|Home | About | Journals | Submit | Contact Us | Français|
Myeloid-derived suppressor cells (MDSCs) bear characteristics of precursors for both M1 and M2 macrophages. The molecular mechanism underlying the differentiation into M1 and M2 macrophages and the relationship of this differentiation to anti-tumor responses remains largely undefined. Herein we investigate the potential function of paired immunoglobulin-like receptor B (PIR-B), also knows as leukocyte immunoglobulin-like receptor subfamily B member 3 (LILRB3) in MDSC differentiation and its role in tumor-induced immunity. Our studies indicated that MDSCs genetically ablated for PIR-B (Lilrb3−/−) underwent a specific transition to M1-like cells when entering the periphery from bone marrow, resulting in decreased suppressive function, regulatory T cell activation activity, primary tumor growth, and lung metastases. Activation of toll-like receptor (TLR), signal transducers and activators of transcription 1 (STAT1), and nuclear factor-kappa B (NF-κB) signaling in Lilrb3−/− MDSC promoted the acquisition of M1 phenotype. Inhibition of the PIR-B signaling pathway promoted MDSC differentiation into M1 macrophages.
Myeloid-derived suppressor cells (MDSCs) play an important role in tumor-induced immune tolerance. In tumor bearing animals and cancer patients, MDSCs accumulate in lymphoid organs such as the spleen and bone marrow, as well as within the tumor, and throughout the peripheral blood circulation (Serafini et al., 2006). These cells were originally identified with the phenotype of Gr-1+CD11b+.(Serafini et al., 2004). Recently, two distinct sub-populations, Ly-6G+Ly-6Clow granulocytic and Ly-6G−/lowLy-6Chigh monocytic MDSCs have been described (Peranzoni et al.) (Movahedi et al., 2008) (Youn et al., 2008). Functionally, MDSCs suppress T-cell responses through productions of inducible nitric oxide synthase (iNOS) and arginase 1 (ARG1), which are induced upon stimulation with interferon-γ (IFN-γ) and interleukin-4 (IL-4) or IL-13, respectively (Kusmartsev et al., 2000) (Sinha et al., 2005c) (Sinha et al., 2005b); (Huang et al., 2006; Marigo et al., 2008). More recently, we showed that CD115+Gr-1+Ly6Chi monocytic MDSCs promoted T regulatory cell (Treg cell) induction and expansion in tumor bearing mice (Huang et al., 2006).
In the tumor microenvironment, tumor associated macrophages (TAMs) constitute the majority of tumor-infiltrating leukocytes. Two distinctive TAM sub-populations have been defined. Classical, or M1, macrophages are characterized by the expression of high amounts of iNOS and tumor necrosis factor-α (TNF-α), whereas, alternatively activated, M2 macrophages typically produce ARG1 and IL-10 (Umemura et al., 2008). At the tumor site in wild-type mice, TAMs are predominately M2-like macrophages, which are the cells primarily responsible for suppressing T cell-mediated anti-tumor responses and promoting tumor progression, metastasis, and angiogenesis (Guruvayoorappan, 2008; Mantovani et al., 2009; Mantovani et al., 2002; Sica et al., 2008). M1 macrophages, in contrast, exhibit, a tumoricidal effect (Sinha et al., 2005a; Sinha et al., 2005c).
Monocytic MDSCs and TAMs share several characteristics, such as expression of the monocyte and macrophage markers F4/80 and CD115, as well as inducible expression of iNOS and ARG1 (Lewis and Pollard, 2006; Mantovani et al., 2009) (Umemura et al., 2008). Accumulating evidence suggests that, upon entering tumor tissues, MDSCs may differentiate into TAMs, leading to elevated IL-10 production, inhibition of T-cell responses, and promotion of angiogenesis (Yang et al., 2004). However, the mechanism behind regulation of MDSC differentiation remains unclear (Sinha et al., 2007) (Kusmartsev and Gabrilovich, 2005; Otsuji et al., 1996; Umemura et al., 2008; Yang et al., 2004).
Mouse paired immunoglobulin-like receptors (PIRs), also known as leukocyte immunoglobulin-like receptors (LIR) or Ig-like transcripts (ILT) in humans, are expressed on macrophages, neutrophils, mast cells, and B cells (Kubagawa et al., 1999). The PIR family includes two molecules - PIR-A and PIR-B, (the latter encoded by the gene Lilrb3). They are type I transmembrane receptors, which bind to major histocompatibility complex (MHC) class I glycoproteins (Takai, 2005). PIR-A also associates with the Fc receptor common γ chain (FcRγ) at its transmembrane region. Intracellularly, the PIR-A-FcRγ complex phosphorylates immunoreceptor tyrosine-based activation motifs (ITAMs) and delivers activating signals, while PIR-B phosphorylates immunoreceptor tyrosine-based inhibitory motifs (ITIMs) and delivers inhibitory signals (Kubagawa et al., 1999). In PIR-B deficient mice, hyperresponsiveness has been observed in a variety of different cell types. In a transplantation model, PIR-B deficient mice show accelerated, lethal graft versus host disease (GVHD) due to the up-regulation of PIR-A and enhanced production of IFN-γ (Nakamura et al., 2004).
We hypothesized that PIRs may regulate MDSC function and differentiation in the tumor environment. In the present study, we found that in PIR-B deficient tumor-bearing mice, Gr-1+CD115+ monocytic MDSCs exhibited enhanced expression of iNOS and tumor necrosis factor alpha (TNF-α), preferentially acquiring an M1-like phenotype in the spleen and at the tumor site, and that tumor growth was significantly retarded. Furthermore, adoptively transferred MDSCs from Lilrb3−/− tumor-bearing mice inhibited the growth of pre-existing tumors and lung metastases. These functional differences were correlated with up regulation of the signal transducers and activators of transcription 1 (STAT1) and nuclear factor-kappa B (NF-κB) signaling pathways. Our findings demonstrated that the signaling balance between PIR-A and PIR-B is a key factor in the regulation of MDSC differentiation into M1 or M2 phenotypes, and has an impact on MDSC's functions in both immune suppression and promotion of tumor progression.
We have shown that monocytic MDSCs suppress T-cell proliferation directly and induce Treg cell expansion in tumor-bearing mice (Huang et al., 2006; Pan et al.). We also found that expression of M1-associated iNOS production by MDSCs was not required for Treg cell expansion (Huang et al., 2006). Here we tested whether M2-associated genes, such as ARG1 and IL-10, were required for Treg cell expansion as mediated by MDSCs. Whereas the addition of iNOS inhibitor did not significantly affect antigen-specific Treg cell proliferation in the presence of MDSCs, addition of ARG1 inhibitor, anti-IL-10, or the use of IL-4Rα-deficient MDSCs significantly abrogated the Treg cell expansion mediated by MDSCs (Fig. 1A and 1B; WT vs. ARG1 inhibitor P = 0.01, wild-type (WT) vs. anti-IL-10 P = 0.0016, WT vs. Il-4rα−/− P = 0.0012). This indicates that M2-associated effector mechanisms employing IL-10 and IL-4 or IL-13-dependent ARG1 are required for enhancement of MDSC-mediated Treg cell proliferation.
To determine whether PIR-B is involved in MDSC-mediated Treg cell activation, we utilized wild type (WT) and Lilrb3−/− Lewis Lung Carcinoma (LLC) tumor-bearing mice. Monocytic MDSCs were purified separately from bone marrow and spleen of WT and Lilrb3−/− tumor-bearing mice and suppressive function was evaluated. MDSCs from the bone marrow of WT and Lilrb3−/− tumor-bearing mice both suppressed OTII T-cell proliferation to similar degrees in response to ovalbumin (OVA) peptide (P = 0.18) (Fig. 2A), while, splenic Lilrb3−/− MDSCs had decreased suppressive activity when compared to WT splenic MDSCs (P < 0.05). Similarly, Treg cell induction by bone marrow derived MDSCs was not significantly different between Lilrb3−/− and WT MDSCs (21.5 ± 0.6 vs. 21.1 ± 2.1), whereas decreased Treg cell induction was observed when comparing Lilrb3−/− to WT splenic MDSCs, (6.0 ± 0.5 vs. 10.7 ± 0.6, P<0.001)(Fig. 2B).
Based on these results, we hypothesized that MDSC function and differentiation may be regulated by PIR-B, especially once MDSCs migrate from the bone marrow to peripheral lymphoid organs and the tumor microenvironment. We also compared phenotypic differences between MDSCs from WT and Lilrb3−/− mice. In WT tumor-bearing mice, splenic MDSCs exhibited a higher ARG1 activity (P < 0.001) and IL-10 (P < 0.01) secretion but a lower iNOS activity (P < 0.001) and TNF-α secretion (P < 0.001), suggesting MDSCs preferentially differentiate into an M2-like phenotype in WT mice and an M1-like phenotype in Lilrb3−/− mice (Fig. 2C). Lilrb3−/− splenic MDSCs expressed lower quantities of the M2 markers, CD36, CD206, Tie2, and CCR2, but higher amounts of the M1 marker, CCR7, than WT MDSCs (Fig. 2D, E). We also measured interferon-γ receptor (IFN-γR) and interleukin-4 receptor (IL4-R) expression, whose signaling promotes M1 and M2 polarization, respectively. IL-4R expression was lower, while IFN-γR was higher in Lilrb3−/− MDSCs (Fig. 2D). Although no differences in IL-10R expression were observed between WT and Lilrb3−/− MDSCs (Fig. 2D), significantly lower IL-10 expression by Lilrb3−/− MDSCs could account for the M1-oriented phenotype (Fig. 2C). However this difference was not observed with bone marrow-derived MDSCs (Fig. S1A). MDSCs isolated from tumor bearing Lilrb3−/− mice have also lost their ability to induce Treg cells and suppress T cell proliferation in vivo in OTII transgenic mice (Fig. S1B).
These results indicate Lilrb3−/− MDSCs favor an M1, over M2, phenotype upon exiting the bone marrow in tumor-bearing mice. To determine whether myeloid growth and differentiation factors, such as macrophage colony stimulating factor (M-CSF) and granulocyte-macrophage colony stimulating factor (GM-CSF), could dictate macrophage polarization in vitro, Gr-1+CD115+ monocytes purified from bone marrow and spleen of Lilrb3−/− or WT naïve mice were cultured in the presence of M-CSF or GM-CSF. In the presence of M-CSF, both bone marrow and splenic Lilrb3−/− monocytes preferentially underwent M1-like differentiation, expressing significantly higher amounts of iNOS and TNF-α, with decreased arginase and IL-10 production, when compared to WT monocytes (Fig. S1C). With GM-CSF, Lilrb3−/− bone marrow monocytes produced significantly more iNOS and TNF-α, while no difference was found in arginase and IL-10 production (Fig. S1D). These results suggest preferential acquisition of an M1-like phenotype by naïve Lilrb3−/− monocytes. Furthermore, M-CSF may play a role in polarization of WT macrophages to both M1 and M2 phenotypes whereas GM-CSF may only promote the M1 oriented phenotype.
Since in vivo suppressive functions of Lilrb3−/− MDSCs were weaker than WT, we examined whether PIR-B deficiency has any impact on tumor growth. The rate of tumor growth was significantly slower in Lilrb3−/− mice compared with WT mice over time (Fig. 3A). To determine whether this effect was correlated with the anti-tumor M1 phenotype observed in Lilrb3−/− MDSCs, we directly isolated MDSCs from these tumors. Tumor-infiltrating Lilrb3−/− MDSCs expressed significantly higher amounts of iNOS (P < 0.05), TNF-α (P < 0.01), IL-12 (P < 0.01), and IL-1-β (P < 0.01), and significantly lower amounts of ARG1 (P < 0.01) and IL-10 (P < 0.01) when compared with WT MDSCs, as assessed by quantitative real-time polymerase chain reaction (PCR) (Fig. 3B). We also examined expression of several M2 surface markers, such as CD36, CD206, Tie2, and IL-4R (Guruvayoorappan, 2008; Kodumudi et al.; Mantovani et al., 2009; Mantovani et al., 2002; Sica et al., 2008), by flow cytometry. Lilrb3−/− tumor-infiltrating MDSCs expressed lower amounts of CD36, CD206, Tie2 and IL-4R than WT MDSCs (Fig. 3C). These results were confirmed by immunofluorescent staining of LLC tumor tissue (Fig. 3D). CCR7 was used as a marker for M1 macrophages and CD206 as an M2 marker with CD11b co-staining to identify MDSCs. CD206 was highly expressed in tumor tissue from WT mice, while CCR7 was dominant in tumor tissue from Lilrb3−/− mice. Both markers co-localized with CD11b+ myeloid cells, indicating that MDSCs in WT tumor-bearing mice had an M2 phenotype, while those from Lilrb3−/− mice bore more resemblance to M1 macrophages. These observations were further confirmed by assessment of the macrophage marker F4/80, and iNOS (M1) and arginase 1 (M2) expression.
Sorted Lilrb3−/− MDSCs isolated directly from tumor infiltrating leukocytes (TILs) also had higher direct cytolytic activity against parental tumor cells than MDSCs isolated from WT mice (Fig. 3E). This suggests that Lilrb3−/− MDSCs in the tumor exhibit an M1-like phenotype, express higher quantities of pro-inflammatory cytokines, and possess direct cytolytic function, suggesting they might have tumor killing activities.
Since M1 macrophages have been shown to exhibit tumoricidal activities (Sinha et al., 2005b), we assessed the therapeutic potential of Lilrb3−/− Gr-1+CD115+ monocytic MDSCs in subcutaneous and lung metastases tumor models. Upon adoptive transfer, Lilrb3−/− Gr-1+CD115+ monocytic MDSCs significantly retarded the growth of subcutaneous tumors (Fig. 3F, P = 0.014 on day 35), prolonged survival rate (Fig. 3G upper panel, P = 0.0036), and inhibited lung metastases in recipient mice (Fig. 3G, lower panels, P < 0.001). Interestingly, neither WT monocytic MDSCs nor Lilrb3−/− Ly6G+ granulocytic MDSCs exerted significant effects on tumor growth or lung metastases when compared to control mice that did not receive adoptive cell transfer.
We next attempted to elucidate the mechanisms underlying the retarded tumor growth and enhanced anti-tumor responses in Lilrb3−/− tumor-bearing mice. MaFIA (macrophage Fas-induced apoptosis) mice, whose endogenous CD115+ monocytic MDSCs are tagged with green fluorescent protein and can be depleted through the use of cross-linking reagent AP20187 (Pan et al.), were chosen. CD45.1 OT-II T cells were adoptively transferred into MaFIA mice with pre-existing OVA-LLC tumors followed by depletion of CD115+ cells, with and without reconstitution using CD115+ monocytic MDSCs derived from WT or Lilrb3−/− tumor-bearing mice. Ten days after the adoptive transfer, tumor infiltrating leukocytes (TILs) and splenocytes were stained for CD45.1+CD4+CD25+Foxp3+ antigen specific Treg cells. In the CD115+ depleted group, significant decreases in the Treg cell population (Fig. 4A–C, P < 0.0001) and tumor weight (Fig. 4C, P < 0.0001), and significantly increased proliferation of antigen-specific T cells (Fig. 4D, < 0.0001) were observed. In the WT MDSC adoptive transfer group, the presence of MDSCs compensated for the depletion of CD115+ cells leading to expansion of the Treg cell population, increases in tumor weight, and suppression of antigen-specific T cell responses, similar to the mock depletion control group. Interestingly, the Lilrb3−/− MDSC adoptive transfer group had significantly fewer Treg cells in the TIL population (Fig. 4A, B; P < 0.001 in TIL) and spleen (P = 0.022, data not shown), lower tumor weights (Fig. 4C; P < 0.001), and strong antigen-specific T cell proliferation (Fig. 4D; P < 0.001), when compared with mice that had received WT MDSCs. The results were similar to those of the CD115 depleted MaFIA mice which were not reconstituted with MDSCs, indicating that MDSCs from Lilrb3−/− mice can not suppress anti-tumor immunity within the tumor microenvironment. Since M2 macrophages promote tumor angiogenesis, we evaluated the vasculature of tumors from the mice that received WT or Lilrb3−/− MDSC by immunofluorescent staining. Tumor tissue from the WT MDSC transfer group exhibited more prominent vascularity, as demonstrated by increased CD31 staining, compared to tissues from the Lilrb3−/− MDSC transfer group (Fig. 4E).
Lilrb3−/− mice exhibit hyper-responsive B cells, neutrophils, macrophages, and mast cells, which can be attributed to the failure to recruit the tyrosine phosphatase, SHP-1, which delivers constitutive inhibitory signals to janus kinase (JAK)-STAT pathways (Blanchette et al., 2009; David et al., 1995; Dong et al., 2001). In Lilrb3−/− MDSCs, phosphorylation of STAT1 was up regulated, while phosphorylation of STAT3 was significantly decreased upon IL-10 stimulation. This effect is even more profound in the presence of STAT1 activating cytokines, such as IFN-γ, or STAT3 activating cytokines, such as IL-10 (Fig. 5A, B). These differences were also observed under inflammatory conditions. We also examined the toll like receptor 4 (TLR4) pathway, which has been shown to play a role in M1 macrophage differentiation (Mantovani et al., 2004; Martinez et al., 2008). Lilrb3−/− splenic-derived MDSCs demonstrated increased phosphorylation of extracellular signal-regulated kinases (ERKs), p38 mitogen-activated protein kinase (MAPK), and NF-κB upon lipopolysaccharide (LPS) stimulation, with minimal response to IL-13-mediated stimulation as measured by STAT6 phosphorylation (Fig. 5C). Therefore, in the absence of functional PIR-B, TLR4 and IFN-γ inflammatory responses were up-regulated. The enhanced TLR4 and IFN-γ signaling suppressed STAT3 activation and pushed Lilrb3−/− MDSC toward M1 polarization.
Normally, PIR-A and PIR-B are expressed simultaneously on myeloid cells and the specific equilibrium maintained in different cell lineages may be determined by cell activation status (Kubagawa et al., 1999; Tun et al., 2003). PIR-A and PIR-B expression were assessed quantitatively by real-time PCR. Low expressions of PIR-A and PIR-B were detected in WT MDSCs isolated from the bone marrow of tumor-bearing mice. Both PIR-A and PIR-B were induced in splenic WT MDSCs, with PIR-B expression 3-fold higher than PIR-A (Fig. 6A). Interestingly, splenic Lilrb3−/− MDSCs expressed a significantly higher amount of PIR-A compared to WT MDSCs. Surface expression of PIR was assessed by flow cytometry using a currently available antibody that recognizes both PIR-A and PIR-B. As anticipated, due to the lack of PIR-B, weaker staining was observed in splenic Lilrb3−/− Gr-1+CD115+ MDSCs from naïve mice compared to WT splenic MDSCs (Fig. 6B). Surprisingly, a similar mean fluorescence intensity (MFI) was detected in WT and Lilrb3−/− MDSCs isolated from tumor-bearing mice, indicating that PIR-A was increased in splenic Lilrb3−/− MDSCs from tumor-bearing mice compared to their WT counterparts or splenic Lilrb3−/− MDSCs isolated from naïve mice. Therefore, in the presence of chronic inflammation, such as tumor-derived signals, PIR-B may inhibit PIR-A expression; however in the absence of PIR-B signaling, PIR-A may compensate for the lack of PIR-B, resulting in PIR-A being the dominant activation signal.
SHP-1 is an important downstream mediator of PIR-B signaling. To confirm the relationship between downstream PIR receptor signaling and macrophage differentiation, we tested whether the SHP-1 and 2 inhibitor, NSC 87877, could inhibit normal PIR-B function in MDSCs. MDSCs treated with NSC 87877 exhibited a phenotype similar to Lilrb3−/− MDSCs. Significant decreases in ARG1 and IL-10 production were observed both in the presence and absence of IL-13 (Fig. 6C), indicating that cytokine stimulation plays an important role in the phenotypic switch of MDSCs via PIR-B signaling. We also used SHP-1 mutant motheaten viable (mev) mice to confirm that SHP-1 was involved in the control of MDSC differentiation as mediated by PIR-B. As with the results obtained using the SHP-1 and 2 inhibitor (Fig. 6C), SHP-1 mutated (mev) MDSCs exhibited an M1 oriented phenotype (Fig. 6D).
MHC class I is one of the putative ligands for both PIR-A and PIR-B (Takai, 2005). Therefore, we examined the phenotype of MDSCs isolated from β2 microglobulin (β2m) deficient (hence MHC class I deficient) tumor-bearing mice. MDSCs from these mice had a similar phenotype to WT MDSC when exposed to IFN-γ or IL-13 stimulation (Fig. 6C), possibly due to the lack of both positive and negative signaling of PIR. Upon cross-linking with PIR-A and -B antibodies (6C1), WT MDSCs acquired an M2 phenotype, i.e. decreased NO and TNFα production, and increased ARG1 and IL-10 production. This result would be consistent with the M2 oriented phenotype of WT MDSCs isolated from tumor-bearing mice. However, no significant increase in NO and TNFα production was observed in Lilrb3−/− MDSCs in the presence of 6C1, possibly due to high basal productions of NO and TNFα (Fig. S2, P >0.05), consistent with the M1 oriented phenotype of Lilrb3−/− MDSCs. Meanwhile, IL-10 secretion was significantly decreased in Lilrb3−/− MDSCs upon triggering with 6C1. (Fig. S2, P <0.001).
Since PIR-A is associated with the FcRγ chain, we analyzed the macrophage polarization of MDSCs isolated from Fcer1g−/− tumor-bearing mice. Consistent with our hypothesis and data from PIR-B deficient mice, Fcer1g−/− Gr-1+CD115+ MDSCs were M2 oriented, associated with a higher arginase and IL-10 production, increased CD36 and CD206 expression, and lower iNOS, TNFα, and CCR7 expression (Fig. 6D).
Taken together, our results demonstrated that PIR-B plays an important role in the regulation of functions and differentiation of MDSCs. In Lilrb3−/− MDSCs, TLR4 and IFN-γ signaling was heightened whereas STAT3 signaling was suppressed, the combination of which leads to M1 polarization. The M1 phenotype of Lilrb3−/− MDSCs correlated with retarded tumor growth, enhanced ant-tumor responses and decreased Treg activation.
Monocytic MDSCs express myeloid precursor markers. Studies have shown that they possess a phenotype containing elements of both M1 and M2 macrophages. However, upon entering the tumor environment, MDSCs appear to polarize toward an M2 phenotype due to the presence of high amounts of M2 cytokines such as IL-4 and IL13 (Gordon and Martinez; Umemura et al., 2008). The mechanism behind MDSC differentiation and the signals that control this commitment and biological function in tumor-bearing hosts are not well understood. In this study, we found that MDSCs are immature macrophage precursors, which have the potential to differentiate into both the M1 and M2 phenotypes. This switch can be regulated through PIR-B signaling as the cells migrate from the bone marrow into peripheral organs.
We demonstrated that MDSCs from Lilrb3−/− mice exhibit an M1, rather than M2, phenotype in the tumor environment. The M1 propensity of Lilrb3−/− MDSCs is intrinsic and not affected by cytokine profile in the tumor microenvironment since the tumors from Lilrb3−/− mice had higher amounts of IL-4 and IL-13 than those from wild type mice (data not shown). Splenic MDSCs from Lilrb3−/− mice also exhibit an M1 phenotype and diminished immune suppression, as well as Treg cell activating ability. This phenomenon was not observed in bone marrow-derived MDSCs, indicating that in order for this phenotypic switch to occur, MDSCs must receive specific signals leading to a state of differentiation and maturity, which is not attainable within the bone marrow environment. We found that Lilrb3−/− MDSCs expressed significantly higher amounts of PIR-A in tumor–bearing, but not naïve animals. PIR-A signaling may be inhibited by PIR-B, suggesting that in an inflammatory environment, such as that found in tumor bearing mice, PIR-B may dominantly regulate MDSC polarization. Constitutive cis binding between PIR-B and MHC class I on mast cells plays an essential role in the regulation of allergic responses (Masuda et al., 2007). Therefore, the MHC class I expression on Gr-1+CD115+ MDSCs may regulate the activation of PIR-A and -B. When comparing the difference in MHC class I expression between bone marrow derived and splenic Gr-1+CD115+ MDSCs, we found that splenic, but not bone marrow derived PIR-B-deficient MDSCs, expressed a higher amount of MHC-I molecules than their wild-type counterparts. This is consistent with our hypothesis that in the absence of PIR-B signaling, splenic Lilrb3−/− MDSCs have the propensity to acquire an M1 oriented phenotype due to cis binding of MHC class I to PIR-A. Recently, other ligands such as the Staphylococcus aureus has been reported to activate PIR signal (Nakayama et al., 2007), which indicates MHC class I is not the exclusive ligand of PIRs, PIR signaling may also be triggered by different ligands, which need be further investigated.
Differentiation of macrophages to the M1 phenotype depends on activation of the TLR4 and IFN-γ pathways, which in turn activate the ERK, NFκB, and STAT1 pathways. Meanwhile, M2 differentiation depends on activation signals from IL-10, IL-13, and IL-4, which activate the STAT3 and STAT6 pathways (Mantovani and Sica; Mantovani et al., 2002). NF-κB and STAT1 signaling suppresses the activation of STAT3 and STAT6, while activations of STAT3 and STAT6 inhibit NF-κB and STAT1 signaling (Mantovani and Sica). Interestingly, the inhibitory signaling of PIR-B can suppress the activation of STAT1, STAT3, and STAT6, while PIR-A signaling mainly activates the NF-κB and ERK pathways (Hu et al., 2007; Kubagawa et al., 1999).
The counterbalance of numerous activating and inhibitory factors will ultimately influence the differentiation of MDSCs toward the M1 or M2 phenotype. The activation state is dependent on the cytokine milieu present within the inflammatory environment. Since IL-10, IL-13, and IL-4 are constitutively expressed and are the predominant cytokines present in the tumor-bearing host, STAT3 and STAT6 are activated, subsequently activating PIR-B and SHP-1, and leading to the inhibition of STAT1. SHP-1 can also inhibit ERK phosphorylation (Nakata et al.), thereby inhibting NF-κB (p65) phosphorylation (Chen and Lin, 2001). A lack of inhibitory signaling from PIR-B resulted in the activation of PIR-A in MDSCs, leading to the activation of the NF-κB and ERK signaling pathways. The combined effect of activation of NF-κB, ERK, and STAT1 further suppressed STAT3 and STAT6 activation and favored M1 differentiation of MDSCs. Studies have shown that ITAM activation signals can activate the NF-κB, ERK, and Stat1 pathways (Hu et al., 2007; Taylor and McVicar, 1999), but that the p50 subunit of NF-κB inhibits NF-κB activation, and orchestrates M2 macrophage differentiation (Mantovani and Locati, 2009; Porta et al., 2009). These previous findings correlate closely with what we have observed and support our hypotheses.
Our previous studies suggest that MDSCs strongly activate Treg cells in tumor-bearing animals (Huang et al., 2006; Pan et al., 2008). In this study, we demonstrated that M2 phenotypic MDSCs are critical for both in vitro and in vivo Treg cell activation. Lilrb3−/− MDSCs favor the M1, rather than M2 phenotype, which led to less antigen-specific Treg cell activation and angiogenesis in the tumor environment in an adoptive transfer model. Our findings may indicate an intrinsic relationship between MDSCs and Treg cells.
Furthermore, we have analyzed both the pro- and antitumor activities of cancer-related inflammation and how, by shifting the balance from M2 to M1 macrophages, we can elicit a protective antitumor response. Indeed, adoptive transfer of Lilrb3−/− MDSCs significantly retarded primary tumor growth, prolonged survival rate, and inhibited lung metastasis. This implies a possible role for therapeutic regulation of paired immunoglobin like receptors in cancer, autoimmune diseases, and transplantation. PIR-B blockade may prove useful in human therapeutics through the inhibition of Treg cell expansion and angiogenesis in cancer patients
Furthermore, ILT3, the human counterpart of PIR-B, and its soluble form, have been shown to be involved in Treg cell development and GVHD prevention (Vlad et al.). Interestingly, peripheral blood CD14+CD16− monocytes from cancer patients, which are phenotypically equivalent to the mouse monocytic MDSCs, expressed a higher amount of ILT3, but not ILT1 (activating receptor), when compared to the equivalent population from normal donors (Fig. S3).
The PIR signaling pathway plays an essential role in immune suppression and activation through the regulation of MDSC-mediated immune responses. The proper control of PIR function may enable modulation of the innate immune response, an effect that carries great potential for therapeutic translation.
Lilrb3−/− mice were a gift from Dr. Clifford A. Lowell (University of California, San Francisco, CA) and Toshiyuki Takai (Tohoku University, Sendai, Japan). C57BL/6 and BALB/c mice were purchased from the National Cancer Institute (Frederick, MD). CD4+ OVA TCR transgenic (OT-II) mice were a gift from Dr. Julie M. Blander (Mount Sinai School of Medicine, New York, NY). SHP-1 mutant mev/mev mice, β2m-deficient mice and FcRγ-deficient mice were purchased from Jackson Laboratory (Bar Harbor, ME). All animal experiments were conducted in accordance with the animal guidelines of the Mount Sinai School of Medicine.
Treg cells were isolated by CD4+CD25+ Regulatory T Cell Isolation kit (Miltenyi Biotec, Germany). In brief, CD4+ T cells were enriched from spleen by negative selection, followed by staining anti-CD25-PE and conjugation with anti-PE microbeads. CD4+CD25+ cells were purified by positive selection. Treg cells were labeled with CFSE (2 μM in PBS) at 37°C for 10 min followed by washing with complete medium for 3 times. CFSE-labeled, purified CD4+CD25+ Treg cells were cultured with MDSCs, at a 4:1 ratio, in the presence of OVA peptides (0.5 μg/ml) and IL-2 (100 U/ml) for 4 days. Proliferation (CFSE dilution) was assessed by flow cytometry.
OVA peptide (ISQAVHAAHAEINEAGR) was purchased from AnaSpec (Fremont, CA). Mouse anti-Gr-1 allophycocyanin (APC) or fluorescein isothyocyanate (FITC), mouse anti-CD4 FITC, mouse anti-CD115 phycoerythrin (PE) or APC, mouse anti-F4/80-FITC, mouse anti-CD11b-APC or FITC, mouse anti-CD25-APC, mouse anti-Foxp3-PE, mouse anti-CD36 Alexa Fluor 647, mouse anti-Tie2-biotin, mouse anti-CD11c FITC or PE-Cy7, and isotype-matched mAbs were purchased from eBioscience (San Diego, CA), PE-conjugated anti-PIR, anti-phospho STAT3 (pY705) and anti-phospho STAT1 (pY701) from BD Biosciences (San Jose, CA), and anti-CD206-biotin from AbD Serotec (Raleigh, NC).
Mice were sacrificed and the spleens, tibias, and femurs were harvested. After lysis of red blood cells (RBCs), bone marrow cells and splenocytes were fractionated on a Percoll (GE Healthcare, Piscataway, NJ) density gradient as previously described (Huang et al., 2006). Cells banding at 40–50% were labeled as fraction 1, at 50–60% as fraction 2, and at 60–70% as fraction 3. MDSCs were positively selected from fraction 2 by anti-CD115-PE and anti-PE microbeads (Miltenyi Biotec, Auburn, CA) that include the predominant Ly6C high and some Ly6C low population Fig S4. TILs were isolated from tumor tissues as described previously (Ozao-Choy et al., 2009) and selected as above. Cells were sorted to >90% purity prior to use in subsequent experiments.
The suppressive activity of MDSCs was assessed in peptide-mediated proliferation of TCR transgenic T cells as described previously (Huang et al., 2006). Briefly, splenocytes (1×105) from OT-II mice were cultured in the presence of OVA peptides (1 μg/mL) and serial dilutions of irradiated (2000 rad) MDSCs in 96-well microplates. Cells were pulsed with [3H]-thymidine during the last 8 hours of a 3-day culture.
Splenocytes (4×106) from OT-II mice were cultured in the presence of OVA peptides (1 μg/mL) and irradiated (2000 rad) MDSCs (1×106) in 12-well microplates. After 5-days, cells were harvested and stained with fluorochrome-conjugated anti-CD4, anti-CD25, and anti-Foxp3 or isotype controls followed by flow cytometry.
Cytokine concentrations in culture supernatants were measured using mouse IL-10 and TNFα ELISA kits (eBioscience) as per manufacturer's instructions. NO production was measured using the Griess reagent system (Promega, San Luis Obispo, CA) and arginase activity was measured using the Quantichrom™ arginase assay kit (BioAssay Systems, Hayward, CA).
Sorted CD115+ MDSCs were isolated from tumor using MACS bead purification (Miltenyi Biotec) and co-incubated with LLC tumor cells (target cells) at 20:1 and 10:1 for 4 hours. Supernatants were collected for measurement of lactate dehydrogenase release (CytoTox 96 Non-Radioactive Cytotoxicity Assay kit, Promega). Specific killing (%) was calculated as experimental LDH release/maximum LDH release.
In the OVA-LLC MaFIA tumor model, MaFIA mice (Burnett et al., 2004) were implanted intrahepatically with OVA-LLC tumor cells. When tumors reached 7×7 to 9×9 mm2, CD115+ cells were depleted by injection of AP20187, which cross-links the transgenic suicide protein under the transcriptional control of CD115 promoter and induces caspase 8-dependent apoptosis (10 mg/kg body weight; ARIAD Pharmaceuticals, Cambridge, MA). On the same day, sorted WT or Lilrb3−/− MDSCs (5×106 per mouse) were injected intravenously (i.v.) Two days after MDSC transfer, purified CD45.1+ OT-II T cells (5×106 per mouse) were injected via tail vein followed by a second dose of MDSCs two days later. Five days after the last injection of MDSCs, mice were sacrificed. The tumor weight was measured. The presence of tumor-specific (OT-II) Treg cells in the tumor was assessed by flow cytometry. The proliferative response of purified splenic tumor-specific (OT-II) CD45.1+ T cells in the presence of OVA peptides and irradiated naïve splenocytes was assessed by the standard 3H thymidine incorporation assay.
To assess the therapeutic effect of Lilrb3−/− MDSCs, LLC tumor cells (1×105) were injected subcutaneously or via the tail vein into wild-type C57BL6 mice to establish subcutaneous and lung metastasis tumor models, respectively. On days 5, 10, and 15 after tumor cell injection, 5×106 monocytic MDSCs (Gr-1+CD115+) or granulocytic MDSCs (Ly6G+) from wild type or Lilrb3−/− tumor-bearing mice, were adoptively transferred by tail vein injection. Tumor size was followed. In the lung metastasis model. Mice were sacrificed, followed by intratracheal injection of 15% India ink (American MasterTech). Lung tissues were harvested and washed in water for 5 min followed by fixation in Feket's solution (70% ethanol, 3.7% formaldehyde, 0.75 M glacial acetic acid). Lung metastasis was assessed by the weight of lung tissues.
Target cells were homogenized in Trizol reagent (Invitrogen) and total RNA was extracted. DNA was digested using DNAse I (Invitrogen) and reverse transcription was performed using SuperScript™ III First-Strand Synthesis SuperMix for qRT-PCR (Invitrogen). Real-time quantitative PCR was used to determine relative quantities of mRNA (SYBR® Green PCR Master Mix, Applied Biosystems, Foster City, CA) with β-actin as the internal control. Primers were purchased from Integrated DNA Technologies (Coralville, IA).
Immunostaining of LLC tumor tissues from wild type and Lilrb3−/− mice was performed using biotinylated anti-CD206 (Abd Serotec, NC), anti-CD11b (eBioscience, CA), anti-CCR7, anti-F4/80 (Santa Cruz biotechnology, CA), anti-iNOS, and anti-arginase 1 (BD Biosciences, CA) followed by Cy3-conjugated streptavidin, Cy2-conjugated donkey anti-rat and Cy5-conjugated donkey anti-goat (Jackson ImmunoResearch, West Grove, PA) separately. Endothelial cell staining was performed using rat anti–mouse CD31 followed by Cy3-conjugated donkey anti–rat IgG (Jackson ImmunoResearch). Slides were viewed with a Leica DM RA2 fluorescent microscope (Leica Microsystems, Wetzlar, Germany) using an HC PLAN APO lens at 100×0.033 and 40×/0.85 and Klear Mount medium (GBI, Mukilteo, WA). Images were acquired using a Hamamatsu ORCAER digital camera (Minneapolis, MN) Model C4742-80-12AG, and processed with Openlab version 5.02 (Improvision, Waltham, MA).
Samples were run in sodium dodecyl sulfate (SDS)–10% polyacrylamide gels and transferred to nitrocellulose membranes. The membrane was blocked in Superblock blocking buffer (Thermo Scientific, Rockford, IL) and incubated with an appropriate antibody, followed by a secondary antibody conjugated to horseradish peroxidase. The immunoreactive bands were visualized using the ECL system (Thermo Scientific).
Statistical comparisons between groups were made after presenting the data as means with SD and then applying the Student's t test. Survival data were analyzed using the log-rank test. For samples with equal variance, the paired Student's t test for equal variance was used. For samples with unequal variance, Wilcoxian signed-rank test was used for statistical analysis. P < 0.05 was considered to be statistically significant.
We gratefully thank Dr. Clifford A. Lowell and Dr. Toshiyuki Takai for providing PIR-B knock out mice, Ms. Marcia Meseck for her kind editing of this manuscript and Dr. Hong-Ming Hu provided the OVA-LLC cell line. The work was supported in part by grants from the National Cancer Institute, Black Family Stem Cell Foundation, and Pfizer research fund to S. -H. Chen, grant support from Susan G. Komen Breast Cancer Foundation to Ping-Ying Pan
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.