VEGF-B prevents excessive angiogenesis by inhibiting FGF2/FGFR1 pathway

Although VEGF-B was discovered as a VEGF-A homolog a long time ago, the angiogenic effect of VEGF-B remains poorly understood with limited and diverse findings from different groups. Notwithstanding, drugs that inhibit VEGF-B together with other VEGF family members are being used to treat patients with various neovascular diseases. It is therefore critical to have a better understanding of the angiogenic effect of VEGF-B and the underlying mechanisms. Using comprehensive in vitro and in vivo methods and models, we reveal here for the first time an unexpected and surprising function of VEGF-B as an endogenous inhibitor of angiogenesis by inhibiting the FGF2/FGFR1 pathway when the latter is abundantly expressed. Mechanistically, we unveil that VEGF-B binds to FGFR1, induces FGFR1/VEGFR1 complex formation, and suppresses FGF2-induced Erk activation, and inhibits FGF2-driven angiogenesis and tumor growth. Our work uncovers a previously unrecognized novel function of VEGF-B in tethering the FGF2/FGFR1 pathway. Given the anti-angiogenic nature of VEGF-B under conditions of high FGF2/FGFR1 levels, caution is warranted when modulating VEGF-B activity to treat neovascular diseases.


INTRODUCTION
VEGF-B was discovered as a VEGF-A homolog in 1996, and is highly expressed in vascular endothelial cells (ECs) and many other cell types. 1,2While VEGF-B has been reported to be involved in metabolic complications, such as in diabetes, 3 the vascular effect of VEGF-B remains poorly understood thus far.5][6][7] However, surprisingly and notwithstanding its name, many other studies have reported anti-angiogenic and anti-tumor effects of VEGF-B [8][9][10][11] (Supplementary Table 1).For example, VEGF-B has been shown to inhibit tumor angiogenesis 10 and tumor growth 9 in various model systems.Particularly, in cancer patients, high levels of VEGF-B have been shown to be associated with low tumor angiogenesis and better survival, 8,11,12 and low VEGF-B levels are linked to high tumor angiogenesis and poor patient survival. 8,13Importantly, in breast cancer patients, low VEGF-B levels are reported to be related to higher risks of cancer, and high levels lower risks. 11Although many studies from different groups have shown anti-angiogenic and anti-tumor effects of VEGF-B, these studies have been largely ignored thus far, with one of the reasons being that the mechanisms involved remain unknown.Yet, despite the poor understanding of the angiogenic effect of VEGF-B, drugs that inhibit VEGF-B together with other VEGF family members are being used in the clinic to treat patients with various neovascular diseases. 14However, in many types of cancers, such treatments failed to show benefit. 14,15t is therefore urgently imperative to have a better understanding of the angiogenic effect of VEGF-B and the underlying mechanisms.
VEGF-B binds to vascular endothelial growth factor receptor-1 (VEGFR-1), 16 which also binds to PlGF and VEGF-A.Unlike VEGFR2, VEGFR-1 has a very weak kinase activity, which is about ten-fold weaker than that of VEGFR-2.As such, VEGFR1 can often serve as a decoy receptor to trap certain ligands, such as VEGF-A, to suppress angiogenesis under many conditions. 17Noteworthy, in breast cancer patients, high VEGFR1 expression levels have been shown to be associated with better patient survival, while low VEGFR1 levels are linked to adverse cancer phenotypes and poor patient survival. 18,19Moreover, in mice, genetic deletion of Flt1 (encoding VEGFR1) or pharmacological inhibition of VEGFR1 increased adipose angiogenesis. 3Furthermore, VEGFR1 has been shown to inhibit Erk activation in vascular ECs, 17,20,21 which may be at least one of the mechanisms underlying its anti-angiogenic effect.However, it remains thus far unknown whether VEGF-B, as a ligand of VEGFR1, plays a role in the inhibitory effect of VEGFR1.
The fibroblast growth factor 2 (FGF2, also known as basic FGF) is a heparin-binding growth factor expressed in many tissues and cell types, including vascular endothelial cells.FGF2 binds to four receptor tyrosine kinases (FGFR1-4) and acts in a variety of physiological and pathological processes, such as angiogenesis, tumor growth, and development.9][30] Moreover, FGF2 deficiency results in blood vessel degeneration. 27,28FGFR1 is composed of an extracellular region with three immunoglobulin-like domains, a single hydrophobic trans-membrane region, and a cytoplasmic tyrosine kinase domain.FGFR1 is widely expressed, plays important roles in angiogenesis and the phenotype of many types of tumor cells.FGFR1 amplification or upregulation is frequently found in different types of tumors, such as breast cancer, non-small cell lung cancer, ovarian cancer, urothelial carcinoma, and hepatocellular carcinoma.Particularly, in breast cancer, FGFR1 amplification is found in 16-27% of luminal B-type breast cancer patients.Due to its critical role in tumor growth, FGFR1 has been considered as an important drug target for cancer therapy, and multiple small molecule inhibitors against it have been developed.0][31][32][33] Despite the abundant expression and potent angiogenic activities of FGF2 and FGFR1, it is thus far not well understood how the potent angiogenic effect of the FGF2/FGFR1 pathway is finely controlled.
In this study, using comprehensive in vitro and in vivo model systems and approaches, including several lines of knockout mice, we unveil an unexpected novel function of VEGF-B as an anti-angiogenic factor to prevent excess angiogenesis by inhibiting the FGF2/FGFR1 pathway when the latter is highly expressed.Mechanistically, we reveal that VEGF-B binds to FGFR1, induces FGFR1/VEGFR1 complex formation, and inhibits FGF2-induced Erk activation.Indeed, in multiple in vitro and in vivo assays, VEGF-B inhibited FGF2-driven angiogenesis and tumor growth.Thus, our work uncovers a previously unknown and surprising function of VEGF-B in restricting the FGF2/FGFR1 pathway.Given the anti-angiogenic nature of VEGF-B under conditions of high FGF2/FGFR1 levels, caution is warranted when modulating VEGF-B activity to treat different types of neovascular diseases.

VEGF-B binds to FGFR1
Although many independent studies have reported antiangiogenic effects of VEGF-B, [8][9][10][11][12] the mechanisms involved remain unknown.To address this, we screened a phospho-receptor tyrosine kinase (pRTK) antibody array using human retinal endothelial cells (HRECs) to explore in an unbiased manner whether VEGF-B interacts with potentially unknown molecules.Since the VEGF-B 167 isoform is predominantly expressed in most tissues and accounts for more than 80% of the total VEGF-B transcripts, 34 we used VEGF-B 167 in this study and referred to as VEGF-B throughout the whole work.We found that VEGF-B reduced FGFR1 activation (data not shown), which was confirmed by Western blot (Supplementary Fig. 1a, b), suggesting a possible interaction of VEGF-B with FGFR1.We next investigated whether VEGF-B binds to FGFR1.A surface plasmon resonance (SPR) assay showed that VEGF-B binds to FGFR1 with a K D value similar to that of FGF2 (17 nM for VEGF-B and 16 nM for FGF2, Fig. 1a, b), while placental growth factor (PlGF), another VEGFR1-binding member of the VEGF family, showed no binding (Fig. 1c), demonstrating that the binding of VEGF-B to FGFR1 was specific.
VEGF-B mainly binds to FGFR1 domains II and III We further investigated which extracellular domains of FGFR1 VEGF-B binds to.For this purpose, we generated recombinant proteins of different extracellular domains (D) of FGFR1: FGFR1 DI, DII, DIII, DI-II, and DII-III, and tested VEGF-B binding using an SPR assay.These recombinant proteins displayed expected molecular weights as shown by both Coomassie blue staining and Western blots (Supplementary Fig. 2a-e).In addition, the functionality of these recombinant proteins was validated by a cell migration assay using human umbilical vein endothelial cells (HUVECs), which express FGFR1 (Supplementary Fig. 2f, g).We found that FGFR1 DII-III reduced FGF2-induced cell migration, while FGFR1 DI-II did not (Supplementary Fig. 2h, i), consistent with previous observations. 35n SPR assay showed that VEGF-B bound to FGFR1 DII-III with a K D value similar to that of FGF2 (49 nM for VEGF-B and 63 nM for FGF2; Fig. 1d, e).By contrast, VEGF-B, like FGF2, showed very weak binding to FGFR1 DI (K D values: 532 nM for VEGF-B and 245 nM for FGF2, Fig. 1f, g).Indeed, binding assays using FGFR1 DIII and DII single domains alone further confirmed these observations by showing that VEGF-B bound to FGFR1 DIII and DII with K D values similar to those of FGF2 (FGFR1 DIII: 64 nM for VEGF-B and 43 nM for FGF2; FGFR1 DII: 101 nM for VEGF-B and 71 nM for FGF2, Supplementary Fig. 3a-d).These data all demonstrate that VEGF-B mainly binds to FGFR1 domains II and III.

Mapping of VEGF-B binding sites for FGFR1
To map the VEGF-B binding sites for FGFR1, we used eleven synthetic VEGF-B peptides covering the entire protein (Fig. 1h, Table 1) and tested whether they bound to FGFR1.An ELISA assay using FGF2 as a positive control revealed that VEGF-B peptides 10 and 11 covering the C terminus of VEGF-B displayed the highest binding activities similar to that of FGF2 (Fig. 1i), with no binding for peptides 1, 3, 5, and 6, and weaker binding for peptides 2, 4, 7, 8, and 9 (Fig. 1i).Noteworthy, the FGFR1-binding VEGF-B peptides 10 and 11 do not overlap with the known amino acid residues important for VEGFR1 binding 36 (Fig. 1h, blue), suggesting a possibility of simultaneous binding of VEGF-B to FGFR1 and VEGFR1.
VEGF-B competes with FGF2 for FGFR1 binding Since both VEGF-B and FGF2 bind to FGFR1, we used complementary methods to explore whether they compete with each other for FGFR1 binding.First, an SPR analysis demonstrated that FGF2 competed with VEGF-B for FGFR1 binding, while PlGF did not (Fig. 1j).Second, a competitive ELISA showed that VEGF-B, but not PlGF, dose-dependently competed with FGF2 for FGFR1 binding (Fig. 1k).These results thus confirmed each other and suggested a possibility that VEGF-B might inhibit FGF2's function by competing for FGFR1 binding.
VEGF-B induces VEGFR1/FGFR1 complex formation Given that VEGF-B binds to VEGFR1 16 and FGFR1 (Fig. 1a, d, i), we explored whether VEGF-B could induce VEGFR1/FGFR1 complex formation.Immunoprecipitation (IP) followed by Western blot showed that VEGFR1 co-immunoprecipitated with FGFR1 in mouse tissues, such as brain, lung, and heart (Fig. 2a, Supplementary Fig. 4a), demonstrating the existence of a naturally occurring endogenous VEGFR1/FGFR1 complex.Importantly, intravitreal injection of VEGF-B into mouse eyes increased the amount of VEGFR1/FGFR1 complex in the retinae, while PlGF did not (Fig. 2b,  c, Supplementary Fig. 4a).To directly visualize the FGFR1/VEGFR1 complex, we performed an in situ proximity ligation assay using HRECs expressing both VEGFR1 and FGFR1 (Fig. 2d, Supplementary Fig. 4a), and found that VEGF-B treatment increased the association of FGFR1 with VEGFR1, while PlGF did not (Fig. 2e, f, upper panels), demonstrating that the effect of VEGF-B was specific.FGFR1 and VEGFR1 levels were not changed after the treatment of VEGF-B or PlGF in the HRECs (Supplementary Fig. 4b, c).Adding one antibody alone at a time did not induce any complex formation (Fig. 2e, f, middle and lower panels).In addition, in situ proximity ligation assay also revealed that intravitreal injection of VEGF-B into mouse eyes increased VEGFR1/FGFR1 complex formation in mouse retinae, while PlGF did not (red dots, Supplementary Fig. 5a, b).Thus, various in vivo and in vitro assays show that VEGF-B induces FGFR1/VEGFR1 complex formation.
VEGF-B binds to VEGFR1/FGFR1 heterodimer with a high affinity Since VEGF-B induces VEGFR1/FGFR1 complex formation, we subsequently explored whether VEGF-B binds to the VEGFR1/FGFR1 heterodimer.To address this, we generated soluble VEGFR1/FGFR1 heterodimer recombinant proteins containing the extracellular domains (ECDs) of VEGFR1 and FGFR1 connected by a linker (L) and conjugated with a histidine tag (His, Fig. 2g, h).Also, VEGFR1 and FGFR1 ECDs were arranged in different orders, thus resulting in two types of recombinant heterodimers: VEGFR1 ECD-L-FGFR1 ECD and FGFR1 ECD-L-VEGFR1 ECD (Fig. 2g, h).These recombinant proteins displayed expected molecular weights as shown by Coomassie blue staining and Western blot and did not form aggregates (Supplementary Fig. 6a, b).SPR analysis showed that VEGF-B bound to each heterodimer with similar K D values, while VEGF-C showed no binding (Fig. 2i-k), demonstrating that the binding of VEGF-B was specific.The K D values of VEGF-B for VEGFR1 ECD-L-FGFR1 ECD and FGFR1 ECD-L-VEGFR1 ECD were 1.4 nM and 3.7 nM, respectively (Fig. 2i, j).Noteworthy, and importantly, the binding affinities of VEGF-B for the VEGFR1/FGFR1 heterodimers (1.4-3.7 nM) are higher than those for the FGFR1 homodimer (17 nM, Fig. 1a), suggesting that VEGF-B may preferably bind to the VEGFR1/FGFR1 heterodimer.
We further identified the tyrosine residues in the cytoplasmic part of FGFR1 important for the inhibitory effect of VEGF-B on FGF2induced Erk activation.We generated a series of site-directed mutations of six tyrosines (Y) of FGFR1 by replacing each of them with a phenylalanine (F), respectively (Fig. 4a, Y463 to F463; Y583 to F583; Y585 to F585; Y653 to F653; Y654 to F654; Y766 to F766), and expressed each mutant in HeLa cells, which express little endogenous FGFR1. 37While VEGF-B inhibited FGF2-induced Erk phosphorylation in cells expressing wild-type FGFR1 (WT-FGFR1, Fig. 4b, c) and the FGFR1-F766, FGFR1-F654, and FGFR1-F583 mutants (Fig. 4d-g), it failed to do so in cells expressing the FGFR1-F463, FGFR1-F585, and FGFR1-F653 mutants (Fig. 4h-k), suggesting that tyrosine residues Y463, Y585, and Y653 are important for the inhibitory effect of VEGF-B.We further performed immunoprecipitation assays and VEGF-B inhibits FGF2-driven angiogenesis and tumor growth Since VEGF-B inhibits FGF2-induced Erk phosphorylation, we subsequently explored whether VEGF-B affected the angiogenic activity of FGF2 using multiple assays.An in vivo subcutaneous Matrigel assay showed that VEGF-B inhibited FGF2-induced angiogenesis, while PlGF showed no effect, demonstrating that the inhibitory effect of VEGF-B was specific (Fig. 5a, b).Moreover, a monolayer cell migration assay revealed that VEGF-B, but not PlGF, suppressed FGF2-induced migration in both HRECs (Fig. 5c, d) and HUVECs (Supplementary Fig. 14a, b).Furthermore, an EC spheroid assay showed that VEGF-B inhibited FGF2-induced EC sprouting in both HRECs (Fig. 5e-g) and HUVECs (Supplementary Fig. 14c-e), while PlGF had no effect, demonstrating that the inhibitory effect of VEGF-B was specific.
We further verified the in vivo relevance of the anti-angiogenic effect of VEGF-B using murine T241 fibrosarcoma cells expressing a high level of FGF2 (T241-FGF2, Fig. 6a) and investigated tumor angiogenesis and tumor growth in Vegf-b −/− mice (produced by the Jackson Laboratory, https://www.komp.org/,Supplementary Fig. 15a, b).We found that the T241-FGF2 cells formed significantly bigger tumors in Vegf-b −/− mice (Fig. 6b-d) with increased tumor angiogenesis (Fig. 6e, f), even though the T241-FGF2 cells expressed VEGF-B (Supplementary Fig. 15c, d), which might partially compensated for the loss of VEGF-B in Vegf-b deficient mice.Together, multiple in vitro and in vivo assays demonstrated an inhibitory effect of VEGF-B on FGF2-mediated functions.
VEGFR1 and FGFR1 are required for the inhibitory effect of VEGF-B Since VEGF-B binds to VEGFR1 (Flt1) and FGFR1, we examined whether they played a role in the inhibitory effect of VEGF-B using primary lung vascular ECs from Flt1 lox/lox and Fgfr1 lox/lox mice with Flt1 or Fgfr1 deleted, respectively, using Cre adenovirus (Cre-Ad) as confirmed by Western blots (Supplementary Fig. 16a, b).We found that while VEGF-B inhibited FGF2-induced Erk activation in wildtype ECs (Control-Ad, Fig. 7a-d), this inhibitory effect diminished after Flt1 or Fgfr1 deletion, respectively, by Cre-Ad (Fig. 7a-d).

DISCUSSION
In this study, we report our novel finding that VEGF-B, a member of the VEGF family, under conditions of high FGF2/FGFR1 levels, functions as an anti-angiogenic factor by suppressing the FGF2/ FGFR1 pathway.Mechanistically, we discovered that VEGF-B binds to FGFR1, induces FGFR1/VEGFR1 complex formation, inhibits FGF2-induced Erk activation, and thereby suppresses FGF2-driven angiogenesis and tumor growth.Our work unveils a previously unrecognized new function of VEGF-B as an endogenous inhibitor of FGF2/FGFR1-driven functions.
][10][11] For example, in various mouse models, VEGF-B inhibited tumor angiogenesis and tumor growth. 9,10In humans, high VEGF-B levels are associated with low tumor angiogenesis, better survival, and low risks of cancer, 8,11,12 and low VEGF-B levels high tumor angiogenesis, poor survival, and high risks of cancer, 8,11,13 further advocating anti-angiogenic and anti-tumor effects of VEGF-B.Notwithstanding, since the mechanisms underlying these observations are unclear, these critical findings have long been ignored.
In this work, we discovered that VEGF-B acts as an inhibitor of FGF2/FGFR1 when they are abundantly expressed.Indeed, FGF2 and FGFR1 are frequently up-regulated in tumors and play important roles in tumor angiogenesis and tumor growth. 38herefore, in tumors with high FGF2/FGFR1 levels, VEGF-B deficiency could result in greater tumor angiogenesis and tumor growth driven by FGF2/FGFR1.This may be particularly relevant for breast cancer, since FGFR1 amplification is found in 16-27% of luminal B-type breast cancers. 39Given the inhibitory effect of VEGF-B on FGF2/FGFR1, high VEGF-B levels may be beneficial for breast cancer patients by damping FGF2/FGFR1-induced tumor neovasculature and tumor growth. 11Indeed, blocking VEGF-B (together with other VEGF family members) has failed to show any benefit for most types of cancers. 14,15As such, at least one possible explanation among others might be the loss of VEGF-B's inhibition on FGF2/FGFR1 signaling in the tumors treated with VEGF-B-blocking drugs.
When is VEGF-B anti-or pro-angiogenic?Given that VEGF-B can be pro- 2,5,40,41 or anti-angiogenic [8][9][10] (and current study) depending on the specific context, outstanding questions thus exist.Under which conditions is VEGF-B antiangiogenic, and under which pro-angiogenic?And, what controls this switch?Our current study shows that at least one of the determining factors is whether FGF2/FGFR1 are expressed and at what levels.That is -when FGF2/FGFR1 are highly expressed, VEGF-B can be anti-angiogenic by inhibiting the FGF2/FGFR1 pathway (Fig. 7e).However, when FGF2/FGFR1 are not expressed/ low, VEGF-B thus has no/little FGF2/FGFR1 to inhibit but can be pro-angiogenic by promoting blood vessel survival as reported previously by our and other groups 2,4-7 (Fig. 7f).
Indeed, multiple assays, such as EC migration/sprouting and tumor growth assays showed that when FGFR1/FGF2 were highly expressed or when FGF2 was administered, VEGF-B displayed an inhibitory effect, suggesting that under conditions of high FGFR1/ FGF2 levels, VEGF-B can be anti-angiogenic by suppressing FGF2 function (Fig. 7e).FGFR1/FGF2 are highly expressed in developing mouse and rat retinae (Supplementary Fig. 17a, e) 42,43 and many types of tumors, 38,39 where the anti-angiogenic effect of VEGF-B has been observed (Supplementary Table 1).On the other hand, the previously reported pro-angiogenic effect of VEGF-B has been mostly found under conditions characterized by tissue/blood vessel degeneration, where no or low level of FGFR1 is present due to cell/ tissue death, such as in myocardial infarction or degenerated retina (Supplementary Fig. 17b-f) 5,40,41 or blood vessel regression after Fig. 2 VEGF-B induces VEGFR1/FGFR1 complex formation and binds to VEGFR1/FGFR1 heterodimer with a high affinity.a Immunoprecipitation (IP) followed by Western blot showing VEGFR1 co-precipitated with FGFR1 in mouse brain, lung and heart.b, c Immunoprecipitation followed by Western blot showing that VEGF-B, but not PlGF, induced VEGFR1/FGFR1 complex formation in mouse retinae.d Western blot showing the expression of VEGFR1 and FGFR1 in HRECs.e, f Representative images (e) and corresponding quantification (f, n = 10 per group.The experiment was repeated three times) of in situ proximity ligation assays showing that 30 min treatment of VEGF-B (50 ng/ml), but not PlGF (50 ng/ml), induced VEGFR1/FGFR1 complex formation in HRECs (top panel).Adding one antibody alone at a time did not induce any complex formation (anti-FGFR1: middle panel; anti-VEGFR1: lower panel).Blue: DAPI; red: VEGFR1/ FGFR1 complex.Data are mean ± s.e.m., n = 4 for (c) and 10 for (f).For (c) and (f), adjusted p values are from one-way ANOVA followed by Sidak post hoc analysis (number of comparisons, 2).Scale bar: 10 µm.The experiment was repeated three times.g, h Schemes of recombinant VEGFR1/FGFR1 (g) and FGFR1/VEGFR1 (h) heterodimers, each containing the extracellular domain (ECD) of VEGFR1 and FGFR1 connected by a linker (L).i-k SPR results showing VEGF-B binding to the VEGFR1/FGFR1 heterodimer (i) and FGFR1/VEGFR1 heterodimer (j) with similar K D values, while VEGF-C shows no binding (k).The red lines are the RU values at different concentrations of VEGFR1/FGFR1 heterodimers (1.875, 3.75, 7.5, 15, 30, and 60 nM).The black lines are the fitted curves FGF2 withdrawal. 2 Consistently, other groups have also reported that FGFR1 levels are much lower in adult mouse/rat hearts (almost nine folds lower than in embryonic hearts). 44,45e have previously shown that VEGF-B is pro-angiogenic in mouse models of choroidal neovascularization and retinopathy of prematurity. 27][48] Therefore, under such conditions of vascular degeneration, the previously reported anti-apoptotic effects of VEGF-B 6,7 can take place and increase blood vessel survival, as such, acting pro-angiogenic (Fig. 7f).In summary, depending on the specific condition (e.g., high versus low/no FGF2/FGFR expression), the vascular function of VEGF-B can be anti-or pro-angiogenic (Fig. 7e, f).
Mechanism underlying the inhibitory effect of VEGF-B on FGF2/ FGFR1 pathway Our data show that the inhibitory effect of VEGF-B may be mediated by its induction of VEGFR1/FGFR1 complex formation.Baseline levels of VEGFR1/FGFR1 complex were detected in multiple mouse tissues, and VEGF-B treatment increased this complex formation in both cultured endothelial cells and mouse retinae.Noteworthy, VEGF-B binds to VEGFR1/FGFR1 with a higher affinity (K D = 1.4 nM) than that for FGFR1/FGFR1 (K D = 17 nM), suggesting that the inhibitory effect of VEGF-B may be mainly through VEGFR1/FGFR1 complex formation.Indeed, genetic studies showed that VEGF-B inhibited FGF2-induced Erk activation in wild-type ECs with Flt1 and Fgfr1 expression but not in ECs with loss of either Flt1 or Fgfr1, thus supporting a critical requirement of VEGFR1/FGFR1 complex in mediating the inhibitory effect of VEGF-B.Consistently, VEGFR1 has been shown to have an antiangiogenic effect under many conditions by inhibiting Erk activation, 20,21 which could contribute to the inhibitory effect of VEGF-B-induced VEGFR1/FGFR1 complex compared with FGF2induced FGFR1/FGFR1 homodimer.

Possible translational implications
Our findings may have translational relevance.Currently, drugs that block VEGF-B together with other VEGF family members are being used in the clinic to treat patients with various neovascular diseases. 14Given the anti-angiogenic nature of VEGF-B under conditions of high FGF2/FGFR1 levels, caution is warranted to inhibit VEGF-B indiscriminately when treating patients with neovascular diseases.

MATERIALS AND METHODS
Other parts of Materials and Methods are included in the Supplementary Information.
To test whether VEGF-B competes with FGF2 for FGFR1 binding, VEGF-B protein (PeproTech) was immobilized at 6900 RU onto a CM5 chip using an amine coupling kit.A continuous flow (10 µl/ min) of FGFR1-Fc (200 nM, Genscript) with 6.25 nM of BSA, FGF2 (PeproTech) or PlGF (PeproTech), respectively, onto the surface with immobilized VEGF-B was monitored by passing the analytes across the sensor chip.
To test whether the VEGF-B peptides binds to FGFR1, an ELISA assay was used and Fc specific anti-human IgG antibody (I2136, Sigma) in PBS was captured onto a 96-well plate (Nunc maxisorp flat-bottom, 44-2404-21, Invitrogen) overnight at 4 °C followed by blocking with 1% BSA in PBS for 1 h.After incubation for 1 h with 20 nM of FGFR1-Fc protein (Z03223, Genscript), biotinylated FGF2 (200 nM, BFF-H4117, Acro Biosystems) and biotinylated VEGF-B peptides (200 nM, Genscript) were added to the wells and incubated for 2 h.The plates were washed with PBS followed by incubation with an HRPconjugated Streptavidin antibody (N100, Thermo Scientific) for 1 h.Binding signals were developed using the 1-Step TMB ELISA solution (Thermo Scientific) and the optical densities measured at 450 nm.
Competitive FGFR1 binding assay using ELISA First, a standard curve of FGF2 binding to FGFR1 was generated.The plates were coated with FGF2 protein (2.5 µg/ml, PeproTech) for overnight.The wells were washed and blocked with 1% BSA in PBS.Then, serially diluted FGFR1-Fc protein (2 nM to 0.03125 nM, 658-FR, R&D Systems) in PBS containing 1% BSA was incubated in the FGF2-coated wells for 30 min.The wells were washed and incubated with anti-human IgG-HRP to detect the bound FGFR1-Fc using the 1-Step Ultra TMB (3,3',5,5'-tetramethylbenzidine)-ELISA substrate solution (34028, Thermo Scientific).The absorbance was measured at 450 nm.The data points were fitted by saturation binding using a GraphPad Prism software (San Diego,  (100 ng/ml, 10 min treatment) inhibits FGF2 (50 ng/ml)-induced Erk activation in WT ECs (Control-Ad).This inhibition is lost upon Fgfr1 deletion (Cre-Ad).For (b) and (d), adjusted p values were from one-way ANOVA followed by Sidak post hoc analysis (number of comparisons, 2).The experiments were repeated three times.e, f Scheme illustrating the context-dependent effects of VEGF-B.Under conditions of high FGF2/FGFR1 levels, VEGF-B can be anti-angiogenic by inhibiting the FGF2/FGFR1 pathway, such as in tumors characterized by abundant FGF2/FGFR1 expression (e).Under conditions of low/no FGF2/FGFR1 expression, such as in tissue/blood vessel disintegration (e.g., myocardial infarction or blood vessel regression after FGF2 withdrawal), VEGF-B can be pro-angiogenic due to its known anti-apoptotic and survival effects (f).Therefore, depending on FGFR1/FGF2 levels, VEGF-B can be anti-or pro-angiogenic depending on the specific condition

Fig. 1 Table 1 .
Fig.1VEGF-B binds to FGFR1 and competes with FGF2 for FGFR1 binding.a-c Surface plasmon resonance (SPR) results showing VEGF-B binding to FGFR1 with a K D value (17 nM, a) similar to that for FGF2 (16 nM, b), whereas PlGF, another VEGFR1-binding member of the VEGF family, does not bind to FGFR1 (c).The lines from bottom to the top represent different concentrations of FGFR1-Fc: 12.5, 25, 50, 100, 200, and 400 nM, respectively.d, e SPR results showing VEGF-B binding to FGFR1 DII-III with a K D value (49 nM, d) similar to that of FGF2 (63 nM, e).The red lines are the resonance unit (RU) values at different concentrations of FGFR1 DII-III (25, 50, 100, 200, 400, 800 nM).The black lines are the fitted curves.f, g SPR results showing a very low binding affinity of VEGF-B (532 nM, f) and FGF2 (245 nM, g) to FGFR1 DI.The red lines are the RU values at different concentrations of FGFR1 DI (18.75, 37.5, 75, 150, 300, and 600 nM).The black lines are the fitted curves.h Scheme of the synthetic VEGF-B peptides.Blue: amino acids important for VEGFR1 binding.Red: eight cysteines forming the cysteine knots.i ELISA results showing the binding of VEGF-B peptides to FGFR1.n = 3. One-way ANOVA followed by Sidak post hoc analysis (number of comparisons against FGF2, 11) was used.Adjusted p values are <1.0E−10 for peptides 1-9, 1.0E−6 and 5.6E−3 for peptides 10 and 11.The experiment was repeated three times.j SPR results showing FGF2 competing with VEGF-B for FGFR1 binding, while PlGF does not.Kruskal-Wallis test with Dunn post hoc analysis (number of comparisons, 3) was used.k ELISA results showing VEGF-B dose-dependently competing with FGF2 for FGFR1 binding, while PlGF does not.Data are mean ± s.e.m.The experiment was repeated three times.Two-way ANOVA was used.**p < 0.01, ***p < 0.001

Fig. 6
Fig. 6 VEGF-B inhibits FGF2-overexpressing tumor growth and tumor angiogenesis.a Western blot showing the abundant expression of FGF2 in murine T241 fibrosarcoma cells (T241-FGF2).b Tumor growth curves showing that the T241-FGF2 cells formed bigger tumors in Vegf-b −/− mice than in wild-type littermates.P value was from two-way ANOVA followed by LSD multiple comparisons test.c Images showing that the T241-FGF2 cells formed bigger tumors in Vegf-b −/− than in wild-type mice.Scale bar: 1 cm.d Shown are weights of the tumors from (c).P value was from Mann-Whitney test.e Representative images showing that T241-FGF2 tumor angiogenesis was higher in Vegf-b −/− mice than in wild-type littermates.Scale bar: 20 µm.f Quantifications of tumor blood vessel densities in (c).P value was from Mann-Whitney test.All data are mean ± s.e.m.The experiment was repeated twice

Fig. 7
Fig. 7 VEGFR1 and FGFR1 are required for the inhibitory effect of VEGF-B.a, b Western blot showing that VEGF-B (10 min treatment) inhibits FGF2-induced Erk activation in WT ECs (Control-Ad).This inhibition is lost upon Flt1 deletion (Cre-Ad).c, d Western blot showing that VEGF-B(100 ng/ml, 10 min treatment) inhibits FGF2 (50 ng/ml)-induced Erk activation in WT ECs (Control-Ad).This inhibition is lost upon Fgfr1 deletion (Cre-Ad).For (b) and (d), adjusted p values were from one-way ANOVA followed by Sidak post hoc analysis (number of comparisons, 2).The experiments were repeated three times.e, f Scheme illustrating the context-dependent effects of VEGF-B.Under conditions of high FGF2/FGFR1 levels, VEGF-B can be anti-angiogenic by inhibiting the FGF2/FGFR1 pathway, such as in tumors characterized by abundant FGF2/FGFR1 expression (e).Under conditions of low/no FGF2/FGFR1 expression, such as in tissue/blood vessel disintegration (e.g., myocardial infarction or blood vessel regression after FGF2 withdrawal), VEGF-B can be pro-angiogenic due to its known anti-apoptotic and survival effects (f).Therefore, depending on FGFR1/FGF2 levels, VEGF-B can be anti-or pro-angiogenic depending on the specific condition