Introduction

Neuropilin 1 (NRP1) is a multifunctional transmembrane protein that is expressed abundantly on the surface of a range cell types, where it binds class 3 semaphorins (SEMA3), heparan sulfate and vascular endothelial growth factors (VEGFs) (Goshima et al., 1999, Mamluk et al., 2002, Gu et al., 2002). VEGFA is the main activator of VEGF receptor 2 (VEGFR2) signalling in endothelial cells (ECs) and its binding to NRP1 promotes heterocomplex formation between NRP1 and VEGFR2 (Soker et al., 2002). Thus, NRP1 has been designated as a co-receptor of VEGFR2 and is known to modulate VEGFR2 signalling. The cytoplasmic domain of NRP1 contains a C-terminal SEA motif, a PDZ binding domain that mediates binding to synectin (GIPC1) and in-turn regulates the endocytic trafficking of both NRP1 and VEGFR2 (Cai and Reed, 1999, Naccache et al., 2006, Horowitz and Seerapu, 2012, Lanahan et al., 2010). Accordingly, removal of the NRP1 cytoplasmic tail delays VEGFR2 endocytosis following VEGFA binding, leading to enhanced surface retention and reduced phosphorylation of tyrosine (Y)1175 in VEGFR2 (Lanahan et al., 2013). NRP1 thus acts as an important modulator of VEGFR2 activation and its downstream signalling pathways. However, reports have also suggested that VEGFA transduces biological responses in ECs via NRP1, in a VEGFR2-independent manner (Roth et al., 2016).

Global NRP1 knockout in mice results in lethality at E10-E12.5 due to abnormal vessel sprouting in major organs and impaired yolk sac vascularization (Kawasaki et al., 1999). Moreover, NRP1 overexpression leads to excessive vessel growth and promotes leaky and haemorrhagic vessels (Kitsukawa et al., 1995). Endothelial-specific knockout of NRP1 in mouse embryos only show mild embryonic brain defects, while in the adult vasculature no gross abnormalities are evident (Lanahan et al., 2013). Several reports also link increased NRP1 expression in tumours to a poor prognosis for survival (Miao et al., 2000, Kawakami et al., 2002, Hong et al., 2007). Furthermore, in diseases such as age related macular degeneration, NRP1 has been linked with increased neovascularization and vessel hyperpermeability, likely due to enhanced signalling downstream of VEGFA (Raimondi et al., 2014, Fernández-Robredo et al., 2017).

VEGFA-induced vascular permeability is mediated through the Y949 and Y1173 phosphosites of VEGFR2. The phosphorylated Y949 presents a binding site for TSAd (T-cell specific adaptor), which in turn binds SFKs (Src family kinases) that phosphorylate and promote internalization of junctional proteins such as VE-cadherin (Matsumoto et al., 2005, Sun et al., 2012, Li et al., 2016, Jin et al., 2022). Concurrently, pY1173 phosphorylates PLCϒ, leading to Ca2+/Protein kinase C (PKC) activation of endothelial nitric oxide synthase (eNOS), contributing to Src activation and phosphorylation of VE-cadherin (Sjoberg et al., 2023). The role of NRP1 in VEGFA-mediated permeability has been controversial. Studies employing in vivo and in vitro models have found NRP1 to be a positive regulator of VEGFA-mediated vascular permeability, or to lack effect, dependent on tissue or experimental setup (Fantin et al., 2017, Acevedo et al., 2008, Wang et al., 2015, Becker et al., 2005, Pan et al., 2007, Cerani et al., 2013). Additionally, treatment of ECs with CendR peptides, which bind and induce internalisation of NRP1, increases permeability in a NRP1-dependent, but VEGFR2-independent manner (Roth et al., 2016). Therefore, the exact role of NRP1 in VEGFA-VEGFR2 mediated permeability remains unclear. NRP1 is a promising target in a number of pathologies where vascular dysfunction is exacerbative. Understanding the relationship between NRP1 and VEGFA-VEGFR2 signalling is thus of potential therapeutic benefit.

Here, we aimed to further uncover the role of NRP1 in VEGFA-mediated vascular permeability. Using mice with global or EC-specific loss of NRP1 expression, we identify a tissue-specific role for endothelial NRP1 in the modulation of VEGFA/VEGFR2 signalling regulating vascular leakage. We find that endothelial-specific loss of NRP1 (Nrp1 iECKO) can both increase and decrease VEGFA-mediated vascular leakage, dependent on the vascular bed. NRP1 is widely expressed and is an important constituent of peri-vascular cells in some, but not all, tissues and vessel subtypes. The consequence of global loss of NRP1 expression demonstrates that peri-vascular NRP1 can modify VEGFA/VEGFR2-induced vascular leakage and steer the tissue-specific role of endothelial NRP1 in barrier integrity. Collectively, these data reveal that the relative expression levels of NRP1 between perivascular cells and ECs acts as a tissue-specific modifier of EC VEGFA/VEGFR2 signalling upstream of vascular leakage.

Results

NRP1 regulates VEGFA-mediated permeability in an organotypic manner

NRP1 has been studied extensively as a co-receptor of the VEGFA/VEGFR2 pathway, and its ability to modulate VEGFR2 signalling in developmental and pathological angiogenesis has been clearly established. However, results concerning the role of NRP1 in VEGFA-mediated vascular permeability have remained contradictory (Fantin et al., 2017, Acevedo et al., 2008, Wang et al., 2015, Becker et al., 2005, Pan et al., 2007, Cerani et al., 2013). Nrp1^fl/fl; Cdh5^CreERT2 (Nrp1 iECKO) mice were used to study the EC-specific role of NRP1 in VEGFA-induced vascular leakage. Following tamoxifen administration, a 75% reduction in NRP1 protein levels was achieved in lung tissue, chosen for analysis due to the high EC content (Figure 1A and B). Furthermore, using immunofluorescent stainings, reduction in NRP1 levels could be observed in the ear and back skin vascular beds with the loss of EC NRP1 (Supp. figure 1A and B). Using a highly sensitive assay combining intravital microscopy and atraumatic intradermal microinjection in the ear skin of Nrp1 iECKO mice, VEGFA-induced vascular leakage was significantly increased following the loss of NRP1 (Figure 1C and D) (Honkura et al., 2018). Additionally, we measured qualitative aspects of vascular leakage including lag period (time from stimulation to leakage onset) and degree of leakage (rate of dextran extravasation). These analyses showed that ECs in Nrp1 iECKO mice respond faster to stimulation (Figure 1E) and exhibit more profuse leakage at each site (Figure 1F), indicative of increased disruption of EC-EC junctions. These data suggest that NRP1 exerts a stabilizing role by suppressing VEGFA-induced endothelial junction disruption in the ear dermis. This finding is in contrast to reports showing NRP1 to be a positive regulator of VEGFA-mediated vascular leakage (Fantin et al., 2017, Acevedo et al., 2008, Becker et al., 2005).

Figure 1:

One possible explanation for these differences compared to previous reports is that the role of NRP1 may be organ dependent, in line with the growing insights into the distinct properties of ECs in different vessel types and vascular beds (Augustin and Koh, 2017). To investigate a potential organotypic role for NRP1 in EC biology, we studied the consequence of Nrp1 knockout on the barrier properties of ECs in different vascular beds.

Initially, selected organs, chosen based on prior experience of their different EC barrier properties (Aird, 2007, Augustin and Koh, 2017, Richards et al., 2022), were investigated for their susceptibility to VEGFA-induced vascular leakage via the systemic administration of fluorescent dextrans with or without VEGFA. After 30 minutes, organs were collected and acute vascular leakage assessed microscopically or following solvent-based extraction of dextran and fluorescence spectroscopy. All organs analysed, except for the liver, showed vascular leakage after VEGFA stimulation (Supp. figure 2A - E). The role of endothelial NRP1 in these organs was further studied in Nrp1 iECKO mice, showing significantly decreased VEGFA-mediated vascular leakage in the trachea and back skin (Figure 1G and H, Supp. figure 3). Kidney, skeletal muscle and heart in contrast did not show a significant difference in VEGFA-mediated vascular leakage following the loss of EC-expressed NRP1 (Figure 1I and J).

Figure 2:

Figure 3:

These data collectively suggest that endothelial NRP1 has an organ-specific role in controlling VEGFA-mediated vascular permeability. Importantly, endothelial NRP1 is a positive regulator of VEGFA mediated permeability in the trachea and back skin but a negative regulator in the ear skin.

Global inactivation of NRP1 reduces VEGFA-mediated vascular permeability

Our results here, where NRP1 may play positive (trachea and back skin), negative (ear skin) and passive (kidney, skeletal muscle and heart) role in VEGFA-induced vascular leakage, were collected using EC-specific knockout mice. In previous studies, mouse models of both EC-specific deletion and a globally expressed NRP1 C-terminal deletion mutant were employed (Fantin et al., 2017, Roth et al., 2016). We thus set out to investigate whether global inactivation of NRP1 using an anti-NRP1 blocking antibody might differently modify the VEGFA-induced leakage response. Wild-type C57Bl/6 mice were locally treated in the ear dermis with an anti-NRP1 antibody via intradermal injection, followed by intravital assessment of VEGFA-induced vascular permeability. Interestingly, NRP1 blockade reduced vascular leakage in the ear dermis when compared to isotype control (Supp. figure 4A). This is in contrast to the EC-specific loss of NRP1 in Nrp1 iECKO mice, which resulted in enhanced VEGFA-mediated vascular leakage (Figure 1C-F). We similarly investigated the effect of blocking NRP1 on the back skin and tracheal vasculatures. Treatment with anti-NRP1 antibody was found to reduce VEGFA-induced leakage in both tissues following local and systemic administration of VEGFA respectively (Supp. figure 4B and C).

Figure 4:

To verify the findings using neutralizing anti-NRP1 antibodies, we crossed Nrp1^fl/fl with Actb^CreERT mice to generate global Nrp1 knock-out mice (Nrp1 iKO). Efficient reduction of NRP1 protein was confirmed in lung lysates (Figure 2A and B). Complete loss of NRP1 could also be seen in the ear and back skin of these mice using immunofluorescent staining (Supp. figure 1). Intravital imaging of the ear dermis of Nrp1 iKO mice showed that global loss of NRP1 reduced VEGFA-induced vascular leakage, in contrast to the phenotype seen in Nrp1 iECKO mouse ear skin and in agreement with blocking antibody data (Figure 2C and D). Furthermore, global Nrp1 knock-out resulted in a ∼75% reduction in VEGFA-mediated vascular leakage also in the back skin and tracheal vasculatures, similar to that seen in Nrp1 iECKO mice (Figure 2E and F).

These data illustrate the heterogenous nature of EC signalling regulation in different vascular beds, and show that NRP1 modulates VEGFA signaling in an organotypic and cell-specific manner. Notably, we demonstrate that VEGFA-induced leakage in the back skin and trachea is indistinguishable between endothelial-specific and global Nrp1 KO mice, whereas in the ear skin there is a contrasting phenotype.

NRP1 is heterogeneously expressed in perivascular cells

NRP1 is known to form a heterocomplex with VEGFR2 upon VEGFA binding. Interestingly, VEGFR2/NRP1/VEGFA can assemble as a juxtacrine trans complex, where NRP1 and VEGFR2 are expressed on the surface of adjacent cells, as well as a cis complex, where NRP1 and VEGFR2 are expressed on the same cell (Koch et al., 2014). Importantly, even though the kinetics of trans VEGFR2/NRP1 complex formation is slow, these complexes are stable and produce a distinct signalling output compared to the cis configuration (Koch et al., 2014). Thus, NRP1 presented in cis or trans produces differential VEGFR2 signalling output upon VEGFA binding. Given the above findings we reasoned that peri-endothelial distribution of NRP1 could be an important modifier of EC VEGFR2 signalling and possibly explain organotypic differences observed in Nrp1 iECKO mice.

To investigate the pattern of NRP1 expression we employed a Pdgfrβ promoter-driven GFP (Pdgfrβ-GFP) mouse to visualize PDGFRβ-positive pericytes. To visualize ECs, and the localization of NRP1 in the ear skin and back skin from these mice, tissues were immunostained for CD31 and NRP1, respectively. NRP1 expression could be seen in both endothelial and perivascular cells, which was lost after global Nrp1 deletion (Figure 3A and Supp. Figure 1). Both arteriolar (Figure 3B) and capillary (Figure 3C) PDGFRβ-positive perivascular cells expressed NRP1. The relative level of perivascular/EC expression of NRP1 was higher in the ear skin compared to back skin. In venules, the low expression of NRP1 in perivascular cells in the back skin resulted in a 5-fold difference in the perivascular/EC NRP1 ratio, when comparing the ear skin to back skin (Figure 3D).

These data thus show that NRP1 is expressed in perivascular cells of the ear skin and back skin. We find however that the ratio of NRP1 expression between ECs and perivascular cells differs between ear skin and back skin, and between different vessel types. In ear skin the perivascular/EC ratio is higher compared to the back skin, which may support a higher ratio of trans NRP1/VEGFR2 relative to cis complexes in the ear skin.

NRP1 distribution modifies VEGFA-mediated signalling and vascular leakage

NRP1 modifies VEGFR2 signalling by controlling its internalisation and intracellular trafficking (Bayliss et al., 2020, Ballmer-Hofer et al., 2011). The presence of NRP1 in trans however, modifies this dynamic and retains VEGFR2 at the cell surface for longer time periods, altering its signalling output (Koch et al., 2014).

We thus wished to investigate whether perivascular NRP1 expression might interact with and alter VEGFR2 signalling upstream of vascular leakage, and explain the above described organotypic effects of EC NRP1 loss. For this purpose, Nrp1 iECKO mice were crossed with homozygous Vegfr2^Y949F/Y949Fmice, to produce Vegfr2^Y949F/Y949F;Nrp1 iECKO mice that are deficient in both EC NRP1 expression and VEGFA-mediated vascular leakage ability. In these mice, loss of NRP1 protein was efficiently established in lung tissue (Figure 4A and B). Analysis of VEGFA-mediated vascular leakage in the ear dermis showed that the increased leakage, induced by the loss of EC NRP1, was abrogated by the loss of the VEGFR2 phosphosite Y949, known to be required for activation of Src in response to VEGFA (Figure 4C and D). These analyses suggest that, in the ear dermis, EC NRP1 negatively regulates VEGFA-induced vascular leakage by modulation of VEGFR2 activation and phosphorylation.

We next sought to determine the relationship between NRP1 and VEGFR2 Y949 signalling in the trachea and back skin vasculatures. Leakage in Vegfr2^Y949F/Y949F;Nrp1 iECKO mice was induced by systemic VEGFA and fluorescent 2000 kDa dextran administration in both Nrp1 iECKO and Vegfr2^Y949F/Y949F;Nrp1 iECKO mice. As we observed previously, loss of EC NRP1 reduced vascular leakage in the trachea and back skin by 75%, an effect that was enhanced further by the VEGFR2 Y949F mutation (Figure 4E and F). Importantly, leakage was similarly inhibited in the Vegfr2^Y949F/Y949F model as in the combined Vegfr2^Y949F/Y949F;Nrp1 iECKO mice in both the trachea and back skin. These data show that NRP1 presented in trans by perivascular cells in the ear dermis can signal through VEGFR2 to induce vascular leakage, and that VEGFR2 Y949 is vital for mediating NRP1’s effects on endothelial junctions.

Perivascular NRP1 modifies VEGFA-induced signalling

VEGFA/VEGFR2 signalling mediates phosphorylation of VE-Cadherin on Y685, required for vascular permeability (Orsenigo et al., 2012, Wessel et al., 2014). We have also recently shown that PLCγ is an important mediator of VEGFA-induced vascular leakage by allowing the production of nitric oxide through eNOS, which in turn modified Src by tyrosine nitration, required for full Src activity (Sjoberg et al., 2023). To clarify the impact of perivascular NRP1 on VEGFA-induced signalling we studied VE-Cadherin and PLCγ phosphorylation in Nrp1 iECKO and Nrp1 iKO mice (Figure 5). Perivascular NRP1 expression in the ear dermis of Nrp1 iECKO mice was lost in Nrp1 iKO mice (Supp. figure 1). In agreement with the enhanced vascular leakage in the ear dermis of Nrp1 iECKO mice, VE-Cadherin phosphorylation at Y685 was enhanced following intradermal VEGFA injection (Figure 5A). No such induction was seen in Nrp1 iKO mice (Figure 5B). In the back skin, VE-Cadherin pY685 levels were induced by VEGFA in the control but not in the Nrp1 iECKO, nor the iKO models (Figure 5C and D), in accordance with the permeability phenotype (see Figure 1). Similarly, PLCγ phosphorylation was potentiated in the ear skin of Nrp1 iECKO but not Nrp1 iKO mice (Supp. figure 5A and B). Moreover, in the back skin loss of either endothelial or global NRP1 led to a reduction in PLCγ phosphorylation following intradermal VEGFA stimulation (Supp. figure 5C and D).

Figure 5:

Taken together these data show that NRP1 is expressed by perivascular cells in a tissue-specific manner and that perivascular NRP1 can be an important modulator of VEGFA-signalling. In the ear dermis, NRP1 is expressed by both endothelial and perivascular cells, forming cis and trans interactions with VEGFR2 respectively. Whilst the cis EC interaction predominates, loss of EC NRP1 promotes trans NRP1/VEGFR2 interaction and potentiates VEGFA-induced signalling regulating vascular leakage (Figure 6). Meanwhile, due to a paucity of perivascular NRP1 in the trachea and back skin vasculatures, loss of EC NRP1 leads to a loss of NRP1:VEGFR2 complexes, attenuating VEGFA-regulated vascular leakage.

Figure 6:

Discussion

Here we find that NRP1 expression by perivascular cells is an important modulator of VEGFA signalling regulating endothelial junction stability. Our data suggest that NRP1 presented in a juxtacrine manner can potentiate certain VEGFR2 signalling upstream of vascular leakage. NRP1 is expressed by a number of cells adjacent to blood vessels, including PDGFRβ-positive pericytes. However, the abundance of these cells and their level of NRP1 expression differ between vascular beds. Loss of EC-specific NRP1 thus has a differential effect on VEGFA-mediated vascular leakage in different vascular beds, dependent on the ability of ECs to form trans NRP1/VEGFR2 complexes.

Numerous studies have implicated endothelial NRP1 as a positive regulator of permeability (Fantin et al., 2017, Becker et al., 2005, Acevedo et al., 2008). However, multiple studies argue against this, with NRP1 blockade or deletion having no effect on VEGFA-induced vascular leakage in the skin and retina (Pan et al., 2007, Cerani et al., 2013). Here we find that the function of endothelial NRP1 in VEGFA-mediated vascular permeability varies in different vascular beds. In keeping with previous data, we find that loss of NRP1 reduces VEGFA-induced vascular leakage in the back skin and trachea. Meanwhile, endothelial NRP1 appears to have no effect on VEGFA-induced vascular leakage in kidney, skeletal muscle and heart. In the ear skin, however, loss of NRP1 surprisingly leads to an increase in VEGFA-induced vascular leakage. This appears to be through enhanced VEGFR2 Y949 signalling, as mutation of this phosphosite normalises the enhanced leakage seen after loss of NRP1. Previous publications have described NRP1 as being pivotal for VEGFR2-induced Src activity via recruitment of Abl kinase (Fantin et al., 2017). How exactly VEGFA-VEGFR2 signals to Abl, and whether this requires the Y949 phosphosite, however is unknown. Further work is required to understand exactly how NRP1 modifies VEGFR2 activity and how this differs between different vascular beds.

As others have noted and we see here, NRP1 is expressed by perivascular cells, including PDGFRβ positive pericytes (Bartlett et al., 2017, Wnuk et al., 2017). However, we find that perivascular NRP1 expression and its ratio to EC expression differs between different vascular beds, being higher in the ear skin than back skin. When the trans NRP1/VEGFR2 complex predominates, as in the ear skin of Nrp1 iECKO mice, our data indicate that VEGFA/VEGFR2 signalling impacting junction stability is potentiated. This is in keeping with reports showing that NRP1 presented in trans can modify VEGFR2 signalling kinetics, resulting in prolonged PLCγ and ERK2 phosphorylation (Koch et al., 2014). However how such prolonged kinetics might affect the VEGFR2 signalling pathways upstream of vascular leakage has not been previously explored. In addition, juxtacrine NRP1 interactions have been shown to have important functions in cell types other than ECs. For example, NRP1 expressed on microglia is important for the trans-activation of PDGFRα on oligodendrocyte precursor cells, resulting in enhanced proliferation and myelin repair (Sherafat et al., 2021). Furthermore, NRP1 has been posited as a potential immune checkpoint inhibitor (Sarris et al., 2008). NRP1 is highly expressed by regulatory T cells, as well as antigen presenting dendritic cells, and the formation of trans NRP1 homodimers has been suggested to stabilise regulatory T cell function and promote immune tolerance. Finally, numerous studies have described the expression of NRP1 on tumour cells, which can modify EC behaviour and alter their angiogenic potential (Koch et al., 2014, Miao et al., 2000, Hu et al., 2007, Shahrabi-Farahani et al., 2016). This may partly be due to the ability of NRP1 to bind VEGFA and thus provide a pool of tumoural VEGFA towards which ECs migrate. In tumours, NRP1 can also form juxtacrine interactions with VEGFR2 on ECs upon VEFGA binding and modify intracellular EC signalling compared to NRP1 action in a cis manner (Koch et al., 2014, Morin et al., 2020, Morin et al., 2018).

One possible mechanism by which NRP1 presented in trans might alter VEGFA/VEGFR2 signalling, when compared to cis, is via modification of VEGFR2 trafficking. The adapter molecule synectin is an important intracellular binding partner of NRP1 and facilitates myosin IV-mediated endocytosis (Cai and Reed, 1999, Naccache et al., 2006). A recent study also found that NRP1 interacts with members of the ERM family of intracellular proteins, which act as intermediaries between the plasma membrane and actin cytoskeleton and that have also been associated with endocytosis and intracellular trafficking (Ponuwei, 2016, Gioelli et al., 2022). NRP1 may thus function not just as a co-receptor, but also as a linker to mediate the intracellular trafficking of cell surface receptors. This idea is supported by studies showing that NRP1 co-ordinates PDGFRα internalisation and its intracellular signalling, as well as the trafficking of integrin α5β1 during cell adhesion (McGowan and McCoy, 2021). Furthermore, it has recently been shown that NRP1 binds to VE-Cadherin and regulates its cell surface turnover, a role that has been proposed to at least partly explain its role in EC barrier integrity (Gioelli et al., 2022, Bosseboeuf et al., 2023). In response to VEGFA the cytoplasmic domain of NRP1, which binds synectin, has been found to be crucial in mediating NRP1’s positive regulation of leakage (Fantin et al., 2017). Furthermore, mutant mice lacking the NRP1 cytoplasmic domain have impaired arteriogenesis due to delayed VEGFR2 endocytosis and resulting reduction in phosphorylation at Y1175 (Lanahan et al., 2013). Interestingly, the C-terminal tail is also crucial for the stability of trans NRP1-VEGFR2 complexes (Koch et al., 2014). A central function of NRP1 is thus to mediate VEGFR2 internalisation and trafficking following VEGFA stimulation. From our data we hypothesise that NRP1 presented in trans delays VEGFR2 internalisation and potentiates VEGFR2 signalling upstream of vascular leakage. Indeed, a recent study found that proper internalisation of VEGFR2 is required to shutdown VEGFR2 Y949 signalling through increasing the stoichiometric ratio of cell surface phosphatases that dephosphorylate Y949 (Corti et al., 2022). Our data are compatible with the hypothesis that VEGFR2 held at the cell surface via juxtracrine NRP1 interaction is capable of enhancing VEGFR2 Y949 signalling and resulting disruption of EC junctions, but due to the rapid turnover of VEGFR2 phosphorylation, this is technically very challenging to conclusively demonstrate. We hypothesize however that NRP1 expression within the perivascular microenvironment could be an important modifying factor of EC signalling and function, and might play an important role in mural cell function during vascular development and quiescence.

Combined, we identify a differential function of endothelial NRP1 in VEGFA-mediated vascular leakage in different vascular beds. Different levels and patterning of perivascular NRP1 expression has the potential to modify VEGFR2 signalling and alter VEGFA-induced vascular hyperpermeability. Further elucidation of NRP1 expression patterns in different tissues and how the VEGFA/NRP1/VEGFR2 complex presented in cis or trans can differentially modify cellular signalling will allow to better understand how NRP1 may serve in therapeutic applications in a distinct manner in different tissues.

Acknowledgements

We thank the BioVis imaging facility, Uppsala University, for expert services. We appreciate expert advice from Dr. Konstantin Gängel. This study was supported by the Swedish Research Council (2022-00896), the Knut and Alice Wallenberg foundation (KAW 2020.0057 and KAW 2019.0276), Fondation Leducq Transatlantic Network of Excellence Grant in Neurovascular Disease (17 CVD 03) to LCW. MR was supported by Olle Engkvist (218-0057) SSMF (201912) and an EMBO long-term fellowship (ALTF 923–2016).

Author contributions

Conceptualization, M.R. and L.C.-W.; methodology, M.R. and S.P.; investigation, M.R., S.P., Y.S.; writing, M.R., S.P. and L.C.-W.; funding acquisition; M.R. and L.C.-W.; resources, L.C.-W.

Conflict of interest

M.R. is now an employee for AstraZeneca (Biopharmaceuticals R&D, Cambridge, UK). All of this work was performed at Dept. of Immunology, Genetics and Pathology, Beijer and Science for Life Laboratories, Uppsala University, Uppsala, Sweden. No funding or support was received from AstraZeneca. Other authors declared no conflict of interest.

Methods

Animals

In vivo animal experiments were carried out in accordance with the ethical permit provided by the Committee on the Ethics of Animal Experiments of the University of Uppsala (permit 6789/18). Nrp1^flox/flox mice contain loxP sites flanking exon 2 and were obtained from The Jackson Laboratory (005247). This strain was crossed with Cdh5(PAC)-CreER^T2 mice (kind gift from Dr. Ralf Adams, Max-Planck Institute Münster) or Tmem163Tg(ACTB-cre)2Mrt mice to generate endothelial-specific Nrp1 iECKO mice and global Nrp1 iKO mice respectively. Mice possessing a knock-in mutation in the VEGFR gene, Kdr, at the Y949 phosphosite have been described previously (Li et al., 2016). These mice were crossed with Nrp1 iECKO mice to generate Vegfr2 ^Y949F/Y949F; Nrp1 iECKO mice. Pdgfrb-GFP reporter mice (Tg(Pdgfrb-EGFP)jn169Gsat/Mmucd, Stock number 031796-UCD) have been described previously (Jung et al., 2018). Mice were maintained in ventilated cages with group housing (2–5 per cage) and access to water and feed ad libitum. Male and female mice age 8-18 weeks were used. Each experiment was conducted on tissue from at least three age-matched animals representing individual biological repeats. Sample size (number of acquired images / movies and number of mice) were chosen to ensure reproducibility and allow stringent statistical analysis. To induce Cre recombinase-mediated gene recombination tamoxifen (SigmaAldrich, T5648) formulated in peanut oil was injected intraperitoneally (80 mg/kg) for 5 consecutive days. The mice were allowed to rest for 4 days before experiments were conducted.

Intravital vascular leakage assay

Intravital imaging of the mouse ear with intradermal injection has been described previously (Honkura et al., 2018). Briefly, following systemic administration of 2000 kDa FITC (SigmaAldrich, FD2000S) or TRITC Dextran (ThermoFischer Scientific, D7139) by tail-vein injection, mice were sedated by intraperitoneal injection of Ketamine-Xylazine (120 mg/kg Ketamine, 10 mg/kg Xylazine) and the ear secured to a solid support. Mice were maintained at a body temperature of 37 °C for the entire experiment, maximum 90 min. Time-lapse imaging was performed using single-photon microscopy (Leica SP8). For intradermal EC stimulation, a volume of approximately 0.1 μl VEGFA (Peprotech, 450-32), concentration 100 ng/μl, was injected using a sub-micrometer capillary needle. 10 kDa TRITC Dextran (ThermoFischer Scientific, D1817) was used as a tracer. Leakage sites were identified in time-lapse imaging as defined sites of concentrated dextran in the extravascular space.

Miles’ assay

Mice were injected intraperitoneally with pyrilamine maleate salt (4mg/kg, Sigma, P5514) diluted in 0.9% saline, to inhibit histamine release, 30 min prior to tail vein injection of Evans blue (100 μl, 1% Evans blue). Evans blue was allowed to circulate for 10 min before intradermal injection in the back skin of recombinant mouse VEGFA165 (Peprotech, 450-32) (80 ng in 25 μl) or PBS. The skin was excised 30 minutes subsequent to VEGFA injection, placed in formamide over night at 55°C and absorbance at 620nm was measured using a spectrophotometer and normalized to tissue weight.

Systemic VEGFA-induced permeability

To assess EC permeability, all tissues were cleaned of excess cartilage, fat and connective tissue. Skeletal muscle was assessed using the tibialis anterior and liver was assessed using the left lateral lobe. To analyse liver, kidney, skeletal muscle and heart permeability, dextran (30 mg/kg) with or without VEGFA (160 μg/kg) were injected systemically through the tail vein. Thirty minutes later mice were anesthetized with Ketamine/Xylazine before intracardiac perfusion with Dulbecco’s phosphate buffered saline (DPBS). Tissues were dissected, washed in DPBS and incubated in formamide for 48 hr at 55 °C. Dextran fluorescence was then measured using a spectrophotometer and normalized to tissue weight. To assess VEGFA-induced leakage microscopically, mixtures of fixable dextran (FITC, Tdb labs; TRITC, ThermoFischer Scientific, D7139) (30 mg/kg) with VEGFA (160 μg/kg) were injected systemically through the tail vein. Thirty minutes later mice were anesthetized with Ketamine/Xylazine before intracardiac perfusion with DPBS followed by 1% paraformaldehyde. Tissues were then immersed in 1% paraformaldehyde for 2 hr before proceeding with immunohistochemistry. For leakage quantification at least three large tile scan areas (≥1 mm²) were captured for each mouse.

Intradermal VEGFA injection in the ear and back skin

Mice were anesthetised using isofluorane and injected with PBS or VEGFA (Peprotech, 450-32) (20ug in 5 ul for the ear dermis and 80ug in 20ul for the back skin) followed by perfusion using 1% PFA. The ear and back skin samples were excised and further fixed for 1 hour in 1% PFA. Afterwards, the tissue samples were processed for immunostaining.

NRP1 blockade

Anti-NRP1 (R&D systems, MAB59941) or isotype control (R&D Systems, MAB006) were administered intradermally in the ear skin (1 µg) or back skin (5 µg), or systemically (2 mg/kg). 24 hours later VEGFA-mediated vascular leakage was assessed using the intravital vascular leakage assay, Miles’ assay or through systemic VEGFA-induced permeability.

Immunohistochemistry

Tissues were fixed through intracardiac perfusion with 1% paraformaldehyde (PFA) followed by immersion in 1% PFA for 2 hr at room temperature. For whole mount preparation of the ear skin, back skin and trachea removal of excess cartilage, fat and connective tissue tissues was carried out. Tissues were blocked overnight at 4 °C in Tris-buffered saline (TBS), 5% (w/v) Bovine Serum Albumin (BSA), 0.2% Triton X-100. Samples were incubated overnight with primary antibody in blocking solution, followed by washing in TBS, 0.2% Triton X-100 and incubation with appropriate secondary antibody for 2 hr at room temperature in blocking buffer before washing and mounting in fluorescent mounting medium (DAKO). Images were acquired using a Leica SP8 confocal microscope. Commercial antibodies used were: rat anti-CD31 (BD Biosciences, 553370), goat anti-CD31 (R&D Systems, AF3628), goat anti-NRP1 (R&D Systems, AF566), chicken anti-GFP (Abcam, Ab13970), rabbit anti-phospho-PLCγ1 (Tyr783) (Cell signaling technology, 2821s), rabbit anti-PLCγ1 (Cell signaling technology, 2822s), and goat anti–VE-cadherin (R&D systems, AF1002). Secondary antibodies against rat (ThermoFischer Scientific; Alexa 488, A21208 and Alexa 594, A21209), rabbit (ThermoFischer Scientific; Alexa 488, A21206 and Alexa 568, A10042), goat (ImmunoResearch Laboratories, Alexa 647, 705-605-147), chicken (ImmunoResearch Laboratories, Alexa 488, 703-545-155) were used. pVEC Y685 antibody was prepared by immunizing rabbits with phospho-peptides of the corresponding region in mouse VE-cadherin (New England Peptide) and further verified using immunostaining in Cdh5^Y685F/Y685F mice. Primary and secondary antibodies were prepared at a dilution of 1:100 and 1:400, respectively unless otherwise stated.

Western blot analysis

Lungs from mice were removed and snap frozen. Protein was obtained by mechanical dissociation in RIPA buffer supplemented with 50 nM Na3VO4, Phosphatase inhibitor cocktail (Roche 04906837001) and Protease inhibitor cocktail (Roche, 04693116001). LDS sample buffer (Invitrogen, NP0007) and Sample Reducing Agent (Invitrogen, NP0009) were added to the samples and heated to 70 °C for 10 min. Proteins were separated on Nu Page 4–12% Bis-Tris Gel (Invitrogen) in MOPS SDS Running buffer (Invitrogen, NP0001), transferred to PVDF membrane (Thermo scientific, 88518) in NuPAGE transfer buffer (Novex, NP006), 10% methanol and subsequently blocked with 5% BSA in Tris-buffered saline with Tween 20 (TBST) for 60 min. The immunoblots were analysed using primary antibodies incubated overnight at 4 °C and secondary antibodies linked to horseradish peroxidase (HRP) (Cytiva) incubated for 1 hr at room temperature. After each step filters were washed four times with TBST. HRP signals were visualized by enhanced chemiluminescence (ECL) (Cytiva) (1:25000) and imaged with Chemidoc (Bio-Rad).

Primary antibodies targeting GAPDH (Chemicon, MONOCLONAL ANTIBODY374), NRP1 (R&D Systems, AF566) were used at a dilution of 1:1,000.

Image quantification

For confocal images, macromolecular leakage was quantified by measurement of tracer area following image thresholding. Tracer area was then normalized to vessel area. For analysis of vascular and peri-vascular NRP1 expression, masks were generated of the CD31 vascular area and the peri-vascular area 3μm around the vessel (Figure 3A). NRP1 area following thresholding was quantified within, and normalised to these areas. Phosphorylation values were quantified by measurement of phosphorylated vessel area following image thresholding. Phosphorylation area was then normalized to vessel area. Threshold values were kept constant within experiments.

Analysis of leakage from intravital movies has been described previously (Richards et al., 2021). Briefly, leakage sites were identified in time-lapse imaging as defined sites of concentrated dextran in the extravascular space. To quantify these their numbers were normalized to vessel length. To assess lag period the time of the appearance of these sites following injection of stimulus was quantified. To assess the extent of barrier disruption the rate of dextran extravasation specifically at these sites was quantified over time.

All measurements were done with Fiji processing package of Image J2 software.

Statistical analysis

Data are expressed as mean ± SEM. The statistical tests used were the Students’t test and two-way ANOVA. P-values given are from independent samples analysed by two-tailed paired t tests or multiple comparisons obtained from two-way ANOVA. Rate of leakage was compared using linear regression and ANCOVA. All statistical analyses were conducted using GraphPad Prism. A p-value <0.05 was considered statistically significant and significances indicated as p<0.05 (*), p<0.01 (**), and p<0.001 (***). For animal experiments, no statistical methods were used to predetermine sample size. The investigators were blinded to allocation during experiment and outcome assessment.