Research ArticleCancer Immunology

Lipocalin 2 from macrophages stimulated by tumor cell–derived sphingosine 1-phosphate promotes lymphangiogenesis and tumor metastasis

See allHide authors and affiliations

Science Signaling  28 Jun 2016:
Vol. 9, Issue 434, pp. ra64
DOI: 10.1126/scisignal.aaf3241

Building the route to metastasis

Infection triggers the release of the phospholipid S1P, which stimulates macrophages to launch a proinflammatory response, including secretion of the protein lipocalin 2 (LCN2). Dying tumor cells also release S1P. Using human primary cells and mouse models, Jung et al. discovered that a multicellular signaling circuit involving the release of S1P from dying breast tumor cells to stimulate release of LCN2 from tumor-associated macrophages stimulated the release of the lymphangiogenic factor VEGFC from lymphatic endothelial cells, thus promoting the growth of lymphatic vessels in the tumor microenvironment. Inhibiting the pathway suppressed lymphangiogenesis and metastasis in mice. The findings reveal not only therapeutic targets but also a way that therapy-induced tumor cell death and viral or bacterial infection could prime the microenvironment for metastasis (see also the Focus by Rodvold and Zanetti).


Tumor cell–derived factors skew macrophages toward a tumor-supporting phenotype associated with the secretion of protumorigenic mediators. Apoptosing tumor cells release sphingosine 1-phosphate (S1P), which stimulates the production of lipocalin 2 (LCN2) in tumor-associated macrophages and is associated with tumor metastasis. We explored the mechanism by which S1P induces LCN2 in macrophages and investigated how this contributed to tumor growth and metastasis. Knockdown of S1P receptor 1 (S1PR1) in primary human macrophages and experiments with bone marrow–derived macrophages from S1PR1-deficient mice showed that S1P signaled through S1PR1 to induce LCN2 expression. The LCN2 promoter contains a consensus sequence for signal transducer and activator of transcription 3 (STAT3), and deletion of the STAT3 recognition sequence reduced expression of an LCN2-controlled reporter gene. Conditioned medium from coculture experiments indicated that the release of LCN2 from macrophages induced tube formation and proliferation in cultures of primary human lymphatic endothelial cells in a manner dependent on the kinase PI3K and subsequent induction of the growth factor VEGFC, which functioned as an autocrine signal stimulating the receptor VEGFR3. Knockout of Lcn2 attenuated tumor-associated lymphangiogenesis and breast tumor metastasis both in the breast cancer model MMTV-PyMT mice and in mice bearing orthotopic wild-type tumors. Our findings indicate that macrophages respond to dying tumor cells by producing signals that promote lymphangiogenesis, which enables metastasis.


Macrophages exhibit a great heterogeneity and functional plasticity in response to microenvironmental stimuli. They are key players during inflammation, contributing not only to the initiation but also to the resolution phase. Their highly diverse functional potential has been considered crucial for the outcome of injury and highlights their pivotal role in maintaining tissue integrity. Macrophages are also associated with malignancies if their activation is not properly controlled, most commonly during chronic inflammatory diseases like cancer, or if they are reeducated by the tumor itself to support tissue vascularization and, thus, foster tumor growth (1).

It became apparent that not only tumor cell–intrinsic alterations promote tumor progression but also the tumor microenvironment. Tumor cells produce and secrete a number of factors that create a tumor-supportive microenvironment that, in turn, supports tumor cell proliferation, angiogenesis, and metastasis (2). It also contributes to evasion from immune surveillance. In many tumors, macrophages are found in high numbers, often correlating with a poor patient prognosis. These tumor-associated macrophages (TAMs) are reprogrammed by tumor cells to support their growth instead of attacking them. The complex TAM phenotype is established, at least in part, in response to tumor cell–released factors such as interleukin-10 (IL-10), transforming growth factor–β (TGF-β), prostanglandin E2 (PGE2), or chemokines as well as tumor hypoxia (3). However, signaling pathways provoking the differentiation and polarization of TAM are still not fully understood. It is believed that TAMs show features of an alternatively activated M2 macrophage phenotype (4). M2-polarized macrophages are considered to play a crucial role during tissue repair, which is mainly achieved through phagocytosis of apoptotic cells and the subsequent production of anti-inflammatory mediators. Previous studies from our group indicate that sphingosine 1-phosphate (S1P) is released during apoptotic cell death by sphingosine kinase 2 (SphK2) and participates in coordinating anti-inflammatory responses in macrophages (5). Within the tumor, a continuous interaction of immune cells and dying tumor cells was suggested, whereby macrophages get polarized toward a protumor M2-like phenotype. Apoptotic tumor cells attenuate macrophage cytotoxicity against viable tumor cells (6, 7). The immunomodulatory effect of apoptotic tumor cells might be attributed to specific recognition by macrophages and other phagocytes and the secretion of immunomodulating mediators such as S1P (8, 9). Along this line, a monoclonal antibody against S1P reduced tumor growth and attenuated tumor angiogenesis and tumor cell survival (10), thereby linking S1P production and tumorigenesis. Moreover, we showed that S1P, which is produced as a consequence of apoptotic cell death, elicits lipocalin 2 (LCN2) production in macrophages, which in turn fosters renal repair after ischemia reperfusion injury (11).

LCN2 is a 25-kD protein of the lipocalin superfamily, which is known to be highly present in kidney cells in response to stimuli such as ischemia or nephrotoxic agents. Moreover, application of exogenous LCN2 to epithelial cells promoted proliferation and the induction of early epithelial progenitor markers, suggesting that LCN2 may act as a growth and differentiation marker (12). Our previous data suggested that LCN2-induced renal cell regeneration demands an inflammatory microenvironment (13). Blocking LCN2 production in macrophages reduced the potency of a macrophage-based cell therapy approach in a kidney ischemia/reperfusion injury model, substantiating a pro-proliferative and anti-inflammatory role of LCN2 (14). Along these lines, we previously showed that LCN2 might promote tumor progression (15), which shares common features of tissue regeneration and repair. It has also been proposed that LCN2 modulates inflammatory macrophage responses by activating the nuclear peroxisome proliferator–activated receptor γ (PPARγ) that conveys anti-inflammatory macrophage responses (16, 17). In addition, LCN2 is highly expressed by lipopolysaccharide-tolerant macrophages (18).

On the basis of the pro-proliferative, pro-regenerative, and anti-inflammatory properties of LCN2, we hypothesized that the tumor-supportive microenvironment relies, at least in part, on the presence of LCN2 in TAMs. We aimed at elucidating molecular mechanisms of LCN2 production in primary macrophages in response to apoptotic cell–derived S1P, exploring macrophage-derived LCN2 in tumor development, and studying its impact on tumor neovascularization.


Apoptotic cancer cells induce the production and release of LCN2 in primary human macrophages through the S1P-S1PR1-STAT3 axis

To generate apoptotic cell–conditioned medium (ACM), tumor cells were stimulated with staurosporine (0.5 μg/ml) for 1 hour, washed with phosphate-buffered saline (PBS), and incubated overnight in RPMI. The supernatant of viable tumor cells (VCM) served as control. The production of LCN2 was induced at both the mRNA and protein levels in primary human macrophages cultured in ACM but not VCM from mammary epithelial (MCF10A) or noninvasive (MCF-7) or invasive (MDA-MB-231) breast cancer cells (Fig. 1, A to C). ACM-stimulated macrophages secreted significantly higher amounts of LCN2 compared to both viable and apoptotic tumor cells (fig. S1A compared to Fig. 1, A to C). LCN2 present in the supernatant of macrophage cultures was exclusively derived from macrophages, given that cells were stimulated with ACM for 6 hours and then washed and left in serum-free medium for additional 16 hours. As we previously observed LCN2 induction in macrophages by S1P released from apoptotic tumor cells, and because S1P synthesis during apoptosis is largely dependent on SphK2 (9), we isolated ACM from MCF-7 cells transfected with either control or SphK2-targeted small interfering RNA (siRNA). Macrophages incubated with the supernatant from apoptotic siSphK2–MCF-7 cells had significantly less induction of LCN2 expression than did those incubated with the ACM from scrambled RNA–treated MCF-7 cells (Fig. 1D). Furthermore, the addition of 100 nM S1P to siSphK2 ACM restored the induction of LCN2 in macrophages, which is evident at the levels of mRNA abundance (Fig. 1D, left), promoter activity (Fig. 1D, middle), and protein abundance (Fig. 1D, right). Thus, the data indicate that S1P induces LCN2 expression and secretion in primary human macrophages through a transcriptional mechanism.

Fig. 1 Apoptotic cell–derived S1P induces LCN2 in primary human macrophages.

(A to C) LCN2 expression [by quantitative real-time polymerase chain reaction (qRT-PCR), top] and LCN2 secretion [by enzyme-linked immunosorbent assay (ELISA); bottom] in primary human macrophages cultured for 6 hours in control (“C”) or conditioned medium from apoptotic (ACM) or viable (VCM) (A) MCF10A, (B) MCF-7, or (C) MDA-MB-231 cells, and then washed and incubated in serum-starved RPMI medium for additional 16 hours. (D) LCN2 expression (by qRT-PCR, left), luciferase reporter activity of the LCN2 promoter (middle), and LCN2 secretion (by ELISA, right) in primary human macrophages cultured for 24 hours with conditioned medium from apoptotic control (scRNA) or SphK2-deficient (siSPHK2) MCF-7 cells in the presence or absence of authentic S1P (100 nM) for 24 hours. RLU, relative light units. Data are means ± SD; n = 6 biological replicate (each analyzed in triplicate); *P < 0.05, **P < 0.01, ***P < 0.001 versus control or as indicated.

We went on to investigate whether the S1P receptor (S1PR) was involved by assessing the relative expression of S1PR1 to S1PR4 in primary human macrophages (fig. S1B). The expression of S1PR3 was lower than that of S1PR1, S1PR2, and S1PR4 in primary human macrophages; the expression of S1PR3 was absent in primary mouse macrophages. Therefore, we decided to focus on the role of S1PR1, S1PR2, and S1PR4 in LCN2 production in macrophages. First, we knocked down S1PR1, S1PR2, and S1PR4 in primary human macrophages, the efficiency of which varied from 40 to 60% (fig. S2, A to C). The stimulation of S1PR-deficient macrophages with ACM implicated S1PR1 in the observed LCN2 induction in primary human macrophages at the mRNA (Fig. 2A, top) and LCN2 protein level (Fig. 2A, bottom) in the respective supernatants. Next, we generated primary mouse bone marrow–derived macrophages (BMDMs) from S1PR1fl/fl F4/80Cre/+ (19), S1PR2−/−, and S1PR4−/− mice. Stimulation of these macrophages with ACM confirmed that mainly S1PR1 mediates S1P-induced LCN2 expression (Fig. 2B, top) and release (Fig. 2B, bottom). To study S1P-S1PR1 downstream signaling, we further examined the induction of LCN2 in primary human macrophages after the treatment with 100 nM authentic S1P. LCN2 mRNA expression was induced 2 hours after S1P treatment, declined after 6 hours, and remained low for up to 24 hours (Fig. 2C, top). However, protein amounts increased after 8 hours and remained stable for at least 24 hours (Fig. 2C, bottom). We then analyzed signaling pathways that cause S1P-mediated LCN2 expression in primary human macrophages by using pharmacological inhibition. Macrophages were pretreated with the indicated inhibitors (all at 10 μM) for 1 hour, washed once with PBS, and then stimulated with S1P (100 nM) for 8 hours. S1P-induced LCN2 production (at both mRNA and protein levels) was not blocked by pretreatment with the phosphoinositide 3-kinase (PI3K) inhibitor LY29004, the p38 kinase inhibitor SB239063, or the mitogen-activated protein kinase kinase (MEK) and extracellular signal–regulated kinase (ERK) inhibitor PD98059 (Fig. 2D). In contrast, a significant suppression of LCN2 induction was achieved by pretreatment with the signal transducer and activator of transcription 3 (STAT3) inhibitor STATIC (6-nitrobenzo[b]thiophene-1,1-dioxide). The phosphorylation of STAT3 after S1P stimulation was also suppressed by STATIC (Fig. 2E). To further strengthen the role of STAT3 in LCN2 induction, we performed LCN2 promoter analysis in macrophages. The full-length LCN2 promoter-reporter construct was activated in response to S1P after 3 hours, but the same promoter construct with a point mutation in the putative STAT3 binding site (15) significantly reduced S1P-dependent activation (Fig. 2F).

Fig. 2 S1P induces LCN2 in primary human macrophages through the S1PR1-STAT3 axis.

(A) Macrophages (MΦ) were transfected with siRNA targeting S1PR1, S1PR2, S1PR4, or a scrambled control scRNA and were treated with either supernatant of apoptotic (ACM) or viable (VCM) MCF-7 cells. LCN2 mRNA (upper panel) was quantified by qRT-PCR, and protein was measured by ELISA. (B) Mouse BMDMs (mouse MΦ) were isolated from S1PR1−/−, S1PR2−/−, and S1PR4−/− mice, stimulated with ACM or VCM of MC57G cells. Lcn2 mRNA (upper panel) was quantified, and protein release was measured. (C) LCN2 mRNA (top) and protein release (bottom) in human macrophages stimulated with S1P (100 nM). (D) LCN2 mRNA (top) and protein (bottom) in macrophages pretreated with inhibitors of STAT3 (STATIC), MEK/ERK [PD98059 (PD)], PI3K [LY294002 (LY)], or p38 [SB203580 (SB)] and then treated with S1P. (E) Western blot of phosphorylated STAT3 (pSTAT3) compared to total STAT3 in macrophages. (F) Luciferase reporter assay of a full-length LCN2 promoter construct in comparison to the construct containing a mutated STAT3 binding site. Data are means ± SD; n = 6 biological replicates (in triplicate); *P < 0.05, **P < 0.01, ***P < 0.001 versus control or as indicated.

Macrophage-derived LCN2 promotes lymphangiogenesis behavior in culture

To explore the biological effects of macrophage-derived LCN2, primary human macrophages were incubated with VCM and ACM as described above. The resulting macrophage supernatants, hereinafter called MCMV (that from macrophage cultures treated with VCM) and MCM (that from macrophage cultures treated with ACM), were then tested for their potential to promote angiogenesis and lymphangiogenesis in a culture system. First, we explored how LCN2 affects pluripotent embryonic stem cell differentiation toward blood or lymphatic vessel–like structures by conducting sprouting assays using murine CGR8 stem cells embedded in a collagen matrix. The addition of recombinant murine LCN2 (1 μg/ml) induced sprouting in CGR8 embryonic bodies compared to unstimulated control cells (Fig. 3A). Recombinant murine vascular endothelial growth factor A (VEGFA) and recombinant human VEGFC were used as positive controls (100 ng/ml each), respectively. To test whether LCN2 is the crucial factor in MCM-induced sprouting, we blocked LCN2 in MCM using a specific antibody against LCN2 or supplemented MCMV with recombinant murine LCN2. Blocking LCN2 in MCM reduced the number of sprouts, whereas the addition of LCN2 to MCMV increased the number of sprouts. The cells from the sprouting assays were then analyzed by flow cytometry. Platelet endothelial cell adhesion molecule 1 (PECAM-1) staining indicates endothelial cells and angiogenesis, whereas podoplanin and lymphatic vessel endothelial hyaluronan receptor 1 (LYVE-1) staining indicates lymphatic cells and lymphangiogenesis. Immune cells bearing these markers were excluded by analyzing only CD45 cells. Unlike the positive controls (VEGFA and VEGFC, respectively), addition of recombinant murine LCN2 did not have a significant effect on the number of PECAM-1–positive cells (Fig. 3B), whereas it significantly increased the number of LYVE-1 and podoplanin double-positive cells (Fig. 3C). In contrast to MCMV, MCM increased differentiation toward both blood and lymphatic endothelial cells (LECs) (Fig. 3, B and C). Neutralizing LCN2 in MCM significantly reduced the emergence of LYVE-1 and podoplanin double-positive (lymphatic endothelial) cells, whereas supplementing MCMV with recombinant murine LCN2 enhanced the emergence of LECs. These observations suggest that macrophage-derived LCN2 promotes lymphangiogenesis rather than angiogenesis.

Fig. 3 Macrophage-derived LCN2 induces lymphangiogenesis in vitro.

(A) Sprouted embryonic bodies of CGR8 stem cells embedded in a collagen matrix and stimulated with supernatants from mouse macrophages that had been cultured with ACM (MCM) or VCM (MCMV), MCM supplemented with a neutralizing antibody against LCN2 (MCM_αLCN2), or MCMV supplemented with recombinant LCN2 (MCMV_LCN2). VEGFA, VEGFC, and LCN2 served as positive controls. Scale bar, 50 μm. (B and C) PECAM-1–positive (+) (B) or LYVE-1/podoplanin–positive (C) cells were quantified by flow cytometry. (D) Tube formation, analyzed by the number of master junctions and the average tube length, in LECs stimulated with VEGFC or LCN2. (E) Transwell assay assessing the migration of LECs toward LCN2 or VEGFC. (F) LEC proliferation measured using xCELLigence and analyzed as slope 1/h. (G and H) Schematic of the assay (G) and analysis of proliferation [H; analyzed as in (F)] in LECs cultured in conditioned medium (CM) from macrophages (M) or macrophages (wild type, scRNA-transfected, or siLCN2–transfected) cocultured with MCF-7 cells (Mcocu). C, control LECs. Data are means ± SD; n = 5 biological replicates (each analyzed in triplicate); *P < 0.05, **P < 0.01 versus control or as indicated.

We then analyzed the effect of LCN2 on lymphatic vessel growth using human LECs. LECs dynamically organize into a network of tubes in culture and exhibit characteristics of lymphatic vessel growth in vivo, including proliferation, migration, and sprouting. LEC tube formation was examined after stimulation with recombinant human LCN2 (1 μg/ml) or human VEGFC (500 ng/ml) as a positive control (Fig. 3D, left). LCN2 enhanced tube formation, as measured by an increased number of master junctions and tube length compared to unstimulated control cells (Fig. 3D, right). Moreover, the degree of tube formation was comparable to VEGFC, which was used as a positive control. These findings suggest LCN2 to promote lymphangiogenesis. A Boyden chamber assay revealed that LEC migration was significantly enhanced by LCN2 addition to the lower chamber (Fig. 3E), which was recapitulated by the positive control VEGFC. We then tested whether LCN2 promotes lymphangiogenesis through enhancing proliferation by measuring LEC proliferation in real time using the xCELLigence system. Recombinant human LCN2 induces proliferation of LEC to a similar extent as the positive control VEGFC (Fig. 3F). Given that macrophage-derived LCN2 is induced by apoptotic tumor cell–released S1P and taking into account that S1P itself is a potent lymphangiogenic factor (20), we compared the effects of authentic S1P (500 nM) and recombinant human LCN2 (1 μg/ml) on lymphangiogenesis in cultured LECs. Both S1P and LCN2 induce tube formation of human LEC (fig. S3, A and B), but only LCN2 enhanced LEC migration (fig. S3C) and proliferation (fig. S3D). We further assessed whether this effect was purely dependent on macrophage-derived LCN2 or whether tumor cell–derived LCN2 might also play a role in this setting. Therefore, we cocultured control or LCN2-deficient primary macrophages (fig. S4, A and B) either with control or stable LCN2-deficient MCF-7 breast cancer cells (fig. S4, C and D) and applied the conditioned medium (supernatant) from those cocultures to LECs (Fig. 3G). The supernatant from control macrophage cultures did not enhance LEC proliferation, whereas knocking down LCN2 in macrophages prevented coculture supernatant from inducing proliferation in LECs (Fig. 3H), suggesting that coculture-elicited LCN2 expression in macrophages was crucial for LEC proliferation. The supernatant from stable LCN2-deficient tumor cells had no effect on LEC proliferation (fig. S4E). Furthermore, we tested different cell lines with varying amounts of LCN2 (fig. S1A) on their ability to induce proliferation in LEC and found that none of the tested cell lines induced LEC proliferation (fig. S4F) compared to macrophage-derived LCN2. Together, these findings suggest a pivotal role for TAM-derived LCN2 in promoting lymphangiogenesis by enhancing LEC proliferation.

LCN2 promotes LEC proliferation indirectly through the VEGFC-VEGFR3 axis

We hypothesized that LCN2 might directly induce the expression of lymphangiogenesis-associated genes, such as VEGFC. Therefore, we screened for VEGFC, VEGFD, VEGFR3, HoxD8, and Prox-1 mRNA expression after stimulating LEC with recombinant LCN2 at the indicated concentrations (Fig. 4A and fig. S5, A to D). LCN2 dose-dependently induced VEGFC, reaching maximal effects at 1 μg/ml of recombinant protein, but did not affect the expression of any other lymphangiogenesis-related gene. Therefore, we focused on the induction of VEGFC. We performed luciferase reporter assays to explore whether LCN2 increases VEGFC at the transcriptional level (Fig. 4B). LEC transfected with a full-length VEGFC promoter construct showed enhanced promoter activity upon LCN2 (1 μg/ml) treatment after 24 hours compared to those transfected with a control vector. VEGFC protein measurements using ELISA corroborated a significant response in LCN2-stimulated LECs (Fig. 4C). We then used pharmacological inhibitors to analyze the signaling pathways involved in LCN2-mediated VEGFC expression. A significant reduction in VEGFC at both the mRNA (Fig. 4D) and protein (Fig. 4E) level was achieved only with the PI3K inhibitors LY294002 and wortmannin, suggesting a PI3K-dependent pathway in LCN2-induced VEGFC expression in LECs. Blocking VEGFC or VEGFR3, the VEGFC-specific receptor, with neutralizing antibodies attenuated LEC proliferation (Fig. 4F), further suggesting that the pro-proliferative effect of LCN2 seen in LECs resulted from VEGFC production and autocrine signaling.

Fig. 4 LCN2 promotes lymphatic vessel growth through the VEGFC-VEGFR3 axis.

(A) qRT-PCR for VEGFC mRNA in human LECs in response to LCN2. (B and C) VEGFC promoter activity (B; displayed as RLU) measured by luciferase reporter assay and (C) VEGFC secretion quantified by ELISA. (D and E) VEGFC expression (D) and VEGFC secretion (E) in LECs pretreated with inhibitors of SRC (PP2), p38 (SB203580), PI3K (LY294002 and wortmannin), protein kinase C (PKC) (Gö6976), c-Jun N-terminal kinase (JNK) (SP600125), or MEK/ERK (PD98059) before stimulation with LCN2. (F) LEC proliferation was measured by xCELLigence after stimulation with VEGFC or LCN2 in combination with neutralizing antibodies against VEGFC, VEGFR3, or a control immunoglobulin G (IgG) antibody. (G) LYVE-1 and PECAM-1 staining (represented as % vascular coverage) of Matrigel containing VEGFC or LCN2 alone or in combination with a neutralizing VEGFR3 or the IgG control antibody injected into C57BL/6 mice. Culture medium served as the negative control. Scale bar, 200 μm. DAPI, 4′,6-diamidino-2-phenylindole. (H) Quantification of vascular (VEC) and lymphatic (LEC) endothelial cells using flow cytometry analysis. Data are means ± SD; n = 5 biological replicates [(A) to (F); in triplicate] or mice [(G) and (H); in duplicate for fluorescence-activated cell sorting (FACS) analysis and five different areas per section]; *P < 0.05, **P < 0.01, ***P < 0.001 versus control or as indicated.

To assess LCN2-induced vessel formation in vivo, we performed Matrigel plug assays in C57BL/6 mice using Matrigel supplemented with recombinant LCN2 (2 μg/ml) or recombinant VEGFC (500 ng/ml) as a positive control. We were not able to discriminate single vessels for the quantification of the lymphatic vessel density as a routinely used parameter but rather recognized only parts of a vascular network. Therefore, we analyzed the relative percentage of blood or lymphatic vascular coverage in five fields of 2 mm2 each. LCN2 markedly increased vascularization and significantly enhanced the percent coverage of lymphatic (LYVE-1–positive) but not vascular (PECAM-1–positive) vessels, as assessed by immunofluorescence quantification (Fig. 4G, left) and flow cytometry analysis (Fig. 4H, left) of LYVE-1– or PECAM-1–positive, CD45-negative cells. To check the previously observed effect of LCN2 on the VEGFC-VEGFR3 axis, we performed Matrigel plug assays using Matrigel supplemented with recombinant LCN2 in combination with either a neutralizing antibody against VEGFR3 or the IgG matching isotype control. Immunofluorescence staining (Fig. 4G, right) and flow cytometry analysis (Fig. 4H, right) revealed that blocking VEGFR3 lowered the number of LECs (as well as VECs) per plug. Collectively, these results indicate an LCN2-dependent induction of VEGFC through PI3K that promotes proliferation of LECs by VEGFR3 signaling. In addition, we tested whether LCN2 also had an effect on the angiopoietin (ANG) and tunica interna endothelial cell kinase (TIE) pathway. Therefore, we screened the mRNA expression of ANG1, ANG2, TIE1, and TIE2 after stimulating LECs with recombinant LCN2 at different concentrations (fig. S5, E to H). Results showed a rather specific and dose-dependent effect of LCN2 on the induction of ANG2, whereas the expression of ANG1, TIE1, and TIE2 was unchanged, thus suggesting that modulation of the ANG2-dependent signaling pathway might provide another route of LCN2-dependent lymphangiogenesis.

LCN2-induced tumor lymphangiogenesis correlates with reduced survival and enhanced metastasis

To investigate whether LCN2 affects tumor growth, metastasis, and neovascularization in PyMT breast cancer model mice, we crossed LCN2-deficient mice into the PyMT background to generate female mice that develop spontaneous breast cancer. Tumors were first observed 8 weeks after birth, and tumor development was monitored until sacrifice (when the largest tumor reached 1.5 cm in size). Kaplan-Meier survival curve analysis revealed that LCN2−/−/PyMT mice had significantly better survival than did wild-type PyMT mice (P = 0.0006, χ2 test; Fig. 5A). In addition, the distribution of lung metastases revealed that wild-type PyMT mice had a significantly greater heterogeneous distribution and incidence of metastases (P = 0.0495, χ2 test; Fig. 5B). We further analyzed the percent coverage of PECAM-1–positive (Fig. 5C, left) and LYVE-1/podoplanin–positive (Fig. 5C, right) cells in the tumor by flow cytometry. Results corroborated our in vitro findings regarding the role for LCN2 in promoting lymphangiogenesis. We observed no difference in PECAM-1–positive cells in LCN2−/−/PyMT mice compared to wild-type PyMT mice. However, the amount of LYVE-1 and podoplanin double-positive cells was significantly reduced in LCN2−/−/PyMT mice. These findings were substantiated by immunohistochemical staining for PECAM-1 and LYVE-1 in tumor sections from wild-type PyMT and LCN2−/−/PyMT mice (Fig. 5D, left). Quantification of positively stained areas and the percent vascular coverage revealed a significant reduction of LYVE-1 staining in LCN2−/−/PyMT mice compared to wild-type PyMT mice (Fig. 5D, right).

Fig. 5 LCN2 ablation significantly increases survival and reduces tumor metastasis and lymphangiogenesis in the PyMT breast cancer model.

(A) The appearance of mammary tumors was determined by palpation twice a week and the survival of tumor-bearing mice as Kaplan-Meier survival curve (P = 0.0006, log-rank test, χ2). n = 12 mice each. (B) The distribution of lung metastases was analyzed in wild-type PyMT (n = 12) mice and LCN2−/−/PyMT (n = 8) mice. (P = 0.0495, χ2 test). (C) Whole-tumor homogenates from wild-type PyMT mice (n = 12) and LCN2−/−/PyMT mice (n = 8) were analyzed by flow cytometry (in duplicate) for (left) PECAM-1–positive VECs and (right) LYVE-1 and podoplanin double-positive LECs. (D) Tumors from wild-type and LCN2 knockout mice were embedded in paraffin and analyzed for LYVE-1 and PECAM-1 by immunofluorescence staining. Quantification of LYVE-1 and PECAM-1 is represented as % vascular coverage per field. Data are means ± SD; n = 5 mice, each analyzed in five different areas; *P < 0.05, **P < 0.01, ***P < 0.001 versus wild-type PyMT mice, by Student’s t test.

To evaluate the impact of macrophage-derived LCN2 on tumor neovascularization, we used an orthotopic breast cancer model as depicted in Fig. 6A. Therefore, we isolated tumor cells from both wild-type PyMT and LCN2−/−/PyMT mice (donors) and implanted them into either wild-type or LCN2−/− C57BL/6 mice (recipients). Within this model, the implantation of wild-type and LCN2−/− tumor cells enabled us to discriminate effects of tumor cell–derived LCN2, whereas implantation of wild-type tumor cells into wild-type and LCN2−/− recipients informs stroma cell–derived LCN2 effects. Our results revealed that the implantation of either wild-type or LCN2−/− tumor cells had no effect on the extent of tumor angiogenesis as determined by immunostaining of PECAM-1 (Fig. 6, B and C, left). However, the implantation of either wild-type and LCN2−/− tumor cells resulted in significantly less LYVE-1–positive staining in primary tumors in LCN2−/− recipients (Fig. 6, B and C, right). Findings from the orthotopic model corroborated the results from the PyMT model, in which LCN2-deficient PyMT mice revealed significantly reduced lymphangiogenesis (Fig. 6C). Thus, our results suggest that stroma-derived, but not tumor cell–derived, LCN2 enhances tumor lymphangiogenesis, as depicted in a graphical summary (Fig. 6D).

Fig. 6 Tumor stroma–derived LCN2 promotes tumor lymphangiogenesis.

(A) Schematic of tumor implant assays. Wild-type PyMT and LCN2−/−/PyMT donors were sacrificed, and CD45 and CD326+ tumor cells were isolated by FACS. Tumor cells were implanted into the mammary gland of wild-type mice (n = 6) and LCN2−/− recipient mice (n = 5). (B and C) Immunohistochemistry for PECAM-1 (green) and LYVE-1 (red) in paraffin-embedded tumors from recipient mice. Data are means ± SD; n = 5 mice, each analyzed in five different areas; *P < 0.05, **P < 0.01, ***P < 0.001 versus wild types. (D) Model describing the mechanism of lymphangiogenesis stimulated by intercellular and autocrine signaling among tumor cells, macrophages, and LECs.


This study supports the emerging concept that LCN2 is a critical player during tumor development. We provide evidence that TAMs produce and secrete this growth-promoting mediator in response to their activation by apoptotic cell–released S1P. Furthermore, we unraveled the role of LCN2 in tumor lymphangiogenesis by activating an autocrine VEGFC-VEGFR3 loop in LECs.

Interactions between tumor cells and TAMs are mediated by tumor cell–secreted factors, promoting a protumor phenotype of TAMs. We previously described S1P, which is released from apoptotic tumor cells, as one of the macrophage-polarizing factors supporting tumor progression (21, 22). Here, we defined the downstream signaling pathways involving the induction of LCN2 production in macrophages. The effects of S1P largely depend on one of its five specific G protein (heterotrimeric guanine nucleotide–binding protein)–coupled receptors. S1PR1 seems to be the most abundant one, being highly expressed in various cell types, including macrophages and LECs. A critical role of S1PR1 was previously described for lymphocyte egress and chemotaxis, cell survival, and increased tumor angiogenesis and metastasis (10, 2328). Furthermore, higher amounts of circulating S1P are associated with increased tumor growth (20). Both SphK1 and S1P are crucial for a proper lymphatic vessel development (29). In this sense, targeting of SphK1 using an inhibitor approach efficiently blocked tumor lymphangiogenesis, which was correlated to reduced tumor burden and metastatic spread (20). Here, we observed enhanced lymphangiogenesis in vitro, reflected by the induction tube formation upon stimulation with S1P. However, we did not observe an S1P-dependent induction of migration or proliferation of human LECs, which is in line with previous reports (30).

S1P-S1PR1 signaling was proposed to maintain persistent STAT3 activation and to promote malignant primary tumor growth (31) and premetastatic niche formation (23). Furthermore, persistent STAT3 activation, especially in myeloid cells, is associated with enhanced immunosuppression, proliferation/survival, angiogenesis, and tumor metastasis (23, 32, 33). Here, we observed a critical role for both S1PR1 and STAT3 in enhancing LCN2 production in macrophages, thereby promoting protumor responses in macrophages. Along this line, we previously observed that LCN2 secretion from primary human macrophages is correlated with prominent effects on cancer cell proliferation (15). LCN2 production was significantly higher in ACM-treated macrophages than in tumor cells, both under viable (VCM) or apoptotic (ACM) conditions (fig. S1A). Therefore, it might be hypothesized that the local, TAM-derived LCN2 production adds to their tumor-promoting capacity, especially at the invasive edge. Along this line, we showed that stroma-derived, but not tumor-derived, LCN2 was associated with enhanced lung metastasis in a transplantable, mammary tumor model (34). This corroborates our results regarding tumor lymphangiogenesis (Fig. 6). However, we cannot exclude the possibility that tumor cell–derived LCN2 contributes to the effects, which might be a critical consideration in the development of novel therapeutic approaches.

Detection of LCN2 in tumor cells has been postulated as a poor prognosis marker (3537). Overexpression of LCN2 in noninvasive human MCF-7 breast cancer cells promotes an aggressive phenotype that is associated with increased growth and metastasis (38). Most studies, however, focused on the role of LCN2 in stabilizing matrix metalloproteinase-9 (MMP-9) to explain its link with cancer metastasis, in close association with remodeling of the extracellular matrix, migration, and invasion. Overexpression of LCN2 in breast cancer cells in a xenograft model stabilizes MMP-9 and significantly increases the amount of newly formed CD34-positive vessels within the tumor (39). However, there is also evidence that LCN2 promotes angiogenesis independent of MMP-9 through enhanced VEGFA production due to the activation of ERK1 and ERK2 and the stabilization of HIF-1α (40). These findings suggest a role for LCN2 in tumor neovascularization. Our results corroborate these data, showing that reduced tumor neovascularization was associated with reduced tumor growth and significantly reduced metastasis in LCN2−/−/PyMT mice. However, we observed an LCN2-dependent promotion of tumor lymphangiogenesis through enhanced production of VEGFC in LECs but observed no effect on angiogenesis. Because LCN2 conveys protumorigenic properties, we propose that the induction of lymphangiogenesis as a key event in tumor metastasis relies, at least in part, on LCN2, which is produced by TAM. This speculation was further corroborated by our observations using an orthotopic breast cancer model in vivo and coculture assays in which tumor stroma–derived LCN2 specifically enhanced tumor lymphangiogenesis, whereas angiogenesis was unaffected.

Induction of lymphangiogenesis is one of the driving mechanisms for cancer cell dissemination. Previous studies speculate that tumor cells use the existing lymphatic vascular system in the tumor periphery as a “launching pad” for systemic metastasis (41, 42). In addition, the detection of lymph node metastases is currently considered as evidence of tumor cell transit, and the density of lymphatic vessels in the tumor is correlated with poor prognosis. Furthermore, it was shown that TAMs promote lymphangiogenesis through the production of growth factors, such as VEGFC (43). A strong correlation between VEGFC detection in primary tumors and lymph node metastasis was found in a variety of cancers, including the breast (44). It was speculated that VEGFC might play a crucial role in the progressive growth of human cancers through both autocrine and paracrine mechanisms (43, 4548). At present, we cannot exclude the involvement of TAM-derived growth factors in promoting breast cancer lymphangiogenesis. However, our data suggest a dominant mechanism involving LCN2 by enhancing the induction of VEGFC in LEC, thereby activating an autocrine feedback loop through its specific receptor VEGFR3. It was postulated that migration and proliferation of LEC are mainly regulated through the VEGFC-VEGFR3 axis, thereby contributing to the development of the lymphatic vascular system in the tumor periphery and enhanced tumor metastasis (49). These observations are in line with our results, suggesting a pivotal role of VEGFR3 in LCN2-mediated tumor lymphangiogenesis. We conclude that LCN2 enhances a protumor lymphatic vessel growth through an autocrine feedback loop through the VEGFC-VEGFR3 axis. We propose that LCN2 might specifically enhance VEGFC expression to foster lymphangiogenesis because other related growth factors such as VEGFD were not affected. Furthermore, we have indications that LCN2-induced VEGFC expression is mediated by the PI3K/AKT pathway. To the best of our knowledge, there is still a lack of information on the mechanistic insights of VEGFC induction in LECs, and more experiments are needed to define the exact molecular pathway of LCN2-dependent VEGFC induction, which is part of ongoing research in our laboratory. However, our results are in line with previous reports in tumor cells showing the implication of the PI3K pathway in the regulation of VEGFC (50, 51), pointing to the possibility for a common regulatory pathway in terms of VEGFC regulation. However, another possibility might be the involvement of the ANG-TIE pathway. We found that Ang2 expression was induced in LECs in response to LCN2 (fig. S5F). The balance of these proteins is essential to determine the vascular phenotype. ANG2 is thought to be secreted primarily by endothelial cells at sites of vascular remodeling. Yet, increased ANG2 within the tumor microenvironment is also associated with enhanced lymph node metastasis, especially in breast cancer patients (52). Moreover, in preclinical models, ANG2 promotes tumor cell dissemination through an integrin-dependent pathway (53). Given that ANG2 was also identified as an essential regulator of cell-cell junctions that form during lymphatic development (54, 55), one might speculate a role for ANG2 in tumor neovascularization, specifically that ANG2 destabilizes newly formed blood and lymphatic vessels within the tumor, thereby enhancing vascular permeability as a prerequisite of tumor cell dissemination. Taking the putative involvement of ANG2 in opening tumor blood and lymphatic vessels and the prominent role of LCN2 for tumor metastasis into account, further investigations should examine the link between LCN2 and ANG2 expression in terms of metastatic spread. Regarding cancer progression, we are currently exploring the molecular mechanisms underlying the metastasis-promoting effects of TAM-derived LCN2, considering the importance of LCN2 in tumor lymphangiogenesis, which might be part of the initial cascade leading to cancer metastasis. Understanding molecular and genetic mechanisms that control tumor-promoting genes such as LCN2 in TAMs could offer new perspectives for efficient therapy. Nevertheless, further research is needed to define the diverse biological effects of LCN2 within the tumor microenvironment.



Investigations were conducted in accordance with the ethical standards and according to the Declaration of Helsinki and to the national and international guidelines and have been approved by the authors’ institutional review board.


Wortmannin, SP600125, STATIC, and PD98059 were bought from Sigma-Aldrich. SB203580, Gö6976, and LY294002 were ordered from Alexis Biochemicals. PP2 was purchased from Enzo Life Sciences. Cell culture supplements and fetal calf serum (FCS) were ordered from PAA Laboratories. Hydrocortisone, epidermal growth factor (EGF), insulin, and cholera toxin for MCF10A culture medium were bought from Sigma-Aldrich. Primers were bought from BioMers. Specific antibodies against VEGFC and isotype-matching IgG antibodies came from Santa Cruz Biotechnology, antibodies recognizing LCN2 and VEGFR3 were purchased from R&D Systems, and antibodies against STAT3 and pSTAT3 came from Cell Signaling. Specific antibodies for LYVE-1 (R&D Systems) and PECAM-1 (BD Biosciences) were used for immunofluorescence stainings. The siRNA against SphK2, LCN2, S1PR1, S1PR2, and S1PR4 and scrambled nontargeting siRNA were from Qiagen. Recombinant mouse LCN2 was purchased from R&D Systems, and recombinant human LCN2 was produced in the laboratory by transformation of Escherichia coli with a pGEX-4T3-NGAL plasmid containing an insert for human LCN2 tagged to glutathione S-transferase. The LCN2 expression was initiated by supplementing isopropyl-ß-d-thiogalactopyranoside (Sigma-Aldrich) to the bacterial culture for activation of the lac operon. LCN2 was purified using ProCatch Glutathione Resin (Miltenyi Biotec) standard protocols. Detoxi-Gel Endotoxin Removing Columns (Thermo Scientific) were used for eliminating bacterial endotoxins according to the manufacturers’ instructions. All chemicals were of the highest grade of purity and commercially available.

Primary macrophage generation

Human monocytes were isolated from buffy coats (DRK-Blutspendedienst Baden-Württemberg-Hessen, Frankfurt, Germany) using Ficoll-Hypaque gradients (PAA Laboratories). Peripheral blood mononuclear cells were washed twice with PBS containing 2 mM EDTA and subsequently incubated for 1 hour at 37°C in RPMI 1640 medium supplemented with penicillin (100 U/ml) and streptomycin (100 μg/ml) to allow their adherence to culture dishes (Sarstedt). Nonadherent cells were removed. Monocytes were then differentiated into macrophages with RPMI 1640 medium containing 5% AB-positive human serum (DRK-Blutspendedienst Baden-Württemberg-Hessen, Frankfurt, Germany) for 10 days and achieved about 80% confluence.

Generation of BMDMs

BMDMs were generated as described previously (56). Briefly, bone marrow from 8- to 12-week-old wild-type and S1PR1−/−, S1PR2−/−, and S1PR4−/− mice was isolated, and cells were differentiated directly in six-well plates (3 × 106 cells per well) in the presence of macrophage colony-stimulating factor (M-CSF; 10 ng/ml) and granulocyte-macrophage colony-stimulating factor (GM-CSF; 10 ng/ml) for up to 7 days. At day 4, fresh M-CSF and GM-CSF were added. Macrophages were identified using F4/80 by flow cytometry.

Preparation of conditioned medium

Apoptosis in MCF-7, MCF10A, MDA-MB-231, and SphK2-deficient MCF-7 cells (22) cultured in medium without FCS was induced with staurosporine (0.5 μg/ml; Sigma-Aldrich) for 1 hour. Cells were washed twice with PBS and incubated for another 16-hour period in RPMI 1640 medium. ACM was harvested by centrifugation (1000g, 10 min) and filtered through 0.22-μm pore membranes (Millipore) to remove large particles and apoptotic bodies. VCM was produced accordingly without the addition of staurosporine. To generate macrophage-conditioned medium, primary human macrophages were incubated with the supernatant of tumor cells for 6 hours, washed twice with PBS, and incubated for another 16-hour time period with RPMI medium to generate MCM (macrophage-conditioned medium stimulated with supernatant of apoptotic tumor cells) or MCMV (macrophage-conditioned medium stimulated with supernatant of viable tumor cells) or remained as unstimulated controls (C). Macrophage-conditioned medium was likewise harvested by centrifugation and filtration.

Cancer cells

Human breast cancer cell lines MCF-7 and MDA-MB-231 and mouse MC57G fibrosarcoma cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with high glucose, 15 mM Hepes, and l-glutamine, supplemented with penicillin (100 U/ml), streptomycin (100 μg/ml), and 10% FCS. The breast epithelial cell line MCF10A was cultured in DMEM with high glucose, 15 mM Hepes, and l-glutamine, supplemented with penicillin (100 U/ml), streptomycin (100 μg/ml), hydrocortisone (100 ng/ml), insulin (10 μg/ml), EGF (100 ng/ml), cholera toxin (100 ng/ml), and 10% FCS. Cells were kept in a humidified atmosphere of 5% CO2 in air at 37°C and were passaged every 2 days.

Coculture of primary human macrophages and MCF-7 breast cancer cells

Primary human macrophages were isolated and differentiated as described above. For coculture with tumor cells, MCF-7 cells were harvested by trypsin, washed once with PBS, and resuspended in macrophage medium. Macrophages and cancer cells were cultured in a 1:1 ratio for 48 hours in macrophage medium in a humidified atmosphere of 5% CO2 in air at 37°C. At the end of the coculture, the conditioned medium from the coculture was harvested and stored for further use at −80°C. The remaining MCF-7 cells were removed by applying trypsin-EDTA, and macrophages were further processed for the detection of LCN2 mRNA expression.

Mouse embryonic stem cells

Mouse embryonic stem cells (CGR8) were cultured in Glasgow minimum essential medium with 10% heat-inactivated fetal bovine serum, 2 mM l-glutamine, 50 μM β-mercaptoethanol, and leukemia inhibitory factor (103 U/ml; Merck Millipore). Cells were grown on gelatin-coated culture dishes.

Primary human LEC culture

Primary human LECs were purchased from PromoCell and were kept in Endothelial Cell Growth Medium MV2 (PromoCell), supplemented with the corresponding Supplement mix (PromoCell). Cells were kept at 37°C in a humidified atmosphere with 5% CO2 and were transferred once per week. For experiments, cells were used up to passage number 7. Where indicated, LECs were pretreated with inhibitors of SRC (PP2), p38 (SB203580), PI3K (LY294002 and wortmannin), PKC (Gö6976), JNK (SP600125), or MEK/ERK (PD98059) at 10 μM each for 1 hour, washed once with PBS, and then stimulated with LCN2 (1 μg/ml) for 24 hours.

RNA extraction and quantitative real-time polymerase chain reaction

RNA from primary human macrophages was extracted using peqGOLD RNAPure (Peqlab Biotechnology). Total RNA (1 μg) was transcribed with the Maxima First Strand cDNA Synthesis Kit (Fermentas). qRT-PCR was performed using MyIQ real-time PCR system (Bio-Rad Laboratories) and Absolute Blue qPCR SYBR Green Fluorescein Mix (Thermo Scientific), and primers were listed in table S1. RT-PCR results were quantified using Gene Expression Macro (version 1.1) from Bio-Rad, with 18S for human and TBP for mouse mRNA expression as internal housekeeping gene control.

Enzyme-linked immunosorbent assay

Sample (100 μl each) was applied to an ELISA 96-well plate and incubated overnight at 4°C. After washing, the plate was blocked with PBS/0.05% Tween 20/1% bovine serum albumin (BSA) for 1 hour. Afterward, a specific antibody against LCN2 (R&D Systems) was added. Detection was done by a biotinylated secondary rat IgG antibody (Dako). After additional washing, horseradish peroxidase–conjugated avidin (R&D Systems) was incubated for 45 min, and the color reagent (R&D Systems) was added. Total protein amount in the sample was determined by the Lowry method for calculating the LCN2 amount per milligram of total protein. ELISA for VEGFC was performed according to the manufacturers’ instructions (Abcam). Total protein amount in the sample was determined by the Lowry method for calculating the VEGFC amount per milligram of total protein.

Transfection and luciferase reporter assay

Primary human macrophages were transiently transfected by using JetPrime transfection reagent (Polyplus-transfection). Cells were either transfected with 1 μg of LCN2 full-length promoter construct or the full-length promoter construct with a mutated STAT3 binding site as described before (15), along with 0.2 μg of Renilla luciferase control vector pRL-TK (Promega). After transfection, cells were incubated for 16 hours, medium was changed, and cells were incubated for another 24 hours followed by stimulation with S1P for 3 hours. LECs were transiently transfected by using Viromer RED transfection reagent (Lipocalyx GmbH) according to the manufacturer’s instructions. Cells were transfected with 1 μg of VEGFC full-length promoter construct (provided by B. M. Pützer, University of Rostock, Germany), along with 0.2 μg of Renilla luciferase control vector pRL-TK (Promega). After transfection, cells were incubated for 16 hours, medium was changed, and cells were stimulated with LCN2 for 24 hours. Luciferase activities in cell lysates were measured as light emission after addition of luciferase assay buffer (Promega) with a luminometer (Mithras, Berthold Technologies). Firefly luciferase activity was normalized to Renilla luciferase activity in the lysate. The background obtained from mock-transfected cells was subtracted from each experimental value.

siRNA transfections

siRNAs against LCN2, S1PR1, S1PR2, or S1PR4 (Qiagen) were transfected into primary human macrophages, and SphK2 was transfected in MCF-7 breast cancer cells using HiPerFect transfection reagent (Qiagen) according to the manufacturers’ instructions. SphK2, LCN2, and S1PR knockdown in comparison with siControl nontargeting scrambled RNA (Qiagen) was controlled by mRNA expression by RT-PCR. The knockdown was routinely confirmed by qRT-PCR for each experiment and showed a knockdown efficiency of about 40 to 60%.

Short hairpin RNA transfection in tumor cells

MCF-7 breast cancer cells were stably transduced with a lentiviral short hairpin RNA (shRNA; Mission shRNA, Sigma-Aldrich) against LCN2 (TRCN0000372827) and selected using puromycin. Controls are control virus–transduced MCF-7 cells with a pLKO.1-puro vector containing a nontargeted short hairpin sequence and a puromycin resistance. Knockdown efficiency was controlled by qRT-PCR and ELISA.

Western blot analysis

pSTAT3 and total STAT3 was quantified by Western blot analysis. Briefly, equivalent numbers of macrophages were washed twice with PBS, lysed in radioimmunoprecipitation assay buffer containing 1 × Complete protease inhibitor cocktail tablets (Roche), and sonicated with 10 pulses, followed by centrifugation for 10 min at 16,000g (4°C). Supernatants were denatured with SDS–polyacrylamide gel electrophoresis sample buffer [250 mM tris (pH 6.8), 40% glycerol, 10% β-mercaptoethanol, 8% SDS, 0.02% bromophenol blue] for 10 min at 95°C. Comparable protein concentrations were detected by Lowry protein assay (Bio-Rad). Proteins were separated on SDS–polyacrylamide gels and transferred onto nitrocellulose membranes by semidry blotting. Membranes were blocked with 5% BSA/100 mM tris-HCl,150 mM sodium chloride, 0.01% (v/v) Tween 20 (TTBS) followed by incubation with an antibody recognizing either pSTAT3 or STAT3 in 5% BSA/tris-buffered saline at 4°C overnight. For protein detection, the membrane was incubated with IRDye secondary antibodies (LI-COR) in 5% BSA/TTBS. Proteins were visualized and densitometrically analyzed with the Odyssey infrared imaging system.

Cell migration assays

For Boyden chamber assays, 1 × 105 starved human LECs were seeded in normal growth medium in Transwell inserts (12-well, 8-μm pores, upper compartment; BD Falcon; BD Biosciences) and were allowed to migrate toward LCN2 (1 μg/ml) or S1P (500 nM) in the lower compartment for 6 hours. VEGFC (100 ng/ml) was used as a positive control. Cells were washed once with PBS, subsequently stained using Mayer’s hemalum, counted in at least five different fields, and averaged.

Proliferation assay

LEC proliferation was measured using the xCELLigence RTCA DP instrument (Roche Applied Science). Initially, a background measurement of the detector containing an E-plate insert was performed using 50 μl of serum-free medium incubated for 30 min in the incubator (37°C, 5% CO2). In parallel, LECs were treated with Accutase, quantified, and added to the insert in 100 μl of serum-free medium (10,000 cells per well). LCN2 (1 μg/ml) was added in the presence or absence of a specific antibody against VEGFR3 (3 μg/ml) or VEGFC (3 μg/ml) or an IgG control antibody (3 μg/ml). Recombinant human VEGFC (100 ng/ml; PeproTech) was used as a positive control. Authentic S1P (500 nM) was used to test the effects of S1P on LEC proliferation. The blocking activity of the neutralizing antibodies was tested in combination with the positive control VEGFC. In case of conditioned medium treatment, LEC cells were harvested and quantified as described above. The initial background measurement of the detector was performed using 50 μl of full growth medium incubated for 30 min in the incubator (37°C, 5% CO2). Cells were then either left in 100 μl of normal growth medium or taken up in conditioned medium (either coming from the coculture system or as the supernatant from viable tumor cells). FCS (1%) was used in all experiments, and proliferation was measured as an increase in impedance continuously for a period of 120 hours. Data are presented as the slope per hour of the normalized cell index as a measure for the time-dependent changes in impedance. The RTCA Software 1.2 (Roche Applied Science) was used for both data acquisition and analysis.

Tube formation assay

Tube formation was performed with μ-Slide Angiogenesis (Ibidi) according to the manufacturer’s instructions. Ten thousand LECs per well were treated with either recombinant LCN2 (1 μg/ml), S1P (500 nM), or the positive control VEGFC (500 ng/ml). Images were captured after 6 hours using an inverted microscope (Axiovert 135, Zeiss) and camera (AxioCam MRc, Zeiss). The number of master junctions and the tube length were quantified with the Angiogenesis Analyzer using ImageJ software (

Sprouting assay

Five hundred CGR8 cells per well were embedded in a collagen matrix containing collagen type I (1.8 mg/ml) (BD Biosciences) in a 24-well plate and cultured for 5 days in the absence of leukemia inhibitory factor to allow embryonic body formation. Wells were covered with Iscove medium (negative control) or Iscove medium 1:1 diluted with macrophage-conditioned medium for 5 days. Macrophage-conditioned medium was generated from mouse BMDMs previously incubated with supernatant from either apoptotic (ACM) or viable (VCM) MC57G cells using the above described protocol. Moreover, stem cells were stimulated with either MCMV supplemented with recombinant LCN2 or MCM depleting LCN2 with a neutralizing antibody (R&D Systems). VEGFC (100 ng/ml) and VEGFA (100 ng/ml) were used as positive controls. Sprouting assays were analyzed by flow cytometry.

Matrigel plug assay

Female C57BL/6 mice (10 weeks old) were anesthetized with chloral hydrate (1 μl of a 4% solution per gram of mouse, intraperitoneally) and then injected along the dorsal midline on each site of the spine with 0.5 ml of a Matrigel (BD Biosciences) mixture containing heparin (0.0025 U/ml) and supplemented with recombinant LCN2 (2 μg/ml) in the presence or absence of a neutralizing antibody against VEGFR3 or the isotype control IgG antibody. Recombinant human VEGFC (500 ng/ml) was used as a positive control. After 10 days, mice were sacrificed, Matrigel plugs were removed, and one half of the plugs was embedded in Tissue-Tek (Sakura Finetek), frozen and cryosectioned (16 μm), and processed for LYVE-1 and PECAM-1. Quantification of vessel formation was performed in a similar manner as described before (57), analyzing five different areas of 2 mm2 per plug each (magnification, ×200). The angiogenic and lymphangiogenic response (represented as percent coverage) was analyzed and graded using ZEN 2011 software (Zeiss). The other half of the plugs was digested and used for flow cytometry analysis. Experiments followed the guidelines of the Hessian animal care and use committee.

Flow cytometry

For flow cytometric quantification, single-cell suspensions were generated from disintegrated embryonic bodies by digestion with collagenase I for 30 min at 37°C with further separation using Filcon (BD Biosciences) and were transferred to FACS tubes. Tumor tissues were lysed with the Miltenyi Tumor Dissociation Kit and the GentleMACS (Miltenyi Biotec) using standard protocols, and a single-cell suspension was obtained for staining and quantification using flow cytometry. Unspecific antibody binding to FCγ receptors was blocked with Fc Block Receptor Binding Inhibitor (130-092-575, eBioscience) for 15 min on ice, followed by staining with an antibody cocktail to exclude immune cells and determine the amount of both angiogenesis and lymphangiogenesis markers (Flk 1–PerCP5.5 and CD45-V500 from BD Biosciences; PECAM-1 PE-Cy7, Sca-1–eFluor 605, c-kit APC–eFluor 780, and CD105–eFluor 450 from eBioscience; podoplanin–Alexa Fluor 488 and CD144-APC from BioLegend; and LYVE-1–PE from R&D Systems). Matrigel plugs were digested with Liberase TM Research Grade (Roche) for 30 min at 37°C with continuous shaking followed by digestion with collagenase I (1590 U/ml; Biochrome) and deoxyribonuclease 1 (25 U/ml; Roche) at 37°C with continuous shaking and further separated by 70-μm filters. A single-cell suspension was obtained for staining and quantification using flow cytometry. Unspecific antibody binding to FCγ receptors was blocked with Fc Block Receptor Binding Inhibitor for 15 min on ice, followed by staining with an antibody cocktail to define both angiogenic and lymphangiogenic markers (CD49f PE-CF594 and CD146–Alexa Fluor 488 from BD Biosciences, PECAM-1 PE-Cy7 from eBioscience, and CD45–VioBlue from Miltenyi Biotec). To discriminate viable cells from apoptotic and necrotic cells, samples were stained with 7AAD (7-aminoactinomycin D). All antibodies and secondary reagents were titrated to determine optimal concentrations. CompBeads (BD Biosciences) were used for single-color compensation to create multicolor compensation matrices. For gating, fluorescence minus one (FMO) controls were used. Samples were acquired with an LSR II/Fortessa flow cytometer (BD Biosciences). The instrument calibration was controlled daily using Cytometer Setup and Tracking beads (BD Biosciences). The number of either CD31 or podoplanin and LYVE-1 double-positive cells was quantified using the counting beads previously added to each sample.

MMTV-PyMT breast cancer model

All procedures involving mice followed the guidelines of the Hessian animal care and use committee. Wild-type (Lcn2+/+) and Lcn2−/− mice were crossed into a PyMT background (all C57BL/6). For genotyping, tail tips were lysed with KAPA Genotyping lysis buffer (Peqlab), and resulting DNA solutions were analyzed with PCR amplification using KAPA HotStart Genotyping reaction mix (Peqlab) and standard protocols. Once the tumor reached 1.5 cm, PyMT mice were sacrificed. PyMT tumors were isolated, and their respective weight was recorded. Half of the tissue was lysed with the Miltenyi Tumor Dissociation Kit and the GentleMACS (Miltenyi Biotec) using standard protocols, and a single-cell suspension was obtained for staining and quantification using flow cytometry. The other half of the tumor was embedded in paraffin for immunofluorescence analysis of vessel formation in the tumor. Lung metastases were determined by Mayer’s hemalum (Merck) staining. The appearance of metastases was evaluated in 12 lung sections per mouse, and the distribution of lung nodules was calculated.

Orthotopic breast cancer model

Donors were female C57BL/6 wild-type PyMT and Lcn2−/− PyMT mice. Recipients were female C57BL/6 wild-type and Lcn2−/− mice. The tumor from PyMT donor mice was isolated and dissociated with the Tumor Dissociation Kit (Miltenyi Biotec). Cells were blocked with Fc Block Receptor Binding Inhibitor (130-092-575, eBioscience) and stained with an antibody cocktail containing CD45-VioBlue (130-102-430, Miltenyi Biotec), CD326-PE (130-096-448, Miltenyi Biotec), and 7-AAD (559925, BD Biosciences). CD45 and CD326+ living tumor cells were sorted using FACSAria (BD Biosciences). Tumor cells (5 × 105) were implanted into the mammary gland (no. 4) of recipient mice. The tumor was removed at a size of 1.5 cm and was processed for immunohistochemical stainings.

Statistical analysis

Statistical analysis was performed using Prism software (GraphPad Software Inc.). Values for all measurements were expressed as means ± SD. Each experiment was performed at least three times (independent experiments using three technical replicates). The Kolmogorov-Smirnov test was used as nonparametric test to check for normal distribution. All data sets were analyzed using two-tailed unpaired Student’s t test and were not corrected for multiple testing. For analysis of Kaplan-Meier survival curve and the distribution of metastases, we performed a log-rank test and defined the χ2 P value. P values were considered significant at *P < 0.05, **P < 0.01, ***P < 0.001.


Fig. S1. Tumor cell–derived LCN2 and expression of S1PR1, S1PR2, and S1PR4 in primary human macrophages.

Fig. S2. Knockdown efficiency of S1PR1, S1PR2, and S1PR4 in primary human macrophages.

Fig. S3. Effects of S1P on lymphangiogenesis in culture.

Fig. S4. Effect of tumor cell–derived LCN2 on LEC proliferation.

Fig. S5. Expression of lymphangiogenesis-associated genes in LECs.

Table S1. PCR primers.


Acknowledgments: We thank E. Herrmann (Institute of Biostatistics and Mathematical Modeling, Goethe-University Frankfurt) for critical review of the statistical analysis. Funding: The work was supported by grants from the Fritz Thyssen Stiftung (Az. to M.J.), Goethe-University Frankfurt (Focus Line B to M.J.), Doktor Robert Pfleger-Stiftung (awarded to M.J.), Monika Kutzner Stiftung (to M.J.), and SFB 1039 (TP B04 to B.B.). Author contributions: M.J.: study design, data acquisition, analysis, interpretation, and writing of the manuscript; B.Ö.: data acquisition, analysis, interpretation, and writing of the manuscript; J.M., C.M., S.D., and R.P.: data acquisition, analysis, and interpretation; A.W.: data analysis and interpretation; N.G.: data acquisition; I.F. and B.B.: data interpretation and writing of the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Stay Connected to Science Signaling

Navigate This Article