Blockade of Tumor Necrosis Factor α Signaling in Tumor-Associated Macrophages as a Radiosensitizing Strategy

Radiation therapy is a common cancer treatment usually delivered in a fractionated schedule (daily 1.8–3 Gy) for 6 to 8 weeks. In some selected lung cancers, or limited forms of metastasis, very large doses (one to five treatments of 10–20 Gy) may be used (1). Despite research into the radiobiology and physics of radiotherapy, many patients fail within the irradiated target volume. Currently, the most effective radiosensitizers represent the commonly used chemotherapeutic agents (2). These drugs are not designed as selective radiosensitizers, and clinical improvement reflected by increased local control is achieved at the expense of significant normal tissue toxicity. More selective radiosensitizing agents and new treatment strategies are required to improve the therapeutic index in radiotherapy.

Research on the mechanisms of tumor radioresistance and the development of radiosensitizers focuses on tumor cells or exploiting differences in the oxygenation status of tumors and normal tissue (3). Recently, the tumor-associated stroma has gained attention as an important component of the response to radiotherapy (4, 5) and investigations have centered on the role of the tumor-associated vasculature (6, 7). Several clinical trials are under way to investigate the role of angiogenesis inhibitors in combination with radiotherapy (8–11). The results are preliminary and, although promising, show mixed results in terms of improving radiotherapy outcomes and unexpected toxicities have been reported (11–13). Conversely, little is known about other components of the tumor-associated stroma and the effect of these cells on the response of tumors to radiotherapy.

Macrophages are multifunctional cells of the myeloid lineage, phenotyped as CD14⁺. Macrophages engulf microbes and cell debris, are antigen-presenting cells, and secrete cytokines and chemokines. Macrophage function is described within the context of normal physiologic or pathologic microenvironments. Macrophages are designated as M1 classically activated macrophages or M2 alternatively activated macrophages. Proinflammatory M1 macrophages are activated by lipopolysaccharide and IFNγ, secrete tumor necrosis factor (TNF) α and interleukin (IL)-12, and support T-cell function. M2 macrophages are considered to be anti-inflammatory and immunosuppressive, expressing F4/80, CD11b, and CD206; secrete IL-10, IL-4, and transforming growth factor-β; and downregulate T-cell function (14, 15). Tumor-associated macrophages (TAMϕ) are usually characterized as M2. However, functional overlap exists between characteristics of M1 and M2 macrophages. Additional/expanded definitions of TAMϕ populations have been investigated as determinants of tumor progression and metastasis. These populations include CD11b⁺/Gr1⁺ myeloid suppressor cells as well as Tie2⁺-expressing macrophages (15, 16). Emerging data suggest key roles for these various monocyte/macrophage cells in generating proangiogenic signals during tumor growth.

Radiation was originally reported to prime the antitumor effects of macrophages through production of TNFα and nitric oxide (NO; refs. 17–20). However, Milas and colleagues investigated the association between TAMϕ and the response of tumors to radiotherapy and noted a trend toward decreasing radiocurability with increasing TAMϕ content. These investigators recognized the contradiction between potential beneficial (phagocytic function) and deleterious functions (proangiogenic phenotypes) of TAMϕ (21–23). Recently, Tsai and colleagues (24) reported that irradiated macrophages contributed to tumor growth through increased secretion of cyclooxygenase-2, inducible NO synthase (iNOS), and Arg-1, although no direct mechanisms for promotion of tumor growth were described. Ahn and Brown (25) reported that cells of the myeloid/macrophage lineage contributed to tumor regrowth following radiation by increasing tumor vasculogenesis when tumor angiogenesis was suppressed by prior irradiation of the tumor bed. Collectively, these studies raise the possibility that macrophage populations contribute to tumor radioprotection. Moreover, radiation of the tumor microenvironment might upregulate macrophage-derived cytokines that promote tumor growth, survival, angiogenesis, and vasculogenesis.

TNFα, which has antitumor effects at high concentrations, also promotes tumor angiogenesis, tumor cell survival, and metastases at lower levels (26). A recent report suggested that TNFα mediates the differentiation of monocytes into angiogenic cells that support tumor angiogenesis (27–29). Radiotherapy, although directly inducing tumor cell death, may upregulate proangiogenic and prosurvival factors within the tumor microenvironment. We and others have found that radiation upregulates TNFα production by tumor cells and cells of the myeloid lineage (30, 31). Here, we report that depletion of macrophages by liposomal clodronate before irradiation increases the antitumor effects of ionizing radiation (IR), whereas coimplantation of tumor cells with bone marrow–derived macrophages (BMDMϕ) results in increased tumor radioresistance. The radioprotective effect of BMDMϕ requires functional TNFα/TNFR signaling and induction of macrophage-secreted vascular endothelial growth factor (VEGF). We report that blockade of macrophage VEGF induction by Enbrel (soluble TNF receptor dimeric fusion protein), used in several murine studies, increased tumor radiosensitivity (32, 33). We also report that blocking VEGF with neutralizing antibodies improves the antitumor effects of IR. These data provide a rationale for targeting macrophage populations generally and TNFα-induced VEGF signaling specifically when designing radiotherapeutic strategies.

C57BL/6 mice wild-type (WT), C57BL/6-129S-Tnfrsf1a^tm1ImxTnfrsf1b^tm1Imx/J (TNFR1,2^−/−), B6:-129S-Tnf^tm1Gkl/J (TNF^−/−), and C57BL/6-Tg (UBC-GFP) 30SchaJ (GFP^+/+) breeding pairs were purchased from The Jackson Laboratory. TNF^−/− mice maintained in the 129/SvEv and C57BL/6 backgrounds after six-generation backcross have been previously used to study the effects of TNFα on tumor promotion and antigen presentation (29, 34). B16F1 melanoma cells were cultured as described (35) B16.SIY melanoma cells expressing model antigen SIY, which can be recognized by CD8⁺ T cells in the context of K^b, were cultured as described. The care and treatment of experimental animals was in accordance with institutional guidelines.

Femoral BM cells were obtained from mice and cultured in complete RPMI 1640 supplemented with 10% FCS and 30% pretested conditioned medium from the L929 cell line as a source of macrophage colony-stimulating factor (M-CSF) for the first 5 d followed by complete RPMI 1640 supplemented with 10% FCS and 10 ng/mL recombinant M-CSF (R&D Systems) for additional 5 d. These BMDMϕ were >95% CD14⁺ and >90% CD11b⁺F4/80⁺, analyzed by fluorescence-activated cell sorting (FACS).

Tumor cells (5 × 10⁵ B16F1 or B16.SIY) were inoculated s.c. into the right hind limb. Tumors were measured with calipers and volume was calculated as length × width × depth/2. At days 10 to 12, local radiotherapy (single dose, 20 Gy; fractional doses, 2 Gy × 10) was delivered. In some experiments, B16F1/B16.SIY cells were coinjected with BMDMϕ at a 4:1 ratio. Elsewhere, macrophages were depleted by i.p. or intratumoral administration of liposomal clodronate every 5 to 7 d (36). Cl₂MDP (or clodronate) was a gift of Roche Diagnostics GmbH (37). Other reagents include phosphatidylcholine (Lipoid GmbH) and cholesterol (Sigma). Macrophage depletion was confirmed by FACS and immunohistochemistry of spleen and tumor cryosections with >90% reduction compared with the control. Blockade of VEGF using neutralizing IgG against mouse VEGF-164 (R&D Systems) has been previously described (38). Briefly, goat IgG against mouse VEGF-164 was suspended in PBS and administered via i.p. injection (10 μg/mouse, 3 h before IR and 3 and 8 d after IR). Control mice received goat IgG (Sigma).

Two hundred control or irradiated B16F1 or B16.SIY cells were seeded in 5-cm culture dishes in RPMI 1640 containing 10% FCS with or without 30% spent supernatant collected from either WT, TNF^−/−, or TNFR1,2^−/− BMDMϕ (5 × 10⁶ cells). After 7 d, cells were washed and stained with crystal violet. Colonies with >50 cells were counted.

Spent culture supernatants were collected and incubated with membranes coated with 62 anti-mouse cytokine antibodies (RayBiotech, Inc.) according to the manufacturer's instruction. An antibody labeled with biotin coated on the upper left and lower right corners of the membrane served as positive control. The film was scanned and spots were quantified by densitometry analysis using UN-SCAN-IT gel automated digitizing system software (Silk Scientific). Thirty-two cytokines were quantified using Mouse Cytokine/Chemokine Premixed 32 Plex (Millipore). Median fluorescence intensity from each well was acquired, and the relative concentration of each cytokine/chemokine was calculated.

Tumors were excised, sectioned, and digested in DMEM supplemented with 2% FCS and 1.5 mg/mL collagenase D (Sigma) for 30 min in a 37°C shaking incubator to collect single-cell suspension. Cells were stained with anti-CD11b, anti-CD206, anti-F4/80, anti-TNFR1, and anti-TNFR2 (BioLegend); washed; and analyzed on a LSR flow cytometer. Frequency of CD11b⁺ and F4/80⁺ TAMϕ was analyzed and sorted on a MoFlow to collect TAMϕ. Purity reached >95% CD11b⁺ F4/80⁺ cells.

Cells were lysed with cell lysis buffer containing 20 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 1 mmol/L Na₂EDTA, 1 mmol/L EGTA, 1% Triton, 2.5 mmol/L sodium pyrophosphate, 1 mmol/L glycerophosphate, 1 mmol/L Na₃VO₄, 1 μg/mL leupeptin, and protease inhibitor mixture (Roche Applied Science). Protein concentrations were determined using the Bio-Rad protein assay kit, and 10 μg from each sample were analyzed on 10% SDS-PAGE gels. Protein bands were transferred onto polyvinylidene difluoride membranes and probed with rabbit anti-mouse VEGF (Santa Cruz Biotechnology) for 2 h and then horseradish peroxidase–conjugated anti-rabbit for 1 h. Membranes were washed and immunodetected using ECL kit (Amersham Biosciences). Anti–β-actin-stained membranes were used as loading control.

Tissues were fixed in 10% neutral formalin or 4% paraformaldehyde. Sections were blocked with 5% goat serum and incubated with VEGFR2 antibody (1:200 dilution; Cell Signaling Technology) for 1 h. After washing with TBS, biotinylated anti-rabbit secondary antibody followed by avidin-biotin complex–alkaline phosphatase (ABC-AP; Vector Laboratories) was applied for 30 min each. Vector blue substrate was used for color development. For the following NG2 chondroitin sulfate proteoglycan staining, the slides were washed and blocked with avidin-biotin blocking kit and then 10% goat serum for 10 min. Slides were incubated in 1:100 dilution of NG2 antibody (Millipore) overnight at 4°C. After applying biotinylated secondary antibody and ABC-AP solution, tissues were covered with Vulcan fast red substrate (Biocare Medical) for 15 min. Slides were washed and counterstained with methyl green. Sections were dehydrated in 100% ethanol, cleared by Histoclear (Wards Natural Science), and then mounted with VectaMount (Vector Laboratories). Images were photographed at ×200 magnification using a Zeiss camera operated by Openlab software.

A random-effects model for longitudinal data was used to obtain an overall estimate of the intercept and slope of linear growth for each group. One-way ANOVA with Dunnett's post test was performed using GraphPad InStat version 3.05 (P < 0.05).

Macrophages were depleted by i.p. injection of liposomal clodronate (37) 1 day before injection of 2 × 10⁶ B16.SIY cells and every 5 days thereafter in WT mice. Ninety percent of the macrophages were specifically depleted in the spleen and tumor microenvironment as analyzed by FACS (Supplementary Fig. S1; refs. 36, 39, 40). Twenty Gray was administered when tumors reached 150 to 200 mm³. The combination of liposomal clodronate and 20 Gy significantly delayed tumor regrowth compared with 20 Gy alone (335 ± 207 mm³ versus 3,215 ± 1,849 mm³, day 22; P = 0.002) or liposomal clodronate alone (273 ± 198 mm³ versus 4,987 ± 2,556 mm³, day 18; P = 0.041; Fig. 1A). Liposomal clodronate also enhanced the antitumor effects of IR after tumors were established compared with 20 Gy alone (2,678 ± 1,243 mm³ versus 4,599 ± 889 mm³, day 22; P = 0.071; Fig. 1B). We also injected liposomal clodronate intratumorally and observed that tumor regrowth was significantly delayed compared with IR alone (1,669 ± 749 mm³ versus 5,317 ± 1,322 mm³, day 26; P = 0.037; Fig. 1C).

TNFα is reported to both enhance tumor growth and mediate antitumor effects. To study the role of TNFα produced by macrophages, we cultured BMDMϕ from WT and TNF^−/− mice and coinjected these BMDMϕ with 5 × 10⁵ B16.SIY cells into WT mice. TNF^−/− and WT BMDMϕ exhibit similar levels of cell surface markers with >95% CD14 and 90% CD11b detected by FACS. They also expressed similar levels of iNOS but TNF^−/− BMDMϕ expressed significantly lower levels of Arg-1 (data not shown). The percentages of F4/80⁺ TAMϕ were similar between the WT and TNF^−/− groups, although TNF^−/− cells had diminished levels of surface CD206 staining (Supplementary Fig. S2). Coinjection of WT BMDMϕ with B16.SIY cells significantly accelerated tumor regrowth after 20 Gy compared with 20 Gy alone (1,384 ± 553 mm³ versus 125 ± 36 mm³, day 22; P = 0.030; Fig. 2A and B). The regrowth of irradiated tumors in which B16.SIY cells were coinjected with WT BMDMϕ occurred significantly earlier compared with irradiated tumors in which B16.SIY cells were coinjected with TNF^{−/−−/−} BMDMϕ (1,384 ± 553 mm³ versus 337 ± 261 mm³, day 22; P = 0.037; Fig. 2B and C). Our data show that depletion of macrophages enhances the antitumor effects of IR and coimplantation of macrophages reverses this effect. These results also implicate macrophage-secreted TNFα in the resistance of tumors to IR and suggest that macrophage-derived TNFα and/or TNFα signaling in TAMϕ contributes in part to B16.SIY tumor radioresistance.

To investigate the role of TNFα produced specifically by BMDMϕ, we injected B16.SIY cells into TNF^−/− mice transplanted with WT BM or TNF^−/− BM. FACS analysis showed that the BM was reconstituted with >70% of donor cells (Supplementary Fig. S3). Animals underwent macrophage depletion with liposomal clodronate and tumor growth was compared with controls. Tumor volume in WT BM transplant (BMT) mice was significantly reduced by macrophage depletion (3,067 ± 615 mm³ versus 825 ± 174 mm³, day 60; P = 0.017; Supplementary Fig. S4A). No tumor growth was observed in TNF^−/− BMT mice, showing an essential role for TNFα produced by BMDMϕ. We next injected B16.SIY cells into TNFR1,2^−/− mice after WT or TNFR1,2^−/− BMT with or without macrophage depletion. WT BMT in TNFR1,2^−/− mice promoted tumor growth 30 days earlier than in TNFR1,2^−/− BMT mice, which was significantly delayed when macrophages were depleted (867 ± 64 mm³ versus 4,353 ± 888 mm³, day 45; P = 0.013; Supplementary Fig. S4B). No significant difference was observed in TNFR1,2^−/− mice with TNFR1,2^−/− BMT when macrophages were depleted (65 ± 15 mm³ versus 73 ± 18 mm³, day 45; P = 0.549). In summary, tumors grew more slowly in TNFR1,2^−/− BMT mice, whereas WT BMT permitted tumor growth. However, tumor growth was suppressed when macrophages were depleted. These results suggest that autocrine/paracrine TNFα signaling in BMDMϕ is essential in promoting tumor growth.

We coinjected BMDMϕ from WT or TNFR1,2^−/− with B16.SIY cells into TNFR1,2^−/− mice. Coinjection of WT BMDMϕ significantly accelerated tumor regrowth after 20 Gy compared with IR alone (782 ± 179 mm³ versus 78 ± 14 mm³, day 22; P = 0.003; Fig. 3A and B). The regrowth of irradiated tumors coinjected with WT BMDMϕ was significantly accelerated compared with the response of tumors coinjected with TNFR1,2^−/− BMDMϕ (782 ± 179 mm³ versus 283 ± 157 mm³, day 22; P = 0.041; Fig. 3B and C). These results indicate that intact TNF/TNFR signaling in macrophages is required for accelerated tumor regrowth after IR. We repeated these experiments using fractionated IR and report that the growth of tumors coinjected with WT BMDMϕ was significantly increased compared with tumors coinjected with TNFR1,2^−/− BMDMϕ (384 ± 64 mm³ versus 38 ± 7 mm³, day 22; P = 0.010; Supplementary Fig. S5). These data further suggest that TNFα/TNFR signaling in BMDMϕ mediates radioresistance.

We explored the direct effects of BMDMϕ on B16.SIY tumor cell radiosensitivity and/or growth in vitro. Supernatant from WT BMDMϕ suppressed B16.SIY colony formation (P = 0.015), whereas supernatant from TNF^−/− or TNFR1,2^−/− BMDMϕ had no effect (P = 0.259 and 0.338; Supplementary Fig. S6). Unexpectedly, supernatants from TNF^−/− and TNFR1,2^−/− BMDMϕ cultures increased colony formation in irradiated cells (P = 0.065 and 0.055). Interestingly, the radioprotective effect of TNF^−/− or TNFR1,2^−/− supernatants in vitro differs from the in vivo findings with TNF^−/− or TNFR1,2^−/− macrophages. Supernatant from WT BMDMϕ had no effect on irradiated B16.SIY colony formation (P = 0.890). Supernatant collected from irradiated WT, TNF^−/−, or TNFR1,2^−/− BMDMϕ had no significant effect on either control or irradiated B16.SIY growth. These results suggest that the radioprotective effects of TNFα signaling in BMDMϕ are not exerted directly on tumor cells but likely on nontumor cell constituents of the tumor microenvironment.

In addition to TAMϕ, tumor stroma is also composed of matrix proteins and various cell types, including blood/lymphatic vessels (41). Recent data suggest that TAMϕ support tumor growth by contributing to angiogenesis and/or vasculogenesis (41–44), in part mediated by TNFα. We used a protein array and examined 62 cytokines and chemokines in unirradiated BMDMϕ and BMDMϕ treated with 5 Gy. Unirradiated WT and TNFR1,2^−/− BMDMϕ produced similar cytokine/chemokine levels, including M-CSF, granulocyte CSF, granulocyte macrophage CSF, CCL2, CCL9, IL-6, CXCL2, IL-10, TNFα, and IL-12, and low levels of VEGF. Following 5 Gy, there was a significant increase of VEGF in WT BMDMϕ but not TNFR1,2^−/− BMDMϕ, whereas TNFα was induced in both WT and TNFR1,2^−/− BMDMϕ (Fig. 4A). These results were confirmed by Luminex (Fig. 4B) and suggest that the induction of VEGF by IR is dependent on TNFα/TNFR autocrine/paracrine signaling in BMDMϕ. Our findings support the hypothesis that VEGF production in TAMϕ through TNFα signaling activated by IR might play an important role in tumor vessel repair and tumor regrowth.

We examined if irradiation leads to TNF/TNFR-mediated upregulation of VEGF in tumor macrophages. We injected B16.SIY cells into WT and TNFR1,2^−/− mice and 20 Gy was delivered when tumors reached 150 to 200 mm³. Tumors were excised and digested into single-cell suspensions. CD11b⁺ TAMϕ were sorted and VEGF expression was assayed by Western blot and Luminex. Significantly higher levels of VEGF were detected in CD11b⁺ TAMϕ isolated from tumors grown in WT mice compared with TNFR1,2^−/− mice (P = 0.015; Fig. 5A and B). Increased TAMϕ VEGF from tumors in WT mice was mirrored by increased tumor neovasculature/angiogenesis after IR was visualized in H&E- and VEGFR2-stained tissue sections (Fig. 5C). We quantified functional vascular structures showing intact blood perfusion by the presence of RBCs in VEFGR2⁺ vessels. In tumors grown in TNFR1,2^−/− mice, there were significantly decreased functional vessels after IR compared with either untreated control or tumors grown in WT after IR (P < 0.0001 and P = 0.002, respectively; Fig. 5D). In addition, more surviving tumor cells exhibiting intact morphology were present in irradiated tumors in WT compared with tumors in TNFR1,2^−/−, which showed greater radiosensitivity.

To test this hypothesis that TNFα blockade might be clinically relevant, we injected B16.SIY cells into WT mice and treated them with Enbrel + 20 Gy (32, 33). Enbrel-treated animals had slightly larger tumors than untreated control animals. Mice treated with Enbrel + 20 Gy exhibited a reduction in tumor regrowth compared with 20 Gy alone (153 ± 34 mm³ versus 440 ± 97 mm³, day 28; P = 0.022; Fig. 6A).

The observation that IR induced VEGF production by macrophages via TNFα/TNFR signaling pathways led us to examine the effect of pharmacologic TNFα inhibition and radiation on tumor angiogenesis. As with genetic deletion of TNFR1,2, Enbrel treatment prevented upregulation of VEGF by tumor macrophages in vivo (Fig. 6B and C). We next assessed the effect of IR + Enbrel on neovascularization. Radiation alone led to a significant reduction in microvessel density as measured by VEGFR2⁺ [PBS (24 ± 1.8) versus PBS + IR (4.1 ± 1.4); P < 0.0001]. Enbrel alone led to a modest decrease in microvessel density [PBS (24 ± 1.8) versus Enbrel (13 ± 1); P < 0.01], whereas Enbrel + IR significantly inhibited angiogenesis when compared with all groups (2.75 ± 0.35 versus Enbrel alone, 13 ± 1, P < 0.0001; versus PBS + IR, 4.1 ± 1.4, P < 0.01; Fig. 6D). NG2⁺ pericyte coverage was reduced in tumors treated with Enbrel compared with PBS (84 ± 2.3% versus 98 ± 0.2%; P < 0.001) and further decreased with Enbrel + IR (57 ± 22.4%; P = 0.03, compared with Enbrel alone). Vascular size and morphology were also substantially affected by treatment. For example, Enbrel + IR–treated tumors showed narrow lumens suggestive of vascular collapse. Finally, whereas radiation exposure led to vascular hemorrhage, the addition of Enbrel intensified hemorrhagic necrosis and regression. These data show a profound effect of TNFα depletion + IR on angiogenesis and vascular function. Given that irradiated macrophages upregulated VEGF expression, we next hypothesized that blockade of VEGF would likewise enhance radiosensitivity. Treatment of mice with neutralizing antibodies to VEGF led to increased radiosensitivity in B16F1 (806.1 ± 329.1 versus IR alone, 2,101.8 ± 525.4; P = 0.05) and B16.SIY (0.5 ± 0.1 versus IR alone, 408.5 ± 65.1; P = 0.005) tumors (Supplementary Fig. S7; ref. 38).

We tested the hypothesis that depletion of TAMϕ by liposomal clodronate would improve the antitumor effects of radiotherapy as determined by a delay of tumor regrowth. The effects of macrophage depletion were most marked when macrophages were depleted before tumors were established. TAMϕ have been reported to secrete a variety of cytokines and other proteins that promote tumor growth. Our results show that the intact TNFα/TNFR signaling in TAMϕ is important in the autocrine/paracrine secretion of cytokines by BMDMϕ. TNFα is a proinflammatory cytokine that is cytotoxic to some tumor cell lines and tumors in vivo in high concentrations (45, 46). In our study, we were unable to show that radioresistance was mediated by a direct effect on B16.SIY cells using supernatant from WT BMDMϕ. This raised the possibility that TNFα-stimulated TAMϕ promoted radioresistance by effects on the tumor microenvironment. TNFα at physiologic concentrations has recently been implicated in cancer induction and support of tumor angiogenesis, in part, through the attraction of BM-derived myeloid precursor cells that contribute to tumor blood vessel formation and stabilization (42, 47). We investigated if TNF/TNFR signaling in macrophages regulates the production of angiogenic cytokines by irradiated macrophages. Our data show that inhibition of TNFR signaling by both genetic and pharmacologic means prevents increased production of VEGF by irradiated macrophages. Immunohistochemical analyses show that the combination of TNFα inhibition plus IR led to both significant decreases in neovascularization as well as vascular function as evidenced by reduced pericyte coverage and increased hemorrhage.

A recent report noted that macrophage-secreted VEGF paradoxically slowed tumor growth in part by inducing tumor vessel abnormalities, such as tortuosity and leakiness, resulting in tumor hypoxia, whereas tumor-secreted VEGF “normalized” tumor vasculature (48). Interestingly, when macrophage VEGF was deleted, tumors grew more rapidly due to a more “normalized” vasculature but were also more sensitive to cyclophosphamide. Our results are consistent with this report, showing the Enbrel-treated tumors were actually more sensitive to IR. In contrast to the report of Stockmann and colleagues (48), TNFα blockade did not increase vessel pericyte coverage, suggesting that TNFR signaling in tumor macrophages likely affects several angiogenic pathways in addition to VEGF. Nonetheless, blockade of VEGF upregulation by inhibition of TNFα signaling represents an alternative clinically relevant method to enhance radiosensitivity via targeting of TAMϕ. Paradoxically, supernatant from macrophages defective in TNF signaling through germline deletions in TNF or TNFR1,2 conferred modest radioprotection on B16.SIY cells in vitro through an unknown mechanism. Together, these data suggest that macrophage blockade mediates in vivo radiosensitivity predominantly through effects on the microenvironment. Active TNFα signaling in irradiated macrophages seems necessary for upregulation of macrophage-derived VEGF, resulting in enhanced preservation of the irradiated vasculature.

Tumor angiogenesis is distinguished from postnatal vasculogenesis in that tumor angiogenesis is proposed to occur through endothelial migration and sprouting from preexisting blood vessels, whereas vasculogenesis is proposed to occur by the recruitment of BM cells to the site of tumor angiogenesis or local inflammatory damage. Although vasculogenesis was previously described as direct incorporation of progenitor cells into newly emerging vasculature, it also involves the recruitment of BM-derived angiogenic populations that enhance angiogenesis through paracrine mechanisms. The extent to which tumor neovascularization depends on local endothelial cells or infiltrating angiogenic cells is controversial. Recently, tumor regrowth by local myelomonocytic CD11b⁺ cells was reported to induce tumor vasculogenesis following irradiation through secretion of matrix metalloproteinase-9 in a model where tumor angiogenesis was suppressed by prior irradiation of the tumor bed (25). Additionally TAMϕ secrete a variety of proangiogenic proteins, including IL-8, TNFα, and VEGF. Therefore, both angiogenesis and vasculogenesis may be supported by TAMϕ-derived factors. Whereas our data using Enbrel suggest that pharmacologic inhibition of macrophage TNFR signaling enhances radiosensitivity, our liposomal clodronate data show that reducing macrophage populations in tumors represents another potential antiangiogenic and antivasculogenic strategy. Systemically delivering liposomal clodronate to patients may not be feasible. However, CSF-1 receptor kinase inhibitors may block macrophage differentiation and function, providing an alternative biological tool to inhibit proinflammatory cytokine production from macrophages (49). The availability of these small molecules and monoclonal antibodies targeting either CSF-1R or CSF-1 to deplete TAMϕ number and function may allow us to translate these strategies to promote radiosensitivity in future studies.

We also report that blocking VEGF enhances the effect of IR (38, 50). Although anti-VEGF therapy could block VEGF derived from either vasculogenic myeloid populations or tumor cells, our results confirm that direct VEGF blockade is another therapeutic strategy to increase radiosensitivity. Despite the clear correlation between tumor vascularization and VEGF expression, TNF can also modulate the expression of other antiangiogenic factors, including thrombospondin and angiostatic chemokines such as CXCL10 and CXCL9. Therefore, targeting the TNFα/TNFR signaling in TAMϕ may enhance radiosensitivity through additional pathways beyond VEGF signaling.

In summary, we used both genetic and pharmacologic inhibition of TNFα signaling to study the role of tumor macrophages in promoting tumor radioresistance. Rather than directly affecting tumor cells, active TNFα signaling in irradiated macrophages results in the production of angiogenic cytokines such as VEGF. We have used coimplantation models to show that macrophages are sufficient for this effect and liposomal clodronate to show that macrophages are required. Although liposomal clodronate might have off-target cellular affects, previous studies suggest that liposomal clodronate selectively depletes macrophages as opposed to other hematopoietic cells. Because the effects of liposomal clodronate on other stromal components remain unclear, we also used BM transplantation studies together with liposomal clodronate to study the effects of depletion of BM-derived cells. It nonetheless remains possible that active TNFα signaling in cellular components of the microenvironment might mediate radioresistance as well. Other limitations of our study include (a) the lack of a tissue-specific promoter such as LysM to delete TNFα specifically in myeloid cells, (b) identification of which BMDMϕ populations (M1 versus M2) mediate tumor vascular formation/stabilization following IR, and (c) whether VEGF secreted by macrophages attracts additional local angiogenic cells, which contribute to tumor vascularization and regrowth following radiotherapy. In spite of these caveats, our results suggest that blockade of TNF/TNFR signaling in TAMϕ is an attractive target to improve the efficacy of radiotherapy.

No potential conflicts of interest were disclosed.

Grant Support: National Cancer Institute grant CA111423 (R.R. Weichselbaum) and National Heart, Lung and Blood Institute grant K08HL071938 (K.S. Cohen).

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.