Monocyte heterogeneity and functions in cancer

Abstract Monocytes are innate immune cells of the mononuclear phagocyte system that have emerged as important regulators of cancer development and progression. Our understanding of monocytes has advanced from viewing these cells as a homogenous population to a heterogeneous system of cells that display diverse responses to different stimuli. During cancer, different monocyte subsets perform functions that contribute to both pro‐ and antitumoral immunity, including phagocytosis, secretion of tumoricidal mediators, promotion of angiogenesis, remodeling of the extracellular matrix, recruitment of lymphocytes, and differentiation into tumor‐associated macrophages and dendritic cells. The ability of cancer to evade immune recognition and clearance requires protumoral signals to outweigh ongoing attempts by the host immune system to prevent tumor growth. This review discusses current understanding of monocyte heterogeneity during homeostasis, highlights monocyte functions in cancer progression, and describes monocyte‐targeted therapeutic strategies for cancer treatment.

that identified 2 populations with different patterns of CX3CR1, CCR2, and CD62L expression. 7 Two years later, a landmark study confirmed the presence of CX3CR1 hi and CX3CR1 lo monocyte subsets with significant differences in chemokine receptor expression, adhesion molecule profiles, and trafficking within homeostatic and inflamed tissue. 8 While these initial studies relied on engineered mice with green fluorescent protein inserted into the Cx3cr1 locus, subsequent studies have identified Ly6C as a more specific marker for discrimination of circulating monocyte subsets and reinforced previously reported phenotypic differences. [9][10][11] Human monocyte subsets were also found to display differential expression of CX3CR1 that corresponded with similar phenotypes to those observed in mouse, suggesting that mouse and human monocyte subsets may share similarities. 8 Genome-wide transcriptome profiling 10 and adoptive transfer assays to assess patrolling and response to inflammatory stimuli in immunocompromised mice 12 later reinforced that classical and nonclassical monocytes in mouse and human are counterparts.
Additional surface markers have been useful for improving discrimination of monocyte subsets. We recently used mass cytometry (CyTOF) to demonstrate that inclusion of HLA-DR, CCR2, CD36, and CD11c markers substantially increases the gating purity of human monocyte subsets, particularly CD14 + CD16 + intermediate monocytes. 13 Ly6C, MHCII, CD43, and Treml4 have enabled better identification of murine monocyte subsets, particularly within tissues where Ly6Chi monocytes can down-regulate Ly6C, but maintain low expression of CD43 and Treml4. 14 High expression of Treml4 is also useful for discrimination of intermediate monocytes from Ly6C hi and Ly6C lo populations. 15 The numbers of these markers will likely continue to expand as high-dimensional techniques are used to further profile monocyte transcriptional and proteomic states.
Single-cell transcriptional profiling has challenged our understanding of heterogeneity within well-established immune populations and monocytes are no exception. Deep single cell RNA sequencing (scRNA-Seq) using Smart-Seq2 recently revealed 2 novel human monocyte subsets. 16 While most (about 70%) of CD14 + C16 + monocytes typically classified in the intermediate subset fell within classical (Mono1) and nonclassical (Mono2) monocyte clusters, the remaining CD14 + CD16 + cells formed 2 new rare clusters termed Mono3 and Mono4 (12% of total monocytes). Mono3 highly expresses genes associated with cell cycle, differentiation, and trafficking, while Mono4 expresses high levels of cytotoxic gene signatures including NK and T cell activation genes. To date, no additional studies have validated the existence of these novel subsets, developed methodologies to isolate Mono3 and Mono4, or sought to identify their function.
In mice, scRNA-Seq revealed that monocytes primarily cluster into 4 types, 2 of which overlap with conventionally defined classical Ly6C hi monocytes and nonclassical Ly6C lo monocytes. 11 The other 2 monocyte clusters were primarily Ly6C int monocytes with 1 subset showing intermediate expression of classical monocyte genes and the other expressing MHCII-associated genes and CD209a. The CD209aexpressing subset is likely similar to the previously identified subset of Ly6C + monocytes that express DC-related genes, including those encoding CD209a, fms related tyrosine kinase 3 (Flt3, CD135), and MHCII, and give rise to DCs following exposure to GM-CSF. 17 Although additional studies are needed to confirm and understand the functional roles of new monocyte subsets, these studies indicate that conventional definitions of monocyte subsets will likely be modified and expanded, particularly within the intermediate subset.

MONOCYTE DEVELOPMENTAL ORIGINS AND FATES
Understanding the origin and fates of monocytes is crucial to understanding how these cells respond in cancer settings. Monocytes circulate in peripheral blood during steady state after developing from a lineage-committed bone marrow progenitor, the common monocyte progenitor (cMoP), which was first discovered in mouse. 18 The human cMoP was recently identified in human umbilical cord blood and bone marrow. 19 In both mouse and human, cMoPs are unipotent monocyte progenitors that express the stem cell marker CD117, as well as the C-type lectin CLEC12A and CD64. In contrast to mouse cMoPs, human cMoPs express CD135, a cytokine receptor and early hematopoietic marker.
Monocyte development is regulated by sequential expression of key transcription factors in the following order: PU.1, IRF8, and KLF4, which has been reviewed previously by our group. 20 PU.1 (encoded by Spi1) is a master regulator that is required for development of all myeloid cells and regulates expression of downstream factors such as IRF8 and KLF4. 21 IRF8 is expressed at low levels in hematopoietic stem cells, but up-regulated in common myeloid progenitors, granulocyte-M progenitors (GMPs), and cMoPs, ultimately restricting differentiation toward mononuclear phagocytes and limiting neutrophil generation. 22 Many monocyte-specific genes are directly controlled by IRF8, but IRF8 also induces expression of KLF4, which is required for monocyte development. 23 The precise relationship of cMoPs to other myeloid developmental lineages such as GMPs and M -DC progenitors (MDPs) is not fully understood. GMPs are a heterogeneous population that contains multiple progenitors with various degrees of restriction for differentiation into monocyte, DC, and granulocyte lineages. 24 Refining the definition of conventional human GMPs to include only CLEC12A hi CD64 int cells excludes cells with lymphoid or DC potential and identifies a population of cells that sequentially produces cMoPs and monocytes. 19 However, traditionally defined MDPs (CD115 + CD116-) phenotypically overlap with both GMPs and cMoPs, indicating that current definitions of progenitor populations are likely heterogeneous. In mice, both MDPs and GMPs can give rise to monocytes, with each appearing able to generate phenotypically different monocytes. 25 GMPderived monocytes possess a subset of monocytes with neutrophil-like gene expression and MDP-derived monocytes can express DC-related genes. 25 Future work will likely leverage scRNA-Seq, epigenetic profiling, and more advanced fate mapping strategies to better understand the developmental hierarchy of monocytes and their progenitors.
Following production of cMoPs in the bone marrow, generation of monocyte subsets is generally believed to involve differentiation of cMoPs into classical monocytes and subsequent conversion into nonclassical monocytes in blood circulation, with intermediate monocytes representing a transition state. 9,26 Early work demonstrated this conversion in mice using in situ bead-based labeling and clodronate liposome depletion, 9 but has been confirmed by others using adoptive transfer techniques, 27 complex fate mapping strategies in transgenic mice, 26 and epigenetic mapping. 11 Evidence that this conversion also occurs in humans was recently demonstrated by tracking accumulation of intravascularly administered deuterium-labeled glucose in healthy individuals and through adoptive transfer of human CD14 + monocytes in humanized mice. 28 The development and survival of nonclassical Ly6C lo monocytes was first found to depend on the orphan nuclear receptor Nr4a1. 29 Our laboratory demonstrated that Nr4a1-deficient mice primarily lack nonclassical monocytes and those that are present display abnormal cell cycling, dysfunctional patrolling behavior, and enhanced apoptosis. 29 Nr4a1 is also expressed in T cells and M s, motivating us to use epigenetic approaches to search for more specific regulatory factors specifically active during Ly6C lo monocyte generation. We identified a super-enhancer region upstream of the Nr4a1 gene that binds the transcription factor KLF2 and exclusively regulates Ly6C lo monocyte development. 30 In addition to KLF2, a similar epigenetic approach identified the transcription factor C/EBP as a regulator of the Nr4a1 gene and generation of Ly6C lo monocytes. 11 In agreement with these findings, Ly6C lo monocytes are absent in C/EBP -deficient mice. 31 Ly6C lo monocyte development may also rely on Notch2 expression and interaction with endothelial cells expressing Notch ligand delta-like 1 in bone marrow and splenic vascular niches. 32 Consequently, conversion of Ly6C hi monocytes into Ly6C lo monocytes depends on epigenetic modifications that could be in part regulated by interactions with their surrounding microenvironment.
While most nonclassical monocytes appear to derive from classical monocytes, 11,26 current work does not rule out the existence of an unidentified lineage-restricted progenitor that is able to differentiate into nonclassical monocytes without passing through a classical monocyte intermediary. Indeed, a population of segregated-nucleuscontaining atypical Ly6C lo monocytes that plays key roles in fibrosis develops from a specialized GMP-dependent progenitor without induction of Ly6C expression. 33 After differentiation, classical monocytes rely on CCR2 to exit the bone marrow, as well as traffic into tissues and lymph nodes 34,35 along gradients of CCR2 ligands CCL2, CCL7, and CCL12. CCR2-deficient mice accumulate Ly6C hi monocytes in bone marrow, but the frequency of circulating Ly6C lo monocytes is minimally affected, indicating that CCR2 is less important for their trafficking. 34 Entry into lymph nodes requires L-selectin (CD62L), whose expression is limited to Ly6C hi and Ly6C int monocytes. 35 Classical monocytes have a half-life in blood circulation of less than 1 day in humans and mice during steady state. 26,28 Conversely, nonclassical monocytes display a longer lifespan of 7 days in humans and at least 2 days in mice. Ly6C lo monocytes rely on a unique set of chemokine receptors for trafficking, including CX3CR1 and sphingosine-1-phosphate receptor 5. 36,37 Computational models predict that only 1% of classical monocytes convert into intermediate and subsequently nonclassical monocytes during homeostasis, while the remainder are predicted to extravasate or undergo cell death. 28 This supports a model where classical monocytes rapidly convert into nonclassical monocytes, traffic to tissues or the lymphatic system, or undergo apoptosis, the latter of which has not been extensively explored.
Monocytes have been detected outside of the circulation, including in skin, lung, and lymph nodes 35 and human tumors. 38,39 However, classical monocytes homeostatically entering tissues also give rise to M s and DCs in steady state. 26,40,41 In mice, most M s appear to be embryonically derived and self-renew through local proliferation. 26,42 However, monocytes are also able to repopulate tissue M populations and self-renew, adopt similar gene expression, and perform tissue-specific functions. 42,43 The mechanisms underlying tissue-specific differences in monocyte recruitment remain unclear, but they likely depend on environmental cues and tissue accessibility both during homeostasis and inflammation. Classical monocytes are recruited at higher rates to inflamed tissues and are able to attract other immune cells by secreting cytokines and antimicrobial factors. 14,34 Their differentiation into M s and DCs is regulated by key factors such as CSF1 (M-CSF), GM-CSF, and Flt3 ligand 41 and transcription factors such as IRF4 and MAFB. 44 A proportion of nonclassical monocytes patrol the vasculature in a lymphocyte function-associated antigen (LFA)-1 and CX3CR1dependent manner, 36 although whether nonclassical monocytes that display patrolling activity are transcriptionally distinct from those that do not patrol has not been investigated. Patrolling monocytes scavenge endothelium-derived cellular debris and flag-damaged endothelial cells for disposal by recruited neutrophils in a TLR7-dependent manner. 45 While nonclassical monocytes primarily remain in the vasculature during homeostasis, 35 they appear to be able to extravasate during inflammation, although they may do so at lower rates than classical monocytes. Whether nonclassical monocytes differentiate after exiting vasculature remains unclear, although evidence exists that they are able to give rise to M s with an alternatively activated phenotype and display anti-inflammatory properties in certain contexts. 36,46,47 The fate of nonclassical monocytes during infection, injury, and disease requires further investigation and is likely dependent on both tissuespecific cues and type of inflammatory signal.
One of the primary challenges to studying monocyte function in tissues is the limited availability of technologies that discriminate between monocytes and monocyte-derived cells, including monocytederived M s and DCs. Surface markers such as F4/80, MHCII, and CD11c are up-regulated once monocytes enter tissue, while Ly6C and CD11b are down-regulated, but these markers are often expressed along a continuum between blood monocytes and cells in tissue. 40,48 In

NK cells F I G U R E 1 Recruitment and functions of monocyte subsets to primary tumors and metastatic sites within the lung.
Classical CCR2 + monocytes extravasate from the vasculature into primary tumor sites in response to CCL2. Classical monocytes are capable of producing tumoricidal mediators, but are likely reprogrammed within the tumor microenvironment to limit their cytotoxicity. Tumor-educated monocytes differentiate into TAMs or moDCs. Monocyte-derived TAMs facilitate tumorigenesis by promoting immune suppression (inhibition of CD8 + T cell recruitment/activities and recruitment of Treg), ECM remodeling, angiogenesis, and tumor cell intravasation into the vasculature. Tie-2 + monocytes/M s display pro-angiogenic functions within primary tumors, although a role for nonclassical patrolling monocytes (PMo) in primary tumors remains unclear. In lung metastatic sites, classical monocytes are recruited in a CCL2-dependent manner, promote metastatic seeding, and have similar protumoral effects. PMo home to tumor metastases in a CX3CL1/CX3CR1-dependent manner, where they engulf tumor material and produce chemokines that stimulates recruitment of cytotoxic NK cells. Tumor-derived microparticles within the vasculature expand bone marrow pools of PMo to increase immune surveillance. Tumor immune evasion within primary and metastatic sites requires protumoral signals to outweigh ongoing attempts by the host immune system to prevent tumor growth monocyte biology, but are typically unable to selectively target monocytes without having concomitant and direct effects on other mononuclear phagocytes. Interpretation of these tools becomes even more challenging in inflamed tissues, including cancer, where rates of monocyte recruitment and differentiation are higher. One strategy to reduce challenges associated with definitively discriminating monocytes and monocyte-derived M s/DCs is to refer to all cells with monocytic origins as "monocyte-derived cells". 41,50 This nomenclature would provide unity in the field regarding distinctions between mononuclear phagocyte subsets, particularly in tissues and inflammatory settings.
Throughout this review, we will use this nomenclature where appropriate, specifically when evidence is provided that the cells of interest are derived from monocytes, but their precise classification is unclear.

FUNCTIONS IN CANCER
Monocytes can display diverse functions at different stages of tumor growth and progression (summarized in Figure 1). Phenotypically simi-lar monocytes can even appear to perform opposing roles due to differences in cancer type/tissue of origin, subtle differences in tumor microenvironment, stage of tumor growth, and experimental model (summarized in Table 1). Here, we discuss current evidence of the functions that different monocyte subsets can perform in cancer and highlight topics that need further investigation in monocyte cancer biology.

Monocyte recruitment to tumors
Monocytes appear to be recruited throughout tumor progression, including during early stages of tumor growth 48,51 and establishment of distal metastases. 2,3,52 In multiple models, CCL2 has emerged as the primary mediator of monocyte recruitment. CCL2 expression increases with neoplastic progression in both human and mouse models of colitis-associated colorectal cancer. 53 54 and CCL2 production increases in the spleens of tumor-bearing mice. 55 The relative contribution of splenic monocytes remains unclear, as bone marrow monocytes were shown to have a competitive advantage in migration to tumors in adoptive transfer studies and photoconversion of bone marrow and splenic monocyte reservoirs. 56 These differences could reflect anatomical differences in tumor microenvironment (orthotopic lung tumors vs. subcutaneously implanted lung tumors) or suggest that even low levels of splenic monocyte recruitment can have significant impacts on tumor progression. Additional work is needed to better understand the precise trafficking of monocytes between organs such as bone marrow, blood, and spleen, as well as lymphatics and lymph nodes during tumor progression.

Direct tumoricidal functions
Monocytes appear to have the cellular machinery to directly kill malignant cells by cytokine-mediated induction of cell death and phagocytosis. Most tumoricidal activity has been demonstrated in vitro, thus whether monocyte-mediated killing is part of the in vivo antitumoral response during cancer progression needs further exploration.
Peripheral blood monocytes exposed to IFN-or IFN-produce the protein TRAIL, which is able to induce cell death in TRAIL-sensitive cancer cells. 57 However, many cancer cells are resistant to TRAILmediated apoptosis and TRAIL can instead stimulate secretion of protumoral cytokines such as CCL2 and IL-8. 58 Monocytes can also induce cancer cell death through Ab-dependent cellular cytotoxicity, which both CD14 + and CD16 + monocyte subsets have the capacity for. 59,60 CD16 + monocytes require contact with tumor cells and TNF-signaling for induction of tumor cell cytolysis. 60 Monocytes collected from peripheral blood or ascites fluid of human ovarian cancer patients display reduced capacity for Ab-dependent cytolysis and phagocytosis of tumor cells upon activation in vitro. 59 In renal cell carcinoma patients, peripheral blood monocytes secrete factors that promote tumor cell invasion in vitro. 61 Consequently, malignant cells may be able to coerce monocytes to adopt a phenotype that supports tumorigenesis, 62,63 thereby overpowering any programmed tumoricidal activities. Tumor-derived exosomes also expand bone marrow pools of patrolling monocytes, which appears to initiate an immune surveillance cascade that prevents metastatic seeding. 70

Interactions with lymphocytes
Monocytes and monocyte-derived cells interact with adaptive immunity by directing the recruitment and function of lymphocytes within the tumor microenvironment through paracrine signaling, as well as by serving as Ag-presenting cells. 14

Lymphocyte recruitment
Monocyte recruitment into tumors appears to be negatively associated with infiltration of cytotoxic CD8 + T cells. In renal cell carcinoma patients, circulating monocytes produce higher levels of many inflammatory cytokines and chemokines compared to those in healthy individuals, including TNF-, IL-1 , IL-6, and CCL3, supporting their likely role in modulating downstream immune responses. 61 Inhibition of CCR2 or CSF1R prevents monocyte-derived cell accumulation in murine pancreatic, liver, and melanoma tumors, which is associated with infiltration of more CD8 + T cells and reduced tumor growth. 72,73 Classical monocytes appear to be the primary precursor of immunosuppressive monocyte-derived cells, at least in melanoma tumors, as CCR2 is required for their accumulation. 74 CCR2 does not appear to be required for their immunosuppressive function on a per cell basis and instead iNOS and arginase are required for inhibition of tumor-specific CD8 + T cell proliferation. 74 In further support of CD8 + T cells being downstream targets of classical monocytes, CCR2 inhibition has no effect on growth of hepatocellular carcinomas in mice lacking CD8 + T cells. 73 Interestingly, one report found that CCR2-deficient mice have more tumor-infiltrating CD4 + and CD8 + T cells, but no differences in the T cell composition of tumor-draining lymph nodes. 75 In these studies, growth of subcutaneously implanted melanoma (B16) and lung (3LL) tumors was not affected by CCR2 deficiency, suggesting that some tumors may rely less on monocytic input for generating antitumoral immunity.
Clinically, intratumoral CCL2 immunostaining negatively correlates with the number of CD8 + T cells in patients with hepatocellular carcinoma 73 and pancreatic cancer patients with tumors expressing high CD8 and low CCL2 display significantly better survival. 76 Absence of immunosuppressive monocytic cells appears to be important for response to checkpoint immunotherapies, as the frequency of CD11b + CD14 + HLA-DR lo cells was higher in metastatic melanoma patients who did not respond to anti-PD-1 immunotherapy compared to those that did respond and correlated with overall survival. 77 Within the tumor microenvironment, monocyte-derived cells can also produce factors such as CCL5 that recruit immunosuppressive regulatory T cells (Tregs). 78 69 In human non-small cell lung cancer tumors, CD16 + monocytes are lower in the tumor microenvironment compared to adjacent nonmalignant tissue, which is associated with reduced infiltration of CD16 + NK cells. 38 These findings suggest that the relationship between nonclassical monocytes and NK cells first observed in mouse models of experimental metastasis may be relevant to human solid tumors, although this and interactions with other immune cells requires further investigation.

Ag presentation
Monocytes can differentiate into M s or DCs after entering tissue (discussed below), but they are also able to enter tissues such as skin and draining lymph nodes without differentiation. 35

Angiogenesis
Induction tumors. 92,93 The relationship between circulating Tie-2 + monocytes and intratumoral pro-angiogenic TAMs (including Tie-2 + M s) has not been directly investigated, although Tie-2 + M s also have significant effects on tumor vasculature that enables dissemination of metastatic cancer cells. 94 The protumoral function of Tie-2 + monocytes/M s appears to extend beyond regulating angiogenesis to include IL-10mediated immune suppression. 95 Anti-angiogenic therapies have been successful both in preclinical models and in clinical trials; however, they often fail to produce durable responses. 96 During treatment with anti-VEGF therapies, tumors increase production of CX3CL1, which recruit pro-angiogenic monocyte-derived cells that drive VEGF resistance. 93,97 In patients with colorectal cancer, treatment with the anti-VEGF Ab bevacizumab increases intratumoral and plasma levels of SDF-1 , a leukocyte migratory factor. 98 Refractoriness to anti-VEGF therapy may be prevented by inhibiting monocyte recruitment, which has been demonstrated experimentally using clodronate liposomes and anti-Gr-1 Ab. 99,100 These results indicate that targeting both the vasculature directly and accessory cells such as pro-angiogenic monocytes/M s is a promising strategy for generating more durable responses to anti-angiogenic therapies.  with CD90-expressing tumor cells. 105 In the metastatic lung, a pop-ulation of Ly6C hi CD11b hi cells derived from classical monocytes that appear within 18 h appears to serve as precursors for mature TAMs. 52 Though these cells are morphologically and transcriptionally distinct from both blood monocytes and TAMs, they also appear to be heterogeneous, therefore requiring further investigation. There may be an important spatiotemporal component to monocyte differentiation, as

Differentiation into tumor-associated M s and DCs
TAMs derived from recently arriving monocytes are more frequently found in collagenous stromal tumor regions and later found in perivascular regions, where they regulate vascular permeability and tumor cell intravasation. 90,106 Taken together, these results suggest that the differentiation trajectory of recruited classical monocytes is complex, depending on spatial and temporal heterogeneity within the tumor microenvironment.
Monocyte-derived TAMs are generally believed to be protumoral, enabling both primary tumor growth and seeding of metastatic tumor cells. 2,48 TAMs appear to be phenotypically and functionally different from M s found in healthy tissues, 48 shifting the tumor immune microenvironment to support immune evasion, angiogenesis, and metastatic outgrowth. However, this may be more complex and depend on factors such as spatial positioning within the tumor. For example, a higher density of M s at the tumor front in colon cancer patients was associated with a higher density of lymphocytes and improved survival. 107 For discussion of M functions within the tumor microenvironment, we refer the reader to excellent reviews elsewhere. 108,109 DCs make up a minor population of tumor-infiltrating myeloid cells, with moDCs representing an even smaller fraction. 4 The cues that govern monocyte fate decisions between remaining a monocyte, differentiation into a M or DC, or undergoing programmed cell-death remain are not fully understood.
In a recent study of human breast cancer intratumoral heterogeneity using scRNA-Seq, the authors inferred a monocytic activation gene signature that reflected a trajectory from blood monocytes to intratumoral monocytes, and from that to other myeloid compartments such as DCs and TAMs. 39 Similar types of analyses in both mouse models and human cancers will be critical for improving our understanding of monocyte fates within tumors.

Monocytes as cancer diagnostics
Early diagnosis remains one of the primary challenges in oncology, with most tumors diagnosed at Stage III or IV when therapies are often less effective promoting tumor regression. Current work seeks to identify novel biomarkers that enable earlier cancer diagnosis or improved patient treatment, with peripheral blood representing a desirable sample site due to its ease of access. The absolute frequency of monocytes in peripheral blood is associated with survival in B cell lymphoma 112 and locally advanced cervical cancer. 113 More commonly, the ratio of lymphocytes to monocytes has emerged as a prognostic factor, including for B cell lymphoma, 114  In patients, high tumor expression of CX3CL1 is associated with infiltration of CD8 + T cells and NK cells, as well as better clinical prognosis in gastric and breast cancer patients. 127,128 These results indicate that CX3CR1 may be a promising therapeutic strategy, but additional understanding of CX3CR1 in the context of cancer is needed and targeting this pathway will likely be challenging.
The SDF-1 /CXCR4 pathway may function in a similar manner to CX3CL1/CX3R1 signaling, although CXCR4 is more broadly expressed among hematopoietic cells. 86,129 Inhibition of CXCR4 with the small molecule AMD3100 has improved the efficacy of antiangiogenic therapies, purportedly through blocking recruitment of pro-angiogenic monocytes. 129 inhibition. 140 Recent work demonstrated that the efficacy of CSF1R inhibitors may be limited by a concomitant increase in recruitment of protumoral immunosuppressive neutrophils, which can be prevented by co-administration of a CXCR2 inhibitor. 141 These studies highlight potential obstacles facing clinical translation of CSF1R inhibitors, while demonstrating that combination therapies can help overcome these limitations to improve long-term outcomes.

Additional therapeutic directions
The divergent role of monocyte subsets in cancer progression is per- Results from this clinical trial will be informative for future directions in clinical monocyte transfer therapies.
Targeting monocytes may be also a promising strategy to mitigate adverse events in patients being treated with existing cancer therapies. For example, in a humanized mouse model of leukemia, monocytes are the primary source of IL-1 and IL-6 released during the onset of cytokine release syndrome following CAR T cell therapy. 144 Depletion of monocytes with CD44-directed CAR T cells or clodronate liposomes prevented cytokine release syndrome. 144 The efficacy of CAR T cell therapies can also be limited by concomitant induction of immunosuppressive myeloid cells in some cancers such as sarcomas. 145 Co-administration of all-trans retinoic acid reduces the abundance of immunosuppressive monocytes in blood and enables antitumor immunity by sarcoma-targeted CAR T cells. 145 Radiation is a major source of toxicity for cancer patients treated with radiation therapy and bone marrow-derived monocytes and M s are critical for regeneration of the nervous system after radiationinduced damage. 146 Administration of G-CSF to mice with focalbrain irradiation injuries increases infiltration of monocytes/M s and improves functional neural repair. 146 These examples highlight the potential value of monocyte-directed adjuvant therapies for improving cancer treatment.

FUTURE DIRECTIONS
Although many important functions have been associated with monocytes in the progression of cancer, the cues that regulate their fate and differentiation into protumoral or antitumoral cells remain incompletely understood. For example, classical monocytes primarily differentiate into pathogenic TAMs, yet they can also differentiate into moDCs that are required for effective adaptive immune responses.
Patrolling monocytes potently prevent tumor metastasis within the lung, yet they also appear able to support angiogenesis within primary tumors. Whether these functions are entirely governed by microenvironmental cues within tumors or preprogrammed in specific monocyte subsets remains unclear. A better understanding of how to shift the balance toward monocyte fates that aid in antitumoral immunity will be critical for the design of more effective immunotherapies.