Intravenous Immunoglobulin Promotes Antitumor Responses by Modulating Macrophage Polarization

Macrophages exhibit a huge functional plasticity and can acquire a continuum of polarization states in response to endogenous and nonself stimuli (1–4). Microbe-derived factors, or cytokines like IFN-γ, GM-CSF, or TNF-α, promote in macrophages the acquisition of proinflammatory, bactericidal, tumor-suppressive, and immunogenic activities, a process commonly referred to as classical or M1 polarization and whose hallmark is the ability to release large amounts of IL-12/IL-23 (2). Conversely, cytokines such as IL-4, IL-10, TGF-β, or M-CSF promote anti-inflammatory, scavenging, tumor-promoting, tissue repair, and proangiogenic functions, all of which are grouped under the terms “alternative” or M2 polarization, which endows them with the ability to produce high levels of IL-10 (3–5). M1-polarized macrophages predominate at the initial stages of an inflammatory response, whereas M2-type macrophages drive resolution of inflammation and tissue repair after injury and maintain tissue homeostasis (6). In vivo, the misbalance of macrophage polarization states underlies numerous pathophysiological processes, including tumor development, autoimmune diseases, and chronic inflammatory pathologies (6–8).

The switch between M1 and M2 polarization states is especially relevant in tumor initiation, progression, and dissemination, which are extremely reliant on the presence and polarization state of macrophages within the tumor stroma (tumor-associated macrophages [TAM]) (5). The contribution of macrophages to tumor development is inferred from the poor outcome associated with enhanced levels of M-CSF, the reduced metastasis observed in Csf1^op/op mice (9, 10), and the positive correlation between high content of TAM and a bad prognosis (11). Depending on their polarization status, macrophages can either promote antitumor immune responses or contribute to tumor progression (12). In fact, as tumor progresses, TAM develop an immunosuppressive and protumoral phenotype, which fuels tumor growth, metastasis, and suppression of tumor-specific immune responses (13).

Intravenous Ig (IVIg) is a preparation of polyclonal and poly-specific Igs derived from the plasma of thousands of healthy donors. IVIg therapy is Food and Drug Administration approved for primary immunodeficiencies, immune thrombocytopenic purpura, and Kawasaki´s disease, and is beneficial for multiple sclerosis, rheumatoid arthritis, and systemic lupus erythematosus (14, 15). Previous reports have evidenced that IVIg exerts potent immunomodulatory actions in immunodeficiency syndromes, autoimmune diseases, and infectious processes (16). The molecular and cellular basis for the IVIg immunomodulatory action, including the identity of the biologically active constituents in IVIg and its specific cell surface receptors and target cells, remains to be completely clarified (17–20). In the present report, we demonstrate that IVIg promotes a M2-to-M1 macrophage polarization switch through ligation of activating Fc receptors in both human and mouse macrophages, and that FcγRIII, FcγRIV, and the FcRγ chain mediate the IVIg-induced repolarization of tumor-associated myeloid cells and inhibition of tumor progression and metastasis in vivo.

Human PBMC isolation was carried out, as described previously (21). Human TAM were obtained from the pleural fluid of a metastatic breast adenocarcinoma patient, after obtaining written informed consent, following Medical Ethics Committee procedures (Hospital General Universitario Gregorio Marañón), and using CD14 microbeads, as described (22). Human postnatal thymocytes were isolated from thymus fragments removed during corrective cardiac surgery of patients aged 1 mo to 4 y, after providing informed consent in accordance with the Declaration of Helsinki. Thymocyte cell suspensions were enriched in non-T cells by sheep erythrocyte resetting, as previously described (23). Intrathymic macrophages (>95% CD13⁺CD11c⁺CD14⁺) were obtained from the resulting cell fraction by positive selection using a PE-labeled anti-CD14 mAb and anti-PE microbeads (Miltenyi Biotec). Bone marrow–derived macrophages (BMDM) were obtained, as described previously (24, 25). For activation, macrophages were treated with Escherichia coli 055:B5 LPS (100 ng/ml for mouse macrophages; 10 ng/ml for human macrophages) for 24 h. The B16F10 mouse melanoma cell line (C57BL/6 origin), the mouse MC38 colon carcinoma cancer line, and the highly invasive human BLM melanoma cell line (provided by J. Teixidó, Centro de Investigaciones Biológicas/Consejo Superior de Investigaciones Cientificas) were maintained in RPMI 1640 (MC38) or DMEM (BLM, B16F10) medium supplemented with 10% FCS, at 37°C, in a humidified atmosphere with 5% CO₂. Fully polarized macrophages were exposed to 10 mg/ml IVIg (Privigen; CSL Behring) for 24 h. For Syk inhibition assays, differentiated macrophages were treated with vehicle (H₂O) or piceatannol (100 μM; Calbiochem) 1 h before IVIg treatment. To determine the CD16 contribution, an anti-human CD16 F(ab′)₂ mAb (LSBio, clone 3G8), or an isotype-control mouse IgG1 F(ab′)₂, was used at 20 μg/ml before IVIg treatment.

Total RNA was extracted using the RNeasy Mini kit or AllPrep DNA/RNA/Protein Mini kit (Qiagen) following manufacturer's guidelines. cDNA was synthesized using the Reverse Transcription System kit (Applied Biosystems). Oligonucleotides for selected genes were designed according to the Roche software (Universal Probe Roche library). Quantitative real-time PCR was performed using custom-made panels (Roche Diagnostics) or standard plates on a LightCycler 480 (Roche Diagnostics) or a iQTM5 (Bio-Rad), respectively. An extensive battery of genes differentially expressed between M1 and M2 macrophages was included in our assays (a total of 33 genes, of which 13 were previously identified as M1 marker genes and 20 as M2 marker genes) (21, 26, 27). Assays were made in triplicate, and results were normalized according to the expression level of GAPDH or to the mean of the expression level of endogenous reference genes HPRT, SDHA, and TBP. Results were expressed using the ΔΔ cycle threshold method for quantitation.

Culture supernatants from LPS-treated (24-h) human macrophages were assayed for the presence of cytokines using commercially available ELISA for TNF-α, IL-10, IL-6 (ImmunoTools), IL-12p40, CCL-2 (OptEIATM IL-12 p40 set; BD Pharmingen), and activin A (R&D Systems). LPS-treated mouse macrophage supernatants were tested for Il-10, TNF-α, and Ccl-2 using available ELISA (BioLegend). ELISA was performed following the protocols supplied by the manufacturers.

BLM cells were plated (5 × 10³ cells/well), allowed to adhere for 24 h, and exposed to culture supernatants from untreated or IVIg-treated human macrophages for 48 h. Cell proliferation was evaluated using MTT (Sigma-Aldrich). Complete media was used as control to determine the basal BLM cell proliferation.

For the pulmonary metastasis animal model, 6- to 8-wk-old wild-type (WT) C57BL/6, Fcgr3^−/−, and Fcer1g^−/− mice (B6.129P2-Fcgr3^tm1Sjv/SjvJ and B6;129P2-Fcer1g^tm1Rav/J from The Jackson Laboratory, provided by J. Ochando and S.-H. Chen, Mount Sinai School of Medicine, New York, NY) were used for all experiments. Mice were injected i.v. (tail vein) with 3 × 10⁵ B16F10 melanoma cells in 0.1 ml sterile PBS on day 0, and with 400 μl IVIg (100 mg/ml) or PBS (control) 24 h before tumor cell injection (day −1) and on days 6 and 13. Mice were sacrificed on day 18, and lung surface metastases were counted under a dissecting microscope as black nodules after bleaching in Fekete’s solution. For xenograft studies, 2- to 3-mo-old BALB/c SCID, C57BL/6, Fcgr3^−/−, Fcgr4^−/− (generated by J. Ravetch and bred at Institut Pasteur), and Fcer1g^−/− mice were injected s.c. (lateral thoracic wall) with 2 × 10⁶ BLM cells or 5 × 10⁵ MC38 cells in 0.1 ml PBS/0.1% glucose. IVIg (400 μl, 100 mg/ml) or PBS (400 μl) was injected i.v. on days −1, 7, and 14, and mice were killed 17–21 d after cell tumor inoculation. Mice were inspected daily, and the tumor volume was measured as width² × length/2. All protocols were approved by the Centro de Investigaciones Biológicas/Consejo Superior de Investigaciones Cientificas Ethics Committee.

The surgical procedure was a variant of that described by Chen et al. (28) and Liu et al. (29). Mice were anesthetized with isoflurane 1.5–2% in a mixture of 80% air/20% oxygen, and body temperature was maintained at physiological levels with a heating pad during the surgical procedure and anesthesia recovery. Mice were subjected to permanent focal cerebral ischemia through the distal occlusion of middle cerebral artery (MCA) by ligature of the trunk just before its bifurcation between the frontal and parietal branches with a 9-0 suture, in combination with the occlusion of the ipsilateral common carotid artery. Physiological parameters were not significantly different among the different groups studied. Following surgery, individual animals were returned to their cages with free access to water and food. All the groups were performed and quantified in a randomized fashion by investigators blinded to treatment groups. IVIg (400 μl, 100 mg/ml) or PBS (400 μl) was injected by vein tail 10 min after MCA occlusion. Forty-eight hours afterward, permanent MCA occlusion mice were killed by an overdose of sodium pentobarbital and brain was removed, cut into 1-mm-thick coronal slices, and stained with 2,3,5-triphenyl tetrazolium chloride (1% in 0.1 M phosphate buffer). Infarct volumes were calculated sampling each side of the coronal sections with a digital camera (Nikon Coolpix 990), and the images were analyzed using ImageJ 1.33u (National Institutes of Health, Bethesda, MD). To exclude the brain edema effects, infarct area was corrected by the ratio of the entire area of the ipsilateral hemisphere to that of the contralateral.

Mice received i.p. injections of 350 μg LPS per 25 g body weight 1 h after injection of 400 μl IVIg (100 mg/ml) or PBS. LPS was dissolved prior to injection in PBS at a concentration of 10 mg/ml. Injected animals were monitored for 7 d.

Differences between the experimental groups in in vivo experiments were analyzed by a nonpaired Student t test. In the case of CD11b⁺ isolated cells, quantitative PCR data were analyzed with the REST-2009 software from Qiagen using 5000 permutations. Statistical significance of in vitro generated data was evaluated using a paired Student t test. In all cases, p < 0.05 was considered as statistically significant.

Because IVIg is used off-label for chronic inflammatory diseases (30), in which M1 macrophages critically contribute to pathology, we first investigated whether IVIg modulates the effector functions of proinflammatory M1 macrophages. A 24-h exposure to IVIg led to a significant reduction in the LPS-stimulated release of TNF-α, IL-12p40, and CCL2 from M1 macrophages (Fig. 1A), without affecting their tumor cell growth-inhibitory ability (Fig. 1C, upper panel). IVIg did not overtly alter the gene expression profile of M1 macrophages, as most M1 polarization-specific markers were only weakly modulated in response to the treatment (Supplemental Fig. 1A). The functional modulation induced by IVIg on M1 macrophages is, therefore, compatible with the previously reported anti-inflammatory activity of IVIg and might help explain the clinical improvement of chronic inflammatory diseases by IVIg treatment (14, 17, 19, 31).

The effects of IVIg on macrophages under homeostatic conditions, or in M2-associated pathophysiological processes, had not been addressed before. Unlike M1 macrophages, treatment of M2-polarized macrophages with IVIg for 24 h led to a significant reduction in the LPS-induced CCL2 and IL-10 release and a concomitant enhancement of the LPS-induced production of TNF-α and IL-12p40 (Fig. 1B). Moreover, IVIg-treated M2 macrophage supernatants, like M1 macrophage-conditioned media, inhibited the growth of BLM melanoma cells (Fig. 1C, lower panel), whose proliferation was not affected by IVIg itself. Along the same line, IVIg provoked a dramatic transcriptomic switch in M2 macrophages, as it reduced the expression of M2-specific markers (between 5 and 100 times) and increased the expression of M1-specific markers (10- to 100-fold) (Fig. 1D). Kinetic analysis revealed that the polarization switch of M2 macrophages is already evident 12 h after IVIg addition (Supplemental Fig. 1B). Unsupervised hierarchical clustering confirmed that the gene expression profile of IVIg-treated M2 macrophages resembles that of proinflammatory M1 macrophages (Fig. 1E). These results indicated that IVIg differentially affects the functional and transcriptomic polarization of M1 and M2 macrophages, as it inhibits the production of proinflammatory cytokines by M1 macrophages while promoting the acquisition of a proinflammatory profile in M2 macrophages. Further supporting these results, IVIg increased the expression of most M1-specific markers, downregulated the majority of M2-specific markers on ex vivo isolated CD14⁺ human thymic macrophages (Supplemental Fig. 2A), and reduced the constitutive and LPS-induced expression of IL-10, while potentiating that of TNF-α, from CD14⁺ tumor-associated macrophages (Supplemental Fig. 2B). Altogether, these results demonstrate the ability of IVIg to promote a M2-to-M1 polarization switch in human macrophages.

To unravel the mechanisms underlying the IVIg effect on human macrophage polarization, we initially focused on activating Fcγ receptors, especially on CD16, whose expression is significantly higher in M2 than in M1 macrophages (Fig. 2A, 2B). The contribution of the CD16-encoding FCGR3A gene to the IVIg-mediated functional polarization switch was assessed by using the anti-CD16 3G8 blocking Ab. The 3G8 F(ab′)₂ significantly reversed the IVIg-mediated change in cytokine profile, as it abrogated the IVIg-dependent increase in TNF-α production and inhibited the IVIg-mediated reduction in IL-10 release (Fig. 2C). Moreover, anti-CD16 3G8 F(ab′)₂ impaired the IVIg-induced downregulation of M2-specific marker expression, including that of IL10, as well as the increase of M1-specific markers (Fig. 2D). Taken together, these results suggest that CD16 is involved in the M2-to-M1 phenotypic and functional polarization switch induced by IVIg. In line with the known CD16-dependent Syk phosphorylation, IVIg triggered activation of Syk and its downstream targets Akt, ERK1/2, CREB, and p38MAPK (Supplemental Fig. 3A) (32, 33), but had no effect on the phosphorylation state of SHIP1 (our unpublished observation), a readout for the engagement of the inhibitory CD32B Fcγ receptor (31) or Src homology region 2 domain-containing phosphatase 1 (Supplemental Fig. 3B). The IVIg-induced polarization switch was afffected by the Syk tyrosine kinase inhibitor piceatannol (Fig. 2E, Supplemental Fig. 3C, 3D), although some of its effects were distinct (e.g., IL10) to those seen in the presence of the blocking anti-CD16 Ab. These results indicate that both CD16 and Syk contribute to the M2-to-M1 macrophage polarization switch induced by IVIg in human macrophages, and that other activating Fc receptors, and not only CD16, might mediate the effect of IVIg on IL10 gene expression.

To determine the extent of the relevance of these findings, we next determined whether the IVIg-mediated polarization switch was also observed in mouse BMDM. Like in the case of human macrophages, a 24-h treatment with IVIg significantly enhanced the expression of the paradigmatic M1 markers Nos2, Tnfa, and Cd11c, whereas it inhibited the expression of a large number of M2 polarization markers (Fig. 3A). At the functional level, the LPS-stimulated production of TNF-α by mouse M2 BMDM was significantly enhanced by IVIg, without affecting IL-10 release (Fig. 3B). This IVIg-induced polarization switch took place via Fc receptors, because 1) deletion of the Fcgr3 gene (encoding FcγRIII) inhibited the increase of Nos2 and the decrease of Cbr2, Emr1, and Cd206 triggered by IVIg (Fig. 3A); 2) deletion of the Fcgr4 gene (encoding FcγRIV) inhibited the IVIg-mediated decrease in Hpdg, Ctla2b, Gas6, Stab1, Cbr2, Emr1, Mgl1, and Cd206 (Fig. 3A); and 3) all the IVIg-triggered gene expression changes were blunted in Fcer1g^−/− (encoding FcRγ-chain) M2 macrophages (Fig. 3A). Regarding LPS-cytokine release, the absence of FcRγ-chain or FcγRIV expression completely abolished the IVIg-induced increase in TNF-α release from mouse M2 BMDM (Fig. 3B). Consequently, all the transcriptomic and functional changes triggered by IVIg on M2 BMDM are absent in Fcer1g^−/− macrophages, whereas ablation of the Fcgr3 or Fcgr4 genes has a partial influence on the polarization switch triggered by IVIg. Interestingly, the impairment of the IVIg-induced M2-to-M1 polarization switch in Fcer1g^−/− macrophages might be explained by their skewed basal M2 polarization, illustrated in Supplemental Fig. 4 and also recently reported (34), and that might make Fcer1g^−/− macrophages more reluctant to the polarizing effect of IVIg. Therefore, and as in the case of human macrophages, Fcγ-activating receptors mediate the proinflammatory polarization of mouse macrophages by IVIg.

To assess the in vivo relevance of the above in vitro data, we evaluated the influence of IVIg on three animal models of disease, as follows: a MCA occlusion stroke and LPS-induced sepsis-like mouse models, in which tissue damage correlates with excessive proinflammatory responses (35), and a xenograft tumor model, in which tumor and tumor-associated cells contribute to the establishment of an immunosuppressive environment (36). IVIg injection led to significant increases in the volume of the infarcted area in the stroke mouse model (Fig. 4A) and diminished survival rates after injection of a lethal dose of LPS (Fig. 4B), thus implying that IVIg misbalances innate immune responses toward a more proinflammatory state. Along the same line, and regarding the xenograft tumor model, IVIg significantly reduced tumor volumes after 15 d (Fig. 4C). Altogether, data from the three assayed animal models revealed that IVIg exerts a global proinflammatory response in vivo.

Because tumor metastasis and progression are dependent on the tumor ability to alter macrophage polarization (5, 37) and given the above described effects of IVIg, we hypothesized that IVIg might inhibit tumor growth and/or metastasis by skewing macrophage polarization via engagement of Fcγ receptors. In the MC38 colon cancer xenograft model, IVIg significantly reduced tumor volume in WT (Fig. 5A) and Fcgr3^−/− mice (Fig. 5B), but had no effect in Fcer1g^−/− (Fig. 5C) or Fcgr4^−/− mice (Fig. 5D), indicating that the latter two genes contribute to the antitumor activity of IVIg. Besides, IVIg significantly lowered B16F10 melanoma lung metastasis in WT mice, an inhibitory effect that was absent in Fcgr3^−/− and Fcer1g^−/− mice (Fig. 5E). Importantly, analysis of the tumor-associated CD11b⁺ myeloid cells from MC38 tumor xenografts revealed a significantly increased expression of the M1 polarization-associated markers Cd11c, Ccr7, and Nos2 in tumor-bearing WT mice, but not in Fcer1g^−/− mice (Fig. 5F). Even more, the IVIg-mediated global upregulation of M1-specific markers group seen in WT mice was completely abolished in Fcer1g^−/− mice (Fig. 5F). Therefore, and through engagement of activating Fcγ receptors, IVIg treatment impairs tumor progression (growth and metastasis) and influences the polarization of tumor-associated myeloid cells. The causal relationship between both effects was assessed by determining the ability of IVIg to inhibit tumor growth upon macrophage depletion. As shown in Fig. 5G, clodronate liposome-mediated depletion of macrophages prevented the IVIg-promoted tumor growth reduction in the MC38 colon cancer model. In fact, even the decrease in serum CCL2 caused by IVIg treatment was found to be macrophage dependent (Fig. 5H). Therefore, IVIg alters the polarization of macrophages, whose presence is absolutely required for IVIg to limit tumor growth.

The immunomodulatory action of IVIg has widened the range of pathologies for which IVIg therapy is either approved or has shown benefit (38, 39). In line with its beneficial actions on inflammatory pathologies, we now show that IVIg impairs the effector functions of proinflammatory M1 macrophages. However, we also report the ability of IVIg to cause a M2-to-M1 phenotypic and functional Fc receptor-mediated polarization switch on human and murine macrophages in vitro and in vivo, thus illustrating that the IVIg immunomodulatory effects are dependent on the polarization state of the responding macrophages.

Immunocomplexes have long been known to promote tumor cell killing in a FcγR-dependent manner and to elicit potent inflammatory responses that underlie pathologies such as systemic lupus erythematosus and rheumatoid arthritis (40). From this point of view, the proinflammatory and antitumor nature of the IVIg effect that we report is not unprecedented. Paradoxically, however, high doses of IgG (IVIg) exert beneficial effects on several autoimmune disorders by virtue of their potent anti-inflammatory activity (16, 41). Attempts to explain this apparent contradiction have indicated that the active components within IVIg constitute a minor fraction of the preparation (e.g., immune complexes, sialylated Fc IgG), thus explaining the large doses requirement (14). The results that we now present shed more light on this issue, because IVIg provokes different responses, either pro- or anti-inflammatory, depending on the polarization state of the target macrophage. Whereas IVIg improves inflammatory diseases through impairment of the functional activities of M1 proinflammatory macrophages, the IVIg effects on M2 anti-inflammatory macrophages skew them toward the acquisition of the phenotypic and functional characteristics of M1/proinflammatory macrophages. These last effects are consistent with the proinflammatory actions exerted by Igs or immune complexes (19). Thus, considering the effects of IVIg on both types of macrophages, IVIg exhibits both proinflammatory and anti-inflammatory properties, and the nature of its effects are dependent on the activation/polarization status of the target macrophages, as it enhances the proinflammatory activities of M2 macrophages and limits the proinflammatory actions of M1 macrophages. Therefore, IVIg appears as a potent and versatile immunomodulatory agent that tunes macrophage effector functions in an environment-dependent manner.

In the context of cancer, we have demonstrated that IVIg impairs tumor progression and metastasis in a Fc receptor- and macrophage-dependent manner, and that IVIg alters the expression of polarization markers in CD11b⁺ tumor-associated myeloid cells from WT but not Fcer1g^−/− mice. The coexistence of autoimmune pathologies and cancer has previously provided evidence that IVIg therapy promotes regression of cancer in patients with chronic lymphatic leukemia, Kaposi’s sarcoma, and melanoma (42). Given the importance of the polarization of TAM for tumor progression and dissemination (37), our results establish a sequential link between the IVIg ability to modulate macrophage polarization and its antitumor effect, suggesting that the proinflammatory activities of IVIg might also be therapeutically useful in pathologies such as cancer, in which immunogenic and proinflammatory macrophage functions need to be promoted.

Numerous molecular and cellular mechanisms have been proposed to explain the immunomodulatory activity of IVIg (19). Our results clearly establish macrophages as an absolute requirement for the antitumor effect of IVIg, because macrophage depletion abrogates the inhibition of tumor growth by IVIg. Moreover, in tumor-bearing animals, IVIg inhibited the level of circulating Ccl2, whose tumor-dependent increase was significantly reduced upon clodronate-mediated macrophage depletion. This is of particular significance because CCL2 is known to promote M2 macrophage polarization (43, 44) and CCL2-induced MCPIP1 inhibits proinflammatory cytokine production (45) and augments the expression of the M2-associated IL-10–driving cMaf transcription factor (26, 46). The ability of IVIg to reduce Ccl2 both in vivo (Fig. 5H) and in vitro (Fig. 1A, 1B) might, therefore, contribute to lower the M2/anti-inflammatory environment seen in tumor-bearing animals, thus favoring the generation of antitumor responses. In this regard, others have also shown that IVIg decreases CCL2 levels in whole blood (47) and skin tissue (48).

Besides monocytes/macrophages, cellular targets for IVIg include NK cells, T and B lymphocytes, granulocytes, and endothelial cells (49). Our data also support the idea that cells other than macrophages contribute to the proinflammatory activity of IVIg because the IVIg-induced increase in circulating TNF-α in tumor-bearing animals is not eliminated upon macrophage depletion (Fig. 5H). Because inhibition of tumor growth by IVIg is macrophage dependent, it seems reasonable to conclude that the IVIg-enhanced levels of TNF-α do not significantly participate in the antitumor action of IVIg. However, and given the TNF-α functional activities, it could be hypothesized that IVIg-induced TNF-α might explain the mild/adverse side effects occasionally seen during IVIg therapy.

We thank Drs. J. C. Ochando and Shu-Hsia Chen (Mount Sinai School of Medicine) for providing Fcgr3^−/− and Fcer1g^−/− mice and Dr. Jeffrey Ravetch (The Rockefeller University, New York, NY) for permission to use Fcgr4^−/− mice.

The authors have no financial conflicts of interest.