Distant Relations: Macrophage Functions in the Metastatic Niche

1.1 Macrophage phenotypes and activation influence cancer progression

The mechanisms underlying macrophage recruitment and activation during the immune response provide insights into the potential contributions of macrophages in the tumor microenvironment. Macrophages represent a diverse population of immune cells that are generally classified as tissue resident macrophages or BMDMs (Table 1). Resident macrophages are derived from yolk sac or fetal liver progenitors to aid in embryonic development and ultimately differentiate into several subtypes to perform tissue-specific functions, phagocytose wastes and foreign bodies, and initiate the immune response [13]. BMDMs are typically recruited to tissues as myeloid-derived cells, which fully differentiate within the tissue and can be polarized in response to signals sensed within the tissue. Traditional macrophage polarization is classified as classical “M1” activation or alternative “M2” activation. Typically, M1 polarization is thought to promote pro-inflammatory, tissue destructive functions and M2 promotes anti-inflammatory, tissue repair functions. Although these definitions imply that M1 and M2 macrophages are distinct entities that perform specialized functions, this classification system is now considered an oversimplification since specific phenotypes are distributed along a continuum to allow for a smooth transition between pro-inflammatory, anti-inflammatory/recovery, and homeostatic states [12,14–16].

However, in many diseased states the balance of polarization is skewed towards chronic inflammation (M1-like) or tissue repair (M2-like). In tumors, cancer cells generally promote the maintenance of a more M2 macrophage polarized state to aid in evading the immune response, promoting angiogenesis, releasing pro-tumoral growth factors, and remodeling the local tissue; however, several M1-associated cytokines such as tumor necrosis factor (TNF)-α and interleukin (IL)-6 may be pro- or anti-tumorigenic depending on the tumor type and the environment [17,18]. In fact, many macrophage subpopulations may contribute to tumor progression and metastasis. While tumor associated macrophages are thought to primarily consist of BMDMs, it is often difficult to distinguish resident macrophages from tumor associated macrophages [7]. A recent mass cytometry-based study of macrophage populations isolated from renal cell carcinomas identified up to 17 distinct macrophage populations [19], further highlighting the need to better understand how phenotypic diversity in macrophages influences cancer progression. Since it is likely that a diverse array of macrophage populations at the metastatic site perform a range of functions to facilitate metastatic colonization, it is important to clearly characterize and not over-generalize macrophage phenotypes and functions to understand how they influence metastasis [16].

2.1 Primary tumor secretome impacts bone marrow derived cell recruitment and function in the secondary site

The ability of cancer cells to localize to specific organs is potentially dependent on the ability of the primary tumor to establish bone marrow derived cell clusters—containing progenitors of BMDMs along with other bone marrow-derived cell types—in the pre-metastatic niche [20–22]. For example, systemic signaling produced by intradermally injected Lewis lung carcinoma (LLC) tumors or LLC conditioned media promotes fibronectin expression in the lungs and liver, which are where intradermally injected LLC tumors generally metastasize. Bone marrow derived cells expressing the fibronectin receptor very late antigen (VLA)-4 localize to these fibronectin-rich regions where they promote neovascularization and metastasis in a vascular endothelial growth factor (VEGF)-1 dependent manner. However, in the absence of primary tumor signals bone marrow-derived cell clusters failed to form, resulting in significantly fewer metastases. Similarly, primary tumor signals from highly metastatic B16 melanomas in the same model initiate pre-metastatic bone marrow derived cluster localization to the lungs, liver, testes, spleen, and kidneys, which are consistent with metastatic burden of B16 tumors. Interestingly, LLC cells injected through the tail vein of the B16-educated mouse model seeded in all traditional B16 affected organs rather than just the lungs and liver, which are the normal sites for LLC seeding [20]. These data highlight the importance of soluble factors derived from the primary tumor in establishing the pre-metastatic niche and contributing to organotropic metastases.

Recent studies have elucidated specific factors within the primary tumor secretome that contribute to initiation of the lung pre-metastatic niche [21,22]. In LLC and B16 tumor secretomes, VEGF-A, TNF-α, and transforming growth factor (TGF)-β induce lung endothelial cell secretion of monocyte chemoattractants S100A8 and S100A9 [21]. In addition to directly recruiting bone marrow derived cells, S100A9 may also increase their retention in fibronectin-rich regions by activating β2 integrins—including CD11b—which increases leukocyte affinity to fibronectin in vitro [23,24]. Activation of the endothelium via VEGF-a, TNF-α, TGF-β, S100A8, and S100A9 is associated with increased endothelial permeability, which enhances extravasation of recruited monocytes and cancer cells in vivo [25–29]. In an orthotypically injected MDA-MB-231 mammary tumor model, primary tumor cell-derived lysyl oxidase (LOX) localized to fibronectin-rich tissues in the lung where it cross-linked collagen IV, resulting in increased affinity to CD11b+/VEGFR1+ monocytes. This binding promoted monocyte expression of the collagen cleaving protein, matrix metalloproteinase (MMP)-2, which aided cancer cell extravasation by reducing invasion impedance and recruiting BMDMs via production of chemotactic collagen fragments [22]. While these data highlight some mechanisms of early bone marrow derived cell recruitment and function correlated with fibronectin expression in the lungs, further investigation is required to better understand how primary tumor secretomes impact organotropic pre-metastatic niche formation.

2.2 Exosomes influence macrophage localization and function

In addition to soluble growth factors, primary tumor-derived exosomes have also been shown to impact the pre-metastatic niche. Exosomes are vesicular carriers secreted by cells for intercellular transport, and exosome integrin expression promotes localization to specific organs [30–33]. As highly phagocytic cells, macrophages engulf exosomes and process the cargo allowing distant, primary cancer cells to alter secondary macrophage phenotypes. Tail vein injection of exosomes from organotropic breast (MDA-MB-231—lung, MDA-MB468—lung and liver) and pancreatic cells (BxPC-3 and HPAF-II—liver) into mice revealed that exosomes with α6β4 and α6β1 integrins predominantly localize to the lungs while exosomes with αVβ5 were associated with liver localization. In both cases, these exosomes were found colocalized with macrophages in fibronectin-rich environments [33]. Exposure of macrophages to exosomes has been shown to directly promote their ability to further enhance pro-tumoral immune cell recruitment. Exosomes collected from lung cancer cells lines contain microRNA-21 or microRNA-29a which bound as ligands to toll-like receptor (TLR)-7 in mice (TLR-8 in humans) and induced expression of IL-6 and TNF-α in an NF-κB-dependent manner [34]. Because both TNF-α and IL-6 are cytokines associated with leukocyte recruitment during the immune response and tumor promotion, these studies warrant further investigation into the effects of primary tumor-derived exosomes on MAM function in the metastatic site.

2.3 Primary tumor regulation of resident macrophages in the metastatic site

In addition to enhancing the recruitment of macrophage precursors into the metastatic site, primary tumor-derived factors can also impact tissue resident macrophages within that site. For example, initiation of the bone pre-metastatic niche is often driven by activation of bone resident macrophages—osteoclasts. In normal physiology, osteoclasts resorb bone tissue and act in cooperation with bone-forming osteoblasts to promote bone development [35]. However, signals derived from the primary tumor can activate osteoclasts to promote the formation of osteolytic lesions, which serve as hospitable environments for future cancer cells [36–38]. For example, secreted LOX from hypoxic, estrogen receptor negative (ER-) mammary tumor cells induced the activity of the NFATc1 transcription factor, a key regulator of osteoclastogenesis. By skewing homeostasis towards bone resorption, LOX-induced osteoclastogenesis lead to the formation of pre-metastatic lesions, which serve as hospitable environments for metastatic cancer cells, in both orthotopically transplanted tumor models and in tumor-free mice injected with tumor conditioned media [37]. Similarly, non-small cell lung cancer-derived TGF-β has been shown to induce osteoclastogenesis, leading to increased metastatic burden in murine models [38,39]. TGF-β-induced osteoclastogenesis stimulates the release of additional soluble factors including TGF-β, creating a feed-forward cycle that continues unless osteoblasts can balance the system. However, continuous stimulation by the primary tumor or the arrival of tumor cells can maintain this osteolytic microenvironment by secreting signals that promote further osteoclastogenesis, suppress osteoblast activity, and directly degrade the bone, generating a “vicious cycle” of tumor progression [36].

The primary tumor secretome can also promote accumulation of lung resident alveolar macrophages that generate a pro-tumorigenic pre-metastatic niche for cancer cells. In the MET-1 mammary tumor model, accumulation of alveolar macrophages was found in the lung, which correlated with increased levels of the leukocyte chemoattractant, C5a. This accumulation of alveolar macrophages occurred before the arrival of cancer cells, indicating that primary tumor signals initiate lung secretion of C5a. Furthermore, depletion of alveolar macrophages significantly reduced lung metastasis of MET-1 cells, which correlated with a shift in the immune profiles from tumor-supportive Th-2 cells to tumoricidal Th-1 cells [40]. Further investigation into how the primary tumor initiates C5a production in the lungs may illuminate mechanisms to prevent this pre-metastatic niche initiation.

In addition to soluble factors, primary tumor-derived exosomes can also impact resident macrophages in distant sites. Isolation of exosomes from liver-tropic pancreatic ductal carcinoma (PAN02) cells shows that PAN02 exosomes containing migration inhibitory factor localize to the liver. Upon arrival, Kupffer cells (liver-resident macrophages) engulf the PAN02 exosomes and activate local fibroblasts to secrete fibronectin, contributing to the formation of pre-metastatic nodules where BMDMs localize [32]. These studies collectively suggest that primary tumor-derived factors can impact the function of resident macrophage in distant sites. Additional studies are needed to delineate the mechanisms of resident macrophage regulation and function in these and other distant tissue sites.

3.0 Macrophages Influence Adhesion, Extravasation, and Early Colonization of Cancer Cells

In the secondary site, MAM interactions with arriving cancer cells have been shown to promote metastatic behavior (Table 3). In many cases, the same mechanisms that macrophages use to recruit leukocytes during inflammation, such as enhancing adhesion and transmigration, are used to recruit cancer cells into the metastatic site. MAMs can further aid secondary tumor formation using the same mechanisms they utilize during tissue repair, such as inhibiting tissue-destructive immune response, promoting cellular growth, initiating angiogenesis, and remodeling the matrix. Furthermore, these MAMs can continually recruit BMDMs to maintain the pro-tumoral immune environment. Since MAMs appear to play major roles throughout metastatic progression, understanding the mechanisms through which MAMs drive tumor cell extravasation and early colonization may reveal therapeutic targets.

3.1 The presence of MAMs during extravasation promotes metastatic colonization

Several studies have provided strong evidence that the presence of MAMs during the metastatic extravasation process is critical to successful formation of a metastatic lesion [9–12]. Qian et al. injected breast cancer cells into the tail veins of mice following macrophage depletion with L-clodronate and found that the number of cancer cells arrested in the lung of macrophage-depleted mice dropped significantly over a 36-hour period post injection, suggesting that macrophages aid retention of disseminated cancer cells in the lung. Additionally, doubling times of tumor growth were twice as high in mice with macrophages than those without, highlighting the importance of macrophages in sustaining a pro-tumoral microenvironment. The authors also found that extravascular cancer cells in these models were in direct contact with CD11b+ BMDMs, indicating that adhesion of cancer cells to MAMs aids transmigration of the endothelium [9].

In addition to interacting with disseminated tumor cells, MAMs also prevent tumoricidal cell infiltrate. Headley et al. observed shedding of 0.5–25 um portions of cytoplasm from cancer cells arrested in the vasculature [12]. These mitochondria-containing “cytoplasts” remained within the lung microvasculature and preceded three waves of monocytic cells: neutrophils (~15 minutes), conventional monocytes (15min-6hrs), and non-alveolar macrophages, patrolling monocytes, and dendritic cells (6–12hrs). Of these phagocytic cells recruited to the metastatic site, non-alveolar macrophages and conventional monocytes were mostly responsible for engulfing the cytoplasts and extravasating into the pre-metastatic niche. Engulfment of cytoplasts increased macrophage expression of adhesion and chemotactic receptors important in cancer progression including vascular cell adhesion molecule (VCAM)-1, exosomal surface protein CD63, and natural killer cell adhesion mediator CD155 [12,41–44]. However, in CCR2 knockout models, immunostimulatory CD130+ dendritic cells and CD8+ T cells dominated the cytoplast-engulfing lung infiltrate, suggesting that CCR2+ BMDMs are important in promoting an immunosuppressive environment [12]. Furthermore, inhibiting recruitment of MAMs via the CCR2-CCL2 axis by genetic or chemical means reduced metastatic burden of tail vein injected MET-1, MDA-MB-231, and melanoma cells in the lungs [9–12], demonstrating the importance of this axis in metastatic progression. Taken together, these studies suggest that the presence of MAMs at the time of extravasation aids formation of metastatic lesions through direct interactions with tumor cells and indirect regulation of immune responses.

3.2 Mechanisms by which macrophages enable metastatic extravasation and colonization

Recent studies have explored potential mechanisms through which macrophages contribute to metastatic progression. MAMs expressing CCR2 are directly implicated in metastatic progression [9–12], and CCR2-mediated regulation of CCL3 has specifically been highlighted as a potential therapeutic target [11]. Stimulation of CCL2/CCR2 induces CCL3 in MAMs, which contributes to MAM retention in the metastatic site in a CCR1-dependent manner. Furthermore, the CCL3/CCR1 axis was found to be important for direct binding between macrophages and tumor cells, in part through an α4-integrin dependent mechanism [9,11]. α4 integrins bind to VCAM-1, which is expressed by leukocytes and endothelial cells [45], but can also be overexpressed in cancer cells [46]. Chen et al. showed that mammary fat pad injections of the lung tropic MDA231-LM2-4175 breast cancer cell line in which VCAM-1 levels were knocked down grew at the same rate, but created significantly fewer lung metastases due to decreased binding affinity to macrophages. When modeling extravasation via tail vein injection of MDA231-LM2-4175 cells, reduced VCAM-1 expression did not affect accumulation of cancer cells in the lungs during the week after injection, but significantly decreased formation of metastases. Binding of macrophage α4 integrins to VCAM-1 in breast cancer cells activates downstream PI3K-AKT pro-migratory, pro-survival pathway in the cancer cells, which reduces cancer cell apoptosis and promotes secondary tumor formation [41,42].

Additional macrophage-derived factors, such as VEGF and endothelin-1 (ET-1), have also been implicated in promoting metastatic colonization. In mammary tumor models, monocyte/macrophage VEGF has been shown to contribute to retention of MET-1 mammary tumor cells in the lung. Furthermore, in vitro studies have shown that VEGF promotes endothelial permeability and transendothelial migration of MET-1 cells. Macrophage-derived VEGF has also been linked to cancer cell retention in the lung [10]. Studies using LLC lung cancer and B16 melanoma models suggest that VEGF also acts on VEGFR1-expressing macrophages and endothelial cells to induce production of MMP9 and monocyte chemoattractant, CXCL12, [47]. MMP-9 contributes to tumor progression by increasing vascular permeability [47], degrading the ECM, cleaving growth factors that ultimately aid in tumor proliferation [48], and angiogenesis to nourish the highly metabolic tumor [5]. Furthermore, MAM-induced VEGF-VEGFR1 signaling maintains gradients of the CXCL12 chemokine, which is a strong chemoattractant for monocytes and CXCR4+ tumor cells, contributing to the cycle of BMDM recruitment. [10,20,49]. Macrophage-derived ET-1, a soluble vasoconstrictor, has been shown to contribute to metastatic colonization. Tail-vein injection of muscle-invasive bladder cancer cells that secrete soluble vasoconstrictor, endothelin-1 (ET-1), leads to improved recruitment of BMDMs and ultimately increased metastatic burden compared to ET-1 knock-down controls. Inhibition of macrophage ET-1 receptor ET_aR significantly reduced metastatic burden in mice by preventing downstream production of pro-tumoral cytokines such as IL-6, CCL2, COX-2, and MMPs, which are all associated with BMDM recruitment, activating angiogenesis, and secondary tumor formation [50,51].

Macrophages recruited to secondary locations not only influence extravasation through direct interactions with cancer cells, but can also interact with other cells within the pre-metastatic niche to promote metastasis. As alluded to in previous sections, macrophage-endothelial cell interactions can enhance metastatic potential of cancer cells by increasing endothelial affinity and permeability which aids circulating cancer cell adhesion and transmigration into the tissue [10, 25–29, 47, 52]. Additionally, cancer associated fibroblasts (CAFs) polarize TAMs to perform pro-tumoral functions in the primary site [53]. In the secondary site, CAFs and MAMs collaborate to increase metastatic potential of arriving tumor cells. In both lung and liver metastases, the primary tumor promotes fibroblast secretion of fibronectin, which recruits BMDMs to the secondary site [20,22,32,33]. BMDMs recruited to the liver promote pancreatic ductal carcinoma (PDAC) metastasis by secreting granulin, generating a fibrotic environment by activating hepatic stellate cells into myofibroblasts. These myofibroblasts ultimately increase growth of secondary tumors by secreting periostin that activates Wnt and αVβ3-Akt/PKB signaling pathways in PDAC cells. Depletion of granulin prevented MAM-dependent development of the fibrotic environment and subsequent PDAC metastatic growth while inhibition of periostin significantly reduced metastatic outgrowth of PDAC cells by blocking the CAF/PDAC paracrine loop [54]. Understanding how macrophages promote pro-tumoral functions of other cell types within the pre-metastatic niche—including endothelial cells and CAFs—may illuminate several therapeutic targets that can serve as combination therapy to alleviate metastatic progression.

While these studies highlight the importance of MAM-derived factors in tumor cell extravasation and growth in the metastatic site, given the diversity of mechanisms through which TAMs in the primary tumor site contribute to tumor growth and progression, it is likely that there are additional MAM-related factors that contribute to metastasis. For example, it is widely accepted that TAMs promote intravasation via the CSF-1/EGF paracrine loop [1–8,52]. Specifically, secretion of CSF-1 by cancer cells recruit BMDMs to the primary tumor where they secrete EGF, which contributes to vascular permeability and cancer cell transmigration [52]. Additionally, cancer stem cells have been identified in circulation and may be better suited to infiltrate and colonize the secondary site than other tumor cell populations since they can quickly adapt to new environments [55]. Macrophage crosstalk with cancer stem cells in the primary tumor promotes development, maintenance, self-renewal, and the epithelial to mesenchymal transition of cancer stem cells via several mechanisms which could also promote early tumor development in the secondary site [55,56]. Whether these mechanisms or other known pathways through which primary TAMs function are replicated in the metastatic site remains to be investigated. Understanding the similarities and differences in TAM and MAM function is critical for developing the most effective macrophage-targeted therapies that will target macrophage function in both sites.

3.4 Macrophage-tumor cell hybrids from the primary tumor promote pre-metastatic niche formation

Another potential mechanism by which macrophages aid tumor cell extravasation and colonization is through fusion between TAMs and cancer cells. It has been shown that when macrophages engulf apoptotic cancer cells they incorporate cancer cell DNA into their nucleus. These “hybrids” promote invasion and immune evasion via intrinsic macrophage functions but tumor traits promote increased proliferation [57–59]. Recently, macrophage-tumor cell hybrids were identified in the blood of patients with epithelial cancers such as melanoma (SK-Mel-24, 28, 31) [59], breast (MCF-7, MDA-MB-231), colorectal (HepG2), and ovarian (SKOV-3) cancer [58]. The hybrid cells expressed both M2-like macrophage markers (CD163, CD206, CD204) and epithelial markers (EpCAM, KRT) and were significantly more invasive than either cell type cultured alone [58–59]. Macrophage-melanoma hybrids subcutaneously injected into the hind limbs of nude mice disseminated and produced metastatic lesions in the pancreas, indicating that these hybrids can establish distant metastases [59]. It is hypothesized that these hybrids also contribute to pre-metastatic niche formation, but additional research is needed to understand the roles of macrophage-tumor cell hybrids in pre-metastatic niche formation, extravasation, and colonization and how these mechanisms may be future therapeutic targets.

4.0 Macrophages are a potential therapeutic target in the secondary tumor microenvironment

Numerous strategies are being pursued to target macrophages in the primary tumor, including ablation of TAMs residing in the tumor, impeding further recruitment of BMDMs, and polarizing TAMs toward a tumoricidal phenotype [60,61]. Several of these treatment strategies targeting the CSF1-CSF1R [62], CCL2-CCR2 [63], and VEGF-VEGFR1 [20,52] axis effectively reduce primary tumor burden and metastasis in murine models [11,20]. However, clinical trials targeting TAMs have thus far been less successful [61]. Inhibition of the CSF1/CSF1R pathway decreases metastatic burden in murine models and is generally well tolerated in patients, but is ineffective in reducing recruitment of BMDMs to the primary tumor in clinical trials [61,62]. Recruitment of BMDMs to the primary tumor may be maintained when the CSF1 pathway is inhibited via stromal cell secretion of other BMDM recruitment cytokines. Kumar et al. demonstrated that CSF1 downregulates CAF secretion of CXCL1, which is a granulocyte/monocyte recruiting cytokine. Inhibition of CSF1 alone promotes upregulation of CXCL1 and, therefore, increases recruitment of granulocytes and BMDMs. However, inhibition of CSF1R and CXCL1 receptor, CXCR2, significantly reduces BMDM recruitment and primary tumor growth in mice [64]. These data show the importance of understanding the various mechanisms that contribute to BMDM recruitment and finding the optimal combination of targets to prevent BMDM recruitment to the pre-metastatic niche.

Similarly to CSF1R inhibition, blocking BMDM recruitment via CCL2 has shown promise in pre-clinical studies. In murine models, ablation of CCR2+ MAMs reduces metastatic burden, possibly due to tipping the immune cell infiltrate from pro-tumoral CCR2+ MAMs towards anti-tumoral CD130+ dendritic cells and CD8+ T cells [11,12]. However, because ablating CCR2+ macrophages directly is not currently a clinical option due to necessary homeostatic functions of macrophages, clinical trials have focused on reducing CCL2 levels with anti-CCL2 antibodies. Stage 1 trials with anti-CCL2 monoclonal antibody carlumab demonstrated that it is well tolerated by patients, but carlumab on its own does not consistently suppress free CCL2 levels and ultimately increased CCL2 levels over time [65]. The complications of carlumab treatments are likely due to short-term affinity of carlumab, overcompensation of cells expressing CCL2, and increased sensitivity of macrophages to CCL2 through upregulation of CCR2 [65–67]. For example, some murine models of mammary cancer associate cessation of anti-CCL2 agents with accelerating metastatic progression by increasing influx of monocytes to the primary and secondary tumors. In these models, monocytes/macrophages in the lungs increased metastatic growth of cancer cells in an IL-6/STAT3 dependent manner via VEGF-A driven neovascularization. When both CCL2 and IL-6 are inhibited, there is reduced metastatic burden [67]. While there is much to be understood about how these therapeutics influence macrophage recruitment and function in the secondary site, these data emphasize the importance of not only inhibiting BMDM recruitment, but concurrently targeting MAM functions to effectively target metastasis.

One important consideration for developing approaches that inhibit pro-metastatic macrophage functions is the method of delivery to macrophages in the pre-metastatic tissues. Directly targeting the intrinsic immune-altering functions of macrophages has been investigated for several inflammatory diseases and should be further examined for cancer metastasis [68–70]. As “big eaters,” macrophages have evolved to rapidly phagocytose and process vesicular transport molecules and foreign bodies. This intrinsic behavior corresponds to observations of macrophages phagocytosing a larger percentage of tumor-derived exosomes [32], cytoplasts [12], and man-made drug delivery vehicles [69] than other cell types. By understanding the surface properties of exosomes and cytoplasts that promote organotropic deposition, more precise targeting of macrophages in tumor-specific pre-metastatic niches may be achieved. Loading these targeted delivery vehicles with agents designed to prevent formation of premetastatic lesions, for example, by blocking macrophage activation of fibroblast-secreted fibronectin, inhibiting the CCR2-CCL2 pathway, or preventing macrophage IL-6 secretion, may prevent the influx and activation of MAMs that initiate the cycle of secondary tumor development. Additionally, better targeted drug delivery to potential metastatic organs may also reduce off-target effects of these drugs, since the pathways exploited for cancer progression are necessary for maintenance of homeostasis.