Induction of Pancreatic Differentiation by Signals from Blood Vessels E. Lammert, O. Cleaver, D. Melton

1. Tissues containing endothelial cells induce insulin expression

To assay the role of endothelial cells during endocrine cell differentiation, we used mouse explant recombinants. We show that tissues containing endothelial cells are able to induce insulin expression when cultured with isolated endoderm. Insulin expression is found when we recombine endoderm with either aorta (Fig. 2), 14.5 dpc umbilical cord endothelium (Suppl. fig. 1, A and B) or 8.5 dpc lateral plate mesenchyme (Suppl. fig. 1, C and D). We do not observe insulin in the notochord recombinants. However, because of its ability to induce Pdx1 expression in some mouse explants, the notochord is likely to be part of a step-wise induction of insulin. In this model, the endoderm would undergo anterior-posterior patterning at 7.5 dpc when it first contacts the adjacent germ layers (14), and then receives signals from the adjacent notochord (16) which help to refine the endodermal patterning into organ domains, such as the Pdx1 expression domain. After this pre-patterning, the adjacent endothelium induces pancreatic differentiation, characterized by the expression of endocrine hormones such as insulin.

2. Removal of aortic precursors in Xenopus embryos

To further assay the role of endothelium in pancreas development, we generated frog embryos lacking a dorsal aorta (18) by removal of intermediate mesoderm, which contains precursors to the dorsal aorta (Suppl. fig. 2, A to D) (Suppl. table 1). Embryos with damaged lateral endoderm (as seen by yolk extrusion during the dissection procedure) are discarded prior to analysis. Moreover, at the stage when the dissection is performed, the intermediate mesoderm lies ventrolateral to the pancreatic endoderm, not in immediate contact, precluding the possibility of direct damage to the pancreatic endoderm (Suppl. fig. 2, A to D).

The absence of insulin expression in dissected embryos is not a result of the absence of the excised intermediate mesoderm per se. Removal of the aortic precursors results in the removal of intermediate mesoderm derivatives such as the pronephros (Suppl. fig. 2, E and F). However, in a number of cases, the aorta regenerates due to endothelial cell migration from unmanipulated parts of the embryo and the presence of the aorta is always accompanied by the presence of pancreatic endocrine cells, despite the absence of the intermediate mesoderm (Suppl. fig. 2, G and H).

3. Pdx-VEGF transgenic mice

To investigate endothelial signals during endocrine development, a transgenic vector was constructed, which drives VEGF164 and EGFP in foregut tissues (Suppl. fig. 3, A to C). It was possible to establish a Pdx-VEGF transgenic line, though over half of the transgenic pups died within a few days after birth because expanded foregut tissue led to foregut closure. The transgenic line is characterized by islet hyperplasia and exocrine hypoplasia (Suppl. table 2). Adjacent sections demonstrate that transgenic islets display normal islet morphology. Like control islets, transgenic islets are surrounded by pancreatic polypeptide (PP), glucagon and somatostatin expressing endocrine cells. Blood glucose levels are also approximately normal in Pdx-VEGF transgenic mice. In a series of transient transgenic experiments, 2 out of 5 Pdx-VEGF transgenic animals revealed an ectopic bud-like structure (compare Suppl. fig. 3D and 3G). The bud had high levels of VEGF expression, marked by green fluorescence, and harbored scattered insulin expressing cells (Suppl. fig. 3, D to I). The bud was morphologically distinct from the dorsal and ventral pancreas (Suppl. fig. 3I), which are extensively lobulated organs at this stage.

Materials and Methods

Isolation and culture of mouse tissues

A 1 mg/ml Dispase solution (Gibco BRL) is used to facilitate the dissection of embryos obtained from outbred ICR mice (Taconic). Endoderm and aortic tissue are carefully removed using tungsten knives and are cleansed of any adhering cells by vigorous pipetting in a dispase solution. Tissue explants from the region of somite 1 to 5 are placed either alone or in combination with other tissues on a layer of growth factor-reduced matrigel matrix (Becton Dickinson), overlaid with matrigel and incubated overnight at 37°C with Media 199 (GIBCO BRL). After 1 day incubation, the medium is changed to CS-C medium. Both media are supplemented with 10% fetal bovine serum, 1× penicillin-streptomycin, 1× glutamine (GIBCO BRL) and 1× endothelial cell growth factor (Sigma). Tissue explants are cultured for 5 to 6 days.

Immunohistochemistry and tissue analysis

Mouse tissue is sectioned and stained as previously described (28). The antibodies and lectins used are: anti-insulin (DAKO), anti-C-peptide (Linco Research), anti-VEGF (Sigma), anti-VEGFR2 (29) (kindly provided by Rolf Brekken), anti-VEGFR2 (PharMingen), biotin-conjugated Wisteria floribunda for duct glycosylation (Sigma), anti-CD31/PECAM1 as endothelial marker (PharMingen), anti-VEGFR1 (Santa Cruz Biotech.), anti-Amylase (Sigma), anti-PP (DAKO), biotin- or Cy3-conjugated Donkey secondary antibodies (Jackson ImmunoResearch lab). For Pdx1 staining, heterozygous Pdx1-LacZ +/- mice (9) (kindly provided by Chris Wright) are stained with HistoMark X-Gal Substrate set (KPL).

Xenopus dissection

Removal of intermediate mesoderm is done using tungsten knives and a Gastromaster dissection tool (Xenotek Engineering). Epidermis is peeled away along the trunk, in the region of the aorta precursors, to visualize the bottom of the somites and allow excision of the intermediate mesoderm. Following dissection, the embryos are incubated in 1× MMR solution until stage 39-40 and fixed with MEMPHA (18).

Analysis of the Xenopus pancreas at stage 40 was carried out by counting 30 transverse sections (10 µm) from the posterior otic vesicle to the anterior edge of the pancreas (as assayed by expression of pancreatic markers and morphological boundaries). In unmanipulated embryos, pancreatic markers then span approximately 15 sections (10 µm) from the anterior to the posterior boundaries of the pancreas. This region was examined in both manipulated and unmanipulated embryos. In addition, regions more anterior and posterior to the pancreatic region were scanned for any errant pancreatic cells. When the aorta is completely absent throughout the pancreatic region, we do not observe pancreatic cells in any region of the embryo (as assayed by insulin, XPax6 and XNeuroD).

In situ hybridization

Digoxygenin labeled RNA probes were generated using clones of XPax6, XNeuroD, Xhex (kindly provided by Paul Krieg) and a 1 kb F-spondin cDNA, obtained by RT-PCR with the following primers: 5´-ATGGGACTGATTTTCCAGCC-3´ and 5´-CACTGTCAGTGCCTGCATCC-3´. Whole-mount in situ hybridization is performed as described (18).

Construction of transgenic vectors

The Pdx-VEGF construct consists of the mouse Pdx1 promoter (kindly provided by Chris Wright) driving mouse VEGF164 (kindly provided by Andras Nagy) followed by IRES-EGFP-SV40PolyA (Clontech). Constructs were tested in HIT-T15 insulinoma cells (ATCC) as described (26). Linearized vectors were microinjected into fertilized mouse eggs (C57BL6/CBA).

Supplemental Figure 1. Insulin expression in recombinants of endoderm with tissues containing endothelial cells. Mouse endoderm was isolated from 8.25-8.5 dpc mouse embryos (8-10 somite stage) and recombined with umbilical artery (A, B) or lateral mesoderm (C, D). Adjacent sections were stained for insulin (A, C) and PECAM1 (B, D).

Medium version | Full size version

Supplemental Figure 2. Removal of aortic precursors in the Xenopus embryo. (A) VEGFR2 expression in endothelial cells within the intermediate mesoderm, ventral to the somites, at the position of the future cardinal veins (arrows). These VEGFR2 expressing cells include aortic precursors and later migrate medially to form the dorsal aorta (18). Dorsal prepancreatic endoderm (p) is located in the dorsal endoderm, ventral to the notochord and flanked by the ventral portion of the somites (s). (B) Schematic of tissue layers in (A) shows the relative positions of pancreatic endoderm and aortic precursors (*). (C) H&E stained section of a dissected embryo, immediately after removal of aorta precursors. (D) Schematic of tissue layers in (C) showing the relative position of the pancreatic endoderm and the excised intermediate mesoderm. (E through H) Appearance of pancreatic markers in dissected Xenopus embryos in the absence of intermediate mesoderm. (E, F) Lateral view of whole-mount in situ hybridization for the pronephros marker XPax2, in stage 40 embryos. (E) Unmanipulated embryo displays a normal pronephros (white arrowhead). (F) Dissected embryo displays absence of pronephros after removal of intermediate mesoderm (white arrowhead). (G, H) Transverse sections through stage 40 embryos after removal of the intermediate mesoderm (compare with Fig. 3). (G) Presence of the regenerated aorta (arrowhead) and of the pancreatic marker XPax6 (blue color, arrow). (H) Presence of the regenerated aorta (arrowhead) and of insulin expression (black color, arrow). Dark brown color is due to melanocytes located around somites.

Medium version | Full size version

Supplemental Figure 3. (A to C) Transgenic VEGF expression. (A) Expression of the transgene Pdx-VEGF in insulinoma cells shows that VEGF and EGFP are coexpressed. A comparison of a Pdx-VEGF transgenic pancreas (B) with a nontransgenic litter (C) shows that VEGF is strongly expressed in the dorsal pancreas of transgenic embryos at 15.5 dpc. (D to I) Ectopic insulin expressing cells in the anterior duodenum of Pdx-VEGF transgenic embryos. Gut structures of Pdx-VEGF transgenic embryos (G to I) are compared with nontransgenic littermates (D to F). The normal pancreas is large and highly branched at this stage and extends dorsally behind the stomach out of the plane of focus. The duodenum is located just posterior to the stomach, and the anterior region of the duodenum is indicated (white arrowhead). An ectopic bud-like structure is evident in the anterior duodenum of the Pdx-VEGF transgenic embryo and it expresses the transgene (G, H). Scattered insulin expressing cells are detected on a transverse section through the bud (I, black arrowhead), but not on a transverse section through a nontransgenic duodenum (F).

Medium version | Full size version

Supplemental Table 1. Failure of pancreatic differentiation in the absence of the dorsal aorta in Xenopus embryos. Cells expressing pancreatic or hypochord markers are counted in the pancreatic region of stage 40 Xenopus embryos (n = 20). The P value for all columns P < 0.005 (Student's t-test).
Stage 40 Xenopus embryonic pancreas	Number of insulin- expressing cells	Number of F-spondin- expressing cells
Control embryos (+aorta)	64.5 ± 12.85	15.2 ± 0.41
Dissected embryos (-aorta)	5.5 ± 5.27	15.2 ± 0.40

Supplemental Table 2. Endocrine hyperplasia in Pdx-VEGF transgenic mouse pancreas. Twenty sections per transgenic mouse were analyzed (n = 4). P value for all columns is P < 0.05 (Student's t-test). After overlay of a grid onto the scanned images of pancreatic sections, the islet and acini area was determined by counting the number of stained and unstained squares.
2-month-old mice	% Islet area in 5 m pancreas sections	% Acini area in 5 m pancreas sections	Number of islets in 5 m pancreas sections
Nontransgenic	1.7 ± 0.55	79 ± 3.34	1.53 ± 0.04
Pdx-VEGF	5.2 ± 0.16	11 ± 1.50	5.56 ± 0.18