Tumour microenvironment

The tumour microenvironment is the cellular environment in which the tumour exists and it consists of ECM and diverse types of non-malignant cells, including cancer-associated fibroblasts (CAF), pericytes, macrophages, dendritic cells (DC), mast cells, lymphocytes, endothelial cells and their precursors in tumours (Ref. Reference Catalano4). Among them, immune cells and mesenchymal cell types have been well studied.

Immune cells (e.g. macrophages, DC, lymphocytes and mast cells) are recruited to the TME where they express diverse tumour-promoting signals. Tumour-associated macrophages (TAM) are well-studied immune cell types and are generally known to be immunosuppressive and pro-angiogenic. TAM are one of the myeloid-derived suppressor cells (MDSC) which are a heterogeneous population of myeloid cells with a potential to repress T cell responses (Ref. Reference Bronte20). Specifically, the M2-type TAM overexpress and secrete pro-angiogenic factors such as vascular endothelial growth factor (VEGF), tumour necrosis factor alpha (TNFα), fibroblast growth factor basic (bFGF), matrix metalloproteinase (MMP)-2, -7, -9 and -12, as well as epidermal growth factor (EGF) to facilitate tumour growth, invasion and metastasis (Ref. Reference Stearman21). TAM also induce lymphangiogenesis (Ref. Reference Schoppmann22). Subpopulations of TAM expressing VEGFC/D and vascular endothelial growth factor receptor 3 (VEGFR3) have been identified. It is thought that the TAM-expressed VEGFC/D could induce peritumoural lymphangiogenesis (Ref. Reference Schoppmann22). TAM express immunosuppressive factors, including prostaglandin E2, interleukin 10 (IL10) and transforming growth factor beta (TGFβ) to facilitate escape of the tumour from attack by the immune system (Ref. Reference Pollard23). DC are the most potent antigen-presenting cells (APC) in general, thus they can activate T lymphocytes and trigger antigen-specific anti-tumour immune responses (Ref. Reference Banchereau24). However, the DC recruited to the TME are impaired in their immunity and start to produce pro-angiogenic factors such as TNFα, TGFβ and granulocyte macrophage colony-stimulating factor (GM-CSF; Ref. Reference Sozzani25). Mechanistically, tumour-secreted factors such as IL6 and TNF suppress DC maturation by activating the phospho-signal transducer and activator of transcription 3 (Stat3) pathways or inhibiting the toll-like receptor (TLR) pathways (Ref. Reference Fainaru26). Tumour infiltrating lymphocytes (TIL) and their modifications are also crucial for tumour cell survival in TME (Refs Reference Pages27, Reference Galon, Fridman and Pages28). Although the TIL are supposed to be anti-tumourigenic, TME induces apoptosis of the TIL by secreting tumour exosomes that contain apoptosis ligands (Ref. Reference Valenti29). T lymphocytes activity in TME can also be edited by LV activity. CCL21 expressed by the LV recruits CCR7-positive naive T cells into the lymphatic systems in the tumour stroma where they can be conditioned to be less immune reactive by tumour-secreted cytokines (Ref. Reference Swartz and Lund30). Tumour-secreted TGFβ promotes tumour-associated regulatory T cell (T _reg) and MDSC activities. These cells impair cytotoxic T lymphocytes (CTL) and helper T1 cell functions (Ref. Reference Swartz and Lund30).

Mesenchymal cell types, such as fibroblasts are also pivotal for tumour progression (Refs Reference Madar, Goldstein and Rotter31, Reference Guan and Chen32). Physiological fibroblasts or their precursors are recruited to the TME where several paracrine factors from cancer cells or stromal cells reprogram the fibroblasts to be pro-tumourigenic: these pathological fibroblasts are referred to as CAF (Ref. Reference Cortez, Roswall and Pietras33). CAF can also be derived from pericytes, smooth muscle cells and mesenchymal stem cells in the TME. CAF generally feature rapid self-proliferation and up-regulation of tumour-promoting genes [CXC chemokine ligand 12 (CXCL12); TGFβ; hepatocyte growth factor (HGF)], pro-inflammatory genes (IL6, IL8) and pro-angiogenic genes [MMPs; platelet-derived growth factors (PDGFs)] (Ref. Reference Madar, Goldstein and Rotter31). CAF also confer cancer cells with resistance to RAF inhibitors by HGF production (Ref. Reference Straussman34); with resistance to anti-VEGF therapy by activating the PDGFC signal (Ref. Reference Crawford35). Moreover, CAF modify host immunity (Refs Reference Feig36, Reference Kraman37) and induce cancer cells to undergo EMT (Ref. Reference Yu38).

Organ microenvironment

Metastasis requires multiple steps; tumour cell intravasation into the vasculature at the primary tumour site, recruitment and survival of the tumour cells in the blood during circulation, extravasation into the metastatic organ, and seeding, proliferation, angiogenesis and colonisation in the organ microenvironment (Refs Reference Stracke and Liotta39, Reference Sahai40). The organ microenvironment is the cellular environment in which organ-specific parenchymal and stromal cells exist and tumour cells can metastasise (Ref. Reference Langley and Fidler3). Specific organ microenvironment serves as metastatic niche and determines the extent of tumour cell proliferation, invasion and survival (Ref. Reference Fidler, Kim and Langley41).

According to the ‘seed and soil hypothesis’ by Stephen Paget (Ref. Reference Paget42), metastatic cancer cells function as ‘seeds’ and a particular organ microenvironment serves as the ‘soil’. The ‘soil’ must be prepared and pre-conditioned to properly function as a metastatic niche. A number of studies have shown evidence supporting this hypothesis. For example, orthotopic versus ectopic murine tumour models revealed that cancer cells implanted in ectopic sites grow slowly and metastasise rarely compared with those in orthotopic sites (Ref. Reference Fidler43). Organs with low incidence of neoplasm metastasis have been reported and it has been found that these organs are also associated with poor metastatic niche formation (Ref. Reference Araki44). Some organs that are vulnerable to metastasis such as lymph nodes (LN), bone marrow and lungs are found to promote tumour cell recruitment, immune modification, vascular remodelling and enhanced angiogenesis (Refs Reference Nishida, Tsukushi, Urakawa, Sugiura, Nakashima, Yamada and Ishiguro45, Reference Dos Santos46, Reference Hirasawa47). It has also been reported that tumour metastasis is organ-specific (Refs Reference Lu48, Reference Rigamonti and De Palma49, Reference Lorusso and Ruegg50, Reference Nguyen, Bos and Massague51). Breast cancer mostly metastasise to the bones, lungs, brain and liver (Ref. Reference Lorusso and Ruegg50). Leukaemias metastasise preferentially to certain parts of the nervous system; they metastasise to the leptomeninges, which are the two innermost layers of the membrane that envelops the brain and spinal cord, but rarely to the brain parenchyma (Ref. Reference Nolan and Abrey52). Most head and neck cancers metastasise to the regional LN (e.g. cervical LN) and salivary glands, while distal metastases are uncommon (Ref. Reference Patel53). For example, autopsies of over 4000 patients who died of head and neck squamous cell carcinoma (HNSCC) showed that the fraction of distant metastasis was <1% of local spread (Ref. Reference Allen54). These data suggest that the nature of the organ microenvironment has a significant influence on tumour metastasis. Therefore, a better understanding of the molecular mechanisms that link the organ microenvironment to metastasis should provide ideas for more effective anti-metastatic therapies.

Several mechanistic studies have demonstrated how the organ microenvironments are primed to facilitate cancer metastasis. TGFβ promotes breast tumour cell seeding in the lungs by activating angiopoietin-like four induced tumour cell extravasation (Ref. Reference Padua55). Other mediators (Angiopoietin 2, MMP3, MMP10, TGFβ and TNFα) of vascular remodelling that are important in metastasis to the lungs have also been discovered (Refs Reference Gupta56, Reference Goncharova57). CXCL12 and insulin-like growth factor 1 (IGF1) are two key mesenchymal signals that select seeds from breast tumours to metastasise and colonise the bone (Ref. Reference Zhang58). Tumour-conditioned media (TCM) from triple negative breast cancer cells induced lymphangiogenesis and angiogenesis in regional LN (Ref. Reference Lee, Pandey and Popel59). Increased angiogenesis in tumour-draining LN (TDLN) were observed and correlated with LN metastasis and lower survival in breast cancer patients (Refs Reference Guidi60, Reference Edel, Harvey and Papadimitriou61, Reference Lee, Lee, Koskimaki, Han, Pandey and Popel62). Exosomes are microvesicles derived from tumours and their role in tumour–organ communication for metastatic progression is emerging. Exosomes secreted from melanoma-induced LN angiogenesis and BV remodelling (Ref. Reference Hood, San and Wickline63). Melanoma exosome-mediated cMet and its effects on bone marrow-derived cells and organ niche formation have been reported (Ref. Reference Peinado64). These examples suggest that tumour-draining secretions, organ-residing cells, and their resulting signals and phenotypes can be targeted to improve upon current therapeutic strategies.

Angiogenesis

Angiogenesis, the formation of new BV from pre-existing BV, has been recognised as a target to inhibit primary tumour growth and haematogenous tumour cell dissemination (Ref. Reference Potente, Gerhardt and Carmeliet65). In general, cells have to be located within 150–200 μm of blood capillaries to survive, although the distance can vary depending on the cell type (Ref. Reference Jain66). For this reason, tumours are not able to grow to more than 1–2 mm³ without angiogenesis within the stroma (Ref. Reference Folkman67). Tumour cells under hypoxia express and secrete VEGFA, which is up-regulated by the transcription factor hypoxia inducible factor 1 alpha (HIF1α), to trigger tumour angiogenesis and make the tumour resistant to hypoxia (Ref. Reference Ferrara68). Though other growth factors are also involved in angiogenesis, the signals mediated by VEGFR and neuropilin (NRP) are relatively well understood (Ref. Reference Mac Gabhann69). There are five growth factors in the VEGF family: VEGFA/B/C/D, and placental growth factor (PlGF); and three VEGF receptors: VEGFR1/2/3. Two neuropilins (NRP1/2) are also involved in VEGFR signalling. Among them, angiogenesis is primarily regulated by VEGFA, VEGFR1, VEGFR2 and NRP1. VEGFC/D, VEGFR3 and NRP2 regulate lymphangiogenesis. NRP1 makes a co-receptor complex with VEGFR2, promoting VEGFA-induced VEGFR2 signalling in BECs. The current understanding of how anti-angiogenic therapies targeting VEGFA or VEGFR2 signalling work is that they inhibit BV formation in tumours, thus limiting the blood supply, depriving the tumour of oxygen and nutrients, and hence induce severe hypoxia and necrosis of the tumours leading to an inhibition of tumour growth (Ref. Reference Sarmiento70).

The stromal BV also serve as routes of haematogenous spread of cancer cells from the primary tumour (Ref. Reference Labelle and Hynes71). Haematogenous metastasis has been studied using the tail vein or intracardiac tumour injection models (Refs Reference Rashid72, Reference Zadnik73). In addition to the stromal BV, organ-residing BV are also important for metastasis. They facilitate tumour cell extravasation, seeding, and colonisation. Particularly, permeability of the organ BV (e.g. lung vascular permeability and blood–brain barrier integrity) is crucial for extravasation of tumour cells from the circulating blood (Ref. Reference Padua55). Several studies have shown mechanisms of how the tight junctions between BEC in different organ microenvironments are opened to facilitate extravasation of tumour cells (Refs Reference Arshad74, Reference Irmisch and Huelsken75). Moreover, the enhancement of organ angiogenesis is essential for tumour cell survival and colonisation. It is well established that up-regulation of angiogenesis in the organ microenvironment is synchronised with the increase in oncogenic activation and the mutation of tumour suppressor genes (Ref. Reference Bergers and Benjamin76). For example, activation of the Ras oncogene promotes VEGF expression in tissues and p53 mutation results in reduced expression of the anti-angiogenic factor, thrombospondin 1. However, it is still not clear whether the angiogenic switch is turned on before the actual metastasis to prepare and condition the pre-metastatic niches, or initiated only after tumour cell invasion into the niches (Refs Reference Hood, San and Wickline63, Reference Farnsworth77). In addition to enhanced angiogenesis, remodelling of the blood vasculature in the organ microenvironment also contributes to metastasis. For example, high endothelial venules (HEV) in LN experience morphological changes during LN metastasis. The HEV are modified so that they are more dilated and cancer cells and red blood cells are detected within the lumen rather than the expected immune cells in tongue cancer compared with normal, physiological LN (Ref. Reference Lee78).

Lymphangiogenesis

Tumour cells express and secrete lymphangiogenic factors, including VEGFC/D, angiopoietins, PDGF-BB/AA and bFGF to promote tumour lymphangiogenesis, the formation of new LV in the tumour stroma (Refs Reference Cao15, Reference Von Marschall79, Reference Kesler80). LV are distinct from BV in their anatomy, growth signals, gene expression and functions (Refs Reference Hirakawa16, Reference Mohammed81). The LV lack a well-defined basement and are scarcely covered by pericytes or smooth muscle cells thus are much more permeable than BV (Ref. Reference Cao15). Due to their leaky nature, LV play a role as a reservoir for proteins and cells that have leaked from the BV and function to transport them back from the tissues to the blood circulation (Ref. Reference Alitalo19). Tumour lymphangiogenesis has not been studied as long as tumour angiogenesis primarily because of the relatively recent discoveries of lymphatic markers such as VEGFR3 (Ref. Reference Kaipainen82), lymphatic vessel endothelial hyaluronan receptor 1 (LYVE1; Ref. Reference Banerji83), prospero homeobox protein 1 (Prox1; Ref. Reference Wigle and Oliver84), NRP2 (Ref. Reference Yuan85) and podoplanin (Ref. Reference Schacht86). It is, however, well understood that the tumour lymphatic vasculature serves as the initial route of tumour lymphatic dissemination (Ref. Reference Kesler80). Lymphatic metastasis occurs at least as frequently as BV-mediated metastasis (Ref. Reference Cao15), and in certain kinds of cancer, such as breast cancer, LV are the primary routes of tumour dissemination (Refs Reference Porter87, Reference Perou88, Reference Moskowitz89). Slow lymphatic flow rates and tumour cell survival factors, such as hyaluronic acids in the lymph fluid also support lymphogenous metastasis (Refs Reference Matsuo90, Reference Lipponen91). Tumour LV are connected to the regional LN (Refs Reference Skobe92, Reference Tammela and Alitalo93) and the tumour cell positive or tumour secretion influenced LN are referred to as TDLN (Ref. Reference Swartz and Lund30). These TDLN are considered to be transient reservoirs of metastatic tumour cells, and as a support system for helping the tumour cells make colonies for further metastasis to distant organs (Refs 94, Reference Boughey95). From the lymphatic system, cancer cells can also be transported back to the circulating blood from where they extravasate and invade distant organs (Ref. Reference Sleeman, Cady and Pantel96). Thus blocking tumour lymphangiogenesis can inhibit metastasis by limiting the likelihood of initial tumour dissemination via intra- and peritumoural LV (Refs Reference Jain and Padera97, Reference Achen, Mann and Stacker98). Lymphangiogenesis in the pre-metastatic organs is also central for successful metastasis (Refs Reference Hirakawa99, Reference Harrell, Iritani and Ruddell100, Reference Alitalo and Detmar101). For example, LV in the TDLN are connected with the tumour LV so that they can serve as direct routes for invasion of the tumour into the LN (Ref. Reference Quagliata102). Also, LV in the pre-metastatic organs can modify host immunity, so that cancer cells can escape immune attack (Ref. Reference Thomas103). This pro-tumour immuno-editing activity is regulated by LEC-secreted factors. We discuss this in more detail in the next section. Recently, we developed a new type of metastasis model in which we inject TCM subcutaneously into animals for 2 weeks followed by orthotopic breast cancer cell inoculation (Refs Reference Lee, Pandey and Popel59, Reference Lee104). We saw very rapid formation of metastases in the LN and lungs in this model. At the same time the TCM treatment resulted in enhanced lymphangiogenesis in the pre-metastatic organs (Ref. Reference Lee105). These results suggest that both organ lymphangiogenesis and tumour lymphangiogenesis may be therapeutic targets to efficiently treat metastasis.

BEC- and LEC-mediated signals in physiology

BEC secrete a large number of molecules, including proteins, lipids and metabolites. These molecules are called angiocrine factors (Ref. Reference Nolan106). These factors allow BEC to communicate with other cells and they play a role in maintaining physiological homeostasis, regulating cell fate and host immunity, and promoting tissue regeneration (Ref. Reference Butler, Kobayashi and Rafii18). For example, angiocrine factors balance self-renewal and differentiation of haematopoietic stem cells (HSC) to maintain haematopoietic homeostasis (Ref. Reference Kobayashi107). The blood endothelium also maintains blood fluidity and immunity through crosstalk with platelets (Ref. Reference Ruggeri108). There have been several studies claiming that angiocrine factors are central for tissue regeneration; e.g. angiocrine signals from liver sinusoidal endothelial cells (LSEC) are required for liver regeneration (Ref. Reference Ding109); angiocrine signals induce and maintain the regeneration of lung alveoli (Ref. Reference Ding110).

LV are distinct from BV in that they do not have a well-defined basement membrane, do not carry erythrocytes, and are composed of LEC that are phenotypically different from BEC. Generally LV function as reservoirs for immune cells, antigens, lipids and macromolecules that have leaked from the vascular system and transport lymph fluid back to the circulatory system (Ref s Reference Tammela and Alitalo93, Reference Weech, Goettsch and Reeves111). If the LV are abnormal, the lymph fluid cannot be efficiently drained and results in lymphedema. LV also function in fat absorption and lipid transport and the consumption of fatty foods increases lymph flow in humans (Ref. Reference Turner and Barrowman112). The lymphatic endothelium also expresses several adhesion molecules and chemokine ligands that interact with immune cells to facilitate maintenance of the host immunity. LEC express CC chemokine ligand 21 (CCL21) (Ref. Reference Nykanen113). CCL21 or CCL19 can attract CC chemokine receptor 7 (CCR7) positive DC and APC (Ref s Reference Forster, Davalos-Misslitz and Rot114, Reference Scandella115, Reference Riedl116). These DC and APC are transported into LN where they meet T lymphocytes in the T cell zone: the T lymphocytes and DC are activated and matured in the LN and depart from the LN through the efferent LV. In addition, resident T and B lymphocytes are important in the regulation of lymphangiogenesis in the LN. T lymphocytes negatively regulate lymphangiogenesis via interferon gamma-mediated mechanisms (Ref. Reference Kataru117). On the other hand, B lymphocytes promote lymphangiogenesis in the LN (Ref. Reference Angeli118).

BEC-mediated signals in cancer

As discussed above, tumour angiogenesis is a target for inhibiting tumour growth by blocking the supply of oxygen to the growing tumour. Recently tumour-associated BEC have been revisited with a focus on BEC-secreted factors and their crosstalk with cancer cells and other stromal cells (Ref. Reference Butler, Kobayashi and Rafii18). BEC-mediated signals in cancer are summarised in Figure 2.

Figure 2.

Crosstalk between blood endothelial cells and cancer cells. Tumour cell and blood endothelial cell -secreted factors, ECM components, microRNAs, and membrane bound or soluble receptors can mediate tumour/blood endothelial crosstalk signals to promote tumour cell proliferation, migration, invasion, EMT, and cancer stem cell phenotypes. Question marks represent unknown mechanisms. ADAM17, ADAM metallopeptidase domain 17; EMT, epithelial-to-mesenchymal transition; miRxx, micro RNA xx; mTOR, mammalian target of rapamycin; NRP1, neuropilin1; S1PR1, Sphingosine-1-phosphate receptor 1; Slit2, Slit homolog 2 protein; TGFβ, transforming growth factor beta; VEGFA, vascular endothelial growth factor A; VEGFR2, vascular endothelial growth factor receptor 2.

There have been a number of studies that show that BEC-cancer cell crosstalk regulates cancer cell proliferation and migration. Slit homologue protein 2 (Slit2), which is negatively regulated by Ephrin A2 receptor (EphA2) in BEC, is a breast tumour suppressive angiocrine factor (Ref. Reference Brantley-Sieders119). Tumour BEC express high levels of EphA2. Enhanced EphA2 expression in BEC is associated with low Slit2 expression in the tumour blood endothelium and this predicted poor prognosis in human breast cancer patients (Ref. Reference Brantley-Sieders119). Large-cell lymphoma cell growth and migration were stimulated by conditioned media obtained from liver-, lung-, brain- microvascular endothelial cells (MEC), suggesting that tumour angiocrine factors induce tumour cell growth and migration (Ref. Reference Hamada120). Tetraspanin CD151, which is expressed in breast cancer cells, regulates breast tumourigenesis by interacting with integrins α3β1 and α6β4 on BEC (Ref. Reference Sadej121). Although CD151 did not influence the inherent proliferative potential of breast cancer cells, it positively affected tumour cell proliferation in a cancer cell-BEC co-culture system (Ref. Reference Sadej121). BEC also promoted invasion of oral squamous cell carcinoma and Kaposi's sarcoma through CXC chemokine signalling (Ref. Reference Warner122). CXCL1 and CXCL8 were up-regulated in VEGF-treated EC, and these CXC chemokines initiated the invasion of CXCR2-positive tumour cells (Ref. Reference Warner122). Surprisingly, perlecan-expressing BEC suppressed breast and lung carcinoma tumourigenesis by reducing pro-tumourigenic and pro-inflammatory signalling in the cancer cells. Perlecan silencing in BEC eliminated this tumour suppressive activity of the endothelial cells via increasing IL6 secretion by the cancer cells (Ref. Reference Franses, Baker, Chitalia and Edelman123).

BEC-cancer cell crosstalk regulates tumour angiogenesis as well. Tumour blood endothelium-derived microRNAs influence tumour progression by regulating tumour angiogenesis (Ref. Reference Heusschen124). A number of studies have explored diverse microRNAs, reporting that Let7f, miR17/5p, miR18a, miR19a, miR21, miR31, miR93, miR126, miR130, miR155, miR210 and miR296 are BEC-expressed pro-angiogenic miRNAs and that miR20a, miR92a, miR221, miR222 and miR320 are anti-angiogenic miRNAs (Ref. Reference Heusschen124). Crosstalk between BEC and breast cancer cells also results in reciprocal regulation of some angiogenic factors, including VEGF and angiopoietin 1/2. Co-culture of BEC with breast cancer cells resulted in enhanced VEGF and angiopoietin 2 expression but reduced angiopoietin 1 expression in the breast cancer cells compared with a monoculture of the breast cancer cells. These changes increased vascular sprouting and destabilisation (Ref. Reference Buchanan125). In a recent study, sphingosine-1-phosphate receptor (S1PR1), one of the G-protein-coupled receptors (GPCR) expressed in BEC, regulated tumour vascular permeability and modulated tumour growth (Ref. 126). When S1PR1 signalling was very low or very high, tumour growth was delayed while, when S1PR1 signalling was intermediate, tumour growth was enhanced. The intermediate levels of S1PR1 appear to be important for stabilising tumour BV and promoting tumour growth (Ref. 126).

BEC-cancer cell crosstalk also influences the stem cell-like properties of cancer cells. BEC-derived VEGF inhibited anoikis, a form of programmed cell death, of head and neck cancer stem cells (CSC) by activating protein kinase B (PKB, also known as Akt) signalling pathways (Ref. Reference Campos127). Conditioned media obtained from brain BEC maintained glioblastoma stem-like cell expansion through the mammalian target of rapamycin (mTOR) pathway (Ref. Reference Galan-Moya128). BEC-derived signalling promoted survival and self-renewal of CSC in HNSCC (Ref. Reference Krishnamurthy129). Selective inhibition of tumour-associated BEC by transducing a caspase-based artificial death switch (iCaspase-9) reduced CSC in the HNSCC tumour xenografts (Ref. Reference Krishnamurthy129). The EC-derived soluble form of Jagged 1 generated by the proteolytic activity of ADAM metallopeptidase domain 17 (ADAM17), led to Notch activation in human colorectal cancer (CRC) cells, promoting CSC phenotypes in CRC (Ref. Reference Lu130).

BEC-cancer cell crosstalk also induces EMT in cancer cells. Conditioned media from bovine aortic endothelial cells (BAEC) induced EMT in A549 and PANC-1 tumour cell lines. In that study, neutralising antibody against either TGFβ1 or TGFβ2 did not reverse endothelial-dependent EMT, but simultaneous inhibition of both TGFβ1 and TGFβ2 abolished the EMT (Ref. Reference Kimura131). Three-dimensional (3D) culture of BEC and breast epithelial cells induced EMT in the breast epithelial cells, and basal-like breast tumours contained cells undergoing EMT around the vascular-rich areas of the tumours (Ref. Reference Sigurdsson132).

LEC-mediated signals in cancer

Tumour lymphangiogenesis, initiated by lymphangiogenic signals from tumour cells, is a therapeutic target to inhibit tumour dissemination through the lymphatic system. Tumour-associated LEC, the LEC-secreted factors (also called ‘lymphangiocrine factors’) and their crosstalk with cancer cells and stromal cells have been studied with this possibility in mind. LEC-mediated signals in cancer are summarised in Figure 3.

Figure 3.

Crosstalk between lymphatic endothelial cells and cancer cells. Tumour cell and lymphatic endothelial cell-secreted factors, ECM components, and membrane bound receptors can mediate tumour/lymphatic endothelial crosstalk signals to promote tumour cell migration, recruitment and adhesion. Lymphatic endothelial cell-secreted factors also induce angiogenesis and enhance vascular permeability in pre-metastatic organs. They can also modify host immunity to mediate tumour immune escape by recruiting immature dendritic cells (iDC) and naïve T cells. Question marks represent unknown mechanisms. CCRxx, CC chemokine receptor xx; CCLxx, CC chemokine ligand xx; DARC, duffy antigen receptor for chemokines; HIF1, hypoxia inducible factor 1; iDC, immature dendritic cells; IL6, interleukin 6; IL6R, interleukin 6 receptor; KAl1, Kallmann syndrome 1; LYVE1, lymphatic vessel endothelial hyaluronan receptor 1; LN, lymph node; LE, lymphatic endothelium; NRP1, neuropilin1; Stat3, signal transducer and activator of transcription 3; T _reg, regulatory T cells; CTL, cytotoxic T lymphocytes; T _H1, helper T 1; VEGFC/D, vascular endothelial growth factorC/D; VEGFR3, vascular endothelial growth factor receptor 3 (homodimer); VEGFR 2/3, vascular endothelial growth factor receptor 2/3 (heterodimer).

Crosstalk between LEC and cancer cells can facilitate the recruitment of cancer cells from the primary tumours to the lymphatic system. The lymphatic endothelium expresses CXCL12 and CCL21 chemokines and these chemokines recruit CXCR4- or CCR7-expressing cancer cells by chemoattraction (Refs Reference Kim133, Reference Wiley134, Reference Hwang135). CXCL12 and CCL21 expression was observed in physiological LEC as well where they function to recruit immune cells to maintain host immunity. The chemokines secreted by LEC in highly metastatic oral cancer cell conditioned media was reported to be highly altered compared with that by normal LEC (Ref. Reference Zhuang136), suggesting that there can be tumour-specific LEC-mediated chemokine ligand/receptor mechanisms for tumour cell invasion into the LV. Recently, Lee et al. showed that triple-negative breast tumour-induced LEC secrete CCL5 and recruit CCR5-expressing cancer cells into the lymphatic system, supporting LN metastasis (Ref. Reference Lee104). This recruitment was inhibited by maraviroc, the CCR5 inhibitor with anti-retroviral activity. In that study, tumour secreted IL6-induced Stat3 phosphorylation in LEC resulting in enhanced CCL5 expression and secretion by the LEC. Inhibition of IL6 or Stat3 signalling resulted in reduced CCL5 secretion by the LEC and significant reduction of lymphogenous metastasis (Ref. Reference Lee104). Hyaluronan (HA), which is expressed in cancer cells, also mediates cancer cell adhesion to the lymphatic endothelium by binding to LYVE1 (Refs Reference Banerji83, Reference Du137) and induces tumour cell invasion into the LV. Other adhesion molecules such as CLEVER-1/Stabilin-1, mannose receptor and Thy1 (CD90) have also been reported to contribute to tumour cell adhesion to LV (Ref. Reference Paupert, Sounni and Noel138). A morphological study of the interaction between M21 melanoma cells and LEC showed that tumour cells adhere via pseudopodia to LEC at the site near the intercellular junction. Upon tumour cell adhesion, the LEC junctions between LEC appear to dissociate and allow invasion by the tumour cells (Ref. Reference Ding139). A recent discovery showed that tumour cell derived 15-lipoxygenase-1 mediates LN metastasis by allowing the tumour cells to make circular holes in the lymphatic endothelium. Selective inhibition of 15-lipoxygenase-1 in tumour cells repressed formation of the circular defects in the LV and resulted in reduced LN metastasis (Ref. Reference Kerjaschki140).

LEC-cancer cell crosstalk can also regulate cancer cell proliferation. Interaction of the duffy antigen receptor for chemokines (DARC) in LEC with the Kallmann syndrome 1 (KAI1) ligand from cancer cells reduces cancer cell proliferation and metastatic progression. This model suggests that tumour-expressed KAI1 can have a tumour suppressing activity by interacting with the host LV (Ref. Reference Iiizumi, Bandyopadhyay and Watabe141). LEC conditioned by MDA-MB-231 breast TCM express EGF and PDGF-BB to promote tumour cell proliferation, angiogenesis and pericyte infiltration into the stroma (Ref. Reference Lee, Pandey and Popel142).

Recently LEC-induced immune modification and tumour immune tolerance has been reported (Refs Reference Swartz and Lund30, Reference Swartz143). Tumour cell spreading is mediated by sentinel, TDLN. Hence, the presence of tumour cells in the TDLN is used as an important prognostic indicator of metastasis in several cancers, such as breast cancer and skin cancer. Generally, LN are the primary organs that regulate host immunity, as diverse immune cells are travelling through the LV and the LN; lymphocytes are abundant in the T–B cell regions of the LN, and antigen-specific immune reactions normally occur in the LN. Thus, the LN can establish and maintain immunity against pathogens and possibly tumour cells. However, interestingly, many studies have shown that TDLN do not function as normal LN do against normal inflammation or infection. They instead appear to play a role as transient reservoirs of tumour cells for distal metastasis (Ref. Reference Zheng144). These results suggest that tumour cells or tumour secretions can modify the LN or lymphatic systems in the LN, thus impairing their original functions. Mechanistically, CCL21 expressed by the lymphatic endothelium can recruit CCR7-positive naive T cells into the lymphatic systems in the TDLN and tumour stroma where they are educated to be less immune reactive; similarly, CCR7-positive DC are maintained in their immature state [immature dendritic cells (iDC)] in the tumour and the TDLN. The iDC promotes tumour-associated regulatory T cell (T _reg) activity to suppress CTL (Ref. Reference Swartz and Lund30). Cytokines around the lymphatic endothelium in the various microenvironments also contribute to immune tolerance. High amounts of TGFβ are secreted by the tumour ECM, as the tumour matrix tension caused by interstitial flow influences TGFβ expression. TGFβ inhibits natural killer cell functions. TGFβ can also promote tumour-associated regulatory T cell (T _reg) and MDSC activities: these immune suppressive cell activities impair CTL and helper T1 cell (T _H1) functions, causing tumour cell immune tolerance (Ref. Reference Swartz and Lund30).

Therapeutic strategies to target blood and LV

Multiple strategies to target the tumour and organ microenvironments, particularly the BV and LV are in the clinic and in development. To target the BV in the microenvironment, several drugs, including the anti-VEGF monoclonal antibody bevacizumab, the receptor tyrosine kinase inhibitor sunitinib and the decoy receptor aflibercept (VEGF-trap) are clinically used. Anti-angiogenic monotherapy, such as anti-VEGF treatment, has shown limitation in clinic. This can be partially explained by that even after inhibiting single factor or pathway, other tumour-secreted factors or signals continue to support angiogenesis persistently (Refs Reference Crawford35, Reference Cascone145, Reference Jain146, Reference Kerbel147). New data on the combination of anti-angiogenic therapies and other pathway blockages showed improved outcomes. For example, anti-VEGF therapy with COX-2 inhibition more potently blocked the angiogenesis, compared with just anti-VEGF treatment by blocking COX-2 derived prostaglandin E2 (PGE₂) production. PGE₂ is a VEGF-independent pro-angiogenic factor secreted by COX-2 expressing tumour cells (Ref. Reference Xu148). No drugs are clinically used to target the LV but some are in preclinical development. They mostly target VEGFC/D, VEGFR3 and NRP2. Although anti-angiogenic therapies have shown promising outcomes in reducing tumour size, a large number of patients suffer from tumour metastasis facilitated by tumour LV which current anti-angiogenic agents are not able to inhibit. Specifically in breast cancer, tumour LV are the predominant routes of tumour dissemination. Thus current interest in the application of anti-angiogenic agents as cancer therapeutics has led to strategies combining inhibitors of angiogenesis and lymphangiogenesis with the goal of developing more effective anti-cancer therapies (Refs Reference Lee, Lee, Koskimaki, Han, Pandey and Popel62, Reference Lee105, Reference Lee149, Reference Koskimaki150).

Therapeutic and diagnostic values of the tumour–endothelial crosstalk signals

As described in this review, BEC and LEC secretomes play important roles in tumour progression, suggesting that we can target tumour–endothelial crosstalk signals in addition to targeting conventional tumour angiogenesis and lymphangiogenesis. In our previous study, we found that breast tumour cell secreted IL6 educated LEC in the pre-metastatic organs to express CCL5 and facilitate metastatic breast tumour recruitment into those organs (Ref. Reference Lee104). In that study, we blocked IL6 and CCR5 resulting in dramatic reduction in lung and LN metastasis. These signalling molecules governing tumour–endothelial crosstalk have a diagnostic value as well. Human sample-based bioinformatics analysis of TCGA (The Cancer Genome Atlas) showed higher levels of IL6 and CCL5 mRNAs in highly metastatic human triple-negative breast cancer (TNBC) tumour samples compared with those in less metastatic human oestrogen receptor positive (ER+) tumour samples. Furthermore, TNBC tumours obtained from LN positive patients showed better correlation of IL6 and CCL5 gene expression compared with those from LN negative patients. These results suggest that repurposing the anti-retroviral CCR5 inhibitor (maraviroc) or the anti-rheumatoid arthritis IL6 receptor antibody (tocilizumab) for advanced and metastatic breast cancer may have substantial clinical benefit (Ref. Reference Lee104). Also these repurposed drugs can be combined with conventional anti-angiogenic drugs (e.g. anti-VEGF antibodies) or chemotherapies. For example, we showed that an anti-VEGF antibody exhibited additive anti-metastatic effects when it was combined with maraviroc.