Department of Biochemistry, La Trobe Institute for Molecular Science, La Trobe University, Melbourne, Australia.
Myeloid-derived suppressor cells comprise a heterogeneous population of immature immune cells that expand during the course of cancer progression. These cells adopt an immunosuppressive phenotype that impairs the anti-tumour immune response through modulation of T cells, natural killer cells, dendritic cells and macrophages, as evidenced both in mouse models of cancer and patients. While much attention has been focused on the immunosuppressive roles of myeloid-derived suppressor cells, it is becoming increasingly clear that they can also promote tumour progression and metastasis via other immune-independent functions, including the regulation of angiogenesis and tumour invasiveness. Their arrival at metastatic sites prior to the arrival of tumour cells also contributes to the formation of a favourable pre-metastatic niche. This review will summarise the various roles for myeloid-derived suppressor cells, including new evidence that supports a role in promoting a favourable pre-metastatic environment for bone metastasis.
That cancer progression relies heavily on the interaction between malignant cells and host stromal cells within the tumour microenvironment, including fibroblasts, endothelial cells and immune cells, is now well accepted. Although initially thought to have anti-tumour roles, it is now well known that immune cells can also become influenced by tumour-derived and other microenvironmental factors and adopt a pro-tumorigenic, immunosuppressive phenotype. In the last three decades, much attention has been focused on the roles of myeloid-derived suppressor cells (MDSCs) in this process. Mostly known for their ability to suppress the anti-tumour immune response through modulation of adaptive and innate immune activation, it is becoming increasingly clear that MDSCs have additional immune-independent pro-metastatic roles, including promoting angiogenesis and bone degradation, and that they play an integral part in the formation of a pre-metastatic niche.
MDSCs comprise a heterogeneous population of immature immune cells. In normal physiology, these cells differentiate into macrophages, dendritic cells, granulocytes and neutrophils. However, in cancer, tumour-derived factors such as vascular endothelial growth factor (VEGF), G-CSF, GM-CSF and the pro-inflammatory proteins S100A8 and S100A9 block this differentiation, leading to an expansion of immature myeloid cells in bone marrow, peripheral blood, spleen and tumour.1-4 Murine MDSCs are most often characterised by co-expression of the cell surface markers CD11b and Gr-1, and they can be divided into subclasses based on differential expression of the Gr-1 epitopes Ly6G and Ly6C. Granulocytic MDSCs are CD11b+/Gr-1+/Ly6Cmid/Ly6Ghigh while monocytic MDSCs are CD11b+/Gr-1+/Ly6Chigh/Ly6G– (table 1).2,5,6 In humans however, MDSCs are defined as CD11b+/CD14–/CD33+ or Lin–/HLA-DR–/CD33+.1,2 Given the heterogeneity of these populations, it will be essential in the future to identify markers that further subdivide unique populations. At present, the most conclusive way to define an MDSC population is by demonstrating its functional ability to suppress anti-tumour immune responses.
Numerous studies have now demonstrated that MDSCs accumulate throughout disease progression.2,7 A small number of CD11b+/Gr-1+ cells are found in healthy naïve mice, however these cells are not true MDSCs because they do not exert immunosuppressive effects. During chronic inflammation and tumourigenesis however, these cells fail to differentiate and instead rapidly expand and adopt an immunosuppressive phenotype.2,7 A correlation between MDSC accumulation in blood and metastatic progression has been observed in human patients with breast, colorectal, gastric, brain, prostate, pancreatic, oesophageal, liver, and skin cancers.8-13 For example, in the peripheral blood, MDSCs increase from an average frequency of 1.26% and 1.96% in healthy patients and early breast cancer patients respectively, up to 4.37% (and as high as 25%) in patients with advanced metastatic disease.8 This correlation has been recapitulated in immunocompetent mouse models of cancer, where MDSC numbers in the spleen and peripheral blood correlate closely with tumour burden and metastatic potential.14-17 One of the most exaggerated models of MDSC accumulation is the 4T1 murine model of metastatic breast cancer, in which MDSC levels can increase by >20-fold compared to naïve mice.18 Although less pronounced, this accumulation is also observed in the MMTV-PyMT breast cancer model, with an increase in MDSC accumulation of approximately seven-fold in tumour bearing mice.18 In general, MDSC accumulation is more pronounced in mouse models, since overall tumour and/or metastatic burden is often higher than in patients who have received therapeutic intervention.
As suggested by their name, the most prominent role for MDSCs is their ability to suppress the anti-tumour immune response, as extensively reviewed elsewhere.4,19,20 Both granulocytic and monocytic MDSCs can suppress the proliferation and activation of T cells. MDSCs perform this function through several mechanisms. They overexpress arginase-1, which converts L-arginine to ornithine and urea.4 They also express high levels of the cystine transporter xc– for importing cystine and lack the ASC transporter required for exporting cysteine.21 MDSCs therefore act as a sponge for arginine and cysteine, reducing the availability of amino acids that are essential for protein synthesis and proliferation of T cells. Additionally, granulocytic MDSCs produce high levels of reactive oxygen species and peroxynitrites, and monocytic MDSCs produce nitric oxide through expression of inducible nitric oxide synthase.4,7,22 Peroxynitrites and nitric oxide impair DNA damage repair and protein synthesis and consequently lead to reduced T cell proliferation.
In addition to their effects on T cells, MDSCs interfere with anti-tumour responses by decreasing the prevalence and function of natural killer cells through TGFβ, IL-1β, or natural killer p30-dependent mechanisms.23-25 They are also able to dampen the immune response by triggering the activation and expansion of another immune suppressive population, regulatory T (TReg) cells.26-28 Furthermore, MDSCs can function to skew the polarisation of macrophages to a pro-inflammatory and tumour promoting M2 phenotype,29 and impair the ability of dendritic cells to activate T cells.30
These multiple suppressive functions are likely to have important implications therapeutically. A number of therapeutic approaches for cancer rely on immune activation and therefore may not be efficacious in an immune-suppressed environment populated by MDSCs. A recent strategy has focused on improving the anti-tumour immune response by combining immunotherapies with agents that manipulate the function of MDSCs, either by eliminating them, deactivating them, blocking their formation or forcing their differentiation, as extensively reviewed elsewhere.31,32 One agent trialled in the clinic is all-trans retinoic acid (ATRA), which functions to promote the maturation of MDSCs into mature myeloid cells such as macrophages, dendritic cells and granulocytes. In renal cell and small cell lung cancer patients, ATRA improved anti-tumour immune effects in combination with vaccine therapy.33,34 However, a new study in murine breast cancer models suggests that ATRA promotes the differentiation of MDSCs into CD11b+/Gr-1–/F4/80+ macrophages that are even more immunosuppressive than their precursors.18 This increase in T cell-suppressive ability was associated with enhanced metastatic outgrowth in the lungs of ATRA-treated mice, suggesting that extreme caution should be taken when going forward with this therapeutic approach. Therapies such as 5-fluorouracil or gemcitabine that deplete MDSCs without inducing differentiation, might be more efficacious at preventing tumour progression.32 In order to improve the outcome of combination therapies, more information is needed regarding the diverse functions of MDSCs, both within the primary tumour and in circulation. This will ensure the educated design of clinical trials that use combination therapies in the right setting based on mechanisms that maximise the likelihood of synergy.
In addition to their presence in bone marrow, blood and spleen, MDSCs have been shown to infiltrate into the tumour microenvironment, where they concentrate at the periphery of the tumour and can exert effects on tumour growth.16,18 MDSCs can mediate reduced sensitivity to anti-VEGF therapy,35,36 suggesting an important role in promoting angiogenesis. In fact, it has been shown in models of colorectal cancer and Lewis lung carcinoma that MDSCs not only promote angiogenesis and tumour growth, but actually directly incorporate into the vasculature and express endothelial markers.15 In this study, enhanced tumour growth and angiogenesis were reversed using matrix metalloprotease-9 (MMP9)-deficient MDSCs. Considering MMP9 has been implicated in VEGF activation,37 this study suggests that apart from direct differentiation into endothelial cells, MDSCs are likely to be critical for promoting angiogenesis within the tumour microenvironment via production of MMP9.15 This was supported in studies using mice lacking the MMP inhibitor TIMP2, whereby accumulation of MDSCs with high MMP activity increased angiogenesis and tumour growth rate in the Lewis lung carcinoma model.38 Additionally, in the 4T1 breast cancer model, co-culture of tumour cells with MDSCs from late-stage tumour bearing mice enhanced tumour cell invasion in an MMP-dependent manner, and this was associated with enhanced metastasis to lung.16
The evidence linking MDSCs to angiogenesis is not limited to their expression of MMPs. Other potential promoters of MDSC-induced angiogenesis include regulator of G protein signalling-2 (Rgs2),39 and Stat3, which governs MDSC-induced angiogenesis through upregulation of angiogenic genes such as VEGF, bFGF, IL-1β, MMP, CCL2 and CXCL2.40 Human studies that support this pre-angiogenic role of MDSCs are lacking, however recent work in patients with metastatic paediatric sarcomas has reported a novel subset of human MDSCs called fibrocytes (CD14–/Cd11chi/CD123–) that, apart from their immune suppressive activity, stimulate vessel growth in tube formation assays.41
In addition to their roles in promoting local invasion and angiogenesis within the primary tumour microenvironment, myeloid-derived suppressor cells have also been implicated in the formation of a pre-metastatic niche. The concept of a pre-metastatic niche involves tumour-induced changes to distant organs that make them well suited to support impending metastatic outgrowth. In 2005, Lyden and colleagues first described early events that initiate a pre-metastatic niche, whereby recruitment of hematopoietic progenitor cells into the lung promoted tumour cell recruitment and degradation of the extracellular matrix.42 Studies in the last decade have focused on MDSCs as a subset of bone marrow-derived cells that is critical for establishing this favourable environment due to their roles in both angiogenesis and immunosuppression.
Hypoxia, or low oxygen levels, in the primary tumour has been linked with metastatic progression and is associated with the formation of a pre-metastatic niche at distant sites.43 It has been demonstrated in mouse models of metastasis that hypoxia inducible factor (HIF)-dependent lysyl oxidase (LOX) secreted by hypoxic tumour cells can remodel the extracellular matrix in secondary sites to promote the adhesion of CD11b+ bone marrow-derived cells.43 In support of this, pre-injection of mice with conditioned media from hypoxic breast tumour cells prior to intravenous injection of tumour cells resulted in clustering of granulocytic MDSCs at the terminal bronchioles of the lung and increased lung colonisation, a phenomenon that was not observed using the normoxic conditioned media counterpart.44 In this study, enhanced metastasis was associated with immune suppression, whereby NK cells had reduced cytotoxic activity.
MDSC accumulation in pre-metastatic lungs is also observed in the 4T1 mammary tumour model, where they have been associated with immune suppression via inhibition of interferon-γ (IFN-γ) production.45 In this study, pre-metastatic lungs harbouring MDSCs had increased levels of growth factors, Th2 cytokines (IL-4, IL-5, IL-9 and IL-10) and inflammatory cytokines (IL-1β, SDF-1, MCP-1), which functioned collectively to create a proliferative, immunosuppressive and inflammatory environment optimal for tumour cell growth. Additionally, the vasculature of the pre-metastatic lung was more disorganised and leaky than the normal lung and more amenable to tumour cell extravasation and metastasis.45 This confirmed an important role for MDSCs in vascular remodelling of the pre-metastatic niche. The evidence for a pre-metastatic niche extends beyond the lung in this model. MDSCs also infiltrate into the brain prior to the arrival of 4T1 tumour cells, where they have been implicated in tumour recruitment to the metastatic site.46
Apart from their role in supporting a pre-metastatic niche in soft tissues, recent evidence in breast cancer models suggests that MDSCs also function to promote a favourable microenvironment in bone, a common site of metastasis in patients. When MDSCs from tumour-bearing mice were co-injected with MDA-MB-231 mammary tumour cells into the fat pad of nude mice, there was a 30% decrease in trabecular volume in femurs compared to tumour cells co-injected with naïve myeloid cells.47 This phenomenon was observed prior to the outgrowth of bone metastases, suggesting that bone loss due to MDSCs may be integral in establishing a pre-metastatic environment in bone. To further confirm this, mice were injected intra-tibially with tumour MDSCs or naïve myeloid cells just prior to cardiac injection with MDA-MB-231 cells. Tumour MDSCs enhanced bone tumour growth and this was associated with enhanced bone degradation and an increased osteoclast activity.
Surprisingly, when MDSCs were isolated from the bone marrow of tumour-bearing mice and cultured on dentine slices in the presence of RANK ligand, the cells differentiated into tartrate-resistant acid phosphatase positive (TRAP+) osteoclasts with bone-degrading capabilities. This was confirmed in vivo, where GFP+ MDSCs became TRAP+ after intratibial injection, providing the first evidence that MDSCs can differentiate into osteoclasts in the bone microenvironment.
Other groups demonstrated that this phenomenon also exists in the 4T1 breast cancer model,48 and in multiple myeloma.49 In the 4T1 model, MDSCs that were isolated from the bone marrow of mice bearing bone metastases were able to differentiate into TRAP+ osteoclasts, whereas those isolated from the metastatic lung, lymph node, spleen or blood of tumour bearing mice could not. This highlighted an importance for the bone microenvironment in this process. MDSC-derived osteoclasts also expressed other osteoclast markers such as cathepsin K, MMP9 and carbonic anhydrase-2, and were capable of resorption of bovine cortical bone slices in vitro and of tibias in vivo.49 High levels of nitric oxide were shown to be critical for the differentiation of MDSCs to osteoclasts, as treatment of mice with 1400W, an inhibitor for inducible nitric oxide synthase, attenuated bone damage caused by MDSCs. Considering recent work by Kang and colleagues has revealed that metastatic outgrowth of disseminated breast tumour cells in bone is dependent on osteoclast activation,50 MDSCs may represent an important subset of osteoclast progenitors that is crucial in the establishment of a pre-metastatic niche in bone.
We have provided a brief overview of the diverse functions of MDSCs in tumour growth and metastasis (figure 1). Most of the work to date has focused on the immune suppressive nature of these cells and their ability to extinguish the anti-tumour immune response through modulation of T cell proliferation and activation. There is now clear evidence however, that these cells have other functions in promoting angiogenesis and providing a favourable microenvironment for outgrowth in common metastatic sites such as lung and bone. Additionally, MDSCs have varying roles depending on the microenvironment from which they are derived. Therefore, further characterisation of specific markers that predict functional characteristics of MDSCs is necessary. Among many other factors, the secondary roles for MDSCs seem to be governed in large part by proteolytic enzymes such as MMPs, which are highly expressed in these cells and whose activity is critical for the cell signalling, matrix degradation and bone resorption necessary to promote metastasis. As the understanding of these diverse MDSC subtypes and functions continues to expand, therapies targeting MDSCs in combination with immunotherapy will continue to improve.