Vegfa/vegfr2 signaling is necessary for zebrafish islet vessel development, but is dispensable for beta-cell and alpha-cell formation

The mechanisms underlying zebrafish pancreatic islet vascularization have not been well characterized. We sought to determine the angiogenic factors responsible for islet vascularization and assess whether an absence of endothelial cells affects beta-cell and alpha-cell formation. We used a double transgenic zebrafish Tg(fli1:EGFP; insa:tagRFP) to label endothelial cells and beta-cells, respectively. Beta-cells developed adjacent to endothelial cells and by 72 hours post fertilization (hpf) the zebrafish pancreatic islet was highly vascularized. Zebrafish beta-cells express vascular endothelial growth factors (vegf), vegfaa and vegfab. Double knockdown of vegfaa and vegfab or the primary Vegfa receptors (Vegfr2), kdr and kdrl, resulted in vessel deficient islets. While beta-cell and alpha-cell numbers remained unchanged in vessel deficient islets, insulina expression was downregulated relative to controls. Vegfaa/Vegfab-Vegfr2 signaling is necessary for proper islet vessel development, but not for the initial formation of beta-cells and alpha-cells.

The pancreas contains both endocrine and exocrine components. The exocrine pancreas constitutes the majority of the pancreas and produces digestive enzymes which are delivered to the duodenum. The endocrine pancreas consists of the islets of Langerhans that are scattered throughout the exocrine tissue. The primary function of pancreatic islets is to regulate blood glucose levels through the secretion of hormones. The islet consists of 5 endocrine cells types, the insulin secreting beta-cells, glucagon secreting alpha-cells, somatostatin secreting delta-cells, ghrelin secreting epsilon-cells, and the pancreatic polypeptide secreting PP-cells.
Pancreatic islets are highly vascularized. Studies in mice indicate that reciprocal interactions between endothelial cells and islets are important for proper islet development, maturation, and function 1,2 . During murine embryogenesis, endothelial cells are important in pancreas specification. The maintenance and induction of key pancreatic transcription factors PDX1 and PTF1A is dependent on signals from aortic endothelial cells, without which pancreas development is severely impaired [1][2][3] . In addition to initiating pancreas morphogenesis, endothelial cells also communicate with mature islet cells. These interactions between islet cells and endothelial cells are primarily mediated by vascular endothelial growth factor-A (VegfA) signaling 4 . Lack of islet VegfA in the early murine pancreas or in mature beta-cells results in a significant loss of intra-islet capillaries, impairments in insulin secretion, and glucose intolerance [4][5][6][7][8] .
While the role of endothelial cells on islet development has been well studied in murine models, it is less documented in zebrafish. Zebrafish is an ideal organism to study islet vessel development due to their transparency and rapid ex-utero development. Zebrafish pancreas development shares many similarities with mammals suggesting that studies within this system can have broadly relevant insights 9 . While it has been previously observed that some insulin-expressing cells still develop in cloche mutants which lack endothelial cells 10 , signals involved in zebrafish islet vascularization and its relationship with islet development is not completely understood.
In this study, we used a combination of genetic knockdown and pharmaceutical techniques to assess the role of vegfaa and vegfab in zebrafish islet vessel development and endocrine pancreas formation. We demonstrate Vegf signaling is essential for islet vessel development, but not beta-cell and alpha-cell formation.
To determine if Vegf signaling is required for islet vascularization, we administered a Vegf receptor competitive inhibitor SU5416. Tg(fli1:EGFP; insa:tagRFP) embryos were treated with SU5416 at 12 hpf until imaging at 72 hpf to assess the development of islet vessels. SU5416 treatment from 12 to 72 hpf reduced islet vessel density ( Fig. 2a-d). This reduction in islet vessels may partially be caused by a failure of the sub-intestinal vein to form which partially gives rise to the pancreatic vessels 11 (Supplementary Fig. 1a,b). No significant changes in beta-cell and alpha-cell numbers were observed in the SU5416-treated embryos (25.2 ± 2.6; 18.8 ± 2.8) in comparison to the DMSO-treated (27.5 ± 2.5; 19.2 ± 2.0) and untreated wildtype (27.7 ± 2.9; 18.5 ± 2.4) control embryos at 72 hpf (Fig. 2e,f). Islet architecture was not affected in vessel deficient embryos as the majority of alpha-cells were observed on the islet mantle with beta-cells localized to the islet core in both the control and SU5416-treated embryos.
To test if continued Vegf signaling is needed to sustain islet vessels, we treated embryos with SU5416 at 72 hpf until imaging at 96 hpf. We observed a reduction of islet vessels suggesting that continued Vegf signaling is necessary to sustain islet vasculature ( Fig. 2h-j). No significant changes in beta-cell numbers were observed in these SU5416-treated embryos (26.3 ± 1.8) in comparison to DMSO-treated and untreated controls (28.5 ± 1.6; 28.7 ± 2.3) (Fig. 2g). We also administered SU5416 at 4.5 dpf until imaging at 6 dpf to determine if duct derived secondary islets form in vessel deficient fish. The proportion of fish that developed secondary islets did not change between SU5416-treated (21.4%) and DMSO-treated embryos (25.0%). (Supplemental Fig. 2a,b). We also examined secondary islet formation in SU5416 treated embryos given a γ-secretase Notch inhibitor (DAPT), which has been previously shown to stimulate the appearance of secondary islets in zebrafish 12,13 . From 3 dpf until 4.5 dpf, we administered DAPT or co-administered DAPT and SU5416 to Tg(neuroD:GFP) fish, a transgenic line that marks early pan-endocrine cells, thereby allowing us to capture early secondary islet formation. While there were no significant differences in secondary islet formation in the lower DAPT dose (20 μM) in comparison to untreated and DMSO-treated controls, we did observe a significant increase in the number of secondary islets with 100 μM DAPT ( Supplementary Fig. 2d-j). The addition of SU5416 did not alter the effect of DAPT treatment on secondary islet formation (Supplemental Fig. 2d-j). These results suggest that Vegf signaling is dispensable for early secondary islet formation. www.nature.com/scientificreports www.nature.com/scientificreports/ Combined knockdown of vegfaa and vegfab causes a reduction of islet vessels. To identify potential mediators of signals responsible for islet vascularization, we isolated RFP-positive beta-cells from Tg(insa:tagRFP) larval fish and adult islets using fluorescence-activated cell sorting (FACS). We found that vegfaa was expressed in beta-cells isolated from fish at 2 dpf and 3 dpf. We also detected vegfaa and vegfab in adult isolated beta-cells ( Fig. 3a and Supplemental Fig. 3a-c). We next injected previously validated translation blocking morpholinos against vegfaa and vegfab into Tg(fli1:EGFP; insa:tagRFP) embryos. In non-injected control zebrafish and those injected with scrambled control morpholinos, we observed greater than 7 endothelial cells adjacent to beta-cells, whereas those animals injected with morpholinos against vegfaa had either a mild reduction of islet vessels (between 4-7 endothelial cells adjacent to beta-cells) or a severe reduction of islet vessels (less than 4 endothelial cells adjacent to beta-cells) at 72 hpf (Fig. 3b,d-f). A small percentage of vegfab morpholino www.nature.com/scientificreports www.nature.com/scientificreports/ knockdown embryos exhibit a mild reduction in islet vessels ( Fig. 3b,g). Combined knockdown of vegfaa and vegfab (vegfaa/vegfab) resulted in a more severe phenotype in comparison to the single knockdown embryos suggesting that both vegfaa and vegfab are important for islet vessel development (Fig. 3b,h). In the embryos that demonstrated a reduction or absence of islet vessels, no significant changes in beta-cell numbers were observed in these vegfaa (27.0 ± 2.2), vegfab (25.7 ± 2.5), and vegfaa/vegfab (23.3 ± 2.3) morpholino injected embryos in comparison to the scrambled injected or non-injected controls (25.9 ± 2.9; 25.5 ± 3.6) (Fig. 3c).

Combined knockdown of kdr and kdrl causes a reduction of islet vessels.
We also assessed the role of kdr and kdrl, the primary receptors of VegfA in zebrafish. We found that single morpholino knockdown of either kdr or kdrl in Tg(fli1:EGFP; insa:tagRFP) zebrafish did not affect islet vessels (Fig. 4a,f-i). However, double knockdown of kdr and kdrl resulted in a reduction of islet vessels (Fig. 4a,j). In the kdr/kdrl injected embryos that demonstrated a reduction in islet vessels, no significant changes in beta-cell numbers were observed (25.5 ± 2.9) in comparison to the scrambled and non-injected controls (27.4 ± 2.6; 27.4 ± 2.7) (Fig. 4b).
To determine if beta-cells originating from the dorsal and ventral pancreatic bud form in vessel deficient fish, we performed a previously described label-retaining cell assay to mark dorsal pancreatic bud derived beta-cells 14 .
In this assay, H2B-EBFP mRNA is injected into one cell stage embryos. The H2B-EBFP protein is diluted by cell division, but cells that are quiescent retain the label. The beta-cells originating from the dorsal pancreatic bud have been previously described to be quiescent by 24 hpf and retain the label (H2B-EBFP + ). Ventral bud derived beta-cells arise later in development after multiple rounds of progenitor proliferation and consequently the ventral bud derived beta-cells do not retain the label (H2B-EBFP − ). We found that both dorsal bud derived (H2B-EBFP + ) and ventral bud derived beta-cells (H2B-EBFP − ) are present in the kdr/kdrl injected embryos (Supplemental Fig. 4a-c). Together, these results suggest that Vegfaa/Vegfab-Vegfr2 signaling is necessary for islet vascularization but not required for beta-cell formation regardless of the origin of the beta-cells.
To determine if islet vessels influenced beta-cell maturation, we used FACS analysis to isolate RFP-positive Tg(fli1:EGFP; insa:tagRFP) beta-cells from scrambled injected controls and kdr/kdrl morpholino knockdown embryos at 72 hpf. Expression of insulina (insa) was significantly downregulated in vessel deficient embryos in comparison to scrambled injected controls (Fig. 4c), but levels of beta-cell maturation genes pdx1 and www.nature.com/scientificreports www.nature.com/scientificreports/ neuroD were not significantly altered (Fig. 4d,e). We also observed a trending decrease in insa expression in SU5416-treated embryos compared to DMSO-treated controls (Supplemental Fig. 1c).

A reduction of islet vessels is observed following beta-cell ablation.
To determine if endothelial cells undergo changes during beta-cell ablation, we crossed the following double transgenic line, Tg(-1.2ins:htBid TE-ON ; -1.2ins:H2BmCherry) 15 to Tg(fli1:EGFP) to create the triple transgenic line Tg(-1.2ins:htBid TE-ON ; -1.2ins:H2BmCherry; fli1:EGFP). In this transgenic, the proapoptotic protein tBID is expressed under the control of the tetracycline-and ecdysone-inducible system which upon addition of DOX and TBF results in ablation of beta-cells 15 . We administered DOX and TBF (DOX/TBF) at 3 dpf until 5 dpf. We observed a significant decrease in the number of beta-cells following DOX/TBF administration at all timepoints in comparison to their DMSO-treated controls (Fig. 5a-h). There is no significant decrease in alpha-cell numbers in the DOX/ TBF-treated fish in comparison to their timepoint matched DMSO-treated controls (Fig. 5i). We also observed a significant decrease in the number of endothelial cells at 5 dpf in the DOX/TBF-treated fish in comparison to DMSO-treated controls (Fig. 5a-c,j). However, there is no significant difference in the number of endothelial cells at 8 dpf or 11 dpf in DOX/TBF-treated fish in comparison to their timepoint matched DMSO-treated controls ( Fig. 5d-g,j). These results suggest a decrease in the number of islet endothelial cells after beta-cell destruction, but revascularization of the islet during beta-cell regeneration.

Discussion
In this study, we explored the angiogenic factors responsible for zebrafish islet vessel development and the effect of endothelial cells on pancreatic endocrine formation and maturation.
Beta-cells develop adjacent to blood vessels. Endothelial cells can be found within the islet as early as 40 hpf. The formation of beta-cells adjacent to endothelial cells is similarly observed in murine models where the pancreatic epithelium forms adjacent to embryonic endothelial cells 1 . Administration of the Vegf inhibitor SU5416 from 12 to 72 hpf caused a significant reduction of endothelial cells within the islet, although no significant changes in beta-cell and alpha-cell numbers were observed. Our results are consistent with previous studies which have observed the formation of insulin-expressing cells in vessel deficient cloche zebrafish mutants 10 . No significant changes in the number of secondary islets were found in SU5416-treated fish from 4.5 until 6 dpf in comparison www.nature.com/scientificreports www.nature.com/scientificreports/ to control fish, suggesting that Vegf signaling is also dispensable for early secondary islet formation. Our results differ from murine pancreatic specification which is dependent on signals from the endothelial cells 1,2 suggesting species differences in pancreas specification.
We found that treatment of zebrafish embryos with Vegf competitive inhibitor SU5416 from 72 hpf to 96 hpf after the islet vessels have been established leads to a decrease in islet vessels, suggesting that continuous Vegf signaling is needed to sustain islet vessels. This finding is consistent with observations in murine islets that Vegf signaling is the primary mediator of islet vessel formation and maintenance 4,5 .
To determine which Vegf angiogenic factors are responsible for zebrafish islet vessel development, we performed RT-PCR and found that both vegfaa and vegfab are expressed in the beta-cells as early as 2 dpf, suggesting the recruitment of endothelial cells early on during islet development. Combined knockdown of vegfaa and vegfab caused a more severe phenotype in the reduction of islet vessels in 72 hpf embryos in comparison to the single knockdown of vegfaa or vegfab. This may reflect partial redundancy between vegfaa and vegfab, as has been suggested for intestinal vessel development 16,17 . Despite the reduction of islet vessels in vegfaa/vegfab knockdown animals, there were no significant changes in beta-cell numbers between the vessel deficient and control islets. Similarly, double knockdown of the primary VegfA receptors, kdrl and kdr, resulted in a similar loss of islet vessels, but there were no significant changes in beta-cell numbers in kdr/kdrl knockdown animals in comparison to controls. Single knockdown of kdr or kdrl did not reduce islet vessels suggesting possible overlapping roles between the two receptors. These data are consistent with a previous report which found a more severe phenotype in intestinal vessel development when kdrl um19 mutants were also injected with a kdr morpholino in comparison to kdrl um19 mutants or kdr morpholino knockdown animals 17 . Together, these results suggest that VegfA/Vegfr2 signaling is necessary for islet vascularization but not required for beta-cell formation. In the vessel deficient embryos, we observed a decrease in insa mRNA expression in comparison to control animals, suggesting that islet hormone expression is influenced by Vegfa/Vegfr2 signaling. Similarly, in VegfA deficient mouse islets, a decrease in insulin expression has also been reported 18 . This reduction in insulin expression in the murine VegfA deficient islets is believed to be due to the absence of a vascular basement membrane which promotes insulin gene expression 18 .
In mice, VegfA inactivation during early development results in severe islet hypovascularization and significantly reduces beta-cell mass as a consequence of a decrease in beta-cell proliferation during postnatal stages 7 . While we did not observe any significant differences in beta-cell number between vessel deficient and control zebrafish at 72 hpf, this discrepancy may reflect the fact that beta-cell neogenesis and not proliferation is the main mechanism of beta-cell formation during zebrafish embryonic stages 14,19 . At 72 hpf, the majority of the islet is composed of quiescent dorsal bud derived beta-cells 14,20 . Thus, any putative changes in beta-cell proliferation would be undetected in our zebrafish model during early pancreas development. While there may be a decrease in beta-cell proliferation and mass during later stages of zebrafish pancreas development, an islet cell specific vegfA knockout zebrafish model may be needed to assess this as vegfaa/ab and kdr/kdrl knockdown models develop pericardial edema and die after 72 hpf, and SU5416 application in larval fish (after 72 hpf) for over 52 hours causes high rates of lethality (data not shown). www.nature.com/scientificreports www.nature.com/scientificreports/ Oxygen accessibility may also lead to the species differences during pancreas development in vascular deficient models. In the murine pancreas, oxygen levels affect endocrine cell differentiation 21 . Embryos from pregnant rats exposed to a hypoxic environment (8% O 2 ) prior to the secondary endocrine transition have significantly blunted endocrine differentiation in comparison to embryos from the control pregnant rats exposed to a normoxic environment (21% O 2 ). In vitro, increasing oxygen levels (21%, 60%, and 80%) leads to the induction of endocrine hormone expression in cultured E13.5 rat pancreatic explants in a concentration-dependent manner 21 . As mammalian embryos require a functional cardiovascular system to bring oxygen to tissues, a lack of endothelial cells creates a hypoxic environment and consequently negatively affects endocrine pancreas differentiation. In contrast, given that zebrafish develop ex-utero, the embryos are in a normoxic milieu 22 . Hence, an avascular environment does not lead to hypoxia during early zebrafish development and may explain the formation of endocrine cells despite the lack of endothelial cells in our zebrafish models. Future studies controlling the ambient oxygen content could reveal its role during early endocrine cell formation in the zebrafish.
We also observed a decrease in islet endothelial cells after beta-cell destruction and revascularization of the islet during beta-cell regeneration. This is unlikely due to a bystander affect as alpha-cell number is not significantly reduced in beta-cell ablated fish compared to control fish, but rather this may indicate changes in vasculature organisation during beta-cell ablation. In the adult zebrafish pancreas, changes in islet vasculature after beta-cell destruction and during regeneration have been reported 23 . In other tissues, revascularization of the damaged tissue occurs quickly after insult 24,25 . In cardiomyocytes, this fast revascularization is essential in the regeneration process as it promotes cell proliferation 24 . Whether islet endothelial cells are needed for beta-cell regeneration is unknown. Future studies addressing beta-cell regeneration in an avascular islet environment may reveal novel mechanisms for beta-cell renewal. In addition, identifying beta-cell transcriptome changes during beta-cell regeneration and during beta-cell development in an avascular environment could elucidate the role of endothelial cells in beta-cell formation and renewal.
In summary, the present study demonstrates that Vegfaa/Vegfab-Vegfr2 signaling is dispensable for beta-cell and alpha-cell neogenesis, although blocking Vegfaa/Vegfab-Vegfr2 signaling decreases insulina expression. Our study has identified Vegfaa and Vegfab as potential mediators for islet vessel development. These findings may have implications in zebrafish beta-cell regeneration studies, as we find changes in islet vasculature during beta-cell destruction and regeneration. Interfering with Vegfaa/Vegfab-Vegfr2 signaling during beta-cell regeneration may provide insight into mechanisms required for the endogenous expansion of beta-cell mass in zebrafish following beta-cell ablation.

Study Approval. All experiments were approved by the University of British Columbia Animal Care
Committee and performed in accordance with the Canadian Council on Animal Care guidelines.