Expression and Functional Studies of the GDNF Family Receptor Alpha 3 in the Pancreas

L NIVLET and others Artn/Gfra3 signaling in the 56:2 77–90 Research pancreas

Expression and functional studies of the GDNF family receptor alpha 3 in the pancreas

Laure Nivlet, Joel Herrmann*, Delia Esteban Martin*, Aline Meunier, Christophe Orvain and Ge´rard Gradwohl Correspondence Institut de Ge´ ne´ tique et de Biologie Mole´ culaire et Cellulaire (IGBMC), Institut National de la Sante´ et de la should be addressed Recherche Me´ dicale (INSERM) U964, Centre National de Recherche Scientiﬁque (CNRS) UMR 7104, to G Gradwohl Universite´ de Strasbourg (UdS), 1 Rue Laurent Fries, 67404 Illkirch, France Email *(J Herrmann and D E Martin contributed equally to this work) [email protected]

Abstract

The generation of therapeutic b-cells from human pluripotent stem cells relies on the Key Words identification of growth factors that faithfully mimic pancreatic b-cell development in vitro. " pancreas In this context, the aim of the study was to determine the expression and function of the glial " GDNF family receptor alpha 3 cell line derived neurotrophic factor receptor alpha 3 (GFRa3) and its ligand artemin (Artn) in " artemin islet cell development and function. GFRa3 and Artn expression were characterized by in situ " islets of Langerhans hybridization, immunochemistry, and qRT-PCR. We used GFRa3-deficient mice to study " endocrine progenitor GFRa3 function and generated transgenic mice overexpressing Artn in the embryonic " b-cells pancreas to study Artn function. We found that GFRa3 is expressed at the surface of a subset " parasympathetic neurons of Ngn3-positive endocrine progenitors as well as of embryonic a- and b-cells, while Artn is " sympathetic neurons found in the pancreatic mesenchyme. Adult b-cells lack GFRa3 but a-cells express the " neurogenin3 Journal of Molecular Endocrinology receptor. GFRa3 was also found in parasympathetic and sympathetic intra-islet neurons as well as in glial cells in the embryonic and adult pancreas. The loss of GFRa3 or overexpression of Artn has no impact on Ngn3 and islet cell formation and maintenance in the embryo. Islet organization and innervation as well as glucose homeostasis is normal in GFRa3-deficient Journal of Molecular mice suggesting functional redundancy. Endocrinology (2016) 56, 77–90

Introduction

A network of transcription factors controls the differen- Mastracci & Sussel (2012) and Lodh et al. (2014))we tiation of islet cells during pancreas organogenesis (for a currently lack information on the nature of signals that review, see Cano et al. (2013)). Neurog3 is central in this would eventually control later steps of islet differentiation process as this gene is essential for determining the fate of including endocrine cell fate decision, survival of endo- endocrine cells and the initiation of islet differentiation crine progenitors, islet subtype specification, and matu- programs resulting in different pancreatic endocrine cell ration. This knowledge could be instrumental to improve types including insulin-secreting b-cells (Gradwohl et al. the generation of glucose responsive b-cells in vitro. To that 2000, Gu et al. 2002, Desgraz & Herrera 2009). While a effect, we searched for endocrine progenitor cell surface series of growth factors controlling early steps of pancreas receptors. Gene expression profiling in sorted Neurog3- C organogenesis have been identified (see for a review positive cells from Ngn3EYFP/ E15.5 embryonic pancreas

http://jme.endocrinology-journals.org Ñ 2016 Society for Endocrinology Published by Bioscientiﬁca Ltd. DOI: 10.1530/JME-15-0213 Printed in Great Britain Downloaded from Bioscientifica.com at 09/25/2021 01:55:53PM via free access Research L NIVLET and others Artn/Gfra3 signaling in the 56:2 78 pancreas

(Soyer et al. 2010) revealed an enrichment of the Glial cell collaboration with the Mouse Clinical Institute (ICS; line derived neurotrophic factor receptor alpha 3 (GFRa3) Illkirch). The Artn-2A-mCherry sequence was synthesized mRNA suggesting that GFRa3 and its ligand artemin by GenScript and cloned downstream of a 5.15-kb DNA (Artn) would control the biology of endocrine progenitor fragment containing the mouse Pdx1 promoter and a heat cells. GFRa3 belongs to the glial cell line derived shock protein minimal promoter (hsp68) (Johansson et al. neurotrophic factor (GDNF) family of receptors which 2007). All mouse lines were kept on CD1 or C56BL/6 contains four members (GFRa1–4). GDNF family of backgrounds and experiments supervised by G Gradwohl ligands (GFLs) GDNF, Neurturin (Nrtn), Artn, and Perse- (agreement N8 C67-59 by the Direction des Services phin (Pspn) bind to co-receptors GFRa1–4 and activate Ve´te´rinaires, Strasbourg, France). PAM mice were geno- RET receptor tyrosine kinase (Airaksinen & Saarma 2002). typed by using 50 GCCACTGCCTGCGGCTGTCT 30 and 50 GFLs are mainly known for their role in the develop- CTTGGCGGTCTGGGTGCCCT 30 primers. ment and function of the nervous system (Airaksinen & Saarma 2002) but GDNF is also important for the growth of Real-time PCR the ureteric bud during kidney development (Costantini & Total RNA was isolated from pancreatic buds at E15.5 Shakya 2006) or spermatogonial stem cell renewal using Tri Reagent (Invitrogen). In total, 1 mg of RNA was (Hofmann 2008). In the embryonic pancreas, Gdnf used for DNaseI (Roche) treatment and cDNA synthesis expression has been described in the pancreatic epithelium with the Transcriptor Reverse Transcriptase (Roche). RNA acting as a neurotrophic factor promoting the differen- from Ngn3 sorted cells was isolated by using the RNeasy tiation and migration of neural progenitors. Pancreatic Micro kit (Qiagen). cDNA was synthesized using the inactivation of Gdnf leads to reduced parasympathetic Transcriptor Reverse Transcriptase (Roche) and 200 ng of innervation in the pancreas (Munoz-Bravo et al. 2013). RNA were used for DNaseI treatment. Quantitative PCR Other studies demonstrated that GFRa2 signaling is were performed using Taqman probes. required for parasympathetic islet innervation (Rossi et al. 2005). More surprisingly, exogeneous GDNF induced the In situ hybridization proliferation of pancreatic progenitors in pancreas explant cultures (Munoz-Bravo et al. 2013), and the overexpression E15.5 embryos were ﬁxed in 4% paraformaldehyde in 1! C C of Gdnf in transgenic mice increased pancreatic b-cell mass PBS without Ca2 and Mg2 overnight at 4 8C, transferred (Rossi et al. 2005). Altogether, these data suggest a role of in 20% sucrose in PBS overnight at 4 8C, embedded in GFLs and receptors in pancreatic innervation and endo- Cryomatrix (Thermo Scientiﬁc, Waltham, MA, USA), and

Journal of Molecular Endocrinology crine cell differentiation. However, pancreatic expression frozen on dry ice. Sections of 10 mm were cut with a Leica and function of GFRa3 has not been explored yet. To assess cryostat CM3050S. Brieﬂy, slides were incubated over- the role of GFRa3 and of its ligand Artn in the pancreas night at 65 8C with hybridization buffer (NaCl 11.4 g/l;

we determined their expression. We show that GFRa3is Tris HCl pH 7.5 1.404 g/l; Tris base 0.134 g/l; Na2PO4 2H2O

expressed in subsets of endocrine progenitors and devel- 0.78 g/l; Na2HPO4 0.71 g/l; 0.05 M EDTA; 50% formamide; oping, but not adult, islet cells, except alpha cells. GFRa3is 10% dextran sulfate; 1 mg/ml tRNA; 0.02% BSA; 0.02% also expressed in the embryonic and adult pancreatic Ficoll; 0.02% PVP) containing the Digoxigenin (DIG) neurons and glial cells. Analysis of the phenotype of GFRa3 labeled Gfra3 cRNA probe. Sections were washed in 1! KO mice as well as of transgenic mice overexpressing Artn SSC, 50% formamide, 0.1% tween at 65 8C, and then revealed that Artn/GFRa3 signaling pathway is not equilibrated in MABT solution (Maleic Acid 100 mM pH essential for islet formation, innervation, and function. 7.5; NaCl 150 mM; Tween 0.1%) and incubated in blocking solution (MABT, Tween; Boehringer Blocking Reagent (BM 1096176) 2%; heat inactivated goat serum Materials and methods 20%) at room temperature. Slides were then incubated with anti-DIG antibody coupled with alkaline phospha- Mouse strains and genotyping tase diluted in blocking solution overnight at room C Ngn3eYFP/ mice were described previously (Mellitzer et al. temperature. Samples were then washed in MABT C 2004). GFRa3tLacZ/ mice were generously provided by solution, equilibrated in staining buffer (NaCl 100 mM;

Dr Jeffrey Milbrandt and have been described previously MgCl2 50 mM; Tris pH9.5 100 mM; Tween 0.1%; Levami- (Honma et al. 2002). The promPdx1-Artn-2A-mCherry sole 0.5 mg/ml), and revealed in the same solution (PAM) transgenic mouse line was generated in containing 3.5 ml/ml NBT and 3.5 ml/ml BCIP.

If followed by immunohistochemistry, slides were For measurement of endocrine innervation at P21, washed in PBS and incubated with blocking solution (PBS; one section every 60 mm was stained for hormones and Triton 0.1%; normal goat serum (NGS) 20%) containing tyrosine hydroxylase (TH) or vasoactive intestinal peptide anti-Ngn3 antibody (Guinea Pig, IGBMC, 1/500). (VIP). Randomly chosen 50 islets from each pancreas were Endogenous peroxydases were inactivated by incubating analyzed. Islets area were deﬁned manually and thres-

slides in 0.5% H2O2 in methanol. After washes in PBS; holded using Image J software. Sympathetic innervation C Triton 0.1%, sections were incubated with secondary (TH) is expressed as TH area vs endocrine area and C antibody coupled to HRP diluted in PBS; Triton 0.1% parasympathetic innervation is expressed as VIP puncta and staining revealed using the DAB Peroxydase substrate vs endocrine area. kit (Vector Laboratories, Burlingame, CA, USA). Metabolic studies Immunohistofluorescence Mice of more than 10 weeks were made to fast for 16 h. Embryos or dissected adult pancreas were harvested, fixed, For Oral Glucose Tolerance Test (OGTT), mice received embedded, and sectioned as described above. Primary glucose by intragastric gavage (1 g/kg body weight). For antibodies were diluted in PBS; Triton 0.1%; NDS 5–20%: Intraperitoneal Glucose Tolerance Test (IPGTT), mice GFRa3 (goat, R&D Systems, Abingdon, Oxfordshire, UK, received glucose by i.p. injection (2 g/kg body weight). 1/500), Pdx1 (rabbit, Chris Wright, Vanderbilt University, Circulating blood glucose was measured in tail blood at USA, 1/2000), Ngn3 (Guinea Pig, IGBMC, 1/500), insulin 0, 15, 30, 45, 60, 90, and 120 min using Glucofix Sensor (Guinea Pig, Linco, St. Charles, MO, USA, 1/1000 or mouse, (A.Menarini Diagnostics, Firenze, Italy). Sigma, 1/1000), glucagon (Guinea Pig, Linco, 1/2000 or mouse, Sigma, 1/2000), Artn (goat, R&D Systems, 1/100), Preparation of single cell suspension for FACS sorting Somatostatin (Rabbit, Dako, Glostrup, Denmark, 1/200), Pancreas from E15.5 pancreas were dissected, mechani- PP (Guinea Pig, Linco 1/1000), tyrosine hydroxylase (TH; cally and enzymatically dissociated by a trypsin treatment rabbit, Merck Millipore, Darmstadt, Germany, 1/500), VIP (0.05%) 5–10 min at 37 8C. Trypsin action was stopped by (rabbit, Phoenix Pharmaceuticals, Burlingame, CA, USA, adding DMEM/F12; FCS10%; 3.15 g/l glucose; genta- 1/500), S100beta (rabbit, Merck Millipore, 1/500), TUJ1 mycine. Cells were filtered the first time on an 80-mm (mouse, Covance, Rueil-Malmaison, France, 1/500) and Nylon Mesh (SEFAR, 3A03-0080-102-11), spinned 5 min Nkx6.1 (DSHB, F55A10 1/200). Appropriate secondary at 90 g, resuspended in DMEM/F12; FCS10%; 3.15 g/l

Journal of Molecular Endocrinology antibodies conjugated to DyLight488, DyLight 549, or DyLight glucose; gentamycine and filtered a second time on a 649 (Jackson ImmunoResearch, West Grove, PA, USA, 1/500). 50-mm Nylon Mesh (Wipak Medical R40, 050-47S) before FACS sorting. Cells were sorted by using an FACS Vantage Quantitative analysis SE (Becton Dickinson, San Jose, CA, USA), with a Diva 5.0.3 software. Once sorted, cells were spinned for 5 min at To quantify the number of Ngn3 cells and insulin/ 90 g and RNA extraction performed. glucagon areas, 7-mm cryosections were cut and each fifth section was immunostained for Ngn3, insulin, or Statistical analysis glucagon. Ngn3 quantification is expressed as the number of Ngn3 cells per pancreas. Insulin and glucagon areas Values are presented as mean of S.D.orS.E.M. The P values were normalised to the pancreatic area (DAPI staining). were determined using the two-tailed Student’s t test with Analyses were performed using ImageJ Software. unequal variance. The P!0.05 was accepted as statistically Innervation was measured as described previously significant. (Munoz-Bravo et al. 2013). Briefly at P0, one section every 60 mm was stained for hormones and TUJ1. Pictures were taken using a slide scanner Nanozoomer 2.OHT Results (Hamamatsu, Hamamatsu City, Japan) and analyzed with GFRa3 is expressed in subsets of Ngn3-positive endocrine the Image J software. Tubeness plugin was used to detect progenitors neurites, then skeletonized to obtain neurites length. Results are expressed as total innervation length Gene expression profiling revealed that the mRNA of C normalized to endocrine area. GFRa3 was strongly and specifically enriched in eYFP

K endocrine progenitors vs eYFP cells (fold change (FC)Z expressed in Ngn3-positive pancreatic cells at E12.5 33.25; false discovery rate (FDR)Z0.02) (Supplementary (Supplementary Figure 1, see section on supplementary Table 1, see section on supplementary data given at the data given at the end of this article) and E15.5 (Fig. 1f, C end of this article) purified from Ngn3eYFP/ embryonic yellow arrows). GFRa3 immunosignal is concentrated at (E15.5) pancreas (Soyer et al. 2010). RT-QPCR experiments the cell periphery suggesting cytoplasmic membrane confirmed this strong enrichment of GFRa3 in the islet localization as expected. However, only a subset of Ngn3 lineage (Fig. 1a). Artemin, a secreted peptide of the GDNF cells expresses GFRa3(Fig. 1f green arrows point to K C ligand family, binds to the co-receptor GFRa3 and thereby GFRa3 /Ngn3 cells). activates the receptor tyrosine kinase RET. We thus examined whether Artn, as well as other members of the GFRa3 persists in developing islet cells but in the adult GDNF receptor (GFRs) and ligand (GFLs) families are pancreas only a-cells express the receptor expressed in the embryonic pancreas. We found that Artn and GFRa2 are specifically enriched in the non-endocrine Interestingly, not all GFRa3-positive cells express Ngn3 K (eYFP ) cell population (Fig. 1b and d) while Gdnf is (Fig. 1f red arrows). Double immunostainings for insulin enriched in the endocrine compartment (Fig. 1c). or glucagon suggest that GFRa3-positive/Ngn3-negative We next performed in situ hybridization and immuno- cells represent developing a- and b-cells (Fig. 2a and b fluorescence experiments on pancreas cryosections to white arrows). Accordingly, we found that the transcrip- more precisely determine GFRa3 expression during tion factor Nkx6.1, which becomes restricted to the b-cell pancreas development. We found that not only GFRa3 lineage (Henseleit et al. 2005), is expressed in double mRNA (Fig. 1e), but importantly also GFRa3 protein are positive GFRa3/insulin cells (Supplementary Figure 2A, see

AB1.2 ***30 *** 1.0 25 0.8 20 3 (fold)

0.6 (fold) 15 α

Artn α 0.4 10 E GFRα3/NGN3 F GFR 3/NGN3 GFR 0.2 5

Journal of Molecular Endocrinology + – + –

eYFP eYFP eYFP eYFP

CD1.2 ***18 ** 16 1.0 14 0.8 12

(fold) 10 160 μm (fold)

0.6 2

α 8 Gdnf 0.4 Gfr 6 4 0.2 2

+ – + –

eYFP eYFP eYFP eYFP

Figure 1 GFRa3 receptor is expressed in a subset of endocrine progenitors. (a, b, c (dark arrows). (f) Immunostaining for GFRa3 (red) and NGN3 (green) on C K C and d) RT-qPCR on puriﬁed EYFP and EYFP cells from E15.5 Ngn3eYFP/ cryosections of WT E15.5 pancreas showing that some Ngn3-positive cells pancreas showing that (a) GFRa3 expression is enriched in EYFPC cells and express GFRa3 at the cell membrane (yellow arrows). Data are summarized K (b) Artn is signiﬁcantly enriched in the EYFP cell population. (c) GFRa2 as meanGS.E.M.; nR4 for each conditions; **P%0.01, ***P%0.001. K C C C K expression is enriched in the EYFP cell population while (d) Gdnf is In f yellow, green and red arrows point to GFRa3 /NGN3 ; NGN3 /GFRa3 strongly expressed in EYFPC cells. (e) In situ hybridization for GFRa3 (blue) and NGN3K/GFRa3C cells respectively. In e, dark, blue and brow arrows C C C K K C and immunostaining for NGN3 (brown) on cryosections of WT E15.5 point to GFRa3 /NGN3 , GFRa3 /NGN3 , GFRa3 /NGN3 respectively. pancreas showing expression of GFRa3 in some islet progenitor cells

Figure 2 C GFRa3 receptor is expressed in insulin- and glucagon-positive cells in the GFRa3 (red) expression decreases in (c) insulin cells (green) and (d) embryo and at birth but in adult islets, only a-cells are GFRa3-positive. is maintained in glucagonC cells (green). (e and f) In the adult pancreas, C C (a, b, c, d, e and f) Immunoﬂuorescence on cryosections of E15.5 embryonic GFRa3(red) is (e) not expressed by insulin cells (green) but (f) glucagon C (a and b), newborn (c and d) or adult pancreas (e and f) for GFRa3 (red) cells (green) express the receptor. White arrows point to hormone / insulin (green) and glucagon (green) expressing cells. (c and d) At P0, GFRa3C cells and red arrows point to GFRa3C/hormoneK cells.

section on supplementary data given at the end of this contrast, in the adult pancreas, b-cells are devoid of GFRa3 article). Similarly d and PP-cells express GFRa3 in the (Fig. 2e), as well as d and PP-cells (Fig. 3c and f), while only embryonic pancreas (Fig. 3a and d). After E15.5, GFRa3 a-cells are GFRa3 positive (Fig. 2f and Supplementary labeling is maintained in embryonic islet cells (not Figure 2B). Of note, we observed that adult islets are always shown), and many islet cells remain positive for GFRa3 surrounded by GFRa3-positive cells (Figs 2e, f and 3c, f) in at P0 (white arrows in Figs 2c, d and 3b, e). In sharp a pattern, different from a-cells, but reminiscent of glial

ABC SST / 3 α GFR

40 μm 40 μm 50 μm 3 α

D EF PP / 3 Journal of Molecular Endocrinology α GFR

40 μm 40 μm 50 μm 3 GFR α GFR

E15.5 P0 Ad

Figure 3 GFRa3 is expressed in developing and newborn but not adult somatostatin- by SomatostatinC cells and PPC cells but not in the adult pancreas. C C and PP-cells. (a, b, c, d, e and f) Immunolocalisations on cryosections of WT White arrows point to hormone /Gfra3 cells, red arrows point to the pancreas for GFRa3 (red) and Somatostatin (SST, green, a, b and c), or same double positive cells on picture where the green layer has been Pancreatic Polypeptide (PP, green, d, e and f) at E15.5, P0 and in the adult removed for a better appreciation of GFRa3 staining. mice. In the embryo at E15.5 and in newborn mice (P0) GFRa3 is expressed

cells (see below). Thus, GFRa3 is found at the surface of a 6 ** Epithelium subset of islet progenitors and persists in embryonic islet Mesenchyme cells suggesting these endocrine cells can receive and 5 integrate Artn signals during pancreas ontogenesis. In 4 contrast among adult islet cells, only a-cells express GFRa3. ** 3

Artn transcripts are detected in the pancreatic 2 mesenchyme Relative mRNA level *** ** *** * 1 To further characterize Artn/GFRa3 signaling in the embryonic pancreas, we next decided to identify the α α α cellular origin of Artn ligand. Unfortunately, in situ Ret Gfr 1 Gfr 2 Gfr 3 Gdnf Nrtn Artn Pspn Pdx1 hybridization as well as immunofluorescence experiments fail to detect any expression in the pancreas of mouse Figure 4 embryos (data not shown). We then thought to take Expression of GFRs and GFLs in embryonic pancreatic epithelial or mesenchymal cells. RT-qPCR on E12.5 pancreatic epithelium and LacZ/C advantage of the Artn mouse (Honma et al. 2002)to mesenchyme revealed significant enrichments of Pdx1 (control), Ret, reveal b-galactosidase activity in Artn-expressing cells in GFRa3, and Gdnf in the epithelium and of GFRa2 and Artn in mesenchyme. whole mount embryos. Again, we could not observe any GFRa1 and Pspn are expressed at the same level both in the epithelium and the mesenchyme. Data are summarized as meanGS.E.M.; nR10 for each staining in the embryonic pancreas although labeled cells tissue; *P%0.05, **P%0.01, ***P%0.001. were readily detected in sclerotomes as expected (data not shown). We concluded that Artn expression must be too persists suggesting expression outside the endocrine line- low to be detected with the above tools. However, RT-PCR age. We thus performed immunofluorescence experi- Artn clearly indicated an enrichment of transcripts in non- ments to identify GFRa3-positive cells in Ngn3-deficient endocrine cells (Fig. 1b) suggesting pancreatic expression pancreas. As expected, GFRa3-immunostaining could not of Artn ligand. To determine whether Artn is expressed C be detected in the pancreatic epithelium (Pdx1 cells) of by other non-endocrine, pancreatic epithelial, or mesen- Ngn3-deficient pancreas in contrast to WT (Supplementary chymal cells, we performed RT-PCR at E12.5 (a stage when Figure 3b and c), confirming GFRa3expressionin many GFRa3-cells are detected; Supplementary Figure 1) developing endocrine cells. However, GFRa3-positive in pancreatic epithelia and their surrounding mesench- cells were found embedded in the exocrine tissue, some- Journal of Molecular Endocrinology yme that were enzymatically dissociated. Significant times surrounding acini (Supplementary Figure 3e) and enrichment of Pdx1 expression in epithelia confirmed co-stained for the neuronal marker TUJ1 suggesting that the purity of our samples (Fig. 4). Similarly GFRa3 and Ret GFRa3 is expressed in developing intra-pancreatic are found in the epithelium fraction in agreement with the neurons. GFRa3/TUJ1 double positive neuronal fibers Affymetrix and/or expression data. Importantly, Artn (Fig. 5a, white arrows) are also found close to clusters of transcripts were five times higher in the mesenchymal GFRa3-positive/TUJ1-negative cells (likely developing islet tissue. Of note is that GFRa2 is mesenchymal as well, while a GFRa1 expression was observed in both epithelia and cells) suggesting that GFR 3 also marks neurons innervat- mesenchyme. Gdnf expression was enriched in epithelia, ing endocrine cells. which is coherent with published data (Munoz-Bravo et al. Different types of neuronal cells innervate the adult 2013). Together, our results suggest that mesenchymal pancreas including neurons of the parasympathetic and Artn signals to GFRa3-positive developing endocrine cells sympathetic system acting antagonistically on pancreatic located in the epithelium at E12.5. hormone secretion (Ahren 2000). Stimulation of parasympathetic neurons will promote insulin secretion, while sympathetic neurons activate glucagon secretion and GFRa3 is expressed in pancreatic neuronal and glial cells inhibit insulin secretion (Ahren 2000). We observed RT-qPCR analysis (Supplementary Figure 3a, see section on GFRa3-positive sympathetic neurons (labeled by the supplementary data given at the end of this article) enzyme tyrosine hydroxylase (TH)) both in the endocrine revealed that while GFRa3 transcripts were reduced by (Fig. 5b, white arrows) and exocrine tissues (Fig. 5c, white K K 70% in Ngn3 / E15.5 pancreata (due to the absence of arrows). GFRa3 labeling was rather surrounding the islets the endocrine cells), a significant level of GFRa3 mRNA while intra-islet TH-positive clusters were GFRa3-negative

Figure 5 Localization of GFRa3 in developing and adult neuronal and glial cells. (b) and acinar (c) tissues, as well as by parasympathetic neuronal cells C (a, b, c, d, e and f) Immunoﬂuorescence on pancreatic cryosections at E15.5 (VIP , green) in both endocrine (d) and acinar (e) tissues, and by glial C (a), and at adult stage (b, c, d, e and f) for GFRa3 (red), neuron-speciﬁc class (f) cells (S100b , green). White arrows point to neuronal cells expressing III beta tubulin Tuj1 (Tuj1, green), tyrosine hydroxylase (TH, green), GFRa3C/neuronsC or GlialC cells. Islets (i) are delimited by dashed lines. vasoactive intestinal peptide (VIP, green), and calcium binding protein *cluster of developing islet cells. a, acinar tissue; v, blood vessel; i, islet. C S100b (S100b, green). GFRa3 is expressed by neuronal cells (Tuj1 , green) in Nuclei are labeled with DAPI (blue). White arrows point to double the embryonic pancreas. (b, c, d, e and f) In the adult pancreas, GFRa3 (red) labeled cells. C is expressed by sympathetic neuronal cells (TH , green) in both endocrine

(Fig. 5b). Regarding parasympathetic neurons (labeled by confirming that signaling through this receptor is the vasoactive intestinal peptide (VIP)), we observed impaired. We could not detect any difference in Ngn3 punctuated VIP signal within the islets (Fig. 5d, white expression pattern when comparing GFRa3tLacZtLac/Z and arrows) as well as in the acinar tissue along blood vessels control WT fetal pancreas (Supplementary Figure 4a (Fig. 5e, white arrows) which frequently overlapped with and b). Furthermore, quantification of Ngn3 cell number GFRa3 immunostaining. Finally, GFRa3 also marks peri- did not reveal any significant variation between these two insular and intra-islet Schwann cells (labeled by the genotypes (Fig. 6a). Thus, GFRa3 is not essential for the calcium binding protein S100b)(Fig. 5f, white arrows). generation and/or maintenance of islet progenitor cells. Altogether we found that GFRa3 is expressed in develop- Likewise, immunofluorescence analyses (Supplementary ing pancreatic neurons and persists in the adult where Figure 3c and d) and quantification of glucagon and both the sympathetic and parasympathetic are labeled as insulin hormones areas (Fig. 6b) did not reveal any well as glial cells. significant defect in a-orb-cell development suggesting that GFRa3 is not essential for a/b subtype specification or maintenance. To reveal any effect of the loss of GFRa3- GFRa3-deficient mice do not present any defect in islet cell signaling on gene expression, we performed Agilent development and glucose homeostasis microarrays on WT and GFRa3-deficient embryonic To decipher the role of Artn/GFRa3 pathway in the pancreas. We found that the expressions of only three embryonic and adult pancreas, we studied GFRaa3- genes were mildly affected by the absence of GFRa3 (see deficient mice (Honma et al. 2002). GFRa3tLacZ/tLacZ mice discussion): Nphp3 (FCZK1.43, t testZ0.047); Pkd2l2 are viable and fertile and blood glucose analysis did not (FCZK1.89, t testZ0.000378), and Pou3f2 (FCZ1.56, reveal overt diabetes in adult mice. As expected, GFRa3 TtestZ0.030803). Because the loss of GFRa3 could impair immunostaining is lost in knock out mice (Supplementary pancreatic innervation which has been shown to impact Figure 4 compare a and b panels, see section on islet architecture and function (Borden et al. 2013), we supplementary data given at the end of this article) next examined adult mice. GFRa3-deficient mice did not

ABC4500 0.016 0.04 DE 450 350 3500 0.012 0.03 350 250 2500 0.008 0.02 250

Journal of Molecular Endocrinology 150 1500 150 DAPI area (P0) Hormones area/ 0.004Hormones area/ 0.01 pancreas (E15.5) DAPI area (E15.5) NGN3 cell number/ Blood glucose (mg/dl) 500 Blood glucose (mg/dl) 50 50

WT GKO Ins Gcg WT GKO 0 1530456090120 0 1530456090120 Time (min) Time (min)

FGH0.20 0.03 4.10–4

0.16 3.10–4

0.02 puncta/ 0.12 + 2.10–4 0.08 TH area/ 0.01 hormone area 1.10–4 hormone area (P0)

0.04 hormone area (P21) TUJ1+ fibers lengths/ Number of VIP

WT GKOWT GKO WT GKO

Figure 6 Islet cell differentiation and function as well as endocrine innervation are GFRa3tLacZ/tLacZ (dashed line) adult mice. (f) Quantification of the length of C not impacted in GFRa3-deficient mice. (a) Quantification of Ngn3 cell Tuj1 neuron fibers normalized to hormones area (P0). (g) Quantification number (immunofluorescence) in WT and GFRa3tLacZ/tLacZ (named GKO) of sympathetic innervation expressed as THC area normalized to hormones pancreas at E15.5. (b and c) Quantifications of insulin and glucagon areas (glucagonCinsulin) area (P21). (h) Quantification of parasympathetic C (immunofluorescence) in pancreas from WT and GFRa3tLacZ/tLacZ embryos at innervation expressed as the number of VIP puncta normalized to E15.5 normalized to DAPI area at E15.5 (b) and P0 (c). In c, hormone area hormones (glucagonCinsulin) area (P21). Data in b, c, f and g are in represents glucagonCinsulin immune-positive areas. Black and white bars arbitrary units for quantifications, nZ4 for each genotype. For glucose represent WT and GFRa3tLacZ/tLacZ pancreas in b. (d) Intraperitoneal and tolerance tests, nZ6 for each genotype. (e) oral glucose tolerance tests performed on WT (black line) and

present any obvious defect in islet organization (Supple- progenitors (Pdx1 promoter). Artn cDNA was thus cloned mentary Figure 4i and j), with insulin-positive b-cells downstream of Pdx1 regulatory sequences and in fusion properly surrounded by glucagon-positive a-cells. Further- with the self-cleaving 2A peptide and mCherry fluor- more, both oral (Fig. 6d) and intraperitoneal (Fig. 6e) escent protein (to follow transgene expression) resulting glucose tolerance tests were normal demonstrating that into pPdx1-Artn-2A-mCherry construct or PAM (Fig. 7a). glucose clearance was not affected. Taken together, our We expected that Artn will be secreted from pancreatic loss of function studies demonstrate that GFRa3 signaling progenitors and signal to Ngn3- and hormone-positive is not essential for endocrine cell differentiation, islet cells endocrine cells which express GFRa3 receptor. One formation, and function. founder mouse expressed mCherry in the embryonic pancreas (Fig. 7b) in Pdx1-expressing cells as expected Islet innervation is normal in GFRa3-deficient mice (Supplementary Figure 6a, see section on supplementary data given at the end of this article). Immunofluores- GFRa2 has been shown to be required for proper cence experiments revealed that Artn was specifically parasympathetic innervation of the endocrine pancreas expressed in E13.5 transgenic embryos following the (Rossi et al. 2005). Since GFRa3 is expressed in developing mCherry pattern (Fig. 7c and d) while no Artn protein and adult pancreatic neurons (and glial cells) we assessed was detected in WT embryo (not shown). Careful a pancreatic innervation in GFR 3-deficient mice. We examination of Artn immunostaining showed that Artn C hypothesized that a mild islet innervation defect could is located both in mCherry cells (Fig. 7d, e and f; red eventually not impact IPGTT tests, a hypothesis supported K arrows) and in adjacent mCherry cells (Fig. 7d, e and f; by the observation that while vagal stimulation of insulin C green arrows). Interestingly in mCherry cells, Artn secretion is lost in GFRa2 KO mice, systemic glucose expression is found in the cytoplasm and at the plasma tolerance is normal (Rossi et al. 2005). Endocrine membrane, contrasting with the polarized and membra- innervation matures postnatally (Burris & Hebrok 2007); K nar signal of Artn observed in mCherry cells (Fig. 7d, e we thus examined endocrine innervation at P0 and P21. and f; green arrows). These results could suggest that Immunofluorescence experiments revealed the presence C C Artn is properly produced by mCherry cells and of TUJ1 neuronal cells in GFRa3tLacZ/tLacZ mice at P0 secreted and binds to cells expressing the receptor (Supplementary Figure 4a and b). Careful quantification of explaining the polarized signal. Due to incompatibilities TUJ1 fiber lengths (normalized to endocrine cells) at this of Artn and GFRa3 antibodies, this hypothesis could not stage did not reveal any significant variation (Fig. 6f), be tested. Journal of Molecular Endocrinology suggesting that GFRa3 is not essential for pancreas We next determined whether Artn overexpression had innervation during embryogenesis and at early postnatal an impact on endocrine differentiation at E13.5. Trans- stages. At P21, sympathetic neuronal cells were present in genic mice displayed a normal pattern of Ngn3-positive both islets and exocrine cells in GFRa3tLacZ/tLacZ mice (Supplementary Figure 4c and d) and quantification of cells (Supplementary Figure 6) and quantification did not intra-islet sympathetic fibers did not show any difference reveal any variation of Ngn3 cell numbers in Artn- between control and mutant mice (Fig. 6g). The same overexpressing embryos (Fig. 7g). Similarly, clusters of conclusion was reached for the parasympathetic inner- a-cells and more scattered b-cells were observed as vation (Fig. 6h and Supplementary Figure 5e, f, see section expected at this developmental stage in controls as well on supplementary data given at the end of this article). as in PAM transgenic mice, and their number did not vary Finally, P21 GFRa3-deficient islets were properly sur- (Fig. 7h). Thus, at E13.5 we did not observe any obvious rounded by glial cells (Supplementary Figure 5g and h). change in islet cell development in embryos overexpres- Thus, GFRa3 is not required for islet cell innervation sing Artn. Due to the expression of Pdx1 in b-cells, the neither for the formation of glial cells. transgene was also found in adult islets (Supplementary Figure 6b). However, both islet organization (Supple- mentary Figure 6c) and glucose tolerance (Fig. 7i) were Artn overexpression has no impact on islet cell normal in adult transgenic mice. Similarly, Artn over- development expression did not alter islet innervation (Supplementary We next thought to determine the consequences of Artn Figure 6d, e and f). Taken together, these results overexpression on islet cell development and generated demonstrate that Artn overexpression does not impact a mouse model where Artn is expressed in pancreatic islet differentiation, organization, and function.

PAM 2A mCherry

mPdx1-hsp68 promoter Artn 1 kb

B CD E F

GH I 1000 0.025 450

800 0.020 350

600 0.015 250

400 0.010 150 DAPI area

200Hormones area/ 0.005 50 pancreas (E13.5) Blood glucose (mg/dl) Number of NGN3 cells/ WT PAM Ins Ggc 01530456090 120 Time (min)

Figure 7 Islet cell differentiation and function in transgenic mice overexpressing respectively. (g) Quantiﬁcation of the number of Ngn3-immuno-positive

Journal of Molecular Endocrinology ARTN. (a) Schematic of the transgene construct (PAM) driving expression of cells per pancreas on cryosections of WT and transgenic PAM embryos at Artn-2A-mCherry protein in Pdx1-expressing cells. 2A, self-clivable peptide. E13.5. (h) Quantifications of insulin and glucagon immuno-positive areas (b) Digestive tract from a E13.5 PAM transgenic embryo showing mCherry on cryosections of WT and transgenic embryos at E13.5 normalized to total fluorescence in the pancreas (red arrows). (c and d) Immunostaining for DAPI area. (i) IPGTT performed on WT (black curve) and PAM transgenic Artn (green) and intrinsic mCherry fluorescence (in d) on cryosections of an adult mice (red curve). For quantifications nZ3–4 embryos for each E13.5 transgenic pancreas. (e and f) Higher magnification of inset in d. genotype. For glucose metabolism, nR4 for each genotype. Magnification C C K C Red and green arrows point to mCherry /Artn and mCherry /Artn cells of macroscopic picture is 20! in (b). sp, spleen; d, duodenum.

Discussion receptors. Importantly we provide evidence that GFRa3 can be used as a bio-marker for the immune-isolation of We provide here the ﬁrst description of the expression endocrine progenitor cells. of the GDNF receptor family GFRa3 in the pancreas. During pancreas development, we found that GFRa3 We found that this receptor is expressed at the is expressed in a subset of Ngn3-expressing cells suggesting plasma membrane of endocrine progenitors and develop- that some islet progenitors could receive and integrate ing a- and b-cells in the embryo but only in a cells in adult Artn signaling. It is not clear however why only a islets. In addition, GFRa3 is found in embryonic and adult subpopulation of Ngn3 cells expresses the receptor. pancreatic neurons and glial cells. This expression pattern Different scenarios can explain this pattern. Since GFRa3 suggested a role of GFRa3 and its ligand Artn in the control is found in all developing islet cells we do not believe that of pancreatic islet and nervous system development or GFRa3/Ngn3 double positive cells mark cells that have function. However, both loss and gain of function studies adopted a particular islet sub-type fate. We believe that did not reveal any function in these processes suggesting GFRa3 is not expressed (or not at detectable levels) in functional redundancies of GDNF family of ligands and nascent Ngn3-positive islet progenitors but only in more

mature Ngn3 cells, which could be in agreement with the GFRa1 in developing islet cells was not seen by us or fact that the receptor persists in developing a- and b-cells. reported by others. Therefore, we believe it is unlikely that Thus this specific expression pattern of GFRa3 could GFRa1 compensates the loss of GFRa3, although we cannot reflect a role of Arnt/GFRa3 signaling in the maintenance/ completely exclude it. Similarly, we have no clear evidence survival or maturation developing islet cells. that GFRa2 is expressed outside the neuronal lineage in the Our attempts to identify the source of Artn expressing embryonic pancreas (data not shown) and thus unlikely cells in the embryonic and adult pancreas by various replaces GFRa3, and finally, we could not detect any means failed, only quantitative RT-PCR revealed Artn pancreatic expression of GFRa4. Interestingly ARTN, as expression in the pancreatic mesenchyme at E12.5 well as GDNF and NRTN, are able to bind to other receptors, suggesting that Artn is expressed at very low levels in the such as Syndecan-3 (Bespalov et al. 2011), but it is unknown pancreas. Interestingly Artn2 has also been described whether this receptor is expressed in developing islet cells. in mesenchymal cells in the vicinity of developing Our studies revealed that GFRa3isexpressedin opercular muscle cells in zebrafish (Knight et al. 2011). pancreatic neurons and glial cells in the embryo and in Similarly Gdnf is expressed by mesenchymal cells of the the adult. Sensitive neurons were not investigated. gastrointestinal tract and by mesenchymal cells in the Although both parasympathetic and sympathetic fibers vicinity of the ureteric bud (Hellmich et al. 1996) but, in express the receptor, parasympathetic and sympathetic the embryonic pancreas, Gdnf is restricted to epithelial islet innervation were not affected in GFRa3 KO mice, pancreatic progenitor cells (Munoz-Bravo et al. 2013). In neither was glial cell differentiation or survival impaired. other studies Artn as been described along blood vessels in GFRa2 is expressed by glial cells as well as by parasym- endothelial smooth cells of the developing vasculature pathetic nervous system in the pancreas and islet para- acting as a chemo attractant guidance factor for sym- sympathetic innervation is reduced severely in GFRa2- pathetic fibers (Honma et al. 2002, Damon et al. 2007). deficient mice (Rossi et al.2003, 2005). Thus, GFRa2could In mice, it has been shown that Artn/GFRa3 signaling compensate GFRa3 deficiency in parasympathetic neurons is essential for sympathetic neurons migration and but eventually also in glial cell where the role of GFRa2in survival (Nishino et al.1999, Honma et al. 2002). unknown. Whether GFRa1 is redundant with GFRa3 In zebrafish embryos, it has been reported that GFRa3is in sympathetic fibers remains to be studied. required for myogenesis; loss of function results in reduced We showed above that islet cell development in expression of myogenic factors including the b-hlh transgenic mice overexpressing Artn was unaltered at transcription factor MyoD (Knight et al. 2011). Our studies E13.5. It could be that endogenous Artn is sufficient or that

Journal of Molecular Endocrinology suggest that GFRa3 is dispensable for endocrine cell Artn has no effect on endocrine cell differentiation or formation, survival, and function which is rather striking survival. Alternatively transgenic artemin is not bioactive, given the remarkable expression pattern of this receptor in a hypothesis that we do not favor as in control the endocrine lineage during development and in a-cells experiments we veriﬁed that the artemin-2A-mRFP con- in the adult islets. Thus, either GFRa3 has a non-essential struct promoted neuroblast cell proliferation (data not function in these processes or another GFR can compen- shown). Of note, in similar experiment, it has been sate for the lack of GFRa3. Importantly variable pene- reported that transgenic mice overexpressing GDNF trance of the sympathetic phenotype in GFRa3 or Artn (which binds GFRa1) displayed a higher number of Ngn3 knock-out mice (Honma et al. 2002) indeed suggest cells (Mwangi et al. 2010). We think, as supported by functional redundancy among GFLs and GFRs. As we det- data from Munoz-Bravo et al. (2013), that this does not ected (Affymetrix data) low amounts of GFRa1 transcripts result from a direct trophic effect of GDNF on Ngn3 cells in the embryonic pancreas and that Artn can weakly but rather from the proliferation of Pdx1-pancreatic crosstalk with GFRa1(Saarma & Sariola 1999), we thought progenitors from which Ngn3-cells derive (Gu et al. that this receptor could eventually compensate for the 2002). While the current study was on-going, Blum et al. absence of GFRa3. Although GFRa1 immunostaining was (2014) published the results of a screen for factors that clearly observed in developing pancreatic neurons at reverse b cell de-differentiation (loss of mature b-cell E15.5, we did not detect any expression of this receptor phenotype), a mechanism that might cause diabetes. in the pancreatic epithelium at the same stage (data not Interestingly, they found that after induced islet dediffer- shown). However, Munoz-Bravo et al. (2013) reported entiation in vitro, a small molecule inhibitor of the TGF b GFRa1 expression in multipotent progenitor TIP cells as receptor I, but also artemin, restored the expression of well as in developing acinar cells. However, expression of Urocortin 3 (Unc3), a marker of functionally mature

b-cells. We did not notice any change in Unc3 protein Acknowledgements expression neither in adult islets of GFRa3 KO mice nor We thank Dr Enomoto and Dr Milbrandt for the generous gift of Artn and in pPdx1-Artn-2A-mCherry transgenics suggesting that GFRa3 KO mice. We thank the Mouse Clinical Institute (Illkirch, France) for islet cells are mature in both models (data not shown). the generation of PAM transgenic mice and the members of IGBMC imaging platform and animal facility for their great help and advices, Whether transgenic mice overexpressing artemin members of the lab for their support and scientific discussions. We thank are protected from induced b-cell de-differentiation or D A Cano and A Grapin-Botton for helpful discussion. diabetes remains to be investigated. Interestingly, two out of the three GFRa3 regulated genes we identified, Nphp3 and Pkd2l2 are coding for References proteins related to cilia. Nphp3 whose mutations are Ahren B 2000 Autonomic regulation of islet hormone secretion – associated with multiorgan polycystic disease encodes a implications for health and disease. Diabetologia 43 393–410. protein located in cilia centrosome complex (Leeman et al. (doi:10.1007/s001250051322) Airaksinen MS & Saarma M 2002 The GDNF family: signalling, biological 2014). Recent findings suggest that Pkd2l2 is coding for functions and therapeutic value. Nature Reviews. Neuroscience 3 calcium channels at primary cilia (DeCaen et al. 2013). 383–394. (doi:10.1038/nrn812) Thus, Artn/GFRa3 could be involved in ciliogenesis or cilia Bespalov MM, Sidorova YA, Tumova S, Ahonen-Bishopp A, Magalhaes AC, Kulesskiy E, Paveliev M, Rivera C, Rauvala H & Saarma M 2011 Heparan based signaling. In conclusion, this study revealed a novel sulfate proteoglycan syndecan-3 is a novel receptor for GDNF, receptor expressed at the surface of developing islet cells neurturin, and artemin. Journal of Cell Biology 192 153–169. as well as adult a-cells. However, the role of Artn/GFRa3 (doi:10.1083/jcb.201009136) Blum B, Roose AN, Barrandon O, Maehr R, Arvanites AC, Davidow LS, signaling in islet cell development and function remains Davis JC, Peterson QP, Rubin LL & Melton DA 2014 Reversal of b cell elusive. Indeed, we could not elucidate the precise de-differentiation by a small molecule inhibitor of the TGFb pathway. function of Artn/GFRa3 during islet cell development eLife 3 e02809. (doi:10.7554/eLife.02809.001) Borden P, Houtz J, Leach SD & Kuruvilla R 2013 Sympathetic innervation and there is thus no clear rational justifying to add Artn during development is necessary for pancreatic islet architecture and growth factor in hES cell differentiation protocols to functional maturation. Cell Reports 4 287–301. (doi:10.1016/j.celrep. improve the differentiation and functionality of the 2013.06.019) Burris RE & Hebrok M 2007 Pancreatic innervation in mouse development derived insulin-producing cells. Nevertheless, we could and b-cell regeneration. Neuroscience 150 592–602. (doi:10.1016/j. take advantage of the expression of GFRa3 at the surface neuroscience.2007.09.079) of Ngn3-cells (provided this is the case in human as well) Cano DA, Soria B, Martin F & Rojas A 2013 Transcriptional control of mammalian pancreas organogenesis. Cellular and Molecular Life Sciences to FACS purify and study specific population of 71 2383–2402. (doi:10.1007/s00018-013-1510-2) endocrine precursors cells as it has been shown for other Costantini F & Shakya R 2006 GDNF/Ret signaling and the development of

Journal of Molecular Endocrinology the kidney. BioEssays 28 117–127. (doi:10.1002/bies.20357) receptors such as Ddr1, Disp2 (Hald et al. 2012) or CD133 Damon DH, Teriele JA & Marko SB 2007 Vascular-derived artemin: (Prominin-1) and CD49f (or a6-integrin) (Sugiyama a determinant of vascular sympathetic innervation? American et al. 2007). The combination of such purification tools Journal of Physiology. Heart and Circulatory Physiology 293 H266–H273. (doi:10.1152/ajpheart.00859.2006) could be very useful to isolate bona fide endocrine DeCaen PG, Delling M, Vien TN & Clapham DE 2013 Direct recording precursors in a mixed cell population, a step that might and molecular identification of the calcium channel of primary cilia. promote the maturation, or increase the yield of hES Nature 504 315–318. (doi:10.1038/nature12832) Desgraz R & Herrera PL 2009 Pancreatic neurogenin 3-expressing cells are derived b-cells in vitro. unipotent islet precursors. Development 136 3567–3574. (doi:10.1242/ dev.039214) Gradwohl G, Dierich A, LeMeur M & Guillemot F 2000 Neurogenin3 is Supplementary data required for the development of the four endocrine cell lineages of the This is linked to the online version of the paper at http://dx.doi.org/10.1530/ pancreas. PNAS 97 1607–1611. (doi:10.1073/pnas.97.4.1607) JME-15-0213. Gu G, Dubauskaite J & Melton DA 2002 Direct evidence for the pancreatic lineage: NGN3C cells are islet progenitors and are distinct from duct progenitors. Development 129 2447–2457. Declaration of interest Hald J, Galbo T, Rescan C, Radzikowski L, Sprinkel AE, Heimberg H, The authors declare that there is no conflict of interest that could be Ahnfelt-Ronne J, Jensen J, Scharfmann R, Gradwohl G et al. 2012 perceived as prejudicing the impartiality of the research reported. Pancreatic islet and progenitor cell surface markers with cell sorting potential. Diabetologia 55 154–165. (doi:10.1007/s00125-011-2295-1) Hellmich HL, Kos L, Cho ES, Mahon KA & Zimmer A 1996 Embryonic Funding expression of glial cell-line derived neurotrophic factor (GDNF) This study was supported by the Beta Cell Biology Consortium through suggests multiple developmental roles in neural differentiation and National Institutes of Health Grant U01DK089571 and the Fondation pour epithelial-mesenchymal interactions. Mechanisms of Development 54 la Recherche Me´ dicale. L N was a recipient of a fellowship from INSERM, the 95–105. (doi:10.1016/0925-4773(95)00464-5) Re´ gion Alsace and from the Fondation pour la Recherche Me´ dicale Henseleit KD, Nelson SB, Kuhlbrodt K, Hennings JC, Ericson J & Sander M (file number FDT20130928131). 2005 NKX6 transcription factor activity is required for a-andb-cell

development in the pancreas. Development 132 3139–3149. Munoz-Bravo JL, Hidalgo-Figueroa M, Pascual A, Lopez-Barneo J, (doi:10.1242/dev.01875) Leal-Cerro A & Cano DA 2013 GDNF is required for neural colonization Hofmann MC 2008 Gdnf signaling pathways within the mammalian of the pancreas. Development 140 3669–3679. (doi:10.1242/dev. spermatogonial stem cell niche. Molecular and Cellular Endocrinology 091256) 288 95–103. (doi:10.1016/j.mce.2008.04.012) Mwangi SM, Usta Y, Raja SM, Anitha M, Chandrasekharan B, Parsadanian A, Honma Y, Araki T, Gianino S, Bruce A, Heuckeroth R, Johnson E & Sitaraman SV & Srinivasan S 2010 Glial cell line-derived neurotrophic Milbrandt J 2002 Artemin is a vascular-derived neurotropic factor for factor enhances neurogenin3 gene expression and beta-cell prolifer- developing sympathetic neurons. Neuron 35 267–282. (doi:10.1016/ ation in the developing mouse pancreas. AJP: Gastrointestinal and Liver S0896-6273(02)00774-2) Physiology 299 G283–G292. (doi:10.1152/ajpgi.00096.2010) Johansson KA, Dursun U, Jordan N, Gu G, Beermann F, Gradwohl G & Nishino J, Mochida K, Ohfuji Y, Shimazaki T, Meno C, Ohishi S, Matsuda Y, Grapin-Botton A 2007 Temporal control of neurogenin3 activity in Fujii H, Saijoh Y & Hamada H 1999 GFR a3, a component of the artemin pancreas progenitors reveals competence windows for the generation receptor, is required for migration and survival of the superior cervical of different endocrine cell types. Developmental Cell 12 457–465. ganglion. Neuron 23 725–736. (doi:10.1016/S0896-6273(01)80031-3) (doi:10.1016/j.devcel.2007.02.010) Rossi J, Herzig KH, Voikar V, Hiltunen PH, Segerstrale M & Airaksinen MS Knight RD, Mebus K, d’Angelo A, Yokoya K, Heanue T, Tubingen Screen C 2003 Alimentary tract innervation deﬁcits and dysfunction in mice & Roehl H 2011 Ret signalling integrates a craniofacial muscle module lacking GDNF family receptor alpha2. Journal of Clinical Investigation during development. Development 138 2015–2024. (doi:10.1242/dev. 112 707–716. (doi:10.1172/JCI17995) 061002) Rossi J, Santamaki P, Airaksinen MS & Herzig KH 2005 Parasympathetic Leeman KT, Dobson L, Towne M, Dukhovny D, Joshi M, Stoler J & innervation and function of endocrine pancreas requires the glial cell Agrawal PB 2014 NPHP3 mutations are associated with neonatal onset line-derived factor family receptor alpha 2 (GFRa2). Diabetes 54 multiorgan polycystic disease in two siblings. Journal of Perinatology 34 1324–1330. (doi:10.2337/diabetes.54.5.1324) 410–411. (doi:10.1038/jp.2014.20) Saarma M & Sariola H 1999 Other neurotrophic factors: glial cell line- Lodh S, O’Hare EA & Zaghloul NA 2014 Primary cilia in pancreatic derived neurotrophic factor (GDNF). Microscopy Research and Technique development and disease. Birth Defects Research. Part C, Embryo Today: 45 292–302. (doi:10.1002/(SICI)1097-0029(19990515/01)45:4/5 Reviews 102 139–158. (doi:10.1002/bdrc.21063) !292::AID-JEMT13O3.0.CO;2-8) Mastracci TL & Sussel L 2012 The endocrine pancreas: insights into Soyer J, Flasse L, Raffelsberger W, Beucher A, Orvain C, Peers B, Ravassard P, development, differentiation and diabetes. Wiley Interdisciplinary Vermot J, Voz ML, Mellitzer G et al. 2010 Rfx6 is an Ngn3-dependent Reviews. Membrane Transport and Signaling 1 609–628. (doi:10.1002/ winged helix transcription factor required for pancreatic islet cell wdev.44) development. Development 137 203–212. (doi:10.1242/dev.041673) Mellitzer G, Martin M, Sidhoum-Jenny M, Orvain C, Barths J, Seymour PA, Sugiyama T, Rodriguez RT, McLean GW & Kim SK 2007 Conserved markers Sander M & Gradwohl G 2004 Pancreatic islet progenitor cells in of fetal pancreatic epithelium permit prospective isolation of islet neurogenin 3-yellow ﬂuorescent protein knock-add-on mice. Molecular progenitor cells by FACS. PNAS 104 175–180. (doi:10.1073/pnas. Endocrinology 18 2765–2776. (doi:10.1210/me.2004-0243) 0609490104)

Received in ﬁnal form 3 November 2015 Accepted 17 November 2015 Accepted Preprint published online 17 November 2015 Journal of Molecular Endocrinology