Functional significance of mononuclear phagocyte populations generated through adult hematopoiesis

Review of key features of differentiation and function of mononuclear phagocytes, which comprise the non-resident monocyte/macrophage populations in the body.

Abstract

Tissue homeostasis requires a complete repertoire of functional macrophages in peripheral tissues. Recent evidence indicates that many resident tissue macrophages are seeded during embryonic development and persist through adulthood as a consequence of localized proliferation. Mononuclear phagocytes are also produced during adult hematopoiesis; these cells are then recruited to sites throughout the body, where they function in tissue repair and remodeling, resolution of inflammation, maintenance of homeostasis, and disease progression. The focus of this review is on mononuclear phagocytes that comprise the nonresident monocyte/macrophage populations in the body. Key features of monocyte differentiation are presented, focusing primarily on the developmental hierarchy that is established through this process, the markers used to identify discrete cell populations, and novel, functional attributes of these cells. These features are then explored in the context of the tumor microenvironment, where mononuclear phagocytes exhibit extensive plasticity in phenotype and function.

Introduction

Evidence for innate immunity can be found throughout the plant and animal kingdoms [1]. Macrophages have long been front and center as major components of the innate immune system as a result of their ability to coordinate directional movement with the detection and elimination of foreign entities. Importantly, macrophages also facilitate additional outcomes that reflect the full complement of their functional profile, including tissue repair and remodeling, resolution of inflammation, maintenance of homeostasis, and disease progression [2–6]. The accumulation of mononuclear phagocyte populations with distinct functional capabilities at diverse anatomical sites is a tightly regulated process that is controlled at many levels. Our goal in this review is to incorporate recent advances in monocyte differentiation and trafficking into a framework of established models of hematopoiesis and blood monocyte activity. Our focus will be on cells of the MPS, defined broadly as a heterogeneous population of monocyte lineage-committed cells that comprise the nonresident monocyte/macrophage populations in the body [7–9]. These are distinct from tissue resident macrophages, which have recently been shown to be derived predominantly during embryonic development [10–12]. A number of recent reviews have addressed the origin and maintenance of these tissue resident macrophages [13, 14]; consequently, they will only be covered in this review as they impact the accumulation, differentiation, and/or function of the nonresident MPS populations. As the majority of work characterizing the differentiation and function of mononuclear phagocytes has been performed in murine models, the nomenclature and data described in this review reflect mouse terminology unless noted otherwise.

ESTABLISHMENT OF MONONUCLEAR PHAGOCYTE POPULATIONS THROUGH ADULT HEMATOPOIESIS

MPS: lineage commitment and the acquisition of lineage-specific markers

The concept that tissue macrophages are derived from multiple organs was developed in the 1960s from studies investigating early-stage monocyte differentiation, the kinetics of tissue macrophage repopulation by circulating monocytes, and macrophage proliferation in tissues [15, 16]. Subsequent advancements in the use of molecular genetics and reporter mice have allowed precise fate mapping of monocyte lineage cells, leading investigators to conclude that a large proportion of tissue resident populations, including those in the pancreas, liver, brain, and skin, is derived from embryonic yolk sac progenitors [10, 11]. Langerhans cells in the epidermis were also shown to originate from embryonic tissue, predominantly from the fetal liver [12]. Indeed, there is now strong evidence that the steady-state maintenance of tissue macrophage populations is not a result of the recruitment of circulating macrophages but rather, proliferation of these resident cells in response to CSFs and cytokines (refs. [17, 18]; reviewed in ref. [13]). These data have contributed to a fundamental distinction in macrophage populations and function. Whereas tissue homeostasis is maintained by resident macrophages established during development, macrophages generated from adult hematopoiesis participate predominantly in response to tissue insult and disease resolution. Resident macrophages can be distinguished from macrophages that enter tissues from the circulation because they tend to express higher levels of the G protein-coupled receptor F4/80 [10].

The MPS, comprised of monocyte progenitors in the BM, monocytes in circulation, and terminally differentiated macrophages at peripheral sites, represents a continuum that is dynamically controlled through proliferation, differentiation, maturation, and precise trafficking [7, 8, 19]. Mononuclear phagocytes are members of the myeloid lineage, which also includes neutrophils, DCs, and osteoclasts [20]. Historically, it was thought that the initial decisions of lineage commitment were stochastic, involving the antagonism and coordination of transcription factors present in progenitor cells. This process leads to the proliferation and survival of progressively more committed cell populations in response to specific growth factors and cytokines that drive lineage specificity [21–24]. Identification of the CMP and the MDP provided two important advancements to our understanding of this process [25, 26]. Isolation of the CMP distinguished GMPs and MEPs from cells that were committed to becoming lymphocytes (Fig. 1). The subsequent purification of the more restricted MDP population helped to define a precursor from which monocytic cells and DCs were derived.

Figure 1. The development of mononuclear phagocytes.

Lineage-committed mononuclear phagocytes are generated primarily in the BM through the tightly regulated differentiation of HSCs to progressively more committed cell populations. Key steps in this pathway include the formation of the CMP, GMP, MDP, and cMoP populations. The recently identified cMoP represents the least differentiated, committed monocyte population. CSF-1 has been shown to be a critical regulatory factor in the generation of monocyte populations, both in instructing myeloid cell fate through activation of the transcription factor PU.1 and in promoting the differentiation of GMP to mononuclear phagocytes. The transcription factor MafB functions to repress the activity of CSF-1 in the HSC. Sca-1, stem cell antigen-1.

Two recent findings have fundamentally changed our understanding of the way monocyte populations are generated. First, it was shown that CSF-1 (also known as M-CSF), the primary survival, differentiation, and proliferation factor for monocytes and macrophages [27, 28], can instruct myeloid cell fate through promoter activation of the transcription factor PU.1 [29] (Fig. 1). PU.1 is an essential regulator of myeloid lineage commitment [30–32]. In HSCs, PU.1 expression can be induced upon CSF-1 treatment through a process that appears to require down-regulation of the transcription factor MafB [33]. Whereas MafB is expressed in HSCs, its expression is reduced as the cells become more committed toward the monocyte lineage. A second study revealed that CSF-1 can direct lineage commitment of the GMP toward mature monocytic cells (Fig. 1). Together, these studies show that CSF-1 can function in an instructional capacity at different stages of differentiation [34, 35]. Importantly, these data challenge the stochastic model of lineage commitment [36–38]. Second, a population of highly proliferative cells was identified in the BM and spleen that differentiated into monocytes and macrophages but not DCs or granulocytes. Referred to as cMoP, this population is committed down the monocyte lineage and is distinguished from MDP by the absence of CD135 expression [9, 39, 40] (Fig. 1).

As is the case for lineage commitment, monocyte differentiation is controlled by a complex circuitry involving cytokines, transcription factors, and other molecules important for cell survival and function. The progression of monocyte progenitor to monocyte in the BM marks an important stage of differentiation. Hallmarks of this transition include reduced proliferative capacity, commitment to the monocyte lineage, and the regulated expression of surface antigens, the latter conveying functionality in addition to hierarchy [39]. The relative expression of key molecules used to distinguish monocyte progenitors and committed monocytes is presented in Table 1. CD11b (integrin αM) is up-regulated during differentiation of the cMoP population (Fig. 2). Expression and clustering of CD11b can lead to down-regulation of the transcription factor FoxP1, resulting in enhanced surface expression of CSF-1R due to release of FoxP1-induced repression at the promoter [42]. Ly6C, a GPI-anchored surface glycoprotein, is expressed on these early monocytes but then becomes down-regulated during subsequent differentiation [43, 44]. Ly6C thus helps distinguish monocytes from more differentiated macrophages.

Table 1. Molecules Used to Distinguish Monocyte Populations.

Molecule	Classification	MDP	cMoP	Ly6Chigh monocyte	Ly6Clow monocyte
CD135 (Flt-3; fms-related tyrosine kinase-3)	RTK	+	–	–	–
Ly6C	Glycoprotein	–	+++	+++	+
CD11b	Integrin αM	–	–	+	+
NR4A1	Orphan nuclear receptor	Low/–	Low/–	Low	High
CX3CR1	Chemokine receptor	+	+	+	++
CCR2	Chemokine receptor			+	–
CD62L	Adhesion molecule			+	–

See refs. [39–41].

Figure 2. Monocyte differentiation and trafficking.

Ly6C^high monocytes retain proliferative potential in the BM and can differentiate or transmigrate to the blood. Two populations of blood monocytes persist in steady state and are distinguished by the relative expression of several functionally important molecules. The patrolling nature of Ly6C^low monocytes on the vascular endothelium is an additional characteristic of this subset. Ly6C^high and Ly6C^low populations can infiltrate peripheral tissues under the appropriate stimuli. Macrophages derived through the differentiation of recruited monocytes and those generated through the local proliferation of resident macrophages established during embryonic development comprise the tissue macrophage populations.

Cytokines and cytokine receptors play a key role in monocyte differentiation; the receptors not only help to distinguish monocyte populations experimentally but also dictate functional capability through activation of signaling pathways that regulate motility, chemotaxis, and gene expression [40]. Monocytes express CX3CR1, a prototypical seven-transmembrane G-protein-coupled chemokine receptor. Signaling through CX3CR1 is critical for blood monocyte survival [45]. There is some evidence to suggest that CX3CR1 can promote adhesion by binding to its ligand fractalkine, which is present on the surface of activated endothelial cells; however, the relevance of this activity in vivo needs to be confirmed [46, 47]. The chemokine receptor CCR2 is also expressed on a subset of monocytes, serving as a primary marker for the classical inflammatory monocyte. This molecule is noted primarily for its ability to promote motility toward CCL2 (also known as MCP-1) in response to inflammation at peripheral sites [48–50]. This is not exclusive, as other MCP family members can function as CCR2 ligands as well [51]. CD62L (L-selectin) is expressed in the same population of monocytes that display high levels of Ly6C and CCR2 [52, 53]. The role of L-selectin in mediating leukocyte rolling is best characterized for lymphocytes. However, evidence suggests that it is also functional on monocyte populations [53–56].

It is important to note that, whereas all of these cell-surface proteins are present on monocytes, their expression is not limited to these cells. For example, CD11b can be found on neutrophils, B cell subsets, and NK cells; Ly6C is expressed on CD8⁺ T lymphocytes; and CX3CR1 is expressed on NK cells and T cells [57–60]. This underscores the need to identify additional, more specific markers that can better distinguish monocyte populations that share unique functional attributes.

Ly6C^high and Ly6C^low monocyte subsets in the blood

Two populations of monocytes can be distinguished in the blood based on the level of expression of Ly6C (Fig. 2). Experiments conducted over one decade ago, using reporter mice with a GFP transgene inserted into the CX3CR1 locus (CX3CR1^GFP/+), showed that there were two discrete populations of circulating murine monocytes [41]. Ly6C^highCX3CR1^lowCCR2⁺CD62L⁺ monocytes were short lived relative to Ly6C^lowCX3CR1^highCCR2⁻CD62L⁻ monocytes and preferentially homed to inflamed sites, where in this study, they were seen to differentiate into functional DCs. A small percentage of Ly6C^low monocytes also expressed markers for DCs, although it was not confirmed that these cells were functional. The elevated expression of CX3CR1 on the Ly6C^low subset was shown to be important for tissue engraftment. The separation of monocyte populations, based on Ly6C expression, has gained considerable traction over the last decade. In some instances, these populations have been designated Ly6C⁺ and Ly6C⁻, whereas in others, they are called Ly6C^high and Ly6C^low. For consistency in this review, we will use the Ly6C^high and Ly6C^low nomenclature.

The results from this work provided the impetus for subsequent investigations into the relationship of Ly6C^high and Ly6C^low monocyte populations. Early studies describing Ly6C had suggested that there was a precursor-product relationship between Ly6C^high and Ly6C^low cells, as BMs treated with the differentiation factor CSF-1 resulted in loss of Ly6C expression [61]. More recent studies using monocyte depletion techniques and reporter mice have confirmed that the Ly6C^low monocyte pool is established through differentiation of Ly6C^high monocytes in vivo [62, 63]. With the use of CX3CR1 knock-in fluorescent reporter mice, Yona et al. [64] were able to track steady-state monocyte differentiation as a function of CX3CR1 and Ly6C. These studies showed that the generation of a Ly6C^low blood monocyte pool required the emigration of Ly6C^high monocytes from the BM, followed by differentiation of this population (Fig. 2). Ly6C^high monocytes were shown to have a shorter half-life (20 h) in circulation compared with Ly6C^low monocytes (2.2 days). Upon depletion of Ly6C^high monocytes, however, the half-life of Ly6C^low monocytes became significantly greater (up to 11 days). Combined with results from an earlier study showing that the Ly6C^low population could be reduced by neutralization of the CSF-1R [65], Yona et al. [64] hypothesized that Ly6C^high monocytes function as a CSF-1 sink that limits the availability of this survival factor to Ly6C^low monocytes. However, there is also evidence showing that mature liver and splenic macrophages clear a large percentage of CSF-1 in circulation, leading to the relatively short serum half-life of CSF-1 (∼10 min) [66]. This raises the possibility that Ly6C^high monocytes and mature macrophages may contribute to the regulation of CSF-1 levels in the blood.

Given the substantial, functional diversity exhibited by the Ly6C^high and Ly6C^low circulating monocyte populations, it is reasonable to consider that the MPS has evolved to maintain both subsets concurrently. This is corroborated by the finding that the gene-expression profiles of murine Ly6C^high and Ly6C^low monocyte populations differ significantly between one another but display a high degree of homology with the two parallel monocyte subsets in humans, suggesting an evolutionary conservation within each subset [41, 67].

Function and differentiation of Ly6C^high and Ly6C^low monocytes

The trafficking of Ly6C^high blood monocytes into peripheral tissues is integral to the inflammatory response [50, 68]. This progressive cascade, resulting in the translocation of suspended blood monocytes into peripheral tissues, is dependent on the ability to coordinate extracellular stimuli with signaling pathways governing cell adhesion and migration. The first step in this process requires adhesion and rolling of the monocytes along the surface of activated endothelium through selectin-mediated interactions [49, 56, 69, 70]. The subsequent activation of chemokine receptors, such as CCR2, in response to inflammatory factors, such as CCL2, leads to an increase in the affinity and avidity of integrins present on the surface of the monocytes [71]. Enhanced integrin engagement permits the crawling phenotype required during the initial stages of monocyte migration across the vascular endothelium [72].

Ly6C^low blood monocytes can be distinguished from Ly6C^high monocytes by their migratory capability on the vasculature. Auffray and colleagues [73] first characterized Ly6C^low monocyte activities in the blood using CX3CR1^GFP/+ mice. They documented that this subset seemed to crawl along vessels in the dermis and mesentery in steady state, appearing as if the monocytes were “patrolling” the endothelium (see Fig. 2). Migration along the endothelial barrier was seen to occur counter to blood flow in various patterns and rarely resulted in extravasation. The average rate of movement was 12 μm/min, which is relatively slow in relation to the average size of monocytes (20 μm). Intravital microscopy illustrated beautifully that a significant proportion of the endothelial surface could be covered in as little as 1 h through the patrolling activity of the full complement of adherent Ly6C^low monocytes in the vessel. LFA-1 (integrin αLβ2) and CX3CR1 were critical for this crawling behavior under steady-state conditions [74]. Upon induction of inflammation, the Ly6C^low monocytes were seen to extravasate rapidly into the tissue before the arrival of neutrophils and Ly6C^high inflammatory monocytes.

A second function has now been attributed to Ly6C^low monocytes that is dependent on the patrolling activity discussed above. Upon ligation of TLR7 on the surface of these cells, the dwell time on the endothelium was seen to increase from 9 to 25 min [75]. This resulted in the local accumulation of neutrophils, which initiated endothelial cell killing. The Ly6C^low monocytes then functioned to remove the cellular debris from the area. Preliminary evidence from this study indicated that this process entailed specificity, as TLR7 but not TLR4 ligands resulted in this patrolling activity.

The dichotomy in blood monocytes arising from the presence of distinct populations distinguished by Ly6C expression levels is expanded significantly once the monocytes translocate into tissues. The need to perform diverse effector functions is most likely the driving force behind the expanded profile of tissue macrophage phenotypes, and the source of these differentiated cells remains a topic of intense study. Jakubzick et al. [76] showed that Ly6C^high monocyte populations could be found outside of the vasculature in the lung and skin and that they were capable of migrating from the tissue to lymph nodes via the lymphatics. However, these cells failed to acquire a phenotype consistent with that of mature macrophages during this process, suggesting that the infiltrating Ly6C^high monocytes might not be responsive to environmental cues that drive differentiation. A second study showed that Ly6C^high and Ly6C^low monocytes in the lung could differentiate into pulmonary DCs during steady state and inflammation [77]. However, only the Ly6C^low monocytes were seen to differentiate into lung macrophages, either through direct recruitment or following the loss of Ly6C. Finally, in the fibrotic kidney, recruited Ly6C^high monocytes from the BM underwent a loss in Ly6C surface expression, after which, they persisted as three populations that were distinguishable by pro- or anti-inflammatory transcriptional profiles [78].

The heart has also served as a model to study monocyte trafficking, given that it harbors MPS populations that maintain steady state, assist in the resolution of tissue following damage, and contribute to pathology [79, 80]. A study conducted in the healing myocardium following infarct revealed that recruitment of Ly6C^high cells preceded the arrival of Ly6C^low monocytes [81]. The temporal regulation of this process was mediated through a shift in chemokine expression in the myocardial tissue. During the early phase, elevated levels of the proinflammatory chemokines CCL2 and MIP-1α resulted in the recruitment of Ly6C^high monocytes. These factors were subsequently down-regulated, concomitant with an increase in fractalkine expression, leading to the second phase, marked by Ly6C^low monocyte recruitment. These data illustrate how the specificity of chemokine-mediated signaling impacts monocyte recruitment. Advancements in imaging and nanoparticle research, along with our understanding of the regulatory mechanisms governing monocyte trafficking and macrophage polarization, bolster the potential of harnessing mononuclear phagocyte biology for the treatment of heart disease [82]. However, additional studies are needed to examine the yet-to-be-defined contribution of the two monocyte subsets to cardiac macrophage populations and to determine if similar processes occur in human patients during myocardial infarction.

Clearly, we are in the early stages of understanding the full array of mechanisms that regulate the movement and function of Ly6C^low monocytes on the vascular endothelium at peripheral sites, as well as the signals that induce transmigration of Ly6C^high and Ly6C^low monocytes into the tissue parenchyma. Once in the tissue, the studies described above underscore the critical role played by the microenvironment in regulating the outcome of the differentiation process, the functional capacity of the fully differentiated cells, and the persistence of these cells in the tissue.

Fundamental regulators of monocyte development

Phenotypic plasticity and efficient cell motility are hallmarks of mononuclear phagocytes. The regulation of these attributes is accomplished through numerous cell intrinsic and extrinsic mechanisms. An assortment of cytokines, chemokines, growth factors, lipids, and ECM elements contributes to the aforementioned trafficking of Ly6C^high and Ly6C^low monocytes into peripheral tissues, their differentiation, and the acquisition of macrophage activation states. In addition to these factors, mononuclear phagocyte phenotype and function can be shaped by other cell lineages present in the tissue microenvironment. For example, factors derived from Th₁ and Th₂ lymphocyte subsets are instrumental in driving M1 (inflammatory) and M2 (regenerative) macrophage polarization, respectively [83–85]. The reciprocal relationship, where subsets of macrophages also shape lymphocyte activity, demonstrates the complexity of the pathways that dictate the polarity of innate and adaptive immune cell populations [86, 87].

The dependence on CSFs for differentiation, survival, and proliferation is one constant across all mononuclear phagocytes [88]. This is true in the BM, blood, and peripheral tissues, where regulation of MPS populations is achieved through engagement of cell-surface receptors, subsequent activation of intracellular signaling cascades, and alterations in transcriptional signatures through activation or repression of specific genes. For cells of the MPS, the principal regulatory factors controlling differentiation are CSF-1 and GM-CSF. CSF-1 is ubiquitously expressed in steady state and directs monocyte lineage commitment in the HSC and later, during differentiation when monocytic cells branch from the GMP [33, 34] (Fig. 1). The critical role of CSF-1 in controlling fundamental aspects of macrophage biology, such as survival, proliferation, activation state, and migration, is well established [27, 28]. Functionality is achieved through activation of its cognate RTK, which leads to receptor autophosphorylation and downstream signaling mediated by Src family kinases, PI3K and the ERK pathway [89–91]. The severe abnormalities observed in mice that have impaired signaling through this pathway underscore its prominent role in tissue homeostasis [92]. For example, mice with a homozygous null mutation for Csf1 (CSF1^op/op) exhibit a deficiency in bone reabsorption by osteoclasts (bone macrophages) and display compromised hematopoiesis. Csf1R null mice present with a reduction in microglia (brain macrophages) and Langerhans cells, in addition to the defects observed in op/op mice [93]. The more severe phenotype observed in the receptor knockout is attributed to a potential compensatory role for IL-34, which is an additional CSF-1R ligand [94–97]. CSF-1 has been shown to regulate the persistence of blood monocyte populations as well. In MacGreen mice, which allow fluorescent tracking of all mononuclear phagocytes as a result of the expression of enhanced GFP under the control of the Csf1R promoter [98], systemic treatment with antibodies against the CSF-1R had no significant impact on monocytopoiesis or the overall percentage of GFP⁺CD11b⁺ populations in the BM, blood, and spleen [65]. However, mice that received the antibody exhibited a lower proportion of GFP⁺F4/80^high and GFP⁺GR-1⁻ (GR-1 comprises Ly6G and Ly6C) cells in the blood. Given that the differentiation of monocytes to macrophages correlates with increased expression of F4/80 [43, 99], the authors concluded that treatment with antibody against the CSF-1R blocked the maturation of GR1⁺ monocytes into GR-1⁻F4/80^high monocytes and that signaling through the receptor was necessary for this maturation [43, 61–63, 65].

The role of GM-CSF as a regulator of steady-state myelopoiesis is considerably more ambiguous than CSF-1, as the data indicating that it functions as a proliferative and differentiation factor in vitro do not appear to be recapitulated in vivo [100, 101]. For example, most myeloid cell populations are not compromised significantly in steady state with the ablation of GM-CSF in mice, with the exception of the alveolar macrophage pool [102]. Evidence indicates that GM-CSF plays a more prominent role during inflammation, when its levels are seen to rise [103]. Ligand binding to the GM-CSF transmembrane receptor activates the downstream signaling molecules JAK2/STAT5, ERK, and PI3K [89].

Whereas CSF-1 and GM-CSF induce the differentiation of BM monocyte progenitor cells to macrophages in vitro [104], studies have revealed significant differences between the activities of these two factors. CSF-1 is a more potent and specific monocyte-differentiating factor than is GM-CSF, as GM-CSF also induces the generation of granulocytic myeloid cells. The cytokines produced by BM cells cultured in the presence of these factors also suggest that GM-CSF induces a more inflammatory phenotype, whereas CSF-1 potentiates a more restorative, growth-oriented phenotype [105]. These data and others have contributed to a model, wherein GM-CSF is thought to be expressed predominantly during inflammation and induces the early stages of myeloid differentiation in the BM. Conversely, CSF-1 is ubiquitously expressed and is critical for the differentiation of Ly6C^high monocytes to Ly6C^low monocytes, as well as for tissue-resident macrophage function [103].

In addition to a role for CSF-1 and GM-CSF in monocyte differentiation, recent studies have identified a critical role for the transcription factor NR4A1 (Nur77) in this process. NR4A1, a member of the NR4A subfamily of orphan nuclear receptors, appears to be required for the generation of Ly6C^low monocytes [106] (see Figs. 1 and 2). NR4A1 mRNA is expressed at significantly higher levels in BM Ly6C^low monocytes compared with Ly6C^high, MDP, and CMP populations. Protein levels are also higher in Ly6C^low monocytes relative to the Ly6C^high cohort. Nr4a1^−/− mice harbor significantly lower levels of Ly6C^low monocytes in the BM, blood, and spleen compared with wild-type mice, whereas the number of Ly6C^high monocytes in these tissues is similar in both genotypes. These findings suggest that along with CSFs, NR4A1 is an essential regulatory factor in the generation of Ly6C^low monocytes. The potential, functional implications of NR4A1 expression in these cells have been discussed in a recent review [107].

SOLID TUMORS AS ORCHESTRATORS OF MONOCYTE DIFFERENTIATION AND FUNCTION

The remainder of this review will explore MPS differentiation and function in the microenvironment of solid tumors. We have elected to use this model for three reasons. First, monocytes and macrophages accumulate substantially in many tumor types in response to an array of intracellular signals, soluble factors, and physical cues. Second, these cells exhibit exquisite plasticity of phenotype and function while in the tumor, again, in response to many of these same cues. Third, solid tumors are established after embryogenesis and thus, are subject to accumulation of mononuclear phagocytes that are derived during adult hematopoiesis. Thus, the tumor microenvironment serves as an ideal tissue in which to address the transition from blood monocytes to differentiated macrophages with varied functions and activation states.

Mononuclear phagocyte influence on tumor progression

The MPS in mice and humans has been implicated in all stages of tumor development [108, 109]. Monocytes in circulation respond to tumor-derived factors, adhere and transmigrate across the endothelium, and migrate within the tumor bed (Fig. 3). This is a directed response, dependent on the secretion of chemotactic and differentiation/survival factors by cells within the tumor and the ability of the circulating monocytes to respond effectively to these stimuli [115]. Once in the tumor microenvironment, these nonterminally differentiated monocytes are exposed to factors that regulate differentiation and function. The end result of this process is the accumulation of MPS cell populations at various stages of differentiation, each with unique effector capabilities that have been acquired concomitantly with differentiation [116] (Fig. 3).

Figure 3. The regulation of mononuclear phagocyte accumulation and activity in the tumor microenvironment.

The expression of soluble chemotactic factors by tumor-associated cells recruits circulating monocytes and M-MDSCs, which have the potential to differentiate to TAMs in response to tumor-derived factors. CSF-1 can regulate TAM function in the tumor microenvironment in several ways. First, a CSF-1/EGF paracrine loop is established [110, 111]. Second, CSF-1 has been shown to increase angiogenesis by inducing TAM to express VEGF [112]. Finally, CSF-1 can stimulate TAMs to secrete MMPs, which in turn, can cause the release of sequestered growth factors from the ECM [112, 113]. TAMs are also critical for the establishment of TMEM, which promotes intravasation of the tumor cells [114].

Two monocytic cell populations in the tumor have been the focus of extensive research as a result of their propensity to accumulate in tumor-bearing animal models and cancer patients. These populations represent continuums along the differentiation spectrum, with M-MDSCs described as less differentiated, Ly6C^highLy6G⁻ myelomonocytic cells, whereas TAMs are characterized as well-differentiated cells that have low Ly6C and high F4/80 expression [117].

A comprehensive summary of TAM accumulation in human cancers was completed in 2002 across a range of tumor types [118]. This was followed by a complementary summary of less-differentiated myeloid cells several years later [119]. Together, these studies showed that the systemic and local environment created by the developing tumor had a profound influence on shaping mononuclear phagocyte function. In the majority of cases, the accumulation of MPS populations in the tumor microenvironment was seen to correlate with poor patient prognosis or tumor stage. Differentiation status did not in itself correlate with severity of disease. Instead, the dominant influencing factor was the behavior or function of the myeloid cell populations.

Regulation of tumor progression and metastasis by TAMs

TAMs have been shown to function as regulators of tumor control in some instances and tumor progression in others [120–122]. These are highly phagocytic cells that can be induced to destroy tumor cells under the right circumstances. They also function as APCs, thus initiating adaptive immune responses that provide anti-tumor activities. Conversely, TAMs can also play a prominent role in facilitating tumor growth and metastasis. This occurs largely through the secretion of factors that promote tumor cell growth and migration, angiogenesis, and the generation of structures called TMEM, through which metastatic cells are thought to intravasate (Fig. 3) [114, 123]. Condeelis and colleagues [110, 111] have shown that a paracrine loop can develop between breast carcinoma cells and TAMs involving the secretion of CSF-1 and EGF, respectively. This mutual exchange enhances the proliferation and directed motility/invasion of the tumor cells, while providing differentiation and survival signals to the TAMs. TAMs can also stimulate the release and availability of sequestered growth factors through secretion of MMPs [112, 113]. Finally, the ability of tumors to expand in volume is dependent on adequate oxygen perfusion from the vasculature. TAMs secrete high levels of VEGF, a potent angiogenic factor that facilitates this process [112, 121]. By undergoing an effective angiogenic switch, the tumor is able to generate new blood vessels that facilitate its growth and enhance metastasis [124].

The pro- and anti-tumor functions of TAM cannot be fully appreciated without considering the exquisite plasticity exhibited by these cells in the tumor microenvironment. Several investigators have pioneered the separation of these activities into two polarization states defined by gene-expression profiles, cytokine production, and functional interactions with other stromal cells [86, 125, 126]. TAMs that exhibit a profile consistent with anti-tumor effector functions were termed M1 or inflammatory macrophages. These were opposed by the tumor-promoting attributes of M2 macrophages, which promoted angiogenesis and were characterized as having a “wound-healing” phenotype. Whereas these distinctions were instrumental initially, the field has since moderated the distinction between M1 and M2 subsets. Macrophages are now viewed as having the capability of falling anywhere within the continuum of the activation state spectrum and of dynamically moving from one state to the other [127]. Recent studies have attempted to exploit the plasticity of TAMs with the goal of reprogramming these cells to adopt anti-tumor activities [128–130]. More globally, methods are being evaluated in the clinic that effectively prevent or reduce the recruitment of myeloid cells into tumors [131]. In these ways, TAMs are being considered as potentially robust, therapeutic targets that can be manipulated or removed to achieve greater tumor control.

Regulation of tumor progression and metastasis by MDSCs

MDSC gained prominence over a decade ago with the observation that a significant number of cancer patients had evidence of reduced levels of DCs, coincident with increased proportions of a myeloid population that lacked surface antigens commonly found on more differentiated cells [132]. Referred to as Lin⁻ ImCs, this population was distinguished further from DCs by reduced expression of HLA-DR (MHC II). When peripheral blood was sorted and cultured in factors that supported growth of myeloid and erythroid cell colonies, Lin⁻HLA-DR⁻ cells formed 10 times as many colonies as did the Lin⁻HLA-DR⁺ DC population, suggesting that the Lin⁻HLA-DR⁻ ImC population harbored more progenitor cells. The vast majority of blood ImCs had differentiated beyond the earliest CD34⁺ progenitor cell, however, and roughly one-third expressed CSF-1R. This indicated that a significant proportion was of the mononuclear phagocyte lineage. The ex vivo culture of ImCs in CSF-1 resulted in approximately one-third of the cells displaying the human monocyte/macrophage marker CD14⁺, confirming that a portion of the ImC population was composed of mononuclear phagocytes. Functionally, the HLA-DR⁻ population displayed a significantly reduced ability to stimulate allogenic T cells compared with the HLA-DR⁺ DC population.

Around the same time, a population of suppressive cells was identified in mice that had similar features to the cells described above [133]. In this case, tumor-bearing or vaccine-immunized mice harbored decreased CD8⁺ T cell function and number, which correlated with the accumulation of CD11b⁺Gr-1⁺ myeloid cells in secondary lymphoid organs. Systemic GM-CSF was shown to be responsible for the increased proportion of CD11b⁺GR-1⁺ cells. Termed iMacs, this population included mature myelomonocytic (polymorphonuclear cells and monocytes) and immature myeloid cells. Expression of Ly6C, ER-MP58, and CD31 on various subsets of these cells indicated that they had a less differentiated or progenitor cell phenotype. It was known at the time that cytokines produced by the Th₁ and Th₂ subsets of CD4⁺ T lymphocytes elicited distinct responses in macrophages [134]. Th₁ derived factors, such as IFN-γ and TNF-α, induced macrophages to become potent antimicrobial effectors through the up-regulation of molecules for antigen presentation and T cell costimulation. Conversely, a phenotype consistent with wound healing or the resolution of inflammation was a consequence of mediators such as IL-4 secreted by Th₂ cells. Interestingly, exposure of the CD11b⁺Gr-1⁺ iMacs to Th₁ or Th₂ cytokines produced mononuclear phagocytes with divergent effector functions. Cells cultured in the absence of exogenous cytokine or with IL-4 were suppressive, whereas those cultured in Th₁ cytokines displayed increased MHCII expression and cytolytic activity of allogeneic targets in mixed lymphocyte assays. Several conclusions could be drawn from these studies. First, they confirmed that iMacs were not terminally differentiated cells. Second, although they were lineage committed, CD11b⁺Gr-1⁺ iMacs retained functional plasticity. Finally, the data suggested that the default pathway for these cells was aligned with immunosuppression; an enhanced cytolytic response was only achieved upon treatment with Th₁ cytokines.

Subsequent research has focused on the factors that drive MDSC accumulation, the mechanisms of MDSC suppressor activity, the delineation of MDSC populations, and the differentiation of MDSCs in the context of other myeloid lineage cells. The significance of MDSCs to tumor outcomes appears to stem predominantly from their persistence. A systemic milieu of factors secreted by the tumor microenvironment inhibits MDSC differentiation and potentiates their proliferation and survival, thus propagating a continued state of immune suppression [135, 136]. Whereas MDSCs are also generated during infection, they quickly dissipate once the infection is resolved.

The two main MDSC populations identified to date, G-MDSCs and M-MDSCs both have suppressive capabilities, although they use different mechanisms [137, 138]. They are commonly distinguished by the relative level of Ly6C and Ly6G expression. Isolated cell populations that are CD11b⁺Ly6G^highLy6C^low are more granulocytic, suppress primarily through reactive oxygen species mechanisms, and depend on close interaction with T cells for suppressive activity [139]. CD11b⁺Ly6G⁻Ly6C^high M-MDSCs express high levels of arginase, tend to have greater immunosuppressive strength, function independently of contact with T cells, and frequently localize to the tumor microenvironment. Evidence from murine tumor models suggests that that the proportion of each population is variable. Their role in shaping tumor responses through potentiating angiogenesis or blocking cytotoxic T cell activity is viewed broadly as favorable from the perspective of the tumor. Consequently, investigators are currently exploring ways to reduce the number of myeloid cells in the tumor as a mechanism of tumor control [131, 140].

Reconciling mononuclear phagocyte function with molecular subtyping

A maturational hierarchy has not been established for G-MDSCs and M-MDSCs, and no defined differentiation pathway currently places either cell type in line with other mononuclear phagocytes. As a group, they appear to fall in between monocyte progenitors and differentiated tissue macrophages. This underscores the problem of identifying distinct populations of myeloid cells using markers that are present on multiple cell populations. For example, high expression of Ly6C and low expression of Ly6G on CD11b⁺ cells have been used to characterize populations of lineage-committed, undifferentiated mononuclear phagocytes. However, cells defined in this way can exhibit completely opposite functionality within the tumor, with inflammatory monocytes performing classically anti-tumor activities and M-MDSC functioning in a pro-tumor capacity. Even more confounding is the demonstration that inflammatory monocytes can, in fact, play both roles [141–143].

These examples demonstrate that the field is still in its infancy regarding the identification of functional myeloid cell populations through surface antigen expression. Difficulties in reconciling function with cell-surface markers are exacerbated further by the fact that monocyte differentiation is very fluid. For example, there is considerable evidence showing that M-MDSCs can differentiate in the tumor microenvironment to become TAMs [144–148], and in a recent study examining Ly6C^high monocyte differentiation in situ, four mononuclear phagocyte populations were identified that could be distinguished based on MHCII and Ly6C expression [149]. One approach to reconciling this issue has been through gene profiling. With the use of these techniques, mononuclear phagocyte populations have been identified with similar genetic signatures that possess common effector functions. For example, TEMs are a subset of proangiogenic mononuclear phagocytes that exist in tumors and the circulation [150]. Isolated TEMs from either compartment can promote angiogenesis, suggesting that function is, to some extent, fundamentally encoded in these monocyte populations [151]. The gene-expression profile of TEMs was found to be closest to that of TAMs and less similar to the expression profile of MDSCs, endothelial cells, and peritoneal macrophages. Interestingly, TEMs and TAMs showed responsiveness to Th₂ (IL-4) over Th₁ (LPS+IFN-γ) stimuli [152]. TEMs, which are primarily Ly6C^lowCD62L⁻, also display a similar genetic signature with Ly6C^low blood monocytes derived from tumor-free mice, which suggests that Ly6C^low monocytes and TEMs may function similarly in steady state, as well as in tumor-bearing animals.

CONCLUDING REMARKS

It is well appreciated that vertebrate animals have developed exquisite mechanisms to regulate hematopoiesis and thus, generate the repertoire of innate and adaptive immune cells that are needed for survival. This review has reinforced this concept for monocyte/macrophage lineage cells. We are at an exciting place in mononuclear phagocyte research. Genetic manipulation of murine embryos combined with advances in vivo labeling and intravital imaging have led to new perspectives on how macrophage populations arise [153]. Data generated using these approaches have revealed that tissue-resident macrophages are predominantly populated during embryonic development and that they appear to be able to maintain a self-renewing capacity as fully differentiated macrophages [9, 13, 14]. This effectively allows the expansion of peripheral tissue macrophage populations without precursor monocyte seeding. The field has also gained new insight into the migratory behavior of Ly6C^low monocytes in the vasculature and in the process, has further distinguished two primary murine blood monocyte subsets that are capable of entering tissues at any given time [52, 68]. However, there remain considerable challenges to our understanding of myelopoiesis, brought about largely as a consequence of the functional heterogeneity, plasticity, and complex developmental relationships between cells of this lineage. Future studies are needed to clarify the molecular mechanisms involved in mononuclear phagocyte trafficking and the extrinsic and intrinsic factors that regulate the accumulation of distinct monocyte populations. Additionally, advances in the field will require an increased emphasis on reconciling molecular subtyping with mononuclear phagocyte functions.

AUTHORSHIP

M.F.G. researched and wrote a first draft of this manuscript, revised subsequent versions, and generated the figures. A.H.B. researched and substantially revised the manuscript.

DISCLOSURES

The authors declare no conflicts of interest.