The lymphatic vascular system is a hierarchical network of vessels comprised of blind-ended capillary beds and larger collecting vessels that form a unidirectional system draining most soft tissues of the body. Lymphatic vessels are found in all vascularized tissues in vertebrates except the bone marrow (BM) and brain as larger body size necessitates a secondary vascular system to maintain fluid homeostasis [1]. The initial lymphatic capillaries are specifically designed to absorb fluid; accordingly, they are made up of a single layer of overlapping lymphatic endothelial cells (LEC) [2] with button-like intercellular junctions [3], discontinuous basement membrane, and specialized filaments that anchor these vessels to the extracellular matrix [4]. Lymphatic uptake of interstitial fluid, a critical function to maintain the balance of interstitial fluid pressure (IFP), is facilitated primarily by these anchoring filaments and functional “gap sites” in the LEC monolayer [5]. Lymph entering the initial lymphatic capillaries is propelled forward into the larger collecting lymphatics by cyclical tissue deformation such as body movement, breathing, and skeletal muscle contraction [6]. Lymph propulsion through higher order lymphatic collectors and trunks is facilitated by unidirectional valves and contractions of smooth muscle cells [1]. After being filtered through a series of LNs, most proteins, lipids, and macromolecules making up the lymph are ultimately returned to venous circulation through the thoracic duct that empties its contents into the left subclavian vein.

In addition to maintaining IFP, the lymphatic vasculature regulates lipid adsorption in the gut [7], facilitates transports of hormones and cytokines produced in adipose tissue [8], and performs many immune functions including antigen presentation [9] and immune cell trafficking [10]. The lymphatic vessels serve as the primary pathway to transport tissue-absorbed soluble antigens, antigen-presenting dendritic cells, and lymphocytes into regional LNs, which is the first step for mounting an adaptive immune response [1]. During either sterile (i.e., due to injury or tumor) or pathogen-induced inflammation, the necessity to maintain fluid, protein, and lipid balance, as well as to interact with the immune system, is drastically increased. This necessity is likely the primary driving force for the formation of new lymphatics because the needs of an inflammatory site, particularly, under prolonged conditions, may well exceed the functional capacity of the local preexisting vessels. Thus, generation of new lymphatic vessels is an appropriate tissue response to growing fluid imbalance and leukocyte trafficking demands imposed by chronically inflamed sites and tumors.

In recent years, research of the lymphatic vasculature has been greatly facilitated by the discovery of several proteins expressed primarily on LECs. The six most frequently used markers that discern between blood and lymphatic vessels are: vascular endothelial growth factor receptor-3 (VEGFR-3), Prospero-related homeobox-1 (Prox1), a hyaluronan receptor LYVE-1, a mucin-type transmembrane glycoprotein podoplanin, integrin alpha-9, and neuropilin-2 (NRP2).The earliest to be described was VEGFR-3 [11], a tyrosine kinase receptor that is the key protein that regulates lymphangiogenesis in adult [12]. Throughout adulthood, VEGFR-3 is mainly found in LECs except occasional expression in inflamed [13] and tumor blood vessels [14], circulating LEC progenitors [15,16], activated macrophages [17,18,19,20,21,22,23,24], and some malignant cells [25]. LYVE-1 [26] is one of the most specific markers of the adult LEC with limited expression on sinusoidal blood vessels in the liver [27] and spleen [26]. LYVE-1 is absent in resting macrophages, and its expression in activated ones [19,23,28,29,30] might signify macrophage transdifferentiation to LECP [30]. An additional marker that is expressed predominately in LECs is the transcription factor Prox1. This marker, however, is also expressed in non-endothelial cells such as hepatocytes and neurons [31]. Another specific marker is podoplanin (also known as T1α/podoplanin and D2-40), a protein mainly found on the LEC surface with minor expression in a few other cell types [32]. LEC can also be identified by integrin alpha-9 that is expressed on lymphatic but not blood vascular endothelial cells [33]. However, besides vasculature, integrin alpha-9 has a relatively broad expression pattern in mesenchymal, epithelial and other cell types [33,34,35]. Lastly, neuropilin-2 (NRP2), a neuronal semaphorin receptor [36], is expressed predominantly on LECs with minor expression on veins [37]. NRP2 physically interacts with VEGFR-3 and serves as a co-receptor for VEGF-C/-D [38]. Additionally, NRP2 is upregulated in the tip cells of sprouting lymphatics and is required for sprout initiation during postnatal development [39]. Although none of these markers has an exclusive specificity to LEC, when used in combination, they unequivocally distinguish lymphatic vessels from blood vasculature.

The purpose of this review is to discuss the role of macrophages in adult lymphangiogenesis. To better understand this process, the authors would like first to highlight the main points of the embryonic lymphatic development without going through detailed information about this process available in recent reviews published elsewhere [40,41,42]. Here, we will mainly focus on two aspects of embryonic lymphatic development as they might be recapitulated during adult lymphangiogenesis: (1) transcriptional control of venous-derived LECs; and (2) the potential of non-venous or myeloid cells to differentiate into LECs.

The prevalent view is that LECs originate from endothelial cells of the cardinal vein [43] and that this process begins around the day E8.5 with the expression of transcription factor CouptfII [44]. This is followed by expression of Sox18 at the day E9.0, a transcription factor that signifies the onset of lymphatic differentiation [45]. Sox18 then drives the expression of transcription factor Prox1 that is considered the “lymphatic master switch” indicating its central role in commitment to the lymphatic lineage [46]. Subsequently, the pre-committed ECs continue to gain autonomy through a stepwise process that includes upregulation of lymphatic genes, downregulation of venous markers, and budding from the cardinal vein to form the first lymphatic structures called lymph sacs [41]. Lymph sac-derived LECs migrate and form nascent vessels that after remodeling serve the foundation for the postnatal lymphatic network [43].

Although this model of embryonic lymphangiogenesis is largely supported by multiple studies, some evidence indicates that the peripheral lymphatic vessels and superficial parts of lymph sacs might be of non-venous origin. This was shown in several species demonstrating that embryonic LECs can derive from mesenchymal cells called lymphangioblasts that are of non-venous [47,48,49,50,51] or myeloid origin [52]. For instance, Prox1 and VEGFR-3 positive lymphangioblasts were detected in mesodermal tissue in avian embryos [49,50] where these cells contributed to newly-formed lymphatic vessels in the wing [49,50] and participated in the formation of jugular lymph sacs along with venous-derived LECs [50]. Similarly, murine embryonic mesenchymal cells positive for Prox1, LYVE-1, and macrophage marker F4/80 were detected within the lymph sacs and developing peripheral lymphatic vessels [52]. It is, therefore, conceivable that genetic programs in myeloid or other non-venous LECP might be potentially reactivated in precursors that contribute to lymphangiogenesis in adult.

Chronic inflammatory conditions are typically associated with increased lymphatic vessel density (LVD) (Table 1). This is illustrated by findings in both clinical studies and experimental models of inflammation including psoriasis [54], inflammatory bowel disease [55,56], rheumatoid arthritis [57], atherosclerosis [58], skin irradiation [59], and cancer [60].

For instance, inflammatory lymphangiogenesis shown by increased LVD was detected in patients rejecting renal transplants [61]. Lymphangiogenesis in this situation was attributed to VEGF-C derived primarily from macrophages [20]. This notion is supported by a study in breast cancer patients receiving radiotherapy that showed a significant correlation between the density of VEGF-C⁺ macrophages and LVD [59]. Inflammation-induced lymphangiogenesis has also been shown in many experimental models including those of wound healing [12], corneal injury [62], skin inflammation [63], peritonitis [64], and chronic inflammation in airways [65]. These models reproducibly showed extensive lymphangiogenesis triggered by macrophage-derived inflammatory mediators suggesting that VEGF-C and other products of activated macrophages are major contributors to the postnatal formation of new lymphatics. Additionally, several models have shown inflammation induced qualitative changes in the lymphatic network such as enlarged, dysfunctional vessels [64], and remodeling of VE-cadherin junctions between LECs [66] (Table 1).

Given the fact that inflammation is the primary trigger of pathological lymphangiogenesis, it is not surprising that most known pro-lymphangiogenic factors are either inflammatory cytokines or downstream products of inflammatory pathways activated by transcription factors of the nuclear factor-kappaB (NF-κB) family [69]. The main NF-κB complexes that transmit inflammatory signals are p50/p65 heterodimers or homodimers of these proteins [70]. Some NF-κB-transcribed genes stimulate lymphangiogenesis directly (e.g., vascular endothelial growth factor A (VEGF-A) [71] and VEGF-C [72]) while others (e.g., IL-1β [73], TNF-α [73], and COX-2 [74]) act indirectly by upregulating lymphangiogenic factors. NF-κB proteins are also known to activate the promoter of VEGFR-3 [68], the key inducer of lymphangiogenesis. The central role of VEGFR-3 in generation of new lymphatic vessels was shown by significantly reduced LVD after blockade of VEGFR-3 at inflammatory [65], wound healing [12], and tumor sites [75].

VEGFR-3 can also be regulated by Prox1, a transcription factor that specifies the fate of LEC during embryogenesis [46]. Prox1 appears to perform a similar function in adult endothelial cells as evidenced by up- and downregulation of VEGFR-3 following forced Prox1 overexpression or silencing in blood vascular endothelial cells (BEC) [33] and in LEC [76], respectively. Also, NF-κB synergizes with Prox1 in regulation of VEGFR-3 expression [68], which shows an additional pro-lymphangiogenic mechanism induced by inflammation.

Inflammation-mediated increase in VEGFR-3 is probably needed to increase the responsiveness of preexisting lymphatic vessels to VEGF-C and VEGF-D. This supposition is based on the fact that while these lymphangiogenic factors are present at very high concentrations at the inflammatory site being produced by a variety of recruited and local cells [65,73,77], the level of surface expression of VEGFR-3 in LEC eventually determines the response rate. The high level of VEGFR-3 on LEC surface collectively regulated by Prox1 and NF-κB is therefore crucial for mounting robust lymphangiogenesis in response to inflammatory stimuli.

Another event that enhances inflammatory lymphangiogenesis is generation of mature forms of VEGF-C/-D through proteolytic processing mediated by plasmin or furin present at high concentrations at inflammatory sites [78,79]. Mature VEGF-C/-D have increased affinity to VEGFR-3 and novel ability to bind VEGFR-2 [80] expressed on both LEC and BEC [81]. Binding of mature VEGF-C/-D to VEGFR-2 and VEGFR-3 expressed in LEC leads to formation of heterodimer of respective receptors, an event that was reported to enhance signal transduction and activation of LEC [81,82].

Another important promoter of lymphangiogenesis is VEGF-A, initially thought to be an exclusively angiogenic factor [83,84]. NF-κB potently upregulates VEGF-A [71] whose elevated expression is noted in a variety of chronic inflammatory conditions such as psoriasis [85], rheumatoid arthritis [86], inflammatory bowel disease [87], chronic airway inflammation [88], and cancer [89]. VEGF-A dependent inflammatory lymphangiogenesis was first demonstrated in mice treated with adenovirus encoding this factor [90], and subsequently shown in models of corneal injury [91,92] and skin cancer [93]. Tumor lymphangiogenesis induced by VEGF-A has been shown in a mouse model of T241 fibrosarcoma [94], as well as in MDA-MB-231 and MDA-MB-435 models of breast cancer [95]. Neutralizing VEGF-A substantially reduced LVD and metastasis in the MDA-MB-231 breast tumor model [95]. Evidence from other tumor models showed that VEGF-A can induce both intratumoral [93] and peritumoral [94] lymphatic vessels that, in turn, facilitate lymphatic metastasis. The pro-lymphangiogenic effect of VEGF-A can be mediated in a direct manner through binding to VEGFR-2 as evidenced by proliferation and migration of VEGFR-2-positive LEC in vitro [96]. However, VEGF-A also recruits macrophages [97] that produce high levels of the pro-lymphangiogenic factors VEGF-C/-D [91], thus acting as an indirect enhancer of lymphangiogenesis. Additionally, VEGF-A was shown to increase VEGFR-3 expression in LEC [98] whereas an anti-VEGF-A antibody was shown to inhibit VEGFR-3 expression [95]. These studies suggest that VEGF-A contribution to regulation of VEGFR-3 expression might be an additional mechanism to promote formation of lymphatics in VEGF-rich environment.

Angiopoietin-2 (Ang-2) is another inflammatory mediator with pro-lymphangiogenic activity. Ang-2 and related protein, Ang-1, are ligands for the tyrosine kinase receptor, Tie2 [99], expressed in both BEC and LEC [100]. Although Ang-1 and Ang-2 have antagonistic functions in BEC activation [101], they both play positive roles in induction of lymphangiogenesis [102] as illustrated by the requirement for Ang-2/Tie2 signaling for embryonic lymphatic development [103,104]. Analogous to the complex pro-lymphangiogenic effect of VEGF-A, Ang-2 promotes lymphatic growth by several mechanisms that include direct activation of Tie2, indirect increase in VEGFR-3 expression [95], and enhancement of LEC activities via crosstalk between Tie2 and VEGFR-2 pathways [105]. Ang-2 also activates Tie2-positive macrophages (TEMs), a highly pro-angiogenic subset of circulating myeloid cells that infiltrate tumors and overexpress VEGF-A and MMP-9 [106]. These proteins promote both angio- and lymphangiogenesis suggesting that Ang-2-activated TEMs might contribute to both processes.

Several inflammatory cytokines can promote lymphangiogenesis either directly or indirectly, by upregulating VEGF-C/-D. For instance, IL-7 increases expression of VEGF-D and induces transcription of lymphatic-specific genes such as Prox1, LYVE-1, podoplanin and VEGFR-3 [107]. IL-7 was shown to increase proliferation, migration, and tube formation of endothelial cells in vitro [107] and the growth of LYVE-1⁺ vessels in tumor-containing matrigel plugs in vivo [108]. IL-7 correlates with lymphatic metastasis in breast cancer patients [109] suggesting that this results from IL-7 induced tumor lymphangiogenesis. Another interleukin, IL-3, has also been shown not only to induce LEC proliferation and migration but also to enhance similar effects of VEGF-C [68]. A possible synergistic mechanism of VEGF-C and IL-3 might be mediated through its ability to activate the NF-κB pathway which, in turn, promotes transcription of VEGFR-3 [68].

Other factors that increase inflammatory lymphangiogenesis include fibroblast growth factor (FGF)-2 [62], platelet-derived growth factor (PDGF) [110], insulin-like growth factor-1 and -2 (IGF-1,-2) [111], hepatocyte growth factor (HGF) [112], growth hormone [113], fasting-induced adipose factor (FIAF) [114], and sphingosine-1 phosphate (S1P) [115]. PDGF [110], IGF-1,-2 [111], and growth hormone [113] induce lymphangiogenesis directly, independently of VEGR-3 signaling. In comparison, factors such as FGF-2 [62] and Cox-2 [116] elicit their effects indirectly by upregulating VEGF-C. Currently, however, the pro-lymphangiogenic mechanisms cumulatively regulated by these factors and their cooperation with VEGFR-3 are incompletely understood. Additional studies are needed to better understand cross-talk between VEGFR-3-induced and other signaling pathways activated in the inflammatory environment.

Macrophages are multifunctional immune cells that respond to a wide array of stimuli including, microbial products, inflammatory cytokines, chemokines, and growth factors (for reviews on these topics see [125,126,154,155,156]. Macrophages are strongly recruited to cancers in response to necrosis [149], hypoxia [120], and tumor-secreted chemoattractants [126,155]. They are highly plastic and can differentiate into multiple subtypes depending on signals present in local environment [156]. Activated macrophages are generally categorized as being either M1- or M2-polarized based on phenotypical characteristics and their involvement in type 1 or type 2 inflammation, respectively. Characteristics that discern between M1- and M2-polarized macrophages include effector functions, cytokine production, and expression of chemokine receptors [156]. It should also be noted that sub-division to M1 and M2 classes oversimplifies the conditions in different tumors that may induce sub-categories with M1/M2 mixed or currently unclassified features.

The classically activated M1 phenotype is stimulated by signals associated with microbial infections such an IFN-γ, TNFα, GM-CSF, and the bacterial product, LPS [125]. M1 macrophages are integral to the anti-tumor type 1 inflammatory response as they have a high capacity to present antigens, and produce pro-inflammatory cytokines that activate Th1 lymphocytes [125,126]. Additionally, M1 macrophages release high levels of toxic intermediates (e.g., nitric oxide, reactive oxygen species, TNFα) responsible for killing intracellular parasites and tumor cells [125,156]. Thus, M1-type macrophages are generally considered to be tumoricidal, although some of their factors have pro-angiogenic properties (e.g., TNFα).

Alternatively activated macrophages, referred to as M2-polarized, include several subtypes that generally suppress type 1 inflammation and promote tumor progression [125]. They activate Th2 lymphocytes, and promote wound healing, tissue remodeling, and angiogenesis. TAMs typically, but not always, undergo M2 “like” polarization and thus exhibit many pro-tumorigenic characteristics [125]. TAMs suppress adaptive immunity by secreting factors that suppress Th1 responses along with chemokines that recruit non-cytotoxic T cell subsets [125,126]. Additionally, TAMs release factors that promote tumor cell growth, survival, and migration [125]. Furthermore, TAMs promote angiogenesis and metastasis through production of growth factors such as EGF, βFGF, and PDGF; angiogenic cytokines such as VEGF-A, VEGF-C, and CXCL8; and matrix degrading enzymes such as MMP-2, MMP-7, MMP-9, MMP-12, plasmin, and urokinase plasminogen activator [125,126,157]. Many of the molecules secreted by TAMs are also lymphangiogenic (Table 2) and are discussed in detail later in this review.

Chronic inflammation is a hallmark of breast cancer [187,188] and has been repeatedly linked to increased tumorigenesis [189,190], angiogenesis [191,192], lymphangiogenesis [65,193] and metastatic progression [194,195,196]. Clinical studies in breast, prostate, cervix, and bladder cancers showed that macrophages, master regulators of inflammation, are massively recruited to tumors and correlate with poor patient outcome [126]. Until recently, strong correlation between macrophage infiltrates and metastasis has been primarily explained by TAM-mediated release of pro-angiogenic factors that heighten angiogenesis and increase hematogenous metastasis [125]. However, many of the same proteins can also contribute to lymphangiogenesis, invasion of lymphatic vessels, and lymphogenous metastasis (Table 2). The notion that macrophages promote lymphangiogenesis is supported by clinical studies on cancers of the cervix [22], pancreas [197], lungs [198,199], breast [200], esophagus [201], and melanoma [202]. These studies have shown statistically significant associations between TAM density and tumor LVD, lymphatic invasion, and LN metastasis.

One of the first clinical studies that showed direct correlation between tumor LVD and the density of VEGF-C/-D producing TAMs was performed using specimens of cervical cancer [22]. This study found that tumor LVD correlated with VEGF-C/-D producing TAMs, and that both TAM and LVD densities correlated with LN metastasis [22]. Interestingly, VEGF-C/-D positive monocytes comprised only a fraction (~25%) of total TAMs, co-expressed VEGFR-3 and formed small clusters around lymphatic vessels [22]. The lymphangiogenic role of TAMs has been also shown in studies with pancreatic [197] and lung cancer [198]. These studies considered that TAM lymphangiogenic potential can be affected by their M1/M2 polarization. The pancreatic cancer studies used CD163/CD204 markers to distinguish M2-polarized macrophages from the entire population of CD68⁺ TAMs [197]. M2-polarized CD163/204-positive TAMs were significantly associated with increased LVD (p = 0.018) and decreased overall patient survival (p = 0.018) whereas CD68⁺ TAMs were mainly associated with LN metastasis (p = 0.029). In the study of lung adenocarcinoma, M1-polarized TAMs were distinguished from M2 using double staining for CD68 and iNOS [198]. Overall, 79% of TAMs were M2-polarized and significantly correlated with both peritumoral LVD (p = 0.009) and LN metastasis (p = 0.003) whereas M1-polarized TAMs were not associated with either parameter. However, high intratumoral TAM density, regardless of their subtypes, was associated with a decrease in five-year survival. An independent study of lung adenocarcinoma also showed that TAM infiltration significantly correlated with peritumoral LVD (r = 0.069, p < 0.001) and was associated with LN metastasis (p = 0.037) and reduced patient survival (p = 0.005) [199]. Interestingly, peritumoral but not intratumoral LVD correlated with TAM infiltrates [199] suggesting that macrophages primarily contribute to the lymphatic formation at the tumor periphery. Clinical associations of TAMs with lymphatic invasion [201] and LN metastasis [200,201] were also shown in esophageal [201] and breast cancers [200], although correlation with LVD in these studies was not determined.

Not all studies have found associations between TAMs and LN metastasis [203,204] or LVD [205,206]. For instance, unlike his previous study with cervical cancer, Shoppmann et al. found that in breast cancer neither VEGF-C producing TAMs nor VEGF-C producing tumor cells were associated with LVD [205]. Some discrepancies might relate to heterogeneity of analyzed patient cohorts. It is tempting to speculate that the anatomical location of the tumor in relation to initial or collecting lymphatics, which exhibit different responses to micro-environmental stimuli [207], would influence the degree of tumor lymphangiogenesis. Additionally, discrepancies may relate to the lack of consideration for macrophage subtypes with differential capacity to influence lymphatic formation. For instance, some studies that failed to show association between TAMs and LVD/LN metastasis did not account for M2 polarization that might determine TAM contribution to lymphangiogenesis [197,198]. However, unlike the studies described above, an additional study that distinguished M2-polarized macrophages with the marker CLEVER-1/Stablin-1 found no association between M2 macrophages and podoplanin⁺ LVD [208]. The variable results from studies that focused on M2-polarized macrophages could be due to macrophage plasticity. Depending on the combination of micro-environmental signals, M2 macrophages can polarize differentially into three subtypes (M2a, M2b and M2c) that have distinct immunological functions and molecular profiles [156]. The specific role of these macrophage subtypes in tumor lymphangiogenesis is currently unknown. The other variable that may contribute to discrepancies among experimental studies is the kinetic of the expression of M2 markers that fluctuates upon macrophage activation or interactions with other immune cells [209]. It is therefore, plausible that some discrepancies in the results of these studies might be due to functional dissimilarities in the analyzed TAM sub-groups.

The lymphangiogenic characteristics of TAMs have been demonstrated in many tumor models in which blocking macrophage recruitment or depleting macrophages correlated with decreased LVD and suppressed LN metastasis. For instance, blocking macrophage recruitment to orthotopic pancreatic tumors by anti-PlGF antibody reduced F4/80⁺ TAMs by 74% [210]. Importantly, this treatment resulted in a 75% decrease in LVD (p < 0.005), and a corresponding ~60% decrease in LN metastasis (p < 0.05). Tumor VEGF-C levels and blood vessel density were also decreased following this treatment [210]. Similarly, blockade of M-CSF signaling markedly reduced the recruitment of LYVE-1⁺ TAMs to osteosarcoma causing an 8–10 fold reduction in the density of peritumoral lymphatics, and ~5–6 fold reduction in blood vessel density [211]. This evidence suggests that tumor macrophages promote both angiogenesis and lymphangiogenesis that contribute, respectively, to hematogenous and lymphatic metastasis.

Another experimental approach that helped to examine the role of macrophages in lymphangiogenesis is systemic depletion of macrophages using clodronate liposomes (CDL). CDL depletion of CD11b⁺/LYVE-1⁺ TAMs in a model of ovarian cancer inhibited tumor-induced lymphangiogenesis by 50–75% (p < 0.05) [158]. Similarly, in an orthotopic model of bladder cancer, CDL depletion of VEGFR-3⁺ TAMs caused a statistically significant 74% reduction in LVD, and a similar decrease in lymphatic metastasis [212]. However, in contrast to other studies, elimination of TAMs affected only lymphatic vessels with no change in tumor blood vessel density [212]. In an orthotopic model of pancreatic cancer, CDL treatment resulted in the same extent of inhibition of LVD as anti-PlGF antibody discussed earlier [210], supporting the hypothesis that tumor lymphatic formation is primarily regulated by macrophages.

Not all studies, however, showed clear dependency between recruited TAMs and tumor lymphatics. For instance, tumor LVD was reduced by only 20% (p < 0.01) and LN metastasis was unchanged after CDL depletion of F4/80⁺ TAMs in a Rip1Tag2 insulinoma model [23]. The differences in study results might be due to the ability of CDL to effectively deplete all subsets of macrophages which seems to directly relate to their pro-lymphangiogenic effect. Currently, it is unclear whether a subset responsible for the pro-angiogenic effect overlaps with pro-lymphangiogenic sub-populations. Potentially, these effects can be mediated by distinct populations that might have differential sensitivity to CDL depletion, or be recruited by different chemokines. Alternatively, differential effects of TAMs depletion (or recruitment) on angio/lymphangiogenesis [23,212] could also relate to the variability in the composition of the tumor milieu in individual models [126,213]. In overall, despite some discrepancies in responses to macrophage depletion in various models, most studies provided supportive evidence for the contribution of TAMs to induction of tumor lymphangiogenesis.