalent to 20 mm3), mice were randomized into groups and treated intraperitoneally (i.p.) twice weekly with C44, 2aG4 (4 mg/kg), docetaxel (5 mg/kg), or a combination of both 2aG4 and docetaxel. Because the binding of bavituximab and 2aG4 to phosphatidylserine is dependent on human β2GP1, antibodies were coadministered with β2GP1 (4 mg/kg) in all experiments.
BLI was conducted on the IVIS200 system with the use of Living Image acquisition and analysis software (Caliper Life Sciences). Luciferin (75 mg/kg; Biosynth Chemistry & Biology) was injected s.c., and serial 5-second images were acquired between 5 and 10 minutes after injections. Signal intensity was quantified as the sum of all detected photon counts within a manually selected region of interest that was kept constant at each time point.
Quantification of tumor endothelium with exposed phosphatidylserine
Mice bearing s.c. PC3 or LNCaP tumors of 1 cm in diameter were injected i.v. with 100 μg of bavituximab or rituximab; after 2 hours, these mice were anesthetized and perfused with heparinized saline. Major organs and tumors were removed and snap-frozen, and 8-μm sections were generated from the center of the tumors. Bavituximab-positive vessels were identified with biotinylated goat anti-human IgG followed by Cy2-labeled streptavidin (green). Vascular endothelium was identified with rat anti-mouse CD31 antibody followed by Cy3-labeled goat anti-rat IgG (red). Sections were counterstained with 4′,6′diamidino-2-phenylindole (DAPI) and assessed by fluorescence microscopy. Single images were captured using a Coolsnap digital camera and analyzed with ImageJ Version 1.44s software (Wayne Rasband, NIH) to quantify pixels positive for both CD31 and bavituximab. Five randomly selected 0.079-mm2 fields were analyzed per section, with three sections per tumor and four to five tumors per group.
Blood vessel perfusion analysis
To quantify perfusible tumor vessels in mice, Hoechst 33342 (Sigma-Aldrich) was injected (15 mg/kg) into a tail vein, and 2 minutes later, mice were sacrificed and the tumors were removed and cryosectioned. Ten sections per tumor and five tumors per group were analyzed and perfusion measured over the entire tumor section. The mean percentage of the area of tumor sections that was stained with Hoeschst 33342 dye was quantified using the ImageJ software.
Blood flow quantification
Tumors were sonographically scanned with a Vevo 770 small-animal high-resolution ultrasound scanner equipped with a VisualSonics RMV 708 scan head (VisualSonics). Three-dimensional (3D) images in power Doppler mode were obtained by acquiring two-dimensional (2D) images every 100 μm along the entire length of the tumor. Ultrasound scanner settings were power Doppler transmission frequency, 23 MHz; power gain, 100%, wall filter, 2.5 mm/s; and scan speed, 2.0 mm/s. The percentage of the tumor volume having detectable blood flow was computed using the 3D segmentation tool in the Vevo 770 software package.
Quantification of immune cells in tumor sections
Frozen sections (three sections/tumor; three to five tumors/group) were cut through the tumors at the widest dimension. The sections were stained with antibodies to identify various cell types: TAMs (F4/80+), M1-subtype TAMs (iNOS+ and F4/80+), M2-subtype TAMs (Arg-1+ and F4/80+), NK cells (NKG2A/CE+ and CD49b+), MDSCs (Gr-1+ and CD11b+), transitioning MDSCs (F4/80+ and Gr-1+), granulocytes (Gr-1+, Ly6Ghi, and CD11b−), dendritic cells (CD11b+ and CD11chi), and mature dendritic cells (CD11chi and CD86+). Double and triple staining were carried out with primary antibodies from different species and developed with appropriate Cy2-, Cy3-, or coumarin aminomethylcoumarin acetate–labeled anti-IgG secondary antibodies. The sections were counterstained with DAPI and analyzed by fluorescence microscopy. Digital images (five per section) were collected. For individual colors, the same exposure times and illumination intensity were used for different sections, and the data were collected and processed in the same way to ensure comparability. The area of double-stained cells was computed using ImageJ software and expressed as a mean percentage of the total area of the sections.
Real-time quantitative reverse transcription analyses
Expression levels of mRNA were quantified relative to housekeeping gene 18S rRNA (forward: GCAATTATTCCCCATGAACG; reverse: AGGGCCTCACTAAACCATCC). Gene-specific primers were used to detect the following: iNOS (forward: TTCTGTGCTGTCCCAGTGAG; reverse: TGAAGAAAACCCCTTGTGCT), Arg-1 (forward: AGAGATTATCGGAGCGCCTT; reverse: TTTTTCCAGCAGACCAGCTT), IL-6 (forward: TGATGCACTTGCAGAAAACA; reverse: ACCAGAGGAAATTTTCAATAGGC), IL-12 p40 (forward: AGCAGTAGCAGTTCCCCTGA; reverse: AGTCCCTTTGGTCCAGTGTG), CD206 (forward: GGCAGGATCTTGGCAACCTAGTA; reverse: GTTTGGATCGGCACACAAAGTC), IL-10 (forward: ATCGATTTCTCCCCTGTGAA; reverse: TGTCAAATTCATTCATGGCCT), TNF-α (forward: CCACCACGCTCTTCTGTCTAC; reverse: AGGGTCTGGGCCATAGAACT), TGF-β1 (forward: GGAGAGCCCTGGATACCAAC; reverse: CAACCCAGGTCCTTCCTAAA), Fizz-1 (forward: CCCTTCTCATCTGCATCTCC; reverse: CTGGATTGGCAAGAAGTTCC), Ym1 (forward: TCTGGGTACAAGATCCCTGAA; reverse: TTTCTCCAGTGTAGCCATCCTT), F4/80 (forward: TTACGATGGAATTCTCCTTGTATATCA; reverse: CACAGCAGGAAGGTGGCTATG), CD11c (forward: CTG GATAGCCTTTCTTCTGCTG; reverse: GCACACTGTGTCCGAACTCA), CD40 (forward: GTTTAAAGTCCCGGATGCGA; reverse: CTCAAGGCTATGCTGTCTGT), CD80 (forward: ACCCCCAACATAACTGAGTCT; reverse: TTCCAACCAAGAGAAGCGAGG), CD83 (forward: CGCAGCTCTCCTATGCAGTG; reverse: GTGTTTTGGATCGTCAGGGAATA), CD86 (forward: TGTTTCCGTGGAGACGCAAG; reverse: CAGCTCACTCAGGCTTATGTTTT), VEGF-A (forward: TTTACTGCTGTACCTCCACCA; reverse: ATCTCTCCTATGTGCTGGCTTT), VEGF-B (forward: CCTGGAAGAACACAGCCAAT; reverse: GGAGTGGGA TGGATGATGTC); MHC II (forward: GCGACGTGGGCGAGTACC; reverse: CATTCCGGAACCAGCGCA), Ccl 5 (forward: TGCCCACGTCAAGGAGTATTTC; reverse: AACCCACTTCTTCTCTGGGTTG), and Cxcl11 (forward: GGCTGCGACAAAGTTGAAGTGA; reverse: TCCTGGCACAGAGTTCTTATTGGAG).
Isolation of TAMs and monocytic MDSCs
For TAMs, tumors were minced into small pieces and incubated for 15 minutes at 37°C with collagenase type II (0.5 mg/mL), collagenase type IV (0.5 mg/mL), and DNase I (0.01 mg/mL). Tumor pieces were mechanically dissociated using a gentleMACS dissociator (Miltenyi Biotec Inc.). The dissociated cells were collected, red blood cells (RBC) were lysed, and 2.5 × 108 cells were transferred into tissue culture flasks. After 2 hours at 37°C, nonadherent and loosely adherent cells were washed away. The adherent cells were detached with Accutase, and TAMs were extracted using anti-F4/80–biotin followed by anti-biotin magnetic beads (Miltenyi Biotec, Inc.). TAMs were 95% F4/80+ by fluorescence-activated cell sorting (FACS) analysis. Monocytic MDSCs were isolated from single-cell suspensions of spleen cells from tumor-bearing mice, by positive selection with anti-CD11b–coated magnetic beads, negative selection with anti-Ly6G-biotin followed by anti-biotin magnetic beads, and a final positive selection with anti-Gr-1–coated beads (Miltenyi Biotec, Inc.). Ninety percent of the cells were CD11b+, Ly6Glo, and Ly6Chi by FACS analysis.
Freshly isolated TAMs and MDSCs were lightly fixed for 5 minutes in 1% methanol-free formaldehyde (Polysciences, Inc.) in PBS. Cells were stained with both 2aG4 and rabbit anti-Alix (H-270; Santa Cruz Biotechnology, Inc.), or irrelevant control antibodies, followed by 12-nm gold-labeled donkey anti-mouse IgG and 6-nm gold-labeled donkey anti-rabbit IgG. Cells were processed for transmission electron microscopy (TEM) using standard methods. Ultrathin sections were viewed on a TEM Tecnai electron microscope.
Inhibition of prostate cancer growth in mice
Combining 2aG4 with docetaxel treatment improved the therapeutic activity on prostate tumors in mice beyond that achievable with either drug alone. Both castration-resistant (PC3) and castration-sensitive (LNCaP) tumors responded to the combination treatment. Treatment of mice bearing established orthotopic PC3 tumors with the combination reduced the bioluminescence intensity of the tumors by 42-fold as compared with mice treated with the C44 control antibody (Fig. 1A and B). 2aG4 alone and docetaxel alone reduced the bioluminescence intensity by approximately 3- and 7-fold, respectively. These differences were confirmed by the weights of the genitourinary tract plus tumor at the end of the experiment (Fig. 1C). The combination treatment was not more toxic to mice than treatment with docetaxel alone; animals in both groups lost 15% more body weight than did the C44-treated controls (Fig. 1D). Physical signs were the same in both groups. 2aG4 treatment alone was not toxic.
The combined regimen of 2aG4 and docetaxel inhibited the growth of LNCaP tumors in mice to a similar extent. When combined with castration, the triple combination caused major regressions of large established tumors and prevented progression to castration-resistant disease (Supplementary Fig. S1A and S1B).
Exposure of phosphatidylserine on tumor blood vessels and amplification by docetaxel
In untreated mice, 22% and 21% of tumor vessels had exposed phosphatidylserine in the PC3 (Fig. 2A) and LNCaP tumors, respectively (Supplementary Fig. S2A). We have previously shown that phosphatidylserine-positive tumor endothelial cells seem to be viable: They lack cytoplasmic and nuclear markers of apoptosis, are morphologically intact, and the vessels transport solutes and blood (42). The percentage of phosphatidylserine-positive tumor vessels was increased markedly 48 hours after administration of a single dose of docetaxel with peak levels of 61% in PC3 tumors (Fig. 2A) and 56% in LNCaP tumors (Supplementary Fig. S2A). No staining was detected with an isotype-matched control antibody, rituximab. Blood vessels in normal tissues (brain, heart, small intestine, large intestine, leg muscle, liver, lung, kidney, and testis) remained phosphatidylserine-negative irrespective of docetaxel treatment. Except for cells in and around necrotic regions, tumor cells were phosphatidylserine-negative in animals not injected with docetaxel. After treatment with docetaxel, 15% to 30% of the tumor cells became phosphatidylserine-positive (data not shown).
Destruction of tumor vasculature by 2aG4
Treatment with 2aG4 caused vessels in PC3 tumors (Fig. 2B) and LNCaP tumors (Supplementary Fig. S2B) to become denuded of vascular endothelium. Collagen IV in the surrounding basement membrane remained, but there were no endothelial cells. Blood vessels in tumors from C44-treated mice were intact. Vascular damage was also evident from the reductions in vascular density, perfusion, and blood flow in PC3 and LNCaP tumors of mice treated with 2aG4 alone or in combination with docetaxel. After 2 weeks of treatment, the combination regimen reduced vascular density in PC3 tumors by 88% (Fig. 2C) and in LNCaP tumors by 84% (Supplementary Fig. S2C), whereas 2aG4 alone reduced vascular density in PC3 tumors by 51% and in LNCaP tumors by 60%, relative to C44 control groups (P < 0.001). Docetaxel alone had little effect on vascular density. Perfusion studies with the fluorescent DNA-binding dye, Hoechst 33342, showed that treatment with 2aG4 alone reduced the mean area of PC3 tumor sections occupied by the dye from 16% (C44 control group) to 8.5% and, in combination with docetaxel, to 2.5% (Fig. 2C). Similar reductions in perfusion were observed in LNCaP tumors (Supplementary Fig. S2C). Using Doppler 3D ultrasound measurements, we showed that the volume of PC3 tumors with detectable blood flow was reduced by 2aG4 treatment from 3.8% to 1.7% (P < 0.0001) and, when combined with docetaxel, to 0.8% (Fig. 2D).
Vascular damage is caused by M1-like TAMs generated by 2aG4 treatment
Immunohistochemical analyses of PC3 and LNCaP tumors from mice treated with 2aG4 alone or in combination with docetaxel revealed that vascular damage was caused by TAMs (F4/80+, green) that congregated around CD31+ tumor blood vessels (red; Fig. 3A; Supplementary Fig. S3A). Only CD31+ remnants of most of the vessels remained. TAMs were the only cell type present around damaged tumor blood vessels; there were no NK cells, granulocytes, or transitioning MDSCs (F4/80+ and Gr-1+). Almost all TAMs were costained with F4/80 and iNOS, and they lacked Arg-1, indicating that they were of the M1-like phenotype (Fig. 3B and Supplementary Fig. S3A). In sharp contrast, in mice treated with C44 or docetaxel alone, TAMs were costained with F4/80 and Arg-1 but not iNOS, indicating they were predominantly of the M2-like phenotype (Fig. 3C and Supplementary Fig. S3B). The M2-like TAMs were less abundant and scattered throughout the tumor interstitium; they were not associated with the vessels. A small fraction of TAMs were of mixed phenotypes, most likely representing transition states between M2 and M1.
The shift in polarity of TAMs from being predominantly M2-like to predominantly M1-like was confirmed by quantifying the mean area of tumor sections that double-stained for markers of M1 (F4/80+ and iNOS+) and M2 (F4/80+ and Arg-1+) TAMs. In PC3 tumors treated with C44, the mean area of sections occupied by M2-like TAMs was 2.5%, as compared with 1.4% for M1-like TAMs (Fig. 3D). In contrast, in tumors treated with 2aG4, the mean area occupied by M1-like TAMs increased to 9.4%, whereas the area occupied by M2-like TAMs decreased slightly to 1.3%. The M1:M2 ratio increased from 0.55 to 7.0, an increase of 12.7-fold (Fig. 3D). Similar changes were observed for LNCaP tumors treated with 2aG4, in which the M1:M2 ratio increased 18-fold (Supplementary Fig. S3C). These changes were supported by quantitative reverse-transcription PCR (qRT-PCR) studies on F4/80+ TAMs isolated from dissociated PC3 tumors from 2aG4- or C44-treated mice. An increase in mRNA-encoding multiple M1 markers and a decrease in mRNA-encoding multiple M2 markers were observed after 2aG4 treatment (Fig. 3E). Among the increased M1 markers were the T-cell costimulatory molecules, CD80, CD86, CD40, and MHC class II, indicating that 2aG4 treatment caused the TAMs to acquire the ability to present antigens. Among the decreased M2 markers were VEGF-A and -B, indicating that 2aG4 treatment reduced the ability of TAMs to induce tumor angiogenesis. Treatment with 2aG4 also switched the production of cytokine mRNA from immunosuppressive (TGF-β and IL-10) to immunostimulatory (TNF-α and IL-12).
Next, we determined whether 2aG4 treatment could enhance NO production by TAMs and induce direct tumoricidal activity. TAMs isolated from 2aG4-treated PC3 tumors synthesized NO, whereas those from C44-treated tumors did not (Fig. 3F). TAMs from the 2aG4-treated tumors efficiently killed tumor cells in vitro, whereas those from C44-treated tumors did not (Fig. 3G). Thus, 2aG4 treatment generated tumoricidal M1-like TAMs.
Treatment with 2aG4 decreases MDSCs, increases TAMs and mature dendritic cells, and shifts the balance of cytokines in the tumor microenvironment from immunosuppressive to immunostimulatory
Sections of PC3 or LNCaP tumors from 2aG4- or C44-treated mice were stained for MDSCs (CD11b+ and Gr-1+), TAMs (F4/80+), and mature dendritic cells (CD11chi and CD86hi). TAMs did not costain for Gr-1, indicating the absence of transitioning MDSCs. The mean area of tumor sections occupied by these cells was quantified and is shown together with representative sections in Fig. 4A and Supplementary Fig. S4A. Treatment of PC3 tumors with 2aG4 decreased the mean area of tumor sections occupied by MDSCs by 8-fold, increased the mean area occupied by TAMs by 3-fold, and by mature dendritic cells by 23-fold (Fig. 4A; P < 0.001). Similar changes were observed for LNCaP tumors (Supplementary Fig. S4B). FACS analyses showed that 2aG4 treatment increased the percentage of CD11b+ cells in PC3 tumors that expressed costimulatory molecules MHC class II, CD40, CD80, and CD86 by 2- to 4-fold, providing further evidence that 2aG4 treatment drives the maturation of APCs (Fig. 4B). qRT-PCR and ELISA analyses showed that 2aG4 treatment of PC3 tumors shifted the balance of cytokines in the microenvironment from predominantly immunosuppressive (IL-10- and TGF-β–dominated) to predominantly immunostimulatory (IL-12-, TNF-α–dominated; Fig. 4C), further confirming the data shown in Fig. 3E. Taken together, these data indicate that 2aG4 treatment reactivates antitumor immunity at multiple levels.
Treatment with 2aG4 induces repolarization and activation of TAMs by binding to cell-surface phosphatidylserine in an Fc-dependent manner
A possible explanation for the large increase in M1:M2 ratio in 2aG4-treated tumors was that 2aG4 directly induces TAM repolarization. To test this, TAMs were isolated from PC3 tumors from untreated mice and incubated with 2aG4 in vitro. FACS analyses showed that TAMs cultured for 4 days in the presence of 2aG4 switched phenotype from predominantly M2-like (F4/80+ and Arg-1+) to predominantly M1-like (F4/80+ and iNOS+) that secreted NO (Fig. 5A and B). This effect was not seen with the F(ab′)2 fragment of 2aG4, indicating that the switch is dependent on Fcγ receptors. TAMs cultured in the presence of C44 did not switch phenotype. qRT-PCR analyses confirmed that TAMs cultured in the presence of 2aG4 had increased mRNA-encoding iNOS, inflammatory cytokines (IL-12 and TNF-α), and T-cell costimulatory molecules (CD80, CD86, and MHC class II) and decreased mRNA-encoding Arg-1, immunosuppressive cytokines (IL-10 and TGF-β), and VEGF-A (Fig. 5C).
We next determined whether TAMs have exposed phosphatidylserine. FACS analyses showed that freshly isolated TAMs from PC3 tumors bound to 2aG4 specifically (Fig. 5D). Electron microscopy studies revealed that 2aG4 does not bind directly to the plasma membrane of TAMs but to microvesicles attached to the cell surface (Fig. 5E). The microvesicles ranged in diameter from 100 to 500 nm, and they lacked the exosomal marker, Alix. Forty percent of the microvesicles carried one or more 2aG4-labeled gold particles. Control C44-labeled gold particles did not bind microvesicles or TAMs (data not shown).
Treatment with 2aG4 in vitro induces MDSCs differentiation into M1-like macrophages and dendritic cells
A possible explanation for the large decrease in MDSCs in 2aG4-treated tumors was that 2aG4 directly induces MDSC differentiation. To test this hypothesis, freshly isolated monocytic MDSCs from spleens of tumor-bearing mice were cultured with 2aG4, F(ab′)2 fragment of 2aG4, or C44 control antibody for 5 days. 2aG4-treated monocytic MDSCs differentiated into macrophages and dendritic cells (Fig. 6A). After 5 days, only 10% of the 2aG4-treated MDSCs retained their MDSC phenotype (Gr-1+ and CD11b+) as compared with 60% of cells treated with C44 or F(ab′)2 fragment of 2aG4. Fifty percent of cells in the 2aG4-treated cultures had a macrophage phenotype (CD11b+ and F4/80+) and 30% of cells had a dendritic cell phenotype (CD11b+ and CD11chi; Fig. 6A). Neutrophils (F4/80−, Gr-1+, and CD11bLo) did not accumulate. Cells in 2aG4-treated cultures synthesized high levels of NO (Fig. 6B). These effects were not seen with cells cultured with the C44 antibody or the F(ab′)2 fragment of 2aG4, indicating that the induction of differentiation is dependent on Fcγ receptors. Cells in 2aG4-treated cultures synthesized high levels of inflammatory cytokines, IL-6, TNF-α, and IL-12, but low levels of IL-10, consistent with the phenotype of M1-like macrophages (Fig. 6C). In contrast, cells in C44-treated cultures synthesized low levels of inflammatory cytokines, IL-6, TNF-α, and IL-12, but high levels of IL-10, consistent with the phenotype of M2-like macrophages (Fig. 6C). qRT-PCR analyses confirmed that monocytic MDSCs cultured in the presence of 2aG4 had increased expression of mRNA-encoding macrophage marker (F4/80), dendritic cell marker (CD11c), iNOS, inflammatory cytokines (IL-12 and TNF-α), and T-cell costimulatory molecule (CD86), and decreased expression of mRNA-encoding Arg-1, and immunosuppressive cytokine (TGF-β; Fig. 6D). These findings indicate that 2aG4 promotes the differentiation of MDSCs into M1-like macrophages and dendritic cells.
We next determined whether monocytic MDSCs have exposed phosphatidylserine. FACS analyses showed that freshly isolated monocytic MDSCs from tumor-bearing mice bound to 2aG4 specifically (Fig. 6E). Electron microscopy studies revealed that 2aG4 does not bind directly to the plasma membrane of MDSCs but to the microvesicles attached to the cell surface (Fig. 6F).
Recent evidence indicates that exposure of phosphatidylserine in the tumor microenvironment contributes to the immunosuppressed state of tumors (36, 43, 44). This suggests that the benefit of chemotherapy, radiotherapy, and other treatments that trigger tumor cell apoptosis is undermined by the increase in local tumor immunosuppression caused by phosphatidylserine exposed on dying tumor cells and their microvesicles (25, 43, 45). Here, we show that treatment of tumor-bearing mice with a phosphatidylserine-targeting antibody counteracts the tumor immunosuppression caused by chemotherapy and activates innate tumor immunity.
We show that 2aG4 functions at several levels to restore tumor immunity: (i) TAMs, which are predominantly in an immunosuppressive M2-like state in untreated or docetaxel-treated tumors, become tumoricidal M1-like TAMs; (ii) highly immunosuppressive monocytic MDSCs in the tumor become depleted, whereas their M1-like TAM and dendritic cell progeny increases; and (iii) immature dendritic cells in tumors become mature and express T-cell costimulatary molecules, indicating their potential to function as APCs. It should be noted that as this study was carried out in immunodeficient animals, the impact of Tregs and other memory T cells on the phenotype-switch of the myeloid infiltrate cannot be assessed. However, our previous studies have shown that 2aG4 treatment allows dendritic cells to present tumor antigens and generate glioma-specific cytotoxic T cells in an immunocompetent rat glioma model (36). These results suggest that 2aG4 treatment reactivates both innate and adaptive tumor immunity.
The M1-like TAMs generated by 2aG4 treatment caused the destruction of tumor endothelium, vascular shutdown, and tumor cell death. M1-like TAMs were the only cell type observed in contact with intact and disintegrating vascular endothelium. Most likely, M1-like TAMs bind via activating Fcγ receptors to the antibody-coated endothelial cells and kill them by antibody-dependent cell-mediated cytotoxicity. Indeed, we have previously shown that macrophages lyse 2aG4-coated vascular endothelial cells in an Fc-dependent manner in vitro (36). 2aG4 does not mediate direct lysis of phosphatidylserine-expressing endothelial cells by complement (mouse or human). We attempted to deplete TAMs by systemic administration of liposomal clodronate but found, as have others (46), that TAMs were not depleted, most likely because of rapid liposome clearance by the liver and spleen. In addition to their vascular-damaging action, we show here that the M1-like TAMs synthesize NO and efficiently kill PC3 tumor cells in vitro, suggesting that they have direct tumoricidal activity in vivo.
One of the mechanisms by which 2aG4 induces TAM repolarization to an M1-like state is by binding to phosphatidylserine on the cell surface of TAM in an Fc-dependent manner. Our electron microscopy studies show that phosphatidylserine on the cell surface of TAM is due to the presence of phosphatidylserine-expressing microvesicles. It is likely that the phosphatidylserine-expressing microvesicles bind to phosphatidylserine receptors on the cell surface, sending signals that maintain TAMs in an anti-inflammatory M2-like state, similar to apoptotic cells (19–21). We hypothesize that 2aG4 binds to the microvesicles and ligates activating Fcγ receptors on the same cell or adjacent cells, sending signals that override the anti-inflammatory phosphatidylserine receptor signal and activate M1 differentiation (Fig. 7). The identity of the phosphatidylserine receptors on TAMs responsible for the anti-inflammatory signals is unknown, but it could be Tim3 (30). Tim4-positive cells have been observed to bind microvesicles (47). Thus, 2aG4 could also block the binding of phosphatidylserine-positive microvesicles to phosphatidylserine receptors on TAMs in vivo. Although we cannot rule out the possibility that M1 progenitors are recruited from the blood, our data indicate that repolarization of resident M2-TAMs is the primary mechanism.
The reduction in MDSCs and the overall increase in the number of TAMs and mature dendritic cells in 2aG4-treated tumors suggest that differentiation of MDSCs is also inhibited by phosphatidylserine-expressing microvesicles in the tumor microenvironment. We speculate that 2aG4 binds to and stimulates MDSCs differentiation into TAMs and dendritic cells. We isolated monocytic MDSCs (CD11b+, Ly6Glo, and Ly6Chi) from the spleens of tumor-bearing mice (48), and cultured them with 2aG4 in the absence of additional growth factors. We found that 2aG4 treatment in vitro induces the differentiation of monocytic MDSCs into M1-like macrophages and dendritic cells (Fig. 6). These progeny secreted NO and their cytokine profile switched from immunosuppressive to immunostimulatory. Electron microscopy studies confirmed that MDSCs, like TAMs, are phosphatidylserine-positive due to the presence of phosphatidylserine-expressing microvesicles on their surface (Fig. 6). It is likely that dendritic cells are also prevented from maturing in tumors by the exposed phosphatidylserine, as dendritic cells have exposed phosphatidylserine and annexin 5A facilitates dendritic cell maturation in vitro (49). 2aG4 may stimulate dendritic cell maturation in tumors by a mechanism analogous to that shown in Fig. 7.
Bavituximab, the human chimeric version of 2aG4, in combination with chemotherapy, is being used to treat patients with cancer in randomized clinical trials (38–40). Impressive antitumor effects in patients with cancer have been obtained with other antibodies that enhance tumor immunity, including anti-PD1, anti-PD-1L, and ipilimumab (anti-CTLA-4; ref. 50). Unlike these antibodies, which inhibit negative feedback pathways in immune cell activation, bavituximab seems to act by reversing the immunosuppressive effects of exposed phosphatidylserine in the tumor microenvironment, resulting in activation of local tumor immunity and damage to tumor vasculature.
Disclosure of Potential Conflicts of Interest
X. Huang and P.E. Thorpe are consultant/advisory board members of Peregrine Pharmaceuticals, Inc. No potential conflicts of interest were disclosed by the other authors.
Conception and design: Y. Yin, X. Huang, K.D. Lynn, P.E. Thorpe
Development of methodology: Y. Yin, X. Huang, P.E. Thorpe
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Yin, X. Huang, K.D. Lynn, P.E. Thorpe
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Yin, X. Huang, K.D. Lynn, P.E. Thorpe
Writing, review, and/or revision of the manuscript: Y. Yin, X. Huang, K.D. Lynn, P.E. Thorpe
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Huang, P.E. Thorpe
Study supervision: X. Huang, P.E. Thorpe
The authors thank Dan Ye, Shuzhen Li, and Janie Iglehart for technical assistance and Drs. Alan Schroit, E. Sally Ward, and Rolf A. Brekken for discussions and comments on the article.
This work was supported by a sponsored research agreement with Peregrine Pharmaceuticals, Inc., Department of Defense grants PC05031 (to P.E. Thorpe) and PC080475 (to Y. Yin), an NIH-supported Small Animal Imaging Research Program at University of Texas Southwestern (U24CA126608), and the Gillson Longenbaugh Foundation.
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.