Interplay between fat cells and immune cells in bone: impact on malignant progression and therapeutic response

Abstract

Bone is a frequent site of metastasis from several organs and a host for cancer cells that have originated from the bone marrow itself. Skeletal malignancies are extremely difficult to treat. This is largely due to the complex, heterogenous nature of the bone marrow microenvironment and the dynamic interplay between the tumor cells and multiple cell types within the marrow niche. One cell type whose dominant role in supporting tumor progression and therapy evasion is being increasingly recognized are bone marrow adipocytes (BMAs). BMAs are metabolically active endocrine cells that supply lipids, hormones and adipocytokines to the neighboring cells. Their numbers in bone marrow increase with age, obesity and in response to certain treatments and metabolic conditions. BMAs have been shown to directly promote tumor growth through a transfer of lipids, upregulation of lipid transporters, modulation of tumor metabolism and activation of adaptive stress mechanisms in the cancer cells to promote survival. Less is known, however, regarding how adipocyte interactions with other cell types in the bone tumor microenvironment support malignant progression. This review summarizes recent findings on the crosstalk between BMAs and immune cells in bone. We focus specifically on how adipocyte-mediated changes in the immune milieu impact the tumor cell survival and therapy response. We also discuss how adipocyte-immune cell interactions alter bone homeostasis to support malignant progression in a context of current therapeutic strategies for skeletal malignancies.

1. INTRODUCTION

Bone is a complex and dynamic tissue comprised of a variety of resident cell types working together to maintain homeostasis (Fornetti, Welm, & Stewart, 2018). Bone is also a site of metastasis from a number of solid tumors (Hernandez, et al., 2018; NCI, 2022) and a host of several hematologic malignancies (Kane, 2020; Lane, Scadden, & Gilliland, 2009). Cancers residing in skeletal sites disrupt normal bone remodeling, cause significant morbidities such as pain, fractures, and hypercalcemia, and are notoriously difficult to treat. Tremendous research efforts have been dedicated into understanding the molecular mechanisms driving progression and therapy evasion in bone. Earlier studies into the mechanisms of tumor-induced bone disease have established that tumor cells colonizing the bone hi-jack normal functions of osteoclasts and osteoblasts to support a “vicious cycle” of metastasis (Mundy, 1997, 2002). The resulting development of bone-targeted agents, such as bisphosphonates, antibodies against receptor activator of nuclear factor kappa-Β ligand (RANKL), or radium 233, showed promise in reducing the incidence of skeletal-related events and delaying the progression of bone pain, but proved to be mainly palliative (Body, et al., 2019; R. Coleman, 2016; R. E. Coleman, 2001; Parker, et al., 2013), underscoring the need for additional treatment options. Over the last 15 years, a growing body of work has emerged demonstrating that, in addition to osteoclasts and osteoblasts, tumor cells cooperate with other cell types in the bone marrow niche in a concerted effort to survive under harsh conditions and evade therapy (Atkinson & Delgado-Calle, 2019; J. D. Diedrich, Herroon, Rajagurubandara, & Podgorski, 2018; Fornetti, et al., 2018; Hardaway, Herroon, Rajagurubandara, & Podgorski, 2014; Lo & Lynch, 2018; Stucker, Chen, Watt, & Kusumbe, 2020; Xiang & Gilkes, 2019). Among these cell types, bone marrow adipocytes (BMAs) and immune cells have gained the most attention as mediators of tumor growth and aggressiveness (J. D. Diedrich, et al., 2018; Falank, Fairfield, & Reagan, 2016; Hardaway, et al., 2014; Lo & Lynch, 2018; E. Morris & Edwards, 2016; Sinder, Pettit, & McCauley, 2015; Xiang & Gilkes, 2019). Tumor cells residing in skeletal sites have been shown to directly engage in a cross-talk with abundantly present BMAs, which allows for metabolic adaptation to the metastatic niche (J. D. Diedrich, et al., 2016; Falank, et al., 2016; Herroon, et al., 2019; Herroon, Rajagurubandara, Diedrich, Heath, & Podgorski, 2018; Herroon, et al., 2013; E. V. Morris, et al., 2020; Shafat, et al., 2017). It is also well-understood that tumor cell interactions with immune cells in bone can result in an immunosuppressive microenvironment and a reduced response to treatment (Janssen, Ramsay, Logsdon, & Overwijk, 2017; Kahkonen, Halleen, & Bernoulli, 2021; Xiang & Gilkes, 2019). Less is known, however, how marrow adiposity and direct BMA-immune cell crosstalk affect the immune landscape in bone and what are the impacts of this crosstalk on the tumor and its response to therapy. In this review we will discuss the current knowledge on tumor-supportive roles of immune cells in the bone tumor microenvironment. We will describe how immune cell phenotypes and metabolic functions are modulated by bone marrow adipose tissue and its crosstalk with bone-residing tumor cells. Finally, we will discuss options of targeting immunosuppressive immune cells and modulating immunometabolism in adipocyte-rich bone tumor microenvironment as potential treatment options for bone-trophic tumors.

2. I SUPPORT YOU - YOU SUPPORT ME: TUMOR-IMMUNE CELL CROSSTALK IN THE BONE MICROENVIRONMENT

Bone marrow is a unique immune regulatory organ, comprised of multiple immune cell subsets with specific functionalities. Although widely accepted to serve as a key defense mechanism against cancer cells, the immune system can also play tumor-supportive functions, both in the primary site and especially in the metastatic niche (Janssen, et al., 2017; Xiang & Gilkes, 2019). Here, we will discuss five specific immune cell types: macrophages, myeloid-derived suppressor cells, neutrophils, T cells, and B cells in the context of their tumor-supportive functions within the bone microenvironment.

A. Bone marrow macrophages

Macrophages are myeloid-derived phagocytic cells with diverse functions in physiological and pathological processes involving lipid, glucose, amino acid, and iron metabolism (Biswas & Mantovani, 2012; Kahkonen, et al., 2021; Lo & Lynch, 2018). Several populations of macrophages exist in the bone tumor microenvironment, each with potential to impact the colonization and growth of skeletal tumors. These include: bone-residing macrophages, termed “osteomacs” or osteal macrophages (Soki, et al., 2015); bone-resorbing osteoclasts, classically viewed as resident macrophages (Gu, Yang, & Shi, 2017; Sinder, et al., 2015); and inflammatory macrophages derived from monocytes that can polarize into proinflammatory M1 and anti-inflammatory M2 phenotypes (Gu, et al., 2017; Lo & Lynch, 2018). Osteoclasts and osteomacs are typically present at the tumor-bone interface (Ihle, et al., 2019; A. C. Wu, et al., 2016), and have been extensively reviewed elsewhere (Maurizi & Rucci, 2018; Pettit, Chang, Hume, & Raggatt, 2008; Soki, et al., 2015; Teitelbaum, 2000).

In contrast to resident macrophage populations, recruited macrophages are found to infiltrate the tumor and take on the tumor-associated macrophage (TAM) phenotype (Sinder, et al., 2015; X. Tang, 2013; A. C. Wu, et al., 2016). Several tumor-secreted factors contribute to macrophage recruitment and polarization in the bone tumor microenvironment. Tumor-derived Colony Stimulating Factor 1 (CSF-1), whose levels increase with standard of care therapies such as radiation and Androgen Deprivation Therapy (ADT), drives CSF-1R-positive macrophages into an CD206, arginase 1 (ARG-1)- and IL-10- positive anti-inflammatory phenotype, and promotes their survival (Escamilla, et al., 2015; J. Xu, et al., 2013). Anti-CSF-1R therapies have shown success in preclinical models of metastatic prostate and breast cancers (DeNardo, et al., 2011; Escamilla, et al., 2015; J. Xu, et al., 2013) and are being investigated in several clinical trials (Lo & Lynch, 2018; Ordentlich, 2021). Chemokine (C–C motif) ligand 2 (CCL2) secreted by prostate tumors has been shown to promote progression in bone through recruitment of endothelial cells, osteoclasts, and osteoblasts (X. Li, et al., 2009; Mizutani, et al., 2009). CCL2 interaction with its receptor, CCR2, favors a switch to M2 (TAM) phenotype, and reduces the production of pro-inflammatory cytokines (Sierra-Filardi, et al., 2014). Preclinical studies have shown that CCL2 is induced in multiple myeloma cells, promotes M2 recruitment and polarization, and consequently protects myeloma cells from chemotherapy-induced apoptosis (R. Xu, et al., 2019).

B. Myeloid-derived suppressor cells

Myeloid-derived suppressor cells (MDSCs) are a heterogenous population of immature myeloid cells (Gabrilovich & Nagaraj, 2009). Two established subpopulations of these cells include mononuclear MDSCs (M-MDSCs) and MDSCs with polymorphonuclear morphology (PMN-MDSCs). PMN-MDSCs share markers and morphology with neutrophils, but are phenotypically and transcriptionally distinct from them (Ostrand-Rosenberg, 2021). MDSCs accumulate in primary and metastatic tumors, secrete ARG-1, upregulate nitric oxide synthase and TGF-β, and play immunosuppressive functions by inhibiting T cells and activating regulatory T cells (Tregs) (Gabrilovich & Nagaraj, 2009; B. Huang, et al., 2006). MDSCs are considered a prognostic indicator in bone-trophic cancers, such as tumors of the breast, lung and multiple myeloma (D’Amico & Roato, 2015). In bone, MDSCs have been shown to differentiate into osteoclasts and promote bone destruction in preclinical models of multiple myeloma and metastatic breast cancer (Danilin, et al., 2012; Sawant, et al., 2013; Zhuang, et al., 2012). PMN-MDSCs have been recently shown to infiltrate stromal areas of bone metastatic lesions in prostate cancer patients and contribute to tumorprogression and immune evasion (Wen, et al., 2020), as well as promote resistance to therapy in multiple myeloma (Romano, et al., 2018).

C. Neutrophils

Neutrophils, like other myeloid cells, are a plastic, heterogenous population highly influenced by other cell types in the bone tumor microenvironment (M. Wu, Ma, Tan, Zheng, & Liu, 2020). Their retention and homing to bone marrow is regulated by the CXCR4/CXCL12 axis (Martin, et al., 2003). The role of neutrophils in cancer, and specifically in bone metastatic disease, is context-dependent and not well-understood. Neutrophils can polarize into a cytotoxic, short-living anti-tumor (N1) phenotype or a long-living, immature, pro-tumor (N2) phenotype, which exhibits low cytotoxicity; the latter is predominantly regulated by TGF-β (Fridlender, et al., 2009; Masucci, Minopoli, & Carriero, 2019), and potentially by IL-35 through induction of G-CSF and IL-6 (Zou, et al., 2017). Growing evidence suggests that during metastatic progression, neutrophils can acquire different, even opposing, phenotypes in response to their microenvironment (Sagiv, et al., 2015; M. Wu, et al., 2020), some of which might be mediated by neutrophil -derived proteolytic enzymes, such as Cathepsin G and Protease 3 (PR3) (M. Wu, et al., 2020). The interaction of neutrophil Cathepsin G with tumor cell RAGE was shown to be important for cytotoxicity against the tumor cells in vitro (Sionov, et al., 2019), while PR3 effects on tumor-derived RAGE promote metastasis of prostate cancer to bone (Kolonin, et al., 2017). In a preclinical model of prostate cancer, Ly6G⁺ neutrophils recruited to early bone lesions - appear to have tumor-killing properties that transition to tumor-supportive roles in the late stages of the disease (Costanzo-Garvey, et al., 2020). It is currently not clear whether this corresponds to Cathepsin G and PR3 activities. Similar to MDSCs, pro-tumor, immunosuppressive effects of neutrophils appear to be largely due to ARG1 production and effects on T-cells (Sippel, et al., 2011; M. Wu, et al., 2020). Tumor-supplied G-CSF was shown to modulate myeloid differentiation in bone marrow and produce T-cell suppressive neutrophils to support breast cancer progression in PyMT mouse model (Casbon, et al., 2015). High density neutrophils were also shown to play immunosuppressive functions in multiple myeloma by activating STAT3 signaling (Romano, et al., 2020).

D. T cells

T cells represent a significant portion of mononuclear cells in bone marrow and play critical roles in tumor immunity within the metastatic niche (Xiang & Gilkes, 2019). Two major types of T-cells in bone marrow include cytotoxic CD8+ T cells, which are critical mediators of tumor destruction, and CD4+ T cells, involved in immunosuppression (Xiang & Gilkes, 2019). Effector CD8+ and CD4+ T cells differentiate into memory T cells under control of IL-7 (Becker, Coley, Wherry, & Ahmed, 2005; Tokoyoda, et al., 2009), and both circulating IL-7 levels and memory T cells have been shown to be increased in cancer patients as compared to healthy subjects, and are especially abundant in patients with bone metastases (Al-Rawi, Rmali, Watkins, Mansel, & Jiang, 2004; M. Feuerer, et al., 2001). CD4+ T cells support tumor progression in bone through secretion of IL-17, a known stimulator of osteoclastogenesis (Sato, et al., 2006). T helper cells (Th17 cells) and T regulatory cells (Tregs), both subtypes of CD4+ T cells, contribute to tumor-driven bone destruction and metastatic progression through activation of RANKL signaling (K. W. Kim, Kim, Kim, Cho, & Lee, 2015; Kotake, et al., 1999; W. Tan, et al., 2011; Van Bezooijen, Papapoulos, & Löwik, 2001). Tregs produce TGF-β, which is a major factor in the immunosuppressive response (Marie, Liggitt, & Rudensky, 2006), and are found at an increased number in the bone marrow of prostate cancer patients with bone metastases (Zhao, et al., 2012). A study comparing T cells in peripheral blood versus bone marrow of multiple myeloma patients revealed that T cell subsets in bone marrow at the tumor site are strikingly altered, with a significant accumulation of effector T cells and a decrease of effector memory T cells within the CD+ T cell population (Zelle-Rieser, et al., 2016). Although these phenotypes generally suggest an anti-cancer immune response, the fact that these cells also show an augmented expression of checkpoint molecules such as PD-1, CTLA-4, CD160 and 2B4, reduced cytokine levels, and a decreased capacity to proliferate and degranulate, suggests ineffective immune surveillance (Zelle-Rieser, et al., 2016). More studies are needed on molecular mechanisms contributing to T cell defects in the bone tumor microenvironment, as they are likely contributors to immune escape in bone-resident cancers.

E. B cells

B lymphocytes are classically considered as positive regulators of immune responses (LeBien & Tedder, 2008), but act as a double-edged sword in cancer (Fridman, et al., 2021). High intratumoral B-cell density is associated with good prognosis in a number of cancers due to the well-recognized roles for B-cells in mediating T-cell responses and antibody-mediated cellular cytotoxicity (ADCC) (Fridman, et al., 2021; He, et al., 2014). A recent systematic study of bone metastasis tissues from prostate cancer patients revealed a dramatic decrease in B cells and B cell progenitors, which is consistent with poor responses of prostate tumors to immune checkpoint blockade (Kfoury, et al., 2021). On the other hand, tumor-infiltrating B cells have been shown to support prostate cancer progression through the secretion of lymphotoxin and activation of NF-κB and STAT3 signaling (Ammirante, Luo, Grivennikov, Nedospasov, & Karin, 2010; Woo, et al., 2014). B-cells have been found to promote lymphangiogenesis in lymphoma and melanoma mouse models (Ruddell, Harrell, Furuya, Kirschbaum, & Iritani, 2011), and accelerate bladder cancer metastasis through secretion and upregulation of IL-8 (Ou, et al., 2015). A subset of IL-10-producing B cells, called regulatory B cells (Bregs), can suppress the anti-tumor immune response by inhibiting effector cells (Lund, 2008). Bregs were shown to inhibit NK-mediated ADCC against multiple myeloma cells, contributing to therapy evasion (L. Zhang, et al., 2017). Notably, PD-L1⁺ B cells were recently reported to be upregulated in advanced melanoma patients, further enriched in bone metastases compared to primary tumors, and shown to suppress T cell responses in a PD-L1-dependent manner (H. Wu, et al., 2020).

3. THE POWERS OF FAT: ADIPOSITY-DRIVEN CHANGES IN THE BONE MARROW IMMUNE MILIEU

A. Immunomodulatory function of adipose tissue: lessons from WAT

It is well-recognized that the immune system responds to specific metabolic cues and that any changes in the metabolic state correlate with altered immune function (Ferrante, 2013). Adipose tissue has the potential to expand and contract, and the impact of this dynamic phenomenon on the immune cell population has been best studied in white adipose tissue (WAT) (Lumeng, DelProposto, Westcott, & Saltiel, 2008; Ostrand-Rosenberg, 2021; Weisberg, et al., 2003), more specifically in subcutaneous and visceral (mesenteric, epididymal and perirenal) fat depots (J. Diedrich, Gusky, & Podgorski, 2015). In the healthy state, the adipose tissue homeostasis is maintained by Tregs, Th2 cells and natural killer (NKT) cells, which interact with other regulatory cells, such as alternatively activated anti-inflammatory M2 macrophages and eosinophils (Q. Wang & Wu, 2018). With WAT expansion, the immune homeostasis is disrupted by accumulating CD4+ and CD8+ T cells (D. L. Morris, et al., 2013; Rausch, Weisberg, Vardhana, & Tortoriello, 2008), specifically Th1, Th17, and cytotoxic lymphocytes (CTLs), as well as classically activated pro-inflammatory M1 macrophages (Lumeng, et al., 2008; Martinez, Sica, Mantovani, & Locati, 2008). Chronic, low-grade inflammation associated with WAT expansion in humans is also a driver of MDSC accumulation, which contributes to metabolic dysfunction and drives the immunosuppressive phenotype of adiposity-associated cancers (Ostrand-Rosenberg, 2021).

T-cell accumulation in WAT is thought to precede infiltration of adipose tissue macrophages (ATMs), the most abundant immune cell population in WAT, comprising up to 50% of total cells (Ferrante, 2013; Weisberg, et al., 2003). Defined as cells that co-express F4/80 and CD11b in mice, or F4/80 and CD68 in humans, ATMs are phenotypically dynamic and responsive to cues from adipose tissue microenvironment (D. L. Morris, Singer, & Lumeng, 2011). WAT expansion is commonly associated with the phenotypic switch from M2 to M1 macrophages (Lumeng, et al., 2008; Martinez, et al., 2008), which, based on the M2 phenotype being associated with TAMs, would be expected to be anti-tumorigenic (Sica, et al., 2008). However, recent emerging data suggest that neither ATMs nor TAMs can be classified using a simple M1/M2 model, as this model doesn’t capture their diversity and plasticity in function (Ferrante, 2013; Pan, Yu, Wang, & Zhang, 2020).

Increased adiposity has been shown to reprogram ATMs into a pro-inflammatory, metabolically activated (MMe) phenotype that supports tumorigenesis (Tiwari, et al., 2019). Furthermore, when compared to monocyte-derived macrophages, human ATMs from obese subjects secrete angiogenic factors, chemokines, cytokines, proteases, and growth factors such as CCL2, CXCL1, IL-1α, IL-1β, IL-6, and FGF2, that are also typically secreted by TAMs (Mayi, et al., 2012). Similar to ATMs, MDSCs, whose primary functions in the tumor microenvironment is to suppress anti-tumor immune responses by inhibiting activated T-cell proliferation, appear to play similar roles in the adipose tissue. MDSC deletion in vivo leads to an accumulation of T cells and M1-type ATMs in WAT (Stivers, et al., 2020). One characteristic feature of expanding WAT is the presence of crown-like structures (CLS) of macrophages around dying adipocytes, typically linked to adipose tissue inflammation (Cinti, et al., 2005; Iyengar, et al., 2016), and recently associated with poor breast cancer prognosis (Chang, Eslami, Ennis, & Goodwin, 2021). CD8+ cells participate in CLS formation along with ATMs, and depletion of CD8+ T-cells in obese mice results in a reduction in the number of CLS and M1 macrophages, while the opposite is observed with adoptive CD8+ T-cell transfer (Nishimura, et al., 2009). Formation of CLS is associated with adiposity-driven inflammation, altered adipokine levels and metabolic dysfunction and speaks to the importance of the adipocyte-immune cell cross-talk in metabolic homeostasis (Weisberg, et al., 2003).

B. BMAT, immune cell regulation and inflammaging

Bone marrow adipose tissue (BMAT) accounts for up to 70% of bone marrow volume and over 10% of total adipose tissue in healthy adults (Suchacki, et al., 2020). Similarly to WAT, BMAT expansion can be driven by aging, obesity, diabetes and hormonal dysfunction; however, in stark contrast to WAT, BMAT volume also paradoxically increases with calorie restriction and decreases with weight recovery (Bredella, et al., 2009; Cawthorn, et al., 2014; Devlin, 2011; Fazeli, et al., 2013; Scheller & Rosen, 2014), Furthermore, certain therapeutic interventions, such as thiazolidinediones, radiation and chemotherapy can also augment marrow adipocyte content (Devlin & Rosen, 2015; Naveiras, et al., 2009; Pop, et al., 2017; Rzonca, Suva, Gaddy, Montague, & Lecka-Czernik, 2004; Veldhuis-Vlug & Rosen, 2018). BMAT is functionally and transcriptionally distinct from WAT (Scheller & Rosen, 2014; Suchacki, et al., 2020), and in both rodents and humans, exists in two forms: constitutive (cBMAT), located in the distal skeletal sites, and regulated (rBMAT), enriched in proximal, central and endosteal skeletal sites (Scheller, Cawthorn, Burr, Horowitz, & MacDougald, 2016; Scheller, et al., 2015; Scheller & Rosen, 2014). Although both cBMAT and rBMAT subtypes morphologically resemble white adipocytes, they differ in lipid profiles, adipokine production, and response to fasting or lipolytic stimuli (Bredella, et al., 2021; Scheller, et al., 2016; Scheller, et al., 2015; Suchacki, et al., 2020). The two depots also differ in lipid saturation levels: human and rat cBMAT adipocytes have an increased proportion of unsaturated fatty acids, and are enriched in palmitoelate, a lipokine implicated in the regulation of insulin signaling, whereas rBMAT adipocytes are enriched in palmitate (Scheller, et al., 2015), a TLR4 agonist that promotes macrophage activation and lipid metabolism-associated inflammation (Tzeng, Chyuan, & Chen, 2019). Notably, lipid saturation levels in rBMAT actually resemble levels found in WAT (Scheller, et al., 2015), suggesting potential similarities between WAT and rBMAT in regulating immune cell function, which needs further exploration.

BMAs derive from bone marrow stromal cells (BMSCs), which are a significant source of immunoregulatory molecules, including PGE2, HGF, TGF-β, and IDO, that can profoundly affect immune cell function and phenotype (reviewed in (Benova & Tencerova, 2020)). The composition of immune cells in bone marrow has been shown to be modulated by obesity, dietary interventions, and exercise, conditions that differ in their impact on BMAT volume (Benova & Tencerova, 2020). Diet-induced obesity and a corresponding BMAT expansion in mice, were reported to increase overall bone marrow cellularity by 20–25%, but despite total increases in all cell types, appeared to have no observed impact on the proportion of specific cell lineages, with an exception of lymphocyte populations increasing by 10–18% (Trottier, Naaz, Li, & Fraker, 2012). A number of other studies have linked obesity with a decrease in lymphocytes, but an increase in monocyte and neutrophil populations, and reported a reversal of these trends with exercise and dietary restrictions (reviewed in (Benova & Tencerova, 2020)). The obesity and a 62% increase in marrow adiposity due to long-term (7 months) exposure of mice to high fat diet, was shown to reduce numbers of B lymphocytes in bone marrow by 52% (Chan, Adler, Green, & Rubin, 2012). Intriguingly, murine adipocytes accumulating in bone marrow under dietary restriction were shown to promote T cell accumulation and enhance protection against infections and tumors (Collins, et al., 2019). How the specific metabolic phenotype of adipocytes and their location in specific areas of bone marrow alters the immune population, especially in humans, is not well understood and needs further studies.

BMAT expansion with age or obesity is a consequence of a shift in BMSC differentiation in favor of adipocyte, which alters the production of secretory factors, and thus also the function of neighboring cells (Benova & Tencerova, 2020). It is well established that BMAT modulates hematopoietic activity and supports differentiation of myeloid and lymphoid cells from hematopoietic stem cells, and these modulatory functions can be exacerbated during aging (Aaron, Costa, Rosen, & Qiang, 2022; Hardaway, et al., 2014). BMAT expansion with age has been shown to induce low-grade chronic inflammation, termed ‘inflammaging’, which, through secreted lipid species and pro-inflammatory cytokines, disturbs normal bone homeostasis (Aaron, et al., 2022). Inflammaging in the bone marrow is thought to decrease immune cell survival factors and be a major contributor to the decline in the efficiency of the immune system (Pangrazzi, et al., 2017). The overexpression of cytokines such as IL-15 in bone marrow during inflammaging (Pangrazzi, et al., 2017) is known to contribute to the generation and maintenance of CD8+CD28- T cells (Feng, et al., 2018), which is a hallmark of immunosenescence, an age-related alteration to the immune system which reduces the ability to respond to new antigens and leads to the accumulation of memory T cells (Aiello, et al., 2019).

Aging BMAT is a source of pro-inflammatory cytokines and adipokines that inhibit B lymphopoiesis and skew hematopoietic cell populations into the myeloid lineage (Bilwani & Knight, 2012). Expanding BMAT induces MDSCs, which regulate B cell responses and inhibit B cell lymphogenesis through the secretion of IL-1 (Kennedy & Knight, 2015). Augmented levels of pro-inflammatory factors in aging BMAT also promote ROS production, which consequently limits the capacity of bone marrow stromal cells to support hematopoietic niche and contributes to cellular senescence (Khatri, et al., 2016; Pangrazzi, et al., 2017). Since macrophages are intimately involved in initiating and resolving inflammation, they’re also heavily impacted by aging. In mice, aging bone marrow has been associated with the M1 macrophage phenotype, exhibiting reduced phagocytic functions and impaired cytokine production (O. H. Kim, et al., 2017). On the other hand, studies from our laboratory showed that both age and obesity increase invasiveness of murine bone marrow macrophages, a process linked to increased expression of COX2 and CCL2 (Hardaway, et al., 2014). Clearly, BMAT is an emerging critical player in aging with clear implications for immune functions in bone marrow.

4. ADIPO-KINES OF CHANGE: IMMUNOMODULATORY FUNCTIONS OF ADIPOKINES

Adipocytes can assume roles of highly functional immune cells by releasing anti- and proinflammatory cytokines and adipokines (Blaszczak, Jalilvand, & Hsueh, 2021). Here, we highlight 5 immunomodulatory adipokines produced by BMAs that play documented roles in immunosuppression and are the likely players in the adipocyte-immune-cell-tumor cell cross-talk that drives malignant progression in bone.

A. Leptin

Leptin is a well-characterized hormone secreted from adipocytes with key roles in the array of physiological functions that include pro-inflammatory responses, angiogenesis, bone metabolism and lipolysis (Abella, et al., 2017). As a critical sensor of energy metabolism, leptin is becoming the cornerstone of an emerging field of immunometabolism. Leptin expression in adipose tissue contributes to the pro-inflammatory phenotype seen in obese individuals (Obradovic, et al., 2021). Accordingly, leptin-deficient mice have an elevated production of anti-inflammatory cytokines and lose their proinflammatory phenotype (Du, et al., 2020). Since the leptin receptor (LEPR or Ob-R) is expressed across the innate and adaptive immune system, this adipokine is directly involved in mediating immune cell activity (Francisco, et al., 2018).

The effects of leptin on macrophages have been well-documented, but are not entirely understood (Francisco, et al., 2018). Leptin has been shown to promote monocyte proliferation, induce phagocytic activity and secretion of pro-inflammatory factors by macrophages, and serve as a trophic factor to prevent apoptosis. Treatment of human macrophages in culture with recombinant leptin polarizes them to M2-phenotype, based on surface marker expression, but leads to secretion of M1-typical cytokines (Acedo, Gambero, Cunha, Lorand-Metze, & Gambero, 2013). Intriguingly, leptin deficiency in mice was shown to shift macrophage polarization from M1 to M2 by inducing anti-inflammatory effects of mast cells (Zhou, et al., 2015) (Figure 1). On the contrary, a recent study by Dudzinski et al. demonstrated that acute leptin treatments in vivo repolarize TAMs to M1 phenotypes and cooperate with PD-1 blockade immunotherapy (Dudzinski, et al., 2021).

Figure 1. Immunomodulatory roles of bone marrow adipocytes.

The bone marrow milieu contains immune cells, which are alternatively activated during adipocyte-immune cell crosstalk. These cells include macrophages (M2-like), Tregs (regulatory T cell), myeloid-derived suppressor cells (MDSCs), neutrophils, and Bregs (regulatory B cell). M2-like macrophages are maintained in the adipose tissue by adipokines (leptin resistin, visfatin, and adiponectin) and adipocyte-supplied CCL2 and COX2. TREM2 receptor activity in lipid-associated and ARG-1 positive macrophages correlates with increased expression of lipid transport and immunosuppression genes. TREM2 also plays a role in osteoclast maturation. Lipid uptake mediated by lipid transporters and active PPAR signaling leads to a metabolic switch in macrophages, B cells, and T cells to FAO, which favors the M2 phenotype and increases the frequency of Bregs and Tregs in the bone microenvironment. Decreased leptin and chemerin can induce Treg proliferation, while resistin and adiponectin promote Treg expansion through IRF-1. Leptin and decreased PPARy signaling promote Th1 polarization of T cells. Like macrophages, B cells, and T cells, MDSCs take up lipids via fatty acid transporters, causing a metabolic switch to FAO to support their immunosuppressive functions, promoting Treg expansion and M2 activation. Neutrophil activation in the bone marrow occurs by adipocyte-supplied G-CSF and IL-6 and MDSC-supplied TGF-B, which can promote polarization to a pro-tumor N2-like phenotype. Adenosine receptor signaling induces neutrophils to produce IL-10, which leads to suppression of T cell response, and dually adenosine can increase TAM accumulation. Resistin contributes to the retention of neutrophils in the bone marrow and reduces their escape. In contrast to M2-like macrophages, Bregs are increased by reduced levels of leptin, limiting B cell lymphogenesis and response. Overall, adipocytes interact with immune cells to facilitate an immunosuppressive environment. FAO = Fatty acid Oxidation

In addition to modulating macrophage function and phenotype, leptin plays key roles in adaptive immunity by regulating T and B cell populations. Leptin dose-dependently promotes proliferation of naïve CD4+ T-cell proliferation and drives CD4+ T-cell polarization into Th1 phenotype (Figure 1). Leptin deficiency induces proliferation of Tregs in adipose tissue (Markus Feuerer, et al., 2009) (Figure 1). Accordingly, the reduced number of Tregs in obese individuals correlates with plasma levels of leptin (Wagner, et al., 2013). Dendritic cells (DCs) generated under leptin-free conditions induce Treg or Th17 cells more efficiently than DCs generated in the presence of leptin (Moraes-Vieira, et al., 2014). In addition to its effects on T-cells, leptin maintains B-cell homeostasis in mice by inducing Bcl-2 and cyclin D1, thus promoting cell cycle entry and protection from apoptosis (Lam, Wang, Ko, Kincade, & Lu, 2010). Leptin is also critical for B-cell development as demonstrated by reduced numbers of pro-B, pre-B, and immature B cells, and augmented numbers of mature B cells with low leptin levels under fasting conditions in mice, a phenomenon reversed by central leptin administration (M. Tanaka, et al., 2011). B cell responses are reduced by increased leptin levels associated with obesity (Frasca, et al., 2016), which has potential implications for B cell population in BMAT. Leptin levels positively correlate with obesity-associated cancers in humans, including tumors of the breast and multiple myeloma (R. Liu, Gao, Lv, Zhai, & He, 2022), and are considered a risk factor for advanced prostate cancer (Hardaway, et al., 2014).

B. Adiponectin

Adiponectin is an adipocyte-derived hormone, whose levels, in contrast to other adipokines, are reduced in obesity, and inversely correlate with tumor progression and metastatic disease (Fowler, et al., 2011; Philp, et al., 2020; Pinthus, et al., 2008; Pu & Chen, 2021). Adiponectin is actively involved in modulating the immune system through its receptors AdipoR1 and AdipoR2, found on many immune cell types (H. M. Choi, Doss, & Kim, 2020; Polito, et al., 2018). Adiponectin binding to receptors present on macrophages affects their proliferation and polarization. A study by Stinj et al. demonstrated that adiponectin not only polarizes murine macrophages to more M2-like phenotype, but M2 macrophages stimulated by adiponectin increase the expression of anti-inflammatory factors such as IL-10 (Stijn, Kim, Lusis, Barish, & Tangirala, 2015) (Figure 1). Adiponectin suppresses the expression of pro-inflammatory cytokines, such as IL-1β, IL-6 and TNF-α by macrophages, and induces the levels of the mannose receptors, ARG-1 and Heme Oxygenase 1 (HO-1) (Mandal, Pratt, Barnes, McMullen, & Nagy, 2011; Monks, Irakleidis, & Tan, 2019). This is consistent with the anti-inflammatory properties of this adipokine and its beneficial effects on metabolism (Luo & Liu, 2016); however, how these effects impact metastatic tumor progression in bone is not presently clear.

Adiponectin is considered a negative regulator of T-cell function due to its reported inhibitory effects on T-cell proliferation, reduction in pro-inflammatory cytokine production and induction of apoptosis (Monks, et al., 2019; Wilk, et al., 2011). The specific inhibitory effects of adiponectin on the CD4+ T-cell-mediated production of IFN-γ and IL-17 in mice were demonstrated to be due to restraining T cell glycolysis (Surendar, et al., 2019). Adiponectin receptors are expressed at higher levels on dendritic cells in metastatic breast cancer patients. They have been shown to interfere with T cell clearance of cancer cells in a T-cell-mediated tumor protection model (P. H. Tan, et al., 2014). Interestingly, an in vitro study by Tsang et al. demonstrated that adiponectin treatment to bone marrow-derived dendritic cells decreased the expression of costimulatory molecules and major histocompatibility complex II (MHC II), but enhanced the expression of PDL-1, leading to suppressed T cell response and an augmented expansion of Tregs (Tsang, et al., 2011) (Figure 1).

One paradoxical characteristic of adiponectin is that in contrast to obesity-mediated reduction in circulating levels of this adipokine, caloric restriction, which, just like obesity, is associated with BMAT expansion, leads to increased plasma levels of adiponectin (Cawthorn, et al., 2014), although consequences of this phenomenon remain unknown. It is also unclear what local concentrations of adiponectin are in the bone microenvironment, and whether adiponectin levels differ between constitutive and regulated BMAT. Recent studies by Edwards laboratory have shown that adiponectin levels are downregulated in BMAs by multiple myeloma cells through the secretion of TNF-α (E. V. Morris, et al., 2020). Suppressing the expression of this anti-tumor anti-inflammatory adipokine by the myeloma cells offers a potential mechanism for regulating the bone marrow niche to support tumor progression.

C. Resistin

Resistin is a small secretory molecule best known for mediating insulin resistance in obesity-induced diabetes (Deb, Deshmukh, Ramteke, Bhati, & Bhat, 2021) . The majority of our current knowledge about resistin is based on studies in rodents, however, and the rodent versus human resistin gene and protein sequences share only about 60% identity (Ghosh, Singh, Aruna, Mukhopadhyay, & Ehtesham, 2003), and they vary significantly in function and expression pattern (Tripathi, Kant, Pandey, & Ehtesham, 2020). In rodents, resistin is an adipokine highly expressed and secreted by adipocytes, whereas human resistin is primarily produced by monocytes and macrophages, and its increased circulating levels observed in obesity are thought to be due to an increased infiltration of immune cells into expanding adipose tissue (Deb, et al., 2021; Tripathi, et al., 2020). Interestingly, recent studies by Miggitsch et al. reported that resistin is enriched in purified human femoral adipocytes as compared to subcutaneous WAT (Miggitsch, et al., 2019), suggesting that BMAT might be contributing to resistin expression in bone marrow in addition to macrophages and hematopoietic precursor cells (Lin, et al., 2020).

Human resistin functions as a pro-inflammatory cytokine and an inducer of TNF-α, IL-6, IL-12, and MCP-1 (Tripathi, et al., 2020), all factors with a strong impact on osteoclast function and bone remodeling (Harmer, Falank, & Reagan, 2018; X. Li, et al., 2009; Mizutani, et al., 2009; Osta, Benedetti, & Miossec, 2014). Resistin decreases TRAF-3, an inhibitor of CD40 ligand signaling (Verma, et al., 2003), and a tumor suppressor known to be deleted or highly mutated in many cancers (Zhu, et al., 2018). Resistin has also been implicated in immunosuppression by impairing neutrophil migration and inhibiting the activation of PI3K and Akt (Y. Li, et al., 2021) (Figure 1). Resistin impairs the ability of human monocyte-derived DCs to uptake antigens and produce cytokines, which negatively regulates T cell proliferation and differentiation into Th1 cells (Y. M. Son, et al., 2008) and induces the expansion of Tregs through IRF-1 (Young Min Son, et al., 2010) (Figure 1).

D. Visfatin

Visfatin is an inflammatory adipocytokine first identified in visceral adipose tissue and best known for its compensatory response in diet- or obesity-induced insulin resistance (Moschen, et al., 2007). Visfatin, also called pre-B-cell colony-enhancing factor (PBEF), is a rate-limiting enzyme in NAD ⁺ biosynthesis from nicotinamide, shown to regulate B cell development, apoptosis, and glucose metabolism (Rongvaux, et al., 2008), and to induce pro-inflammatory factors, such as TNF-α, IL-6, IL-8, and NF-kB by activating the insulin receptor-(IR)-Ras-MAPK signaling pathways (Y. Sun, et al., 2012). Similar to resistin, visfatin is produced and released by adipose tissue-derived macrophages rather than adipocytes (Curat, et al., 2006). Visfatin appears to be important for macrophage polarization to the tumor-promoting M2 phenotype in chronic lymphocytic leukemia (CLL) (Figure 1), where it was shown to enhance immunosuppressive phenotype of CD68+ cells and promote their ability to sustain CLL survival (Audrito, et al., 2015).

In murine bone marrow, visfatin is reported to alter the balance between bone resorption and bone formation and to increase the pro-inflammatory phenotype of MSC cells differentiating to marrow adipocytes and osteoblasts (Tsiklauri, et al., 2018). Visfatin-driven increase in bone matrix mineralization combined with a decrease in collagen I expression and induction of pro-inflammatory factors is suggested to be a key factor in impaired bone remodeling and bone fragility (Tsiklauri, et al., 2018). This is of importance for cancers thriving in bone, where dysregulated bone remodeling often supports progression. In multiple myeloma, a cancer highly affected by marrow adiposity (Falank, et al., 2016; Jafari, Fairfield, Andersen, & Reagan, 2021), visfatin has been shown to be a key driver of osteoclast formation and activity, which is a critical component of myeloma progression in bone (Venkateshaiah, et al., 2013). This is in contrast to other reports demonstrating suppressive effects of visfatin on osteoclastogenesis (Baek, et al., 2017; Moschen, Geiger, Gerner, & Tilg, 2010). Bone-modulatory and immunomodulatory functions of visfatin are clearly not fully understood, but the fact that this adipocytokine induces pro-inflammatory transcription factors such as NF-kB, and regulates MAPKs, and PI3K and ROS pathways, indicate its role as a possible cancer driver that needs further investigation (Parimelazhagan, et al., 2021).

E. Chemerin

Chemerin, also known as TIG2 or RARRES2, is an adipokine and a ligand for chemokine-like receptor 1, ChemR23 (CMKLR1), present on dendritic cells, macrophages (Samson, et al., 1998), NK cells (Parolini, et al., 2007) and adipocytes (Goralski, et al., 2007). In addition to its signaling through CMKLR1, chemerin also binds to the atypical non-signaling CCRL2 receptor involved in leukocyte migration (Schioppa, et al., 2020). Chemerin regulates adipocyte metabolism and inflammatory responses (Fatima, Rehman, Baig, & Khan, 2014; Goralski, et al., 2007). Its expression in BMAT has been shown to promote adipogenesis at the expense of osteoblastogenesis (Muruganandan, Roman, & Sinal, 2010), and the negative effects of chemerin on bone architecture in mice have recently been attributed to the inhibition of Wnt/β-catenin signaling and reduced bone formation in combination with activation of RANK signaling and increased bone resorption (Han, et al., 2021), which has implications for cancers residing in bone.

Although reported as a marker of several cancers (Fatima, et al., 2014), the data correlating chemerin levels with tumor progression have been conflicting. Serum levels of chemerin are increased in multiple myeloma and correlate with disease stage, and interaction of adipocyte-supplied chemerin with CMKLR1 and CCRL2 on myeloma cells is thought to contribute to myeloma progression (Westhrin, et al., 2018). Intratumoral levels of chemerin are reported to be reduced in several cancers, including AML, breast, prostate and melanoma, while its circulating levels are increased (Shin, Zabel, & Pachynski, 2018). Tumors expressing chemerin show increased infiltration by NK and T cells, and an increased ratio of NK and T cells to MDSCs and plasmacytoid dendritic cells (pDC) (Shin, et al., 2018) (Figure 1). It has been proposed that tumors might be silencing chemerin to evade immune surveillance while circulating levels might be increased by the host to heighten the defensive mechanisms and mount an anti-tumor response (Shin, et al., 2018). The role of chemerin in modulating host immune responses in the bone tumor microenvironment is currently unclear and calls for investigations.

5. YOU’VE LOST THAT FAT LOVIN’ FEELING: LIPID MODULATION OF IMMUNE CELLS

Multiple cell types within the bone marrow space, including immune cells, are subject to metabolic effects of marrow fat cells. Fat cells store lipids in the form of triglycerides in lipid droplets and break them down into glycerol and free fatty acids in the process of lipolysis (J. D. Diedrich, et al., 2018; Granneman & Moore, 2008). Although marrow adipocytes are less responsive to lipolytic stimuli than WAT adipocytes (Scheller, et al., 2019), their lipid hydrolyzing activity has been shown to be modulated by the neighboring tumor cells, and adipocyte-supplied lipids are known to fuel tumor metabolism (J. D. Diedrich, et al., 2018; J. D. Diedrich, et al., 2016; Panaroni, et al., 2022; Shafat, et al., 2017). The fact that marrow adipocytes engage in lipid-mediated interactions with tumor cells suggests that they also likely participate in a cross-talk with other cell types in the bone tumor microenvironment, including immune cells. It is well recognized that the dysfunctional behavior of tumor-infiltrating immune cells can be driven by the uptake of extracellular lipids or de novo lipogenesis (Al-Khami, et al., 2017; Howie, Ten Bokum, Necula, Cobbold, & Waldmann, 2017; Manzo, et al., 2020; Veglia, et al., 2019). Here, we will focus on how exogenous lipids present in adipose-rich bone tumor microenvironment might be impacting the immune cell phenotype and function.

A. Fatty acid uptake mechanisms

Several membrane proteins have been implicated in the trafficking of lipids, including CD36, SR-A6, CD206, fatty acid-binding proteins (FABPs) and fatty acid transport proteins (FATPs) (Abumrad, Coburn, & Ibrahimi, 1999; Vogel, Brunner, Hajto, Sharif, & Schabbauer, 2022). Through examination of Oncomine cancer databases, we have previously shown that CD36, FABP4 and several FATPs are upregulated in metastatic tissues from prostate cancer patients as compared to primary prostate tumors (Chkourko Gusky, Diedrich, MacDougald, & Podgorski, 2016). We do know that metastatic prostate tumor cells upregulate CD36 and FAPB4 upon exposure to adipocytes (Herroon, et al., 2013), which is reflected in augmented mRNA levels of these transporters in metastatic lesions (Chkourko Gusky, et al., 2016). Substantial literature evidence, however, suggests that specific immune cell types present in the bone microenvironment also upregulate lipid transport proteins to carry out their metabolic functions, and their overexpression likely contributes to overall expression levels detected in metastatic tissues.

CD36, also known as fatty acid translocase (FAT), is a heavy glycosylated type B scavenger receptor, and a pattern recognition receptor with unique capability to transport long chain fatty acids (LCFA) (Chen, Zhang, Cui, & Silverstein, 2022). LCFA binding by CD36 modulates binding and uptake of oxidized LDL (oxLDL) and regulates cellular cholesterol levels (Jay, Chen, Paz, Hung, & Hamilton, 2015). CD36 is expressed on wide variety of immune cells, including macrophages, dendritic cells and subsets of T and B cells and functions both as a peroxisome proliferator-activator receptor γ (PPARγ)-regulated fatty acid transporter and a cell signaling receptor that controls immune cell fate (reviewed in (Chen, et al., 2022)). CD36 is the main receptor responsible for lipid uptake by TAMs (Su, et al., 2020) (Figure 1) and a facilitator of fatty acid transport and immunosuppressive functions of MDSCs (Al-Khami, et al., 2017) (Figure 1). CD36 expression is augmented upon IL-4-mediated alternative activation of macrophages and, conversely, macrophages that are CD36-deficient are also defective in alternative activation (S. C. Huang, et al., 2014). This occurs in tandem with IL-4-driven upregulation of lysosomal acid lipase (LAL) and increased lipolysis, further underscoring the importance of uptake and lipolysis of exogenous lipids in M2 activation. In a similar manner, lipolysis appears to be critical for the release of fatty acids required for oxidation in IL-15-driven memory T cells (O’Sullivan, et al., 2014). CD36-mediated uptake of oxidized lipids was recently reported to suppress the effector functions of CD8+ T cells and induce lipid peroxidation in a p38-dependent manner (S. Xu, et al., 2021). CD36 upregulation is suggested to be necessary for adaptation to increased lipid availability in the tumor microenvironment and for survival and activity of intratumoral Tregs (H. Wang, et al., 2020; S. Xu, et al., 2021). The extracellular form of CD36, known as “soluble CD36”, is thought to be released from multiple cell types, including immune cells and adipocytes, into extracellular vesicles as a novel mechanism of intracellular lipid transfer (Chen, et al., 2022).

SR-A6 (Marco) is a type A scavenger receptor present on a variety of myeloid cells including macrophages and dendritic cells (Vogel, et al., 2022). Marco has been shown to facilitate uptake and accumulation of lipids and its surface expression is highly augmented in lipid-laden adipose tissue macrophages as well as TAMs (Brunner, et al., 2020; Su, et al., 2020). The expression of Marco in CD68-positive TAMs in human PCa specimens was shown to correlate with lipid content, expression of genes associated with lipid metabolism and higher tumor grade (Masetti, et al., 2022). Administration of anti-Marco antibodies to mice bearing subcutaneous PCa tumors, reduced tumor growth and improved docetaxel response (Masetti, et al., 2022). Notably, Marco levels in TAMs were shown to be induced by PCa tumor-supplied IL1β (Masetti, et al., 2022), a cytokine known to be highly augmented by marrow adipocytes in metastatic PCa cells, and involved in increased production of PGE2 and CCL2 in bone tumor microenvironment and reduced sensitivity to docetaxel (Herroon, et al., 2019), This suggests Marco might be an important player in adipocyte-tumor cell-macrophage crosstalk, and its role in a context of bone-trophic tumors needs investigation.

Intracellular shuttling of fatty acids is mediated by FABPs (Furuhashi & Hotamisligil, 2008; Hotamisligil & Bernlohr, 2015). Upregulation of FABP4 and FABP5 in a subpopulation of resident CD8 memory cells is critical for their memory function (Pan, et al., 2017). Inhibition of FABP5 in Tregs induces mitochondrial changes, augments IL-10 production and promotes Treg suppressive activity in vitro (Field, et al., 2020). Loss of FABP4 in murine macrophages attenuates NF-κB signaling, reduces inflammatory cytokine secretion, and polarizes them to anti-inflammatory, alternatively activated M2 phenotype (Steen, Xu, & Bernlohr, 2017). FABP5 was recently reported among 31 abnormally expressed immune-related mRNAs that correlated with B-cell pathways, immunosuppressive microenvironment and unfavorable outcomes in multiple myeloma (Jia, et al., 2021), underscoring the importance of fatty acid transport proteins in immune function.

FATP family comprises 6 members of fatty acid transporters (Black, Ahowesso, Montefusco, Saini, & DiRusso, 2016; Kazantzis & Stahl, 2012). The predominant member in bone marrow derived macrophages is FATP1, an Acyl-CoA Synthetase with affinity for very long fatty acids (Johnson, et al., 2016). Total body knockout of FATP1 in mice protects against HFD-induced obesity, insulin resistance, and intramuscular lipid uptake (Q. Wu, et al., 2006). Deletion of FATP1 specifically in macrophages blunts fatty acid uptake and oxidation, and induces NOS2, while reducing ARG-1 and HO-1 expression, suggesting a switch from M2 to M1 phenotype (Johnson, et al., 2016). Notably, in striking contrast to total body FATP1 knockout, chimeric mice with FATP1-deficient macrophages show an increased susceptibility to obesity and metabolic dysfunction (Johnson, et al., 2016), suggesting a potential role of macrophage FATP1 in maintaining metabolic homeostasis (Figure 1). Similar to alternatively activated macrophages, MDSCs rely on lipids as their energy source (Yan, et al., 2019). FATP2 is reported to be a fatty acid transporter highly induced in PMN-MDSCs from mice and cancer patients (Yan, et al., 2019). Its overexpression in MDSCs has been implicated in the acquisition of an immunosuppressive phenotype through accumulation of arachidonic acid and its major metabolite PGE2, regulation of ROS production, and consequent CD8+ T-cell inhibition (Veglia, et al., 2019; Yan, et al., 2019). Of note, FATP2 expression and lipid uptake appear to be enhanced by tumor-supplied GM-CSF (Yan, et al., 2019), highlighting potential cross-talk between multiple cell types within the bone tumor microenvironment.

Although not a typical lipid scavenger, CD206, also known as mannose receptor, is highly expressed in lipid-laden macrophages (Vogel, et al., 2022). CD206 is an M2 macrophage marker, whose expression is induced by lipid exposure in a manner dependent on the activity of PPARγ. CD206-positive macrophages are thought to play a role in buffering the lipids released from dying macrophages (Hill, et al., 2018; Nawaz, et al., 2017). They are also highly present in metastatic tissues from castrate resistant prostate cancer patients (Zarif, et al., 2019) and in the bone marrow of multiple myeloma patients as compared to healthy subjects or patients presenting with asymptomatic monoclonal gammopathy (MGUS) (Cencini, Fabbri, Sicuranza, Gozzetti, & Bocchia, 2021).

The triggering receptor expressed on myeloid cells 2 (Trem2) is yet another atypical lipid-sensing receptor in macrophages, which is widely studied in microglia in relation to Alzheimer’s Disease, but has only recently been linked to adipose tissue homeostasis (Deczkowska, Weiner, & Amit, 2020; Jaitin, et al., 2019). Trem2 is expressed in a subset of lipid associated macrophages (LAMs) in the expanding adipose tissue and was shown to be involved in LAM response to metabolic dysfunction by contributing to the formation of crown-like structures, reducing inflammation, and preventing adipocyte hypertrophy (Jaitin, et al., 2019; Vogel, et al., 2022). Analysis of transcriptional signature of Trem2-positive LAMs revealed genes involved in lipid metabolism and phagocytosis (CD36, FABP4, FABP5, LIPA, LPL, CTSB, CTSL, CD9) and immunosuppression (LGASLS1, LGALS3) (Jaitin, et al., 2019). Interestingly, Trem2 is also highly expressed by a population of ARG-1-positive TAMs, and genetic ablation of Trem2 in mice results in reduced accumulation of myeloid cells, a decrease in dysfunctional CD8+T cells, and inhibition of tumor growth (Katzenelenbogen, et al., 2020) (Figure 1). Trem2 signaling has been shown to promote macrophage survival under low nutrient and stress conditions (Deczkowska, et al., 2020). Adding to its potential importance for tumor residing in bone, Trem2 appears to be critical for osteoclast maturation, and mice lacking this receptor show impaired osteoclastogenesis and dysregulated bone metabolism (Otero, et al., 2012). Collectively, this places Trem2 as a potential hub linking bone, adipose tissue, and immune microenvironment with metastatic progression, which warrants further investigation.

B. Lipid metabolism regulation by PPARs

Lipids are strong ligands for PPARs, lipid-sensing nuclear receptors comprised of 3 members: PPARα, PPARβ/δ, and PPARγ (Poulsen, Siersbaek, & Mandrup, 2012). PPARs play key roles in adipocyte differentiation and function (Rosen, et al., 2002), regulation of immune cell phenotypes, as well as tumor development, progression and metastasis (Cheng, Yip, Lim, Wahli, & Tan, 2021; Hernandez-Quiles, Broekema, & Kalkhoven, 2021; Szeles, Torocsik, & Nagy, 2007; Tachibana, Yamasaki, Ishimoto, & Doi, 2008), and thus are likely to be at the center of adipocyte-immune cells crosstalk in the bone tumor microenvironment.

Two main PPARs implicated in immune regulation are PPARα and PPARγ. Both have documented roles in alternative activation of macrophages and a switch to a TAM phenotype (Christofides, Konstantinidou, Jani, & Boussiotis, 2021). PPARγ, in particular, activates genes involved in fatty acid uptake and transport (CD36, FABP4), but also the genes involved in lipid storage (perilipins) and metabolism (Angela, et al., 2016; Berger & Moller, 2002). It also directly induces the transcription of ARG-1 and MGL1, genes associated with alternatively activated M2 phenotype (Hernandez-Quiles, et al., 2021) and, accordingly, bone marrow macrophages from mice with conditional deletion of PPARγ have reduced expression of ARG-1 (Odegaard, et al., 2007). One mechanism behind TAM differentiation and their pro-tumorigenic action is cleavage of PPARγ by caspase 1. Truncated PPARγ has the ability to translocate to the mitochondria and inhibit medium-chain acyl-CoA dehydrogenase (MCAD) and fatty acid oxidation (FAO), thereby promoting the accumulation of lipid droplets in TAMs (Niu, et al., 2017) (Figure 1). PPARγ activity is considered to be critical to T cell function regardless of the T cell subtype, although the literature reports on specific effects have been somewhat conflicting. PPARγ-driven fatty acid metabolism controls the phenotype of adipose Tregs residing in the adipose-rich microenvironment (Figure 1), and its activation with thiazolidinediones (TZDs), synthetic PPARγ agonists, in mice upregulates fatty acid transporters (CD36 and FATP2), as well as enzymes involved in FAO, lipolysis, triglyceride synthesis and lipid droplet formation and storage (CPT1A, HSL, DGAT1, PLIN2) (Cipolletta, et al., 2012). PPARγ is thought to have inhibitory effects on T cell proliferation and Th1 responses through binding of nuclear factor of activated T cells (NFAT) (Hernandez-Quiles, et al., 2021; X. Y. Yang, et al., 2000). Curiously, PPARγ interaction with NFAT has also been shown to suppress Th2 cytokine production (Chung, Kang, & Kim, 2003), indicating the link between PPARγ activity and T cell function is complex and likely context and ligand dependent.

PPARα senses fatty acids released by adipocyte lipolysis (Iroz, et al., 2017), and transcriptionally activates fatty acid β-oxidation through induction of fatty acid catabolic enzymes and transport proteins and repression of enzymes involved in fatty acid synthesis (Grabacka & Reiss, 2008). It has been well-established that classical M1 macrophages rely on glycolysis, whereas alternatively activated M2 macrophages have an intact TCA cycle and enhanced OXPHOS driven by fatty acid uptake and subsequent FAO (Batista-Gonzalez, Vidal, Criollo, & Carreno, 2019; Odegaard & Chawla, 2011). Mitochondria-dependent β-oxidation of LCFAs requires the carnitine palmitoyltransferase enzymes CPT1 and CPT2 (Houten & Wanders, 2010). Preclinical studies have suggested that CPT1-mediated FAO is responsible for the activation of M2 macrophages induced by IL-4, a process reversed by the CPT1 inhibitor Etomoxir (Malandrino, et al., 2015). Others have shown that inhibition of M2 polarization by Etomoxir is due to off-target alterations in CoA levels, rather than its effects on FAO (Divakaruni, et al., 2018). There are also reports that FAO is necessary for NLRP3 inflammasome activation in bone marrow macrophages, suggesting M1 rather than M2 polarization (Batista-Gonzalez, et al., 2019). Notably, it is the exogenous saturated fatty acids (SFA) that trigger inflammasome activation by inducing the level of saturated phosphatidylcholine and loss of membrane fluidity in human macrophages, a process reversed by treatment with unsaturated fatty acids (UFA) (Gianfrancesco, et al., 2019).

Just like myeloid cells, T cells change their metabolism and phenotype based on microenvironmental cues and nutrient availability. PPARα regulates T cell response in mice in a gender-specific manner. Its expression is increased by androgens, and PPARα activity represses IFNγ production in male T cells (Christofides, et al., 2021). CD8+ T cells utilize fatty acids to facilitate memory differentiation and subset specialization. CD8+ central memory cells exhibit elevated FAO when compared to CD8+ effector cells, although they favor the use of extracellular glucose to fuel fatty acid and triglyceride synthesis as opposed to uptake of FFA from the environment (O’Sullivan, et al., 2014). However, tissue-resident memory cells comprised of CD4+ and CD8+ T cells have a much higher requirement for fatty acid uptake and FAO than central memory T cells or effector cells, which can be reversed by FAO inhibitors or FABP4/5 deficiency (Pan, et al., 2017). In addition to macrophage and T cell regulation, PPARα expression is induced in CLL cells as compared to normal B cells, and it is thought to protect them from harsh microenvironmental conditions by increasing FAO (Messmer, et al., 2015).

The effects of PPARβ/δ on immune cell phenotype and function are less established. PPARβ/δ deletion in adipose tissue resident macrophages in mice prevents transition from M1 to M2 phenotype, but PPARβ/δ activation does not affect macrophage polarization in humans (Christofides, et al., 2021). PPARβ/δ regulates expression of inflammatory genes such as MCP-1 (CCL2), MCP-3 and IL-1β in macrophages, but its pro- vs anti-inflammatory functions appear to be respectively dependent on binding and dissociation of transcriptional repressors (Lee, et al., 2003). Similar to PPARα, pathways regulated by PPARβ/δ include FAO and mitochondrial respiration, which are key to T cell differentiation, activation and survival (J. M. Choi & Bothwell, 2012). Furthermore, just like PPARα, PPARβ/δ is expressed at higher levels in CLL cells than normal lymphocytes and is thought to promote survival in energetically stressful conditions by detoxifying free radicals produced by mitochondrial oxidation of fatty acids (Y. J. Li, et al., 2017).

C. Lipid metabolism and hypoxia in bone

Bone is naturally hypoxic, and all cell types residing in the bone marrow niche have to tailor their metabolism to adapt to low oxygen concentrations, a process coordinated by hypoxia-inducible factors (HIFs) (J. D. Diedrich, et al., 2018; Mylonis, Simos, & Paraskeva, 2019). Under hypoxic conditions, PPARγ-mediated uptake of fatty acids as well as FABP-mediated transport are under direct control of HIF-1 (Bensaad, et al., 2014; Krishnan, et al., 2009). Since fatty acid catabolism is impaired under hypoxia, cells convert fatty acids to triglycerides and store them as lipid droplets. One key factor involved in lipid storage is Hypoxia Inducible Lipid Droplet Associated (HILPDA) (de la Rosa Rodriguez & Kersten, 2020). HILPDA is induced by both hypoxia and PPARs in macrophages and promotes lipid storage by inhibiting human and murine Adipocyte Triglyceride Lipase (ATGL) as well as the process of lipolysis (Padmanabha Das, et al., 2018), and is thought to serve as a lipid-buffering process to neutralize potentially toxic and reactive fatty acids (van Dierendonck, et al., 2020). Deletion of HILPDA in myeloid cells was shown to trigger heightened inflammatory response through increased production of PGE2 (van Dierendonck, et al., 2022). A recent study evaluating HILPDA expression in The Cancer Genome Atlas (TCGA) database demonstrated its overexpression in 14 tumor types, including the tumors with propensity to metastasize to bone, like breast, prostate, kidney and bladder cancers (C. Liu, Zhou, Zeng, Wu, & Liu, 2021). Importantly, HILPDA overexpression was correlated with TAM infiltration and expression and immunosuppressive genes, such as PD-L1, PD-1, TGF-β1, and TGF-βR, suggesting an important role in immunosuppression.

Hypoxia in the TME can divert the MDSC differentiation into immune suppressive TAMs (Kumar & Gabrilovich, 2014) (Figure 1). TAMs accumulate in the hypoxic areas of the tumor and their metabolism is under direct regulation of HIF-1α, which drives their glycolytic phenotype (Biswas, 2015). Tumor hypoxia via HIF-1α has been shown to promote the expansion of Tregs and induction of PD-L1 on MDSCs, which binds PD-1, induces T cell failure, and perpetuates immuno-suppressive microenvironment (Ben-Shoshan, Maysel-Auslender, Mor, Keren, & George, 2008; Noman, et al., 2014). The hypoxic niche and HIF activity clearly impact progression and therapy response of tumors residing in bone, including multiple myelomas and leukemias (Borsi, et al., 2015; Colla, et al., 2010; Deynoux, Sunter, Herault, & Mazurier, 2016). Studies from our lab have shown that exposure to bone marrow adipocytes induces HIF-1α signaling in bone-residing prostate carcinoma cells and causes a metabolic switch to a Warburg phenotype (J. D. Diedrich, et al., 2016). Notably, these hypoxic and Warburg signatures are reflected in metastatic tissues from prostate cancer patients (J. D. Diedrich, et al., 2016). Whether the marrow fat cells contribute to immunosuppressive TME in bone by directly activating HIF-1α signaling in the immune cells is not well-understood.

6. I GUESS FAT’S WHY THEY CALL IT THE BLUES: BMAT AND CANCER

A. Cancer-associated adipocytes as immunomodulators

Adipocytes within the tumor microenvironment, termed cancer-associated adipocytes (CAAs), engage in multi-directional interactions with tumor cells and immune cells (Q. Wu, et al., 2021). These interactions alter the CAA phenotype, and consequently modulate the TME to support progression. For example, a CAA crosstalk with breast cancer cells promotes an ‘activated’ CAA phenotype, which is associated with an increased expression of pro-inflammatory cytokines, such as IL-6 and IL-1β (Dirat, et al., 2011). The consequences of increased IL-1β production include the induction of IL-17 expression in γδT cells in mice, causing expansion and polarization of neutrophils in the support of progression and metastasis (Coffelt, et al., 2015). Notably, CAA-tumor cell crosstalk also affects IL-1β levels in tumor cells, and IL-1β supplied by prostate carcinoma cells triggers the activation of COX-2/CCL2 signaling in CAAs (Herroon, et al., 2019). This has important implications, as both CCL2 and COX-2, as well as products of COX-2 activity, including PGE2, play key roles in bone metabolism, macrophage recruitment and phenotype regulation, as well as T cell function (Bakos, et al., 2017; Gschwandtner, Derler, & Midwood, 2019; X. Li, et al., 2009; Mizutani, et al., 2009; Sierra-Filardi, et al., 2014; Sreeramkumar, Fresno, & Cuesta, 2012; Takahashi, Uehara, Bando, & Izumi, 2008). Tumor-derived IL-1β has also been shown to promote osteolysis, bone metastatic niche expansion, and increased homing of breast cancer cells to the bone (Tulotta, et al., 2019).

Interaction with tumor cells can induce senescence-associated secretory phenotype (SASP) in CAAs (Fairfield, et al., 2021; Q. Wu, Li, Sun, & Sun, 2020). As initially established by the studies in cancer-associated fibroblasts (CAFs), senescent cells spread the stress response and influence the neighboring cells to support tumor progression (Takasugi, Yoshida, Hara, & Ohtani, 2022). Indeed, SASP promotes the activated CAA phenotype and induces the secretion of cytokines and chemokines, such as CXCL1, CXCL2, IL-6, and IL-8 (Fairfield, et al., 2021). These inflammatory SASP-associated proteins have been shown to contribute to tumor progression in preclinical models of multiple myeloma and metastatic prostate cancer by stimulating osteoclastogenesis and creating a tumor-supportive niche (Fairfield, et al., 2021; Hardaway, Herroon, Rajagurubandara, & Podgorski, 2015; H. Liu, J. He, et al., 2019). Tumor cells interacting with adipocytes induce ROS (Herroon, et al., 2018; Ladanyi, et al., 2018), which activates NF-κB signaling and promotes a DNA damage response in CCAs, with the potential to result in immunosuppressive TME (Nelson, Kucheryavenko, Wordsworth, & von Zglinicki, 2018). Senescence is often associated with a higher glycolytic rate and secretion of lactate (Sabbatinelli, et al., 2019), a phenomenon also observed in CAAs (Q. Wu, et al., 2021). Lactate modulates the immune landscape in the TME by driving polarization of TAMs (Colegio, et al., 2014), suppressing IFN type-I, and modulating adaptive functions by promoting antigen degradation and decreasing cross-presentation (Q. Wu, et al., 2021). Lactate and low pH microenvironment associated with lactate secretion are also thought to regulate Treg phenotype and function by weakening aerobic glycolysis and promoting OXPHOS, which is a metabolic preference of this regulatory cell type (Huber, et al., 2017; Q. Wu, et al., 2021)

Adipocytes engaging in a crosstalk with tumor cells in vitro undergo phenotypic changes, exhibit smaller lipid droplet size, and show a reduced expression of adipocyte-related genes such as adiponectin, FABP4 and resistin (Dirat, et al., 2011; Herroon, et al., 2013). This is likely due to the tumor-induced lipolysis and FABP4-mediated lipid transfer as observed in adipocytes interacting with prostate and leukemia cells (J. D. Diedrich, et al., 2016; Shafat, et al., 2017). The presence of small adipocytes was actually correlated with poor prognosis in AML patients (W. Lu, et al., 2018), as was remodeling of BMAs by leukemia cell-derived GDF15, a ligand for transforming growth factor B receptor type II (S. Yang, et al., 2020).

One key metabolic consequence of adipocyte lipolysis is a production and release of adenosine triphosphate (ATP) (Tozzi, Hansen, & Novak, 2020). When released into the extracellular space, ATP is actively metabolized by a class of catalytic enzymes known as ectonucleotidases (Lohman, Billaud, & Isakson, 2012), including ectonucleotide pyrophosphatase/phosphodiesterases (NPP), produced by adipocytes and involved in the generation of adenosine (Rancoule, et al., 2014). Adenosine is a known negative regulator of the immune system and a driver of an immunosuppressive environment. Inhibition of adenosine receptors, which are ubiquitously expressed in a wide variety of immune cells, can induce secretion of inflammatory cytokines and activate tolerogenic antigen-presenting cells (APCs) (Ohta & Sitkovsky, 2014). In murine neutrophils, adenosine binding to adenosine receptor 2 (A2AR) facilitates the production of anti-inflammatory IL-10, leading to suppression of T and NK cell responses (Cekic, Day, Sag, & Linden, 2014) (Figure 1). Adenosine receptor signaling also polarizes active dendritic cells into a tumor-promoting suppressive phenotype (Cekic, et al., 2014) (Figure 1), and facilitates the differentiation of immunosuppressive Tregs (Ohta & Sitkovsky, 2014). Adenosine treatment of activated macrophages was seen to augment the expression of TAM markers such as ARG-1 and suppress macrophage-induced inflammation (Csóka, et al., 2012). Adenosine in the TME was shown to induce macrophage proliferation and TAM-mediated release of GM-CSF was reported to work synergistically with adenosine to induce macrophage proliferation in vitro (J. Wang, et al., 2021). This indicates that adenosine signaling induced by CAAs creates an immune-suppressive environment in the TME, contributing to tumor cell survival in the bone microenvironment. Targeting adenosine signaling in the TME has not been thoroughly investigated, but provides a promising therapeutic avenue toward turning the immune-cold tumors “hot”.

B. Impact of lipid-mediated ER stress on BMAT and cancer-immune cross-talk

Endoplasmic reticulum (ER) stress response is a complex, evolutionarily conserved process central to protein folding, calcium homeostasis and lipid synthesis and metabolism (H. Wu, Carvalho, & Voeltz, 2018). ER stress is initiated by 3 sensor proteins: IRE1, PERK and ARF6, each responsible for one branch of ER stress response, and can result in an accumulation of misfolded proteins in the ER lumen, triggering unfolded protein response (UPR) and consequently, depending on severity of UPR, either apoptosis or cell survival (M. Wang & Kaufman, 2014). It is well-established that activation of UPR can promote tumor cell survival in the TME (Chevet, Hetz, & Samali, 2015). Indeed, our recent studies have shown that metastatic prostate tumor cells residing in the adipocyte-rich bone marrow have a transcriptomic signature reflective of ER stress/UPR response (Herroon, et al., 2021; Herroon, et al., 2018). The ability of tumor cells to survive in the lipid-rich metastatic niche is at least partially due to the activity of ER chaperone, BIP, and its target UPR genes, SYVN1 and HERPUD1, which are likely involved in the disposal of harmful proteins to counteract ER stress and play cytoprotective roles against lipid-induced oxidative stress (Herroon, et al., 2021). The cross-talk between oxidative and ER stress pathways has also been shown to contribute to the progression and chemotherapy resistance in multiple myeloma (Vincenz, Jager, O’Dwyer, & Samali, 2013; Xiong, Chng, & Zhou, 2021).

In adipocytes, ER stress is known to be activated by the need to synthesize lipid droplets or activate triglyceride storage due to overnutrition (Zha & Zhou, 2012). Free fatty acid-induced ER stress can lead to an induction of inflammatory cascade in adipocytes through PERK activating IKB kinase B (P. Jiao, et al., 2011). ER stress due to adipose tissue expansion in mice has been linked with chronic inflammation, recruitment and activation of resident macrophages, as well as reduced levels of adiponectin (Kawasaki, Asada, Saito, Kanemoto, & Imaizumi, 2012; Zha & Zhou, 2012). One potential contributor to suppressive effects of adipocyte-ER stress on adiponectin levels can be overexpression of ER stress factor CHOP, which has been shown to interfere with the activity of adiponectin promoter (Hosogai, et al., 2007).

The abundance of lipids impacts all cell types in the bone tumor microenvironment, including the immune cells. ER stress is critical to macrophage metabolism and function (Sukhorukov, et al., 2020). ER stress response can be triggered by CD36-mediated oxLDL, which, in turn, drives the upregulation of CD36, perpetuating the lipid uptake (Yao, et al., 2014). ER stress is also thought to contribute to lipid accumulation in macrophages by modulating the activity of ER-resident sterol regulatory element binding proteins 1 and 2 (SREBP1 and 2) (Yao, et al., 2014). An elegant study by Di Conza et al recently demonstrated that tumor-supplied lipids trigger ER stress response in murine macrophages and promote their polarization to the TAM phenotype through the activation of IRE1-XBP1 and IREI-STAT3 signaling (Di Conza, et al., 2021). ER stress has also been implicated for the fate of MDSCs. Increases in CHOP, BIP and XBP1 levels were observed in MDSCs from tumor-bearing mice and cancer patients (Condamine, et al., 2014). The immunoinhibitory roles of MDSCs in the TME were recently attributed to the activity of PERK (Mohamed, et al., 2020). Deletion of PERK in tumor-infiltrating MDSCs transformed them into the myeloid cells driving CD8+ T-cell immunity against the tumor cells. UPR activation has also direct effects on all stages in the life of a T cell (reviewed in (Kemp & Poe, 2019)). Activation of PERK, IRE1α or CHOP results in dysfunctional intratumoral T cells (Cao, et al., 2019; Hurst, et al., 2019; M. Song, et al., 2018).

There is some convincing evidence that ER stress can be transmitted from one cell type to another, which has tremendous implications for a complex microenvironment such as bone marrow. ER stress response is known to lead to transcriptional activation of a proinflammatory cytokine program in the tumor and the immune cells (Mahadevan, et al., 2011). Importantly, factors released by stressed tumor cells can be sensed by the toll-like receptors on macrophages and mount an inflammatory response (Mahadevan, et al., 2011). Tumor cells undergoing UPR have been shown to elicit a transmittable ER stress (TERS) in CD11+ macrophages and dendritic cells (Condamine, et al., 2014; Mahadevan, et al., 2012). A similar reaction occurs in adipose tissue where ER stress induction in adipocytes triggers the inflammatory response (Schenk, Saberi, & Olefsky, 2008). ER stress appears to be front and center in the BMAT-tumor cell-immune cell cross-talk, highlighting the need to identify the transmissible factors and molecular mechanisms they activate to support tumor progression in bone.

7. HAPPILY, N’EVER CANCER: THERAPEUTIC IMPLICATIONS

Therapeutic targeting of tumors residing in bone has been proving to be a daunting task, and despite many advances in research and drug development, treatment options for patients with bone metastases and hematological malignancies, such as myelomas, have been mainly palliative. The ability to target the immune system with checkpoint blockade has been groundbreaking and clearly changed the treatment landscape for many cancers (Allard, et al., 2018), but it has also taught us that some tumors are immunologically “cold”, and don’t respond well to immunotherapy (reviewed in (Bonaventura, et al., 2019)). The limited efficacy of checkpoint inhibition has been particularly apparent in metastatic prostate cancer, a malignancy exhibiting reduced immunogenicity and low mutational tumor burden (Stultz & Fong, 2021). Single-target immunotherapies have also been failing in metastatic breast cancers, multiple myelomas and other bone trophic-cancers (Paul, Kang, Zheng, & Kang, 2018; Topalian, et al., 2019; Vonderheide, Domchek, & Clark, 2017). However, growing understanding of the complexity of the bone microenvironment and the role of the immune system in bone homeostasis and anti-tumor response is opening some potentially promising therapeutic avenues that give hope to patients with skeletal cancers, although there are still many roadblocks in the journey ahead (Figure 2).

Figure 2. New potential therapeutic avenues for targeting adipocyte-immune-tumor cell interactions in bone.

The interplay between fat cells and immune cells and the resulting immune suppression, metabolic adaptation, and activation of the pro-survival mechanisms in the tumor are key contributing factors to malignant progression in the bone tumor microenvironment. These tumor-supportive interactions are driven by four major mechanisms: 1) activation of lipid uptake and storage, 2) changes in lipid catabolism, 3) induction of ER stress and 4) modulation of immune cell recruitment and inflammation. Targeting these mechanisms opens the possibilities for therapeutic strategies that include: inhibitors of lipid uptake or lipogenesis (lipid uptake and storage), PPAR-targeting agents and fatty acid oxidation (FAO) inhibitors (lipid catabolism), inhibitors of PERK or IRE1-XBP1 (ER stress), and adipokine regulators, CSF1R antagonists, CCL2 inhibitors/antibodies, and COX2 inhibitors (immune cell recruitment and activation).

A. Targeting Immunosuppressive cell populations in the bone tumor microenvironment.

Interfering with immune cell infiltration or modulating the immune cell phenotype are two powerful ways of turning the immunosuppressive TME against the tumor. Here, we focus on three immune cell populations – macrophages, MDSCs, and Tregs – that have emerged as key drivers of immunosuppression resulting from BMAT-tumor-immune cell crosstalk (Table 1).

Table 1.

Targeting the immune suppressive environment

Cell Types	Inhibitors	Mode Of Action	Cancers	Effects	References and Clinical Trials
Macrophages (TAMS)	Pexidartinib (PLX3397)	Block activity of CSF-1R	Relapsed or Refractory AML	Reduces M2 polarization and stimulates T cell infiltration to TME	NCT01349049 (Completed with no results)
			Refractory leukemias		NCT02390752 (Recruiting)
			Bone and soft-tissue sarcomas		Mouse models (Fujiwara, et al., 2021)
			Metastatic Breast Cancer		NCT01596751 (Completed with no results)
			Metastatic Prostate Cancer to bone		NCT01499043 (Terminated)
	LY3022855 (IMC-CS4)	Anti-CSF-1R human monoclonal antibody	Advanced breast and prostate		NCT02265536 (Completed with results) (Autio, et al., 2020)
	Sotuletnib (BLZ945)		AML	TAM depletion	(Simonis, et al., 2021)
	Plozalizumab (MLN1202)	Anti-CCR2 humanized monoclonal antibody	Bone metastasis	Blocks M2 macrophage infiltration	NCT01015560 (Completed with no results)
	mNOX-E36	CCL2 inhibitor	AML	Reduces marrow infiltration of TAMs	Mouse Models (Rodrigo Omar Jacamo, 2015)
	Trabectedin (ET-743)	DNA damaging agent	Metastatic Prostate Cancer to bone mCRPC	Depletes M2-likes macrophages in marrow	Mouse models (Jones, et al., 2019) NCT00072670 (Completed with results)(Michaelson, et al., 2012)
			Metastatic Breast Cancer		NCT005580112 (Completed with results)(Delaloge, et al., 2014)
			Soft tissue sarcoma Osteosarcoma		NCT05131386 (Recruiting)
			CLL	Promotes M2 macrophage killing	Mouse Models (Banerjee, et al., 2019)
MDSCs	Entinostat	HDAC inhibitor	Metastatic Breast Cancer	Inhibits immunosuppressive function to re-establish T cell activity	NCT03473639 (Recruiting)
	Tasquinimod	Inhibitor of S100A9	mCRPC	Reducing recruitment of MDSCs in TME	NCT0123431 (Completed with results) (Sternberg, et al., 2016)
			Refractory Myeloma		NCT04405167 (Recruiting)
	CI-8993	Monoclonal antibody VISTA antagonist	Advanced Solid Tumors	Reduced MDSC inhibition of CD8 T cells activity	NCT04475523 (Recruiting)
	Sunitinib	RTK inhibitor	Sarcoma	Depletes MDSCs	NCT00474994 (Completed with results) (George, et al., 2009)
	Plerixafor (AMD3100)	CXCR4 antagonist	Osteosarcoma	Inhibits MDSCs migration to tumor	Mouse models (K. Jiang, et al., 2019)
	BKT-140		MM		NCT01010880 (Completed with no results)
	Ulocplumab (BMS-936564)	Anti-CXCR4 human monoclonal antibody	AML CLL Diffuse Large B-cell leukemia		NCT01120457 (Completed with no results)
			Refractory MM		NCT01359657 (Completed with results)(Ghobrial, et al., 2020)
	AZD5069	CXCR2 antagonist	mCRPC	Suppresses MDSC recruitment	NCT03177187 (Active)
Tregs	Ipilimumab	Anti-CTLA-4 antibody	AML, CML	Depletes Tregs	NCT03912064 (Recruiting)
	Tremelimumab		Sarcomas	Activates effector T cells	(Khan, et al., 2011) NCT02815995 (Active)
			Pediatric Hematological Malignancies		NCT03837899 (Recruiting)
			mCRPC		NCT03204812 (Completed with results)(Subudhi, et al., 2021)
			Hormone positive Metastatic Breast Cancer		NCT03608865
	Basiliximab	Anti-CD25 chimeric mouse-human monoclonal antibody	AML	Depletes Tregs	NCT01842139 (Completed with no results)
	Imatinib	Bcr-abl tyrosine kinase inhibitor	CML		Human Samples & Mouse Models (A. Tanaka, et al., 2020) NCT02977312 (Completed with no results)
	Nimesulide	COX-2 inhibitor	AML	Downregulates IDO1 expression to inhibit Treg differentiation	(Iachininoto, et al., 2013)
	Indoximod	IDO inhibitor		Reduces Treg activation	NCT02835729 (Completed with no results)

NOTE: AML = Acute Myeloid Leukemia, mCRPC = Metastatic Castrate Resistant Prostate cancer, MM = Multiple Myeloma, CLL = Chronic Lymphocytic Leukemia, CML = Chronic Myeloid Leukemia

CSF-1R = Colony stimulating factor 1 receptor, TME = Tumor Microenvironment, CCR2 = C-C Motif Chemokine Receptor 2, CCL2 = CC motif chemokine ligand 2, HDAC = Histone Deacetylase, VISTA = V-domain immunoglobulin suppressor of T cell activation, RTK = Receptor Tyrosine Kinase, CXCR4 = CXC motif chemokine receptor 4, CXCR2 = CXC motif chemokine receptor 2, CTLA-4 = Cytotoxic T-lymphocyte Associated Protein 4, CD25 = Interleukin-2 receptor alpha chain, COX-2 = cyclooxygenase 2 Cylooxygenase-2, IDO = Indoleamine-pyrrole 2,3-dioxygenase

Macrophage-targeting agents

Given the fact that survival and recruitment of macrophages is driven by CSF-1/CSF-1R axis, a number of agents have been developed to target CSF-1R or its ligand CSF-1, both as monotherapies and as combination treatments with chemotherapy as well as other cancer-immunotherapies. Among these agents, an oral tyrosine kinase inhibitor pexidartinib (PLX3397) has undergone the broadest clinical development (Cannarile, et al., 2017). PLX3397 has been tested as a monotherapy in metastatic prostate cancer, breast cancer, leukemias, and several other cancers, but regretfully, has shown mostly limited efficacy and some adverse effects (Cannarile, et al., 2017; Smith, et al., 2020; Trials.gov, 2020). Use of a monoclonal antibody (LY3022855) against CSF-1R in the refractory population of breast and prostate cancer patients was also somewhat disappointing. LY3022855 treatment resulted in reduced numbers of circulating monocytes and showed a reduced expression of myeloid genes in the patients who did exhibit clinical benefit (Autio, et al., 2020); however, overall impact on progression-free survival was very limited. Lacnotuzumab, a CSF-1 neutralizing antibody was tried as a combination therapy with gemcitabine and carboplatin in metastatic triple negative breast cancer patients, but showed no improvements above gemcitabine and carboplatin alone (Kuemmel, et al., 2022). In addition to limited efficacy observed in the tumor, CSF-1R/CSF-1 inhibitors are known to broadly target all macrophage populations in bone, including osteoclasts. Although many preclinical studies have shown inhibitory effects of these agents on bone loss (Mun, Park, & Park-Min, 2020), the use of CSF-1R/CSF-1 inhibitors can affect normal bone remodeling processes and could potentially cause skeletal issues in cancer patients.

Similar to CSF-1R/CSF-1 axis, the CCL2/CCR2 duo is also key to TAM recruitment to TME. Blocking this axis has shown promise in preclinical models of metastatic prostate cancer, especially in combination with docetaxel (Kirk, et al., 2013; Loberg, et al., 2007; Rozel, et al., 2009), which increases CCL2 levels (Qian, et al., 2010) but proved to be ineffective in human trials as a stand-alone treatment or in combination with chemotherapy (Brana, et al., 2015; Sandhu, et al., 2013). A phase 2 clinical trial (NCT01015560) using a humanized mAb to CCR2 (MLN1202, plozalizumab) in bone metastasis patients evaluated effects on tumor cell proliferation, macrophage recruitment and osteoclast maturation, and found limited efficacy, but there has been no result reported to date (Yumimoto, Sugiyama, Mimori, & Nakayama, 2019). More studies are needed to understand the regulation of CCL2/CCR2 axis bone tumor microenvironment, as it involves multiple cell types including the immune cells, bone cells, tumor cells and adipocytes (Fei, Ren, Yu, & Zhan, 2021).

Simply depleting TAMs through inhibition of CSF-1R/CSF-1 or CCR2/CCL2 axis, rather than modulating their immunosuppressive function, may not be the best approach to suppress anti-tumor immunity. Several trials are currently ongoing to examine the effects of reprogramming macrophages from an immune-suppressive to an immune-activating phenotype. Specifically, an oral, selective PI3K-γ inhibitor, IPI-549, is being evaluated in four clinical trials, (NCT03961698, NCT03719326, NCT03980041, NCT03795610). Notably, nanoparticles with IPI-549 were effectively used to repolarize M2 to M1 macrophages and enhance the efficacy of anti-PDI therapy in mouse models of metastatic breast cancer (Y. Song, et al., 2022), highlighting the importance of investigating this mechanism in skeletal tumor progression. DNA damaging agent Trabectedin (ET-743) was shown to selectively deplete M2-like (F4/80+CD206+) macrophages and inhibit skeletal tumor growth in preclinical models of prostate cancer. The same agent was also shown to target PD-L1⁺ monocytes/macrophages, and PD-1⁺ T cells in the preclinical model of CLL (Banerjee, et al., 2019; Jones, et al., 2019). More studies are needed, however, into the potential utility of using Trabectedin in treating bone cancers, as the Trabectedin clinical trial showed modest activity in metastatic CRPC (Michaelson, et al., 2012).

Targeting MDSCs in bone

MDSCs exhibit immunosuppressive activity by blocking the proliferation and activity of T cells, and are a promising target for anti-tumor therapy in bone (Xiang & Gilkes, 2019). Several approaches have been developed to either directly deplete MDSCs or target their immunosuppressive activity (Table 1). Tyrosine kinase inhibitor (Sunitinib) inhibits both hematopoiesis and MDSC production and has shown activity in the sarcoma trial against giant tumor of the bone (George, et al., 2009). Since the CXCR2 axis is important for MDSC migration and immunosuppressive functions, CXCR2 blocking represents an attractive approach towards improving existing and novel immunotherapy strategies (Bullock & Richmond, 2021). CXCR2 antagonist AZD5069 is currently in Phase I/II clinical trial in combination with Enzalutamide for castration resistant metastatic prostate cancer (NCT03177187). In addition to CXCR2 blockers, AMD3100, a CXCR4 inhibitor, has shown synergistic activity with anti-PD1 therapy in the mouse model of osteosarcoma (K. Jiang, et al., 2019). Furthermore, an anti-human CXCR4 antibody Ulocuplumab, in combination with lenalidomide and dexamethasone, showed high response in patients with relapsed/refractory myeloma, making it a promising anti-myeloma drug (Ghobrial, et al., 2020).

A currently ongoing clinical trial is evaluating the efficacy of HDAC inhibitor Entinostat in combination with Capecitabine in metastatic breast cancer (NCT03473639). Entinostat is a histone deacetylases (HDAC) inhibitor shown to reprogram MDSCs through alterations in STAT3 and NF-κB phosphorylation, and downstream effector expression (Sidiropoulos, et al., 2022). HDAC inhibitors, including Entinostat and 5-azacytidine, have also been shown to inhibit the trafficking of MDSCs through the downregulation of CCR2 and CXCR2 (Z. Lu, et al., 2020), and are considered anti-tumor agents in skeletal malignancies. Another novel agent targeting MDSCs is Tasquinimod, currently in clinical trials for relapsed or refractory multiple myeloma (NCT04405167) (Dan T. Vogl, 2020). Tasquinimod targets MDSCs via S100A9 protein and was previously evaluated in chemotherapy-naïve men with mCRPC and was shown to have positive effects on progression-free survival (Sternberg, et al., 2016). Other approaches to target MDSCs could include antibodies against V-domain immunoglobulin suppressor of T-cell activation (VISTA). VISTA is highly expressed on myeloid-derived suppressor cells (MDSCs) in AML patients and was shown to inhibit CD8 T cell response in preclinical model of AML (L. Wang, et al., 2018). VISTA was also identified as an inhibitory checkpoint after anti-CTLA-4 therapy in prostate cancer (Gao, et al., 2017), underscoring the need for further evaluations. Phase I clinical trial of Anti-VISTA antibody in patients with advanced solid tumor malignancies is currently ongoing.

Inhibiting Tregs in bone tumor microenvironment

The therapeutic application of targeting Tregs has been well demonstrated in the clinic. The anti-CTLA-4 antibodies, such as Ipilimumab and Tremelimumab, or anti-PD-1 antibodies, such as Nivolumab, are used to block Treg suppressive functions (Huppert, et al., 2022; S. D. Wang, et al., 2016). However, the response to anti-PD-1 therapy in prostate and breast cancers, two major bone-trophic tumor types, is less than 10% (Huppert, et al., 2022). In fact, only a 4% response was reported in the recent CheckMate 650 trial evaluating the efficacy of Nivolumab and Ipilimumab in combination with Cabazitaxel (NCT02985957). Furthermore, a recent trial combining anti-CTLA-4 and anti-PD1 blockade using Tremelimumab and Durvalumab showed an upregulation in myeloid and neutrophil immune subset transcriptional signatures, and an increased expression of inhibitory immune checkpoints (Subudhi, et al., 2021). This suggests that additional strategies are needed to target immunosuppressive mechanisms driving these cancers.

Tregs express high levels of CD25, and Daclizumab and Basiliximab, two anti-human CD25 mAbs, are currently approved for use in a number of cancers (Xiang & Gilkes, 2019). Daclizumab was shown to reduce circulating CD25^highFOXP3⁺ Tregs after vaccination with cancer antigen peptides (hTERT/survivin) in metastatic breast cancer (Rech & Vonderheide, 2009). A clinical trial combining Basiliximab with vaccine therapy in AML was completed (NCT01842139), but no results were reported. One mechanism of controlling Treg infiltration into the tumor is inhibiting COX-2 and reducing PGE2 levels. An increased production of PGE2 by tumor cells was shown to stimulate the recruitment of Tregs, and blocking PGE2 or COX2 expression in tumor cells reduced bone metastases in preclinical models (Karavitis, et al., 2012). In line with these findings, COX2 inhibitor Nimesulide was shown to inhibit the generation of Tregs by leukemia cells through inhibiting the activity of IDO, suggesting that both COX2 and IDO are required for proper Treg development (Iachininoto, et al., 2013). This is of potential significance, as an IDO inhibitor, Indoximod, is being evaluated in a clinical trial in combination with chemotherapy in AML (NCT02835729), but the results are pending. Other potential agents with immunomodulatory functions against Tregs are bisphosphonates, which are widely used for bone metastases from several cancers (Santini, et al., 2010). The bisphosphonate zoledronic acid has been shown to reduce proliferation and immunosuppressive effects of Tregs and downregulate the expression of CTLA-4 and PD-1 on Tregs (H. Liu, Wang, Chen, Chen, & Lin, 2019; H. Liu, et al., 2016). A decrease in Tregs by zoledronic acid treatment was shown to increase the proliferation of T and NK cells In metastatic patients (Sarhan, et al., 2017). Targeting the RANKL pathway might be another way of inhibiting Tregs in the bone metastatic niche. Tregs have been shown to produce more RANKL in the presence of tumor cells, and the resulting increase in osteoclastogenesis and bone resorption was shown to release growth factors such as TGF-β, which is known to increase the number of Tregs in the bone microenvironment (S. Jiao, et al., 2019). Accordingly, a combination of anti-CTLA-4 and anti-TGF-β therapies has shown to mute Treg numbers in the TME (S. Jiao, et al., 2019). Clearly, Tregs are essential to bone homeostasis and play essential roles in tumor progression in bone. Identifying the pathways most critical to Treg biology is of critical importance for future treatments of skeletal tumors.

B. Immunometabolism as a therapeutic target in bone

While it is well established that dysregulated cellular energetics give cancer cells a growth and survival advantage, the emerging literature is beginning to indicate that tumor progression and escape from immune surveillance are also highly dependent on the tumor-immune cell cross-talk and the metabolic regulation in the TME. The interplay between the tumor and the immune cells drives metabolic reprogramming in both populations and is further complicated by other cell types in the TME, especially adipocytes. Fatty acid-induced lipid metabolic reprogramming in particular affects all cell types in the bone tumor microenvironment, and, although complex, provides a potentially attractive strategy for targeting skeletal tumors.

Targeting fatty acid uptake and transport

Inhibition of exogenous fatty acid uptake by targeting transporters has shown some immunomodulatory effects in several tumor types. Human monoclonal antibody to scavenger receptor CD36 (clone CRF D-2712) was shown to deplete intratumoral Tregs and suppress tumor growth in the mouse model of YUMM1.7, B16-OVA melanoma (H. Wang, et al., 2020) (Table 2). Anti-tumor effects of anti-CD36 antibodies or irreversible CD36 antagonist sulfo-N-hydroxysuccinimidyl (NHS) ester of oleate (SSO) were also demonstrated in ovarian and prostate cancer models, but the effects of these agents specifically on immune cells were not evaluated (Ladanyi, et al., 2018; Watt, et al., 2019). VT1021 is a promising, novel CD36-targeting compound currently in Phase I/II clinical trials for solid tumors (NCT03364400). VT1021 binds to MDSCs and induces the expression of thrombospondin-1 (Tsp-1) in the TME, which blocks the CD47 immune checkpoint and reprograms the CD36 receptor to induce tumor cell apoptosis, inhibit angiogenesis, and cause a phenotypic switch from the M2 to M1 macrophage phenotype (Devalingam Mahalingam, 2021). The advancement of this compound to Phase II/III trial in glioblastoma was recently announced by Vigeo Therapeutics (Therapeutics, 2022), showing promise for broader use as a potential anti-tumor therapeutic in other cancers. CD36 has been recently reported to be dispensable for normal hematopoietic cells, which makes it a potentially exciting target for hematological malignancies (HoudaAlachkar, 2021)

Table 2.

Targeting Immunometabolism

Targets	Agents	Mode of Action	Cancer	Effects on Immune Cells	References and Clinical Trials
Fatty Acid Transporters
CD36	Anti-CD36 antibody	Blocks CD36 ligand binding	Melanoma Model	Depletes intratumoral Tregs	Mouse Models(H. Wang, et al., 2020)
			Metastatic Ovarian Cancer	Not evaluated	Mouse Models (Ladanyi, et al., 2018)
			Prostate Cancer		Mouse Models (Watt, et al., 2019)
	VT1021	Induces expression of Tsp-1, which binds and inhibits CD36	Advanced solid tumors	Reprograms MDSCs to induce Tsp-1	NCT03364400(Active)
FAPT2	Lipofermata	FATP2 Inhibitor	Melanoma	Blocks MDSCs ROS production and enhances PD-L1 blockade therapy	Mouse Models (Adeshakin, et al., 2021)
			Pancreatic Cancer	Inhibit PGE2 that is secreted by PMN-MDSC	Mouse Models (Veglia, et al., 2019)
FABP	BMS309403	FABP4 inhibitor	Ovarian Cancer	Not evaluated	Mouse Models (Mukherjee, et al., 2020)
			Prostate Cancer	Not evaluated	Mouse Models (M. Huang, et al., 2017)
TREM2	PY314	Anti-TREM2	Advanced Solid Tumors	Depletes TAMs and reverses anti-PD-1 resistance	(Binnewies, et al., 2021) NCT04691375 (Recruiting)
Lipid Metabolism Regulators
PPARy	GW9662	PPARgamma antagonists	Breast Cancer	Increased anti-PD-L1/PD-1 blockade therapy	Mouse models(B. Wu, et al., 2018)
			Melanoma	Increases efficacy of PD-L1	Model Models(B. Wu, et al., 2020)
	Rosiglitazone	PPARgamma agonist	Melanoma and LCC	Increases CD8: Treg ratio and enhances immunotherapy	Mouse Models(Goyal, et al., 2018)
	Pioglitazone		CML	Not evaluated	NCT02852486(Active with results)(Rousselot, et al., 2017)
	CDDO-Me	Triterpenoid with PPARgamma activity	Breast Cancer	Reduces Tregs and TAMs	Mouse Models (Ball, et al., 2016)
			Solid Tumors and Lymphoma		NCT00322140 (Completed with no results)
	Efatutazone	PPARgamma agonist	Anaplastic Thyroid Cancer		NCT02152137 (Active with results) (Smallridge, et al., 2013)
PPARa	TPST-1120	PPARa antagonist	mCRPC,RCC,Colorectal Cancer, TNBC	Repolarizes M2 macrophages to M1 phenotype	NCT03829436 (Active)
	Fenofibrate	PPARalpha agonist	Melanoma	Increases efficacy of PD-1 blockade	(Y. Zhang, et al., 2017)
CPT-1	Etomoxir	Irreversible inhibitor of CPT-1	LLC, Colon Cancer	Decreased immunosuppression function of MDSCs	Mouse Models (Hossain, et al., 2015)
			Bladder Cancer		Mouse Models(S. Cheng, et al., 2019)
Lipogenesis Modulators
FAS	Cereulenin	FAS Inhibitor	Ovarian	Prevents immunosuppression effects of DCs	Mouse Models (L. Jiang, et al., 2018)
	TVB-2640		KRAS mutated NSCLC		NCT03808558 (Recruiting)
			HER2+ Metastatic Breast		NCT03179904 (Recruiting)
			Solid tumors		NCT02223247 (Completed with results) (Falchook, et al., 2021)
	EGCG		Lung Cancer		Mouse Models (Relat, et al., 2012)
	Omeprazole		Pancreatic Cancer		NCT04930991 (Recruiting)
			Castration Refractory Prostate Cancer		NCT04337580 (Recruiting)
SREBP-1	Fatostatin	Inhibits SCAP, which is required for SREBP activation	Melanoma	Sensitized TAMs to therapeutic targeting	Mouse Models (C. Liu, et al., 2019)
			Endometrial Carcinoma		Mouse Models (Ma, et al., 2021)
			Prostate Cancer		Mouse Models (X. Li, et al., 2014)

NOTE:LCC = Lewis Lung Carcinoma, CML = Chronic Myeloid Leukemia, mCRPC = Metastatic Castrate Resistant Prostate Cancer, RCC = Renal Cell Carcinoma, TNBC = Triple Negative Breast Cancer, NSCLC = Non-Small Cell Lung Cancer = NSCLC

CD36 = Cluster of Differentiation 36, Tsp-1 = Thrombospondin 1,FAPT2 = Fatty-acid binding protein 2, ROS = Reactive Oxygen Species, PD-L1 = Programmed Cell Death Ligand, PGE2 = Prostaglandin E₂, FABP = Fatty Acid Binding Protein 1, TREM2 = Triggering Receptor Expressed on Myeloid Cells 2, CDDO-ME = Bardoxolone methyl, CPT-1 = Carnitine palmitoyltransferase 1, FAS = Fatty Acid Synthase, KRAS = Kristen rat sarcoma virus, Human epidermal growth factor receptor 2 = HER2, EGCG = Epigallocatechin gallate, SREBP-1 = Sterol regulatory- element binding proteins, SCAP = Sterol regulatory element-binding protein

The regulation of lipid accumulation by MDSCs by targeting FATP2 was recently shown to inhibit tumor progression in a preclinical model of melanoma. Specifically, FATP2 inhibitor Lipofermata was shown to reduce ROS production, dampen ARG-1 and iNOS expression, mute immunosuppressive activities of MDSCs, and improve response to anti-PD-1 therapy (Adeshakin, et al., 2021). Lipofermata was also shown to delay tumor progression in four pancreatic cancer models by inhibiting PGE2 in PMN-MDSCs and reducing their immunosuppressive functions (Veglia, et al., 2019). This is a promising result that offers a possibility of directly targeting the PGE2 axis in FATP2 expressing immune cells without the systemic effects of PGE2 inhibitors. There are also promising preclinical results demonstrating efficacy of targeting FABPs. Small-molecule inhibitor of FABP4 (BMS309403) was shown to significantly reduce tumor burden and increase sensitivity to Carboplatin in a syngeneic orthotopic ovarian cancer model (Mukherjee, et al., 2020). The same inhibitor was shown to inhibit the progression of prostate cancer xenografts (M. Huang, et al., 2017). Given previous studies by our laboratory demonstrating the importance of FABP4 in metastatic progression in bone, studies in immunocompetent models of prostate cancer are needed to evaluate immune cell contribution in this process. In addition to traditional lipid transporters, Trem2 is emerging as a novel target in cancer therapy. PY314, a humanized anti-Trem2 mAb, was shown to promote anti-tumor immunity by depleting the TAM population and enhancing infiltration and effector functions of CD8+ T cells (Binnewies, et al., 2021). This highlights the need of investigating this novel receptor in the context of anti-tumor therapies in bone.

Targeting regulators of lipid metabolism

PPARs are front and center in regulating metabolic homeostasis, have been a subject of drug development for metabolic diseases and cancer, and a number of synthetic PPAR ligands have been developed and evaluated therapeutically (reviewed in (H. S. Cheng, et al., 2019; Christofides, et al., 2021)). There is substantial evidence that PPARγ agonists might be candidates for cancer immunotherapy. Administration of Rosiglitazone to mice exposed to GM-CSF-secreting tumor-cell vaccines (GVAX) exhibited myeloid-driven induction of CD8⁺ T effector to Treg ratio and overall attenuation of Treg induction (Goyal, et al., 2018). An increased anti-tumor activity of GVAX in combination with anti-CTLA-4 blockade was also observed(Goyal, et al., 2018). Another agent with PPARγ activity, triterpenoid CDDO-Me, was shown to relieve immunosuppression in estrogen receptor negative breast cancer model by inducing TAMs and CD4⁺ Foxp3⁺ regulatory T cells (Ball, Shipman, Kim, Liby, & Pioli, 2016). On the flip side, antagonizing PPARγ activity with GW9662 was shown to reduce PD-L1 expression in adipose tissue and to enhance the efficacy of anti-PD-L1 or anti-PD-1 therapies in preclinical models of breast cancer and melanoma (B. Wu, et al., 2018; B. Wu, et al., 2020). Immunomodulatory activity exhibited by both PPARγ agonists and antagonists might be the reason why promising anti-tumor effects of PPARγ ligands in preclinical models have shown no significant efficacy in patient trials (H. S. Cheng, et al., 2019). Similar reasoning might be true for agents targeting PPARα activity. PPARα agonist, Fenofibrate, was shown to induce FAO and reprogram CD8⁺ T cells to enhance the efficacy of PD-1 blockade(Y. Zhang, et al., 2017). In line with the role of FAO in immunosuppression, an irreversible inhibitor of CPT-1, Etomoxir, inhibited tumor progression in preclinical models of colon and lung carcinomas by inhibiting the immunosuppressive effects of MDSCs (Hossain, et al., 2015). On the other hand, selective PPARα antagonist TPST-1120, known to inhibit M2, MDSCs and Tregs from infiltrating into TME, is being tested as a monotherapy or in combination with Nivolumab in several advanced cancers (NCT03829436). This speaks to the importance of PPARs in immunosuppression and underlines the need for more mechanistic investigations.

Investigating immunomodulatory functions of lipogenesis modulators

Fatty acid synthase (FASN) plays crucial roles in cancer and has been shown to contribute to the growth and survival advantage in multiple tumor types (Fhu & Ali, 2020). A number of other FASN inhibitors are currently being evaluated in clinical trials (Table 2). A potent FASN inhibitor, TVB-2640, has shown disease control responses across several tumor types in combination with Paclitaxel (Falchook, et al., 2021). FASN inhibition with Cerulenin was shown to lead to growth regression in subcutaneous and intraperitoneal models of ovarian cancer (L. Jiang, Fang, Wang, Li, & Wang, 2018). Cerulenin was able to restore immunostimulatory functions of tumor-infiltrating dendritic cells to promote T-cell mediated anti-tumor immunity. Studies in multiple immunocompetent tumor models and critical evaluation of immune landscape in patient samples upon FASN inhibition are needed to understand the role of fatty acid synthesis in immune regulation of anti-tumor response.

One of the regulators of FASN expression and activity is SREBP1, a protein highly expressed on immune cells and highly involved in metabolic and immunological processes (Park, Kang, & Im, 2018). Fatostatin, which inhibits SREBP activation by binding to SCAP and preventing SCAP-mediated escort of SREBP1 to Golgi, shows significant anti-tumor efficacy in several preclinical models (X. Li, Chen, Hu, & Huang, 2014; C. Liu, et al., 2019; Ma, et al., 2021). Notably, use of Fatostatin in combination with anti-PD-1 blockade impaired M2-like activation, lowered the threshold for CDT8+ T cell reinvigoration and immunostimulatory anti-tumor TME in B16 melanoma model (X. Li, et al., 2014). This suggests that targeting fatty acid synthesis in TAMs might be a viable strategy to improve response to checkpoint blockade. Whether this strategy is applicable to immune “cold” tumors in bone needs investigation.

C. Targeting ER stress as a novel approach to inhibit immunosuppression

Activation of the ER stress response is intimately connected to T cell differentiation and, therefore critical to T-cell mediated anti-tumor activity (Andrews, Tennant, & Thaxton, 2021). T cell expansion requires T effector cells to greatly increase new protein synthesis and initiate post-translational modifications, which can lead to unsurmountable stress and cell death (Rutkowski & Kaufman, 2004). Activation of the IRE1α–XBP1 arm of ER stress in T cells controls their mitochondrial respiration and limits infiltration into the tumor, and shows anti-tumor activity in the ovarian cancer model (Rutkowski & Kaufman, 2004). Several agents targeting IRE1α endonuclease activity and XBP1 splicing have been developed and showed potent anti-tumor activity in several preclinical models of hematological malignancies (Mimura, et al., 2012; Papandreou, et al., 2011; Ri, et al., 2012; H. Sun, et al., 2016; C. H. Tang, et al., 2014), but have not been tested in bone metastases and not at all in the context of T cell function.

Interestingly, PERK inhibition with first generation PERK inhibitor, GSK26064141, generated complete responses in mice bearing sarcomas when combined with anti-PD-1 blockade (Hurst, et al., 2019) and inhibited tumor progression in the mouse model of multiple myeloma (Bagratuni, et al., 2020). Another study tested GSK26064141 along with a selective PERK inhibitor, AMG-44, and demonstrated striking impairment in immunoregulatory activity of tumor-MDSC, induced expansion of tumor-infiltrating CD8⁺ T cells expressing IFNγ and synergy with anti-PD-L1 therapy (Mohamed, et al., 2020). These studies speak to the role of PERK in T cell dysfunction and introduce opportunities to target ER stress pathways to enhance anti-tumor response.

8. PERSPECTIVES AND UNANSWERED QUESTIONS

The evidence is compelling that dynamic and multidirectional interactions between immune cells, fat cells, and the cancer cells are key to shaping the fate of the tumor in the bone marrow niche. Studies into molecular mechanisms behind bone metastases or hematological malignancies have primarily focused on two-way communications between BMAT and the tumor or immune cells and the tumor; and have not yet considered combining these two fields together to examine how BMAT-immune cell crosstalk impacts progression and therapy response in bone. There is a wealth of literature demonstrating that fat cells, through the secretion of lipids, hormones and adipokines, can modulate the phenotype and metabolism of immune cells. There is also mounting amount of data that tumor energetics and pro-survival signaling are highly regulated by adipocyte-supplied lipids. Understanding how these mechanisms converge in bone to create immunosuppressive microenvironment, could fundamentally change therapeutic approaches for cancers that thrive in adipocyte-rich bone marrow, and especially those that are immunologically “cold” like metastatic prostate cancers.

As we learn more about role of metabolism in immune regulation, and understand the impact of metabolic stressors like hypoxia, ER stress, lipid toxicity and nutrient deprivation on immune cell phenotype and function, new avenues for investigation are emerging, that will help transform the field towards new cancer therapies. There are number of unanswered questions that remain: 1) Can we target lipid metabolism specifically in immune cells to enhance efficacy of immunotherapies?; 2) Are the opposing effects of PPAR ligands on immune cell phenotype and function disqualify PPARs are therapeutic targets?; 3) Is ER stress a viable target for bone-tumor specific therapeutics?; 4) Do we need to take a closer look at the impact of BMAT on adenosine receptor signaling and immunosuppression in bone?; and finally 5) How is inflammaging regulating immune landscape in bone marrow and how that impacts therapeutic approaches for skeletal tumors? Tackling these questions is critical for enhancing our understanding of the complexity of the bone tumor microenvironment and for identifying opportunities for targeted therapeutic interventions centered around BMAT-immune cell-tumor cell crosstalk.

Table 3.

Targeting ER stress

Targets	Agents	Mode of Action	Cancer	Cell Types and effects	Reference
IRE1α/XBP1	STF-083010	Inhibited IRE1 endonuclease activity	Human MM	Inhibited tumor growth	Mouse model(Papandreou, et al., 2011)
			AML	Cytotoxicity against tumor cells	Mouse model(H. Sun, et al., 2016)
	Toyocamycin	Inhibits XBP1 cleavage	MM	Inhibited tumor growth	Mouse model(Ri, et al., 2012)
	B-I09	Inhibits RNase activity of IRE1	CLL	Induces tumor apoptosis	Mouse model (C. H. Tang, et al., 2014)
	MKC-3946	Blocks XBP1 splicing	MM	Induced tumor apoptosis and inhibited growth	Mouse model(Mimura, et al., 2012)
PERK	GSK-2606414 and AMG-44	Dual PERK and RIPK1 inhibitor (ATP competitive) Potent, selective PERK inhibitor	Melanoma	Decreased immunoinhibitory effects of tumor-MDSCs, induced proliferation of tumor-infiltrating CD8+ T cells, and delayed tumor growth	Mouse model(Mohamed, et al., 2020)
	GSK-2606414	Dual PERK and RIPK1 inhibitor (ATP competitive)	Sarcoma	Reduced mitochondrial oxidative stress and supported viability in CD8+ T cells, increased tumor clearance	Mouse model(Hurst, et al., 2019)
			MM	Decreased tumor proliferation and increased apoptosis	Mouse model(Bagratuni, et al., 2020)

NOTE:MM = Multiple Myeloma, AML = Acute Myeloid Leukemia, CLL = Chronic Lymphocytic Leukemia

IRE1 = Inositol-Requiring Enzyme 1, XBP1 = X-Box-Binding Protein 1, PERK = Protein Kinase RNA-Like Endoplasmic Reticulum Kinase, ATP = Adenosine Triphosphate

ABBREVIATIONS:

RANKL

Receptor activator of nuclear factor kappa-B ligand

BMAs

Bone marrow adipocytes

TAM

Tumor-associated Macrophages

CSF-1

Tumor-derived Colony Stimulating Factor 1

ADT

Androgen Deprivation Therapy

ARG1

Arginase 1

CCL2

Chemokine (C-C motif) ligand 2

MDSCs

Myeloid-derived suppressor cells

M-MDSCs

mononuclear MDSCs

PMN-MDSCs

polymorphonuclear morphology MDSCs

Tregs

Regulatory T cells

PR3

Protease 3

Th17

T helper cells

ADCC

Antibody-mediated cellular cytotoxicity

Bregs

Regulatory B cells

WAT

White adipose tissue

ATMs

Adipose tissue macrophages

CLS

Crown-like structures

BMAT

Bone marrow adipose tissue

cBMAT

constitutive bone marrow adipose tissue

rBMAT

regulated bone marrow adipose tissue

DCs

Dendritic cells

CLL

Chronic lymphocytic leukemia

FABPs

Fatty acid-binding proteins

FATPs

Fatty acid transport proteins

FAT (CD36)

Fatty acid translocase

LCFA

Long chain fatty acids

PPARγ

Peroxisome proliferator-activator receptor γ

PGE2

Prostaglandin E2

TREM2

triggering receptor expressed on myeloid cells 2

FAO

Fatty acid oxidation

HILPDA

Hypoxia inducible lipid droplet associated

CAAs

Cancer-associated adipocytes

SASP

Senescence-associated secretory phenotype

ATP

Adenosine triphosphate

ER

Endoplasmic reticulum

UPR

Unfolded protein response

SREBP1/2

Sterol regulatory element binding proteins 1 and 2

PLX3397

Pexidartinib

VISTA

V-domain Ig suppressor of T cell activation

SSO

Sulfo-N-hydroxysuccinimidyl ester of oleate

Tsp-1

Thrombospondin-1