Metastatic conjunctival melanoma cells induce angiogenesis and recruitment of macrophages in a zebrafish model

To explore if conjunctival melanoma cells can trigger macrophage-dependent angiogenesis, we applied a well-described zebrafish tumor xenograft angiogenesis assay, which allows imaging of angiogenetic activity after 16–24 h [29]. In this assay, the growth of the blood vessels from the sub-intestinal vein plexus (SIV) is induced by the engraftment of tumor cells, triggering sprouting or remodeling of the SIV complex without adverse effects on other physiological vessels (Fig. 1a, Supplementary Movie 1). For this purpose, we injected the CoM cell lines CRMM1 and CRMM2 (both derived from a local recurrence) stably labelled with mCardinal (far-red fluorescence) into the perivitelline space (PVS) of Tg(kdrl:EGFPs843; mpeg1:GAL4-VP16gl24; UAS-E1b: NfsB-mCherryi149) zebrafish embryos. This zebrafish transgenic line expresses the green fluorescence protein (EGFP) under the promoter of the kinase insert domain receptor (kdrl, zebrafish homolog of VEGF receptor) [32], which labels endothelial cells, and the fusion protein of E. coli NsfB reductase and mCherry under the macrophage-expressed gene 1.1 (mpeg1.1), thus labelling macrophages. The kinetics of the induced angiogenic response in live embryos was monitored directly after engraftment by time-lapse imaging from 2 h post injection (hpi) to 24 hpi in 15 min intervals. As shown in Fig. 1b, there was a fast recruitment of macrophages to the position of engrafted cells (~ 2hpi) and a pronounced colocalization with cancer cells during the first 24hpi, coinciding with the elongation of the SIV towards the injection foci.

Fig. 1

Metastatic conjunctival melanoma cells induce angiogenesis and recruitment of macrophages in a zebrafish angiogenesis model. a Schematic representation of the angiogenesis assay in zebrafish xenografts. Cancer cells or 2% PVP used as vehicle are injected into the PVS at 2 day post fertilization (dpf) zebrafish larvae Tg(kdrl:EGFPs843; mpeg1:GAL4-VP16gl24; UAS-E1b: NfsB-mCherryi149). In response to a strong angiogenic signal, the SIV complex is deformed towards the injection focus. b Representative images of time-lapses recording the angiogenic response to the SIV from 2 to 24 hpi. c Quantifications of the angiogenic activity induced by engraftment of CoM cells at 24hpi as measured by the angle, total length, and elongation of the SIV complex

To quantify the angiogenic activity induced by engraftment of CoM at 24hpi, the angle, total length, and elongation of the SIV complex were measured (Fig. 1c). Both CRMM1 and CRMM2 significantly decreased the directional angle and increased SIV length and elongation, illustrating the induction of angiogenesis. In comparison, the angle, length, and elongation of the SIV complex of zebrafish injected with vehicle without cells showed no modification despite the initial recruitment of macrophages, which was considered a transient inflammatory reaction to wounding.

In conclusion, these results demonstrate that both of the CRMM1 and CRMM2 cell lines have a high angiogenic activity in the zebrafish xenograft model, correlating with macrophage recruitment.

The angiogenic response induced by engraftment of metastatic CoM cells is macrophage-dependent

To study the functional significance of macrophage recruitment towards engrafted CoM cells during the development of angiogenesis, we used the NsfB nitroreductase (NTR) / Metronidazole (MTZ) ablation system to deplete macrophages in living zebrafish embryos. The zebrafish transgenic larvae mentioned above express NTR in their macrophage lineage. Upon delivery of MTZ in the culture medium, NTR converts this substrate into a cytotoxic agent capable of killing macrophages [33]. First, we optimized the efficiency of macrophage ablation without signs of toxicity. Embryos from 2dpf until 8dpf were exposed to various concentrations of MTZ (Supplementary Fig. 1a). Embryos treated with 10mM MTZ showed higher lethality than controls, indicating that MTZ was toxic at higher concentrations, but lower concentrations had no effect on the survival of embryos. We concluded that 2.5mM MTZ successfully depleted macrophages in embryos from 2dpf to 5dpf without adverse effects and used this concentration to examine to role of macrophages in tumor-induced angiogenesis (Supplementary Fig. 1b and c).

To this aim, we injected embryos with the 2% PVP-containing vehicle as control or mCardinal-labeled CRMM1 and CRMM2 cells at 2dpf and treated the embryos with 2.5mM TMZ from 2hpi. Under these conditions, we imaged and quantified the angiogenetic activity at 24hpi (Fig. 2a and b). In the PVP-injected group, the developmental pattern of SIV was unchanged regardless of the presence or absence of macrophages. As described in Fig. 1b and c, engraftment of CoM cells lines induced macrophage recruitment and angiogenesis, but chemical ablation of macrophages by MTZ resulted in reduction of the angiogenetic response as quantified by measurement of the SIV directional angle, length, and elongation (Fig. 2b). Thus, the presence of macrophages is essential for the induction of angiogenesis by the conjunctival melanoma cell lines CRMM1 and CRMM2 in our model. Importantly, MTZ treatment of wild type embryos (without NTR) engrafted with CRMM1 and CRMM2 cells had no effect on the tumor cell viability and tumor-induced angiogenesis suggesting that MTZ operates strictly via NTR-dependent ablation of macrophages (Supplementary Fig. 1d).

Fig. 2

Chemical ablation of macrophages inhibits tumor-induced angiogenesis in a zebrafish model. a Representative images of xenografted zebrafish larvae, treated with or without 2.5mM MTZ at 24hpi. b Quantification of the angiogenic capacity of cancer cells with and without macrophages. c Xenografted zebrafish material was retrieved and used to detect angiogenetic and inflammatory factors with qPCR. d Expression levels of zebrafish VEGF-A (vegfaa), TGF-β (tgfb1a), and IL-10 (il10) in CRMM1 and CRMM2 engrafted groups. e Expression of the M2 marker IL-4 (il4) in tumor engrafted zebrafish treated with DMSO and MTZ. f Expression of proinflammatory markers, iNOS (nos2a), IL-1β (il1b), and TNF-α (tnfa) in DMSO or MTZ-treated groups

To further explore if zebrafish macrophages recruited to CRMM1 and CRMM2 induced tumors polarized to an M2-like phenotype, which is associated with angiogenesis in mice [34], the expression level of pro-angiogenic, pro-inflammatory, and M2 markers in engrafted zebrafish larvae were quantified by qPCR using zebrafish-specific primers (Fig. 2c-f). We found that pro-angiogenic factors, such as VEGF-A, TGF-β, and IL-10 were overexpressed in the groups injected with CRMM1 and CRMM2 compared to the PVP vehicle controls, but that the expression of these genes was reduced if macrophages were ablated (Fig. 2d). The same pattern was observed with the M2 marker IL-4 (Fig. 2e). In contrast, the expression level of the proinflammatory markers iNOS, TNF-α, and iL-1β was increased in the absence of macrophages (Fig. 2f).

Altogether, these results suggest that angiogenesis caused by the engraftment of metastatic conjunctival melanoma cells is macrophage-dependent, and that recruited macrophages are polarized towards a M2-like stage in the proximity of tumor cells.

Lactate produced by glycolytic cancer cells attracts and polarizes zebrafish macrophages to promote angiogenesis

We then wondered how macrophages are recruited and retained to the site of CoM cell engraftment. We hypothesized that, as initially described by Colegio et al. [25], lactate produced by glycolysis of cancer cells attracts macrophages to facilitate the angiogenetic response. In order to explore this possibility, we first studied if zebrafish macrophages are able to sense lactate as demonstrated for mammalian macrophages. Analysis of the zebrafish single cell atlas from the Miller lab [35] reveled that indeed paralogs of human lactate receptors GPR81 and GPR132 are expressed on zebrafish macrophages (Supplementary Fig. 2 and Table 1). To test this functionally, we directly injected lactate into the hindbrain of 2dpf zebrafish transgenic larvae with fluorescently labelled macrophages, which is free of this cell type at this developmental stage (Fig. 3a). The human cytokine CCL2, a well-known chemoattractant of zebrafish macrophages [36], was used as positive control, while we used PVP solvent to determine the macrophage influx to the local inflammation generated by the injection wound. Injection of CCL2 and lactate significantly increased the number of accumulated macrophages in the hindbrain 3hpi compared to the number of macrophages in the same area of control and PVP-injected embryos, demonstrating that indeed lactate can drive zebrafish macrophage migration (Fig. 3b).

Fig. 3

Lactate and glycolytic 4T1 cells attract zebrafish macrophages. a Schematic diagram of injection of lactate and hCCL2 into the zebrafish hindbrain. b Quantification of the macrophages attracted to the hindbrain under different conditions. c Representative images of the angiogenic effect in the SIV complex with and without treatment with MTZ at 24 hpi. d Quantification of the angiogenic influence of the 67NR and 4T1 cell lines with and without macrophages. e Expression of the zebrafish-derived proangiogenic markers VEGF-A (vegfaa), TGF-β (tgfb1a), and IL-10 (il10)

Next, we tested whether highly glycolytic malignant cells, which preferentially metabolize glucose to lactate even in aerobic conditions, can efficiently attract zebrafish macrophages in vivo. For this, we used two reference isogenic breast cancer cell lines: the highly metastatic, glycolytic 4T1, and the non-metastatic, low-glycolytic 67NR [30]. It has previously been reported that 4T1 cells generate 10-fold more lactate than isogenic non-metastatic 67NR cells [37]. Engraftment of control PVP vehicle and of 67NR cells did not modify SIV growth (Fig. 3c and d). In contrast, the isogenic metastatic variant 4T1 caused a significant angiogenetic response and recruited more macrophages compared to the vehicle PVP control and to 67NR cells. Similar to the CRMM1 and CRMM2 lines, the angiogenetic effect of the engrafted 4T1 line was dependent on macrophages, which were required for the expression of zebrafish angiogenic factors.

The differential angiogenetic phenotype seen by the engraftment of 4T1 and 67NR cells in the zebrafish model suggests that tumor-derived lactate can drive macrophage recruitment and the angiogenic response in this zebrafish angiogenesis model. In addition, we show the capacity of the zebrafish short-term angiogenesis assay to discriminate between highly angiogenic and poorly angiogenic tumor cell lines, thus allowing a critical examination of the angiogenetic properties of the CRMM1 and CRMM2 cell lines.

Conjunctival melanoma cells secrete lactate and maintain their glycolytic properties after xenografting

Given the possible effect of lactate on angiogenesis, we questioned whether CRMM1 and CRMM2 cells also secrete lactate to govern macrophage recruitment and polarization leading to induction of angiogenesis in the zebrafish model. To evaluate the glycolytic properties of CRMM1 and CRMM2, the level of lactate was measured in their supernatant with and without pretreatment with 2-Deoxy-D-glucose (2DG) and with GSK2837808A (GSK), and compared to lactate level in the reference low-glycolytic 67NR and high glycolytic 4T1 cell lines (Fig. 4a-d, Supplementary Fig. 3a-d). 2DG blocks glycolysis by competitively inhibiting Hexokinase 1 (HK1), an enzyme responsible for the phosphorylation of glucose in the first step of glycolysis (Fig. 4a) [38], while GSK inhibits the dehydrogenase A (LDHA), which converts pyruvate to lactate. First, we optimized the concentration of the drugs used so that cell physiology was not affected. Treatment of CRMM1, CRMM2 and 4T1 cells for 24 h with both inhibitors did not influence proliferation or cellular ATP production (Fig. 4b and c, Supplementary Fig. 3b and c). Importantly, under these conditions, the secretion of lactate from the highly glycolytic 4T1 cells and the CRMM1 and CRMM2 cells was significantly reduced to the level of the 67NR non-glycolytic cells, indicating that CoM cells indeed secrete lactate at a level comparable to 4T1 cells (Fig. 4d, Supplementary Fig. 3d).

Fig. 4

CoM cells secrete lactate and maintain their glycolytic properties in xenograft models. a Schematic representation of the glycolysis pathway with key enzymes in the process, showing the effect of 2DG inhibiting the activity of Hexokinase 1 (HK1). b Quantification of cell viability during treatment of increasing 2DG concentrations for 24 h. c The cellular ATP level of 67NR and 4T1, CRMM1, and CRMM2 after 2DG treatment. d 2DG inhibited the lactate production in 67NR and 4T1, CRMM1, and CRMM2 cells. e Expression of the glycolysis-related enzymes HK1, PFK1, PDK, and LDHA in zebrafish 4T1, 67NR, CRMM1 and CRMM2 xenografts

Subsequently, we evaluated the metabolic properties of 67NR, 4T1, CRMM1 and CRMM2 after xenografting. To do so, we dissected zebrafish xenografts and analyzed the expression of glycolytic enzymes in the cancer cells via qPCR, including HK1; phosphofructokinase 1 (PFK1), which is the rate limiting enzyme that converts fructose 6-phosphate to fructose 1–6 bisphosphate and has a vital role in leading endothelial cells (ECs) sprouting; pyruvate dehydrogenase kinase (PDK), which regulates the catalytic activity of pyruvate decarboxylation oxidation, and it further links glycolysis with the tricarboxylic acid cycle and ATP generation; and lactate dehydrogenase A (LDH-A), which controls the final formation and release of lactic acid. As expected, 4T1 expressed significantly higher levels of HK1, PFK1, PDK, and LDHA than 67NR (Fig. 4e). CRMM1 and CRMM2 engraftments showed similar high expression levels as the 4T1 reference. Therefore, we conclude that 4T1, CRMM1, and CRMM2 retained their high glycolytic properties in the zebrafish model.

In all, the consistency between zebrafish macrophage-dependent angiogenesis and glycolytic properties of engrafted cells suggests that the lactate secreted by these cells during aerobic glycolysis may indeed attract macrophages and induces their polarization towards a M2 phenotype, leading to the expression of high levels of pro-angiogenic factors and the induction of angiogenesis in zebrafish.

Supernatant secreted by conjunctival melanoma cells polarizes human macrophages to M2-like and leads to higher expression of proangiogenic factors

To further validate the translational relation between the metabolism of CoM cells and the polarization of macrophages towards their proangiogenic properties as observed in the zebrafish model, we exposed human macrophages derived from the THP-1 monocytic cell line [39] with the conditioned medium collected after 24 h treatment of CoM cells with DMSO solvent, 10 mM 2DG (Fig. 5a), or with 10µM GSK (Supplementary Fig. 4). As a positive control, the differentiated macrophages were directly treated with lactate. We then assessed the expression of polarization and pro-angiogenic gene markers (Fig. 5b). Lactate and the conditioned medium of CRMM1 and CRMM2 cells induced an increase in the expression of the M2 marker CD206 as well as of the pro-angiogenic markers VEGF-A and TGF-β which was ablated by inhibition of the glycolysis in the CoM culture with 2DG or GSK. Consistently, the expression of the M1 marker CD86 was only increased in the macrophages exposed to medium from treated CoM cells.

Fig. 5

Supernatant secreted by CoM cells polarizes macrophages to an M2-like phenotype leading to higher expression of proangiogenic factors. a Schematic diagram of the conditioned medium experimental design. b Expression of CD86, CD206, VEGF-A, and TGF-β of macrophages in conditioned medium models

These results prove that glycolytic CoM cells are able to polarize human macrophages towards a M2 phenotype and increase their proangiogenic capacity.

Inhibition of glycolysis attenuates macrophage-dependent angiogenesis induced by highly glycolytic CoM cells

Since inhibition of glycolysis by 2DG reduced the production of VEGF-A and TGF-β in human macrophages in vitro, we aimed to validate this effect in vivo. Considering the lack of suitable rodent models for metastatic CoM, we employed the versatile zebrafish angiogenic assay previously presented. Cell lines 4T1, CRMM1, and CRMM2 were treated with 10 mM of 2DG for 24 h prior to their engraftment in the zebrafish PVS and imaged by confocal microscopy (Fig. 6a). As previously shown, the highly glycolytic tumors 4T1, CRMM1, and CRMM2 induced macrophage-dependent angiogenesis (Fig. 6b and c). However, treatment of these cells with 2DG before engraftment obstructed the sprouting of new vessels towards these xenografts. As a consequence of 2DG treatment, the SIV angle was flattened, its length and elongation were reduced (Fig. 6c). In contrast, the low glycolytic 67NR cells had similar phenotype as the control PVP.

Fig. 6

Inhibition of glycolysis attenuates macrophage-dependent angiogenesis of CoM cells. a Schematic diagram of 2DG treatment in cancer cells prior to their xenografting in zebrafish larvae. b Representative images of the angiogenesis response of embryos xenografted with CoM lines treated with and without 2DG treatment. c Quantification of the angiogenesis response in these conditions. d Quantification of the pro-angiogenic factors VEGF-A (vegfaa) and TGF-β (tgfb1a) in engrafted zebrafish

Mechanistically, the 2DG treatment decreased the recruitment of zebrafish macrophages towards the engrafted 4T1, CRMM1, and CRMM2 cells as quantified by fluorescence intensity and by its expression marker, matrix metalloprotease 9 (mmp9) (Supplementary Fig. 5). In addition, zebrafish engrafted with untreated and 2DG treated cells were dissected and subjected to expression analysis of pro-angiogenic genes. As shown in Fig. 6d, the high expression of proangiogenic genes like VEGF-A and TGF-β induced by highly glycolytic cells was inhibited by 2DG treatment to the levels caused by engraftment of the low glycolytic 67NR cells. Therefore, inhibition of glycolysis in engrafted CoM cells by 2DG suppressed macrophage recruitment, as well as the M2 polarization responsible for secretion of proangiogenic factors in the zebrafish tumor microenvironment.

These results confirm that lactate secreted by glycolytic CoM cells attracts and polarizes macrophages to drive angiogenesis in the zebrafish model.