letters to nature null . Two different were used to mark mutant clones: yw tumours in Cdx2 mutant mice. Nature 396, 84–87 (1997). SA1 + - 24. Miller, D. J. & Miles, A. and the zootype. Nature 365, 215–216 (1993). FLP122; FRT42D Dll /FRT42D Dp(y )44B or yw FLP122; FRT42D Dll 25. Sa´nchez, L. Casares, F., Gorfinkiel, N. & Guerrero, I. The genital disc of melanogaster. II. SA1 − /FRT42D pwn. In the first , Dll heat-shock-induced clones were Role of the genes hedgehog, decapentaplegic and wingless. Dev. Genes Evol. 207; 229–241 (1997). labelled with the cuticle marker y. As a possible transformation into hindgut 26. Murakami, R. et al. aproctous, a locus that is necessary for the development of the proctodeum in Drosophila embryos, encodes a homolog of the vertebrate Brachyury . Wilhelm Roux Arch. Dev. could not be scored with this system, in the second genotype the marker pwn Biol. 205, 89–96 (1995). labels the Dll+ clones twins of the DllSA1 clones. For Dll clones27 in the genital 27. Gorfinkiel, N., Morata, G. & Guerrero, I. The homeobox gene Distal-less induces ventral appendage SA1 development in Drosophila. Genes Dev. 11, 2259–2271 (1997). disc, larvae of genotype yw FLP122; FRT42D Dll /FRT42D arm–lacZ were 28. Diaz-Benjumea, F. & Cohen, S. M. Serrate signals through Notch to establish a wingless-dependent SA given 1-h heat shocks and Dll clones were marked by the loss of either Dll organizer at the dorsal-ventral compartment boundary of the Drosophila wing. Development 121, protein or lacZ activity. Partial loss of Dll function was studied in trans 4215–4225 (1995). 3 MP 8 combinations of the Dll , Dll and Dll–Gal4 . We studied the require- Acknowledgements. We thank P. Fernandez, T. Kusch, C. Parras, R. Rivera, I. Rodriguez and G. Struhl for ments of the analia for hh, wg and dpp function using the hhts2, wgIL114, dppd12 antibodies, cDNAs and fly stocks; J. Casanova and E. Sanchez-Herrero for comments on the manuscript; S. Gonzalez for helping with the DNA injections; and R. Gonzalez and J. M. Gala´n for their technical help. d14 20 21 and dpp alleles as described . To activate the Hh pathway in the entire This work was supported by grants from the Direccio´n General de Investigacio´n Cientifica y Te´cnica and analia primordium we generated yw; cad–Gal4/UAS–ci; UAS–y flies, in which from the Human Frontier Science Program. An institutional grant from the Fundacion Ramon Areces to the Centro de Biologia Molecular is also acknowledged. E.M. is supported by a scholarship from the the Ci product is expressed in the whole of the analia primordium. The anal Comunidad Auto´noma de Madrid. plates of these flies are marked by the expression of the UAS–y gene, which Correspondence and requests for materials should be addressed to G.M. (e-mail: [email protected]). confers a dark colour. Ectopic cad expression. We used the Gal4/UAS method19.AUAS–cad gene was constructed by inserting the cad complementation DNA (a gift from R. Rivera-Pomar) in the pUAS vector as described8. The UAS–cad construct is function, as it can rescue the mutant of the cad–Gal4 insertions. The Notch signalling controls MS-248 (a gift from C. Parras) and ap–Gal48 driver lines gave the best results. Flies of genotype MS-248/UAS–cad and ap–Gal4/UAS–cad were mounted and pancreatic cell differentiation examined at the light microscope. A˚ sa Apelqvist*†, Hao Li*†, Lukas Sommer‡, Paul Beatus§, X-gal and antibody staining. X-gal staining of adult flies and imaginal discs David J. Andersonk, Tasuku Honjo¶, 8 was performed as described . For adults it is important that they had emerged Martin Hrabeˇ de Angelis#, Urban Lendahl§ shortly before fixing. Imaginal discs were immunostained using the normal & Helena Edlund* procedures for confocal microscopy. The specific antibodies used were rabbit * Department of Microbiology, University of Umea˚, S-901 87 Umea˚, Sweden anti-Byn (a gift from T. Kusch), rabbit anti-Cad (from G. Struhl), monoclonal ‡ Institute of Cell Biology, Swiss Federal Institute of Technology, anti-Dll (from S. Cohen and I. Duncan), rabbit anti-Eve (from M. Frusta) and ETH-Hoenggerberg HPM E38, CH-8093 Zu¨rich, Switzerland monoclonal anti-Abd-B (from S. Celniker and A. Macias). k Division of Biology 216-76, California Institute of Technology, 1201 E. California Blvd., Pasadena, California 91125, USA Received 17 May; accepted 28 June 1999. ¶ Department of Medical Chemistry, Kyoto University Faculty of Medicine, 1. Macdonald, P. & Struhl, B. A molecular gradient in early Drosophila embryos and its role in specifying Sakyo-ku Yoshida, Kyoto 606–8501, Japan the body pattern. Nature 324, 537–545 (1986). # GSF, Institute for Mammalian Genetics, Ingolstaedter Landstrasse 2. Mlodzik, M., Fjose, A. & Gehring, W. J. Isolation of caudal,aDrosophila homeobox-containing gene 1, D-85764 Neuherberg, Germany with maternal expression, whose transcription forms a concentration gradient at the pre-blastoderm stage. EMBO J. 4, 2961–2969 (1985). § Department of Cell and Molecular Biology, Karolinska Institute, 3. Hunter, C. P.& Kenyon, C. Spatial and temporal controls target pal-1 blastomere-specification activity S-171 77 Stockholm, Sweden to a single blastomere lineage in C. elegans embryos. Cell 87, 217–226 (1997). † These authors contributed equally to this work 4. Brooke, N. M., Garcia-Fernandez, J. & Holland, P. W. The ParaHox gene cluster is an evolutionary ...... sister of the cluster. Nature 392, 920–922 (1998). 5. McGinnis, W. & Krumlauf, R. Homeobox genes and axial patterning. Cell 68, 283–302 (1992). The pancreas contains both exocrine and endocrine cells, but the 6. Basler, K. & Struhl, G. Compartment boundaries and the control of Drosophila limb pattern by Hedgehog protein. Nature 368, 208–214 (1994). molecular mechanisms controlling the differentiation of these cell 7. Wu, L. H. & Lengyel, J. A. Role of caudal in the hindgut specification and gastrulation suggests types are largely unknown. Despite their endodermal origin, between Drosophila amnioproctodeal invagination and vertebrate blastopore. Development pancreatic endocrine cells share several molecular characteristics 125, 2433–2442 (1996). 1–5 6,7 8. Calleja, M., Moreno, E., Pelaz, S. & Morata, G. Visualization of gene expression in living adult with neurons , and, like neurons in the central nervous system , Drosophila. Science 274, 252–255 (1996). differentiating endocrine cells in the pancreas appear in a scat- 9. Casares, F., Sanchez, L., Guerrero, I. & Sanchez-Herrero, E. The genital disc of Drosophila 8,9 melanogaster. I. Segmental and compartmental organization. Dev. Genes Evol. 207, 216–228 (1997). tered fashion within a field of progenitor cells . This indicates 10. No¨thiger, R., Du¨bendorfer, A. & Epper, F. Gynandromorphs reveal two separate primordia for male that they may be generated by lateral specification through Notch and female genitalia in Drosophila melanogaster. Wilhelm Roux Arch. Dev. Biol. 181, 367–373 (1977). signalling6,7. Here, to test this idea, we analysed pancreas devel- 11. Epper, F. Three-dimensional fate map of the female genital disc of Drosophila melanogaster. Wilhelm Roux Arch. Dev. Biol. 192, 270–274 (1983). opment in mice genetically altered at several steps in the Notch 12. Cohen, S. M., Broner, G., Kuntter, F., Jurgens, G. & Jackle, H. Distal-less encodes a homeodomain signalling pathway. Mice deficient for Delta-like gene 1 (Dll1)10 or protein required for limb development in Drosophila. Nature 338, 432–434 (1989). 11 13. Frasch, M., Hoey, T., Rushlow, C., Doyle, H. & Levines, M. Characterization localization of the even- the intracellular mediator RBP-Jk showed accelerated differen- skipped protein of Drosophila. EMBO J. 6, 749–759 (1987). tiation of pancreatic endocrine cells. A similar phenotype was 14. Kispert, A., Herrmann, B. G., Leptin, M. & Reuter, R. Homologs of the mouse Brachyury gene are observed in mice over-expressing neurogenin 3 (ngn 3)12 or the involved in the specification of posterior terminal structures in Drosophila, Tribolium, and Locusta. Genes Dev. 8, 2137–2150 (1994). intracellular form of Notch3 (ref. 13) (a repressor of Notch 15. Singer, J. B., Harbecke, R., Kusch, T., Reuter, R. & Lengyel, J. A. Drosophila brachyenteron regulates signalling). These data provide evidence that ngn3 acts as pro- gene activity and morphogenesis in the gut. Development 122, 3703–3718 (1996). 16. Nusslein-Volhard, C., Kluding, H. & Jurgens, G. Genes affecting the segmental subdivisions of the endocrine gene and that Notch signalling is critical for the Drosophila embryo. Cold Spring Harb. Symp. Quant. Biol. 50, 145–154 (1985). decision between the endocrine and progenitor/exocrine fates in 17. Sanchez-Herrero, E., Vernos, I., Marco, R. & Morata, G. Genetic organization of the Drosophila the developing pancreas. bithorax complex. Nature 313, 108–113 (1985). 18. Lecuit, T. & Cohen, S. M. Proximal–distal axis formation in the Drosophila leg. Nature 388, 139–145 In the Notch signalling system, ligand activation leads to intra- (1997). cellular cleavage of the Notch receptor. The activated intracellular 19. Brand, A. H. & Perrimon, N. Targeted gene expression as a means of altering cell fates and generating dominant . Development 118, 401–415 (1993). domain of Notch receptors interacts with the DNA-binding protein 20. Diaz-Benjumea, F. J., Cohen, B. & Cohen, S. M. Cell interaction between compartments establishes the RBP-J␬ to activate expression of the negative basic helix–loop–helix proximal–distal axis of Drosophila legs. Nature 372, 175–179 (1994). (bHLH) HES genes, which, in turn, repress expression of down- 21. Alexandre, C., Jacinto, A. & Ingham, P. W. Transcriptional activation of hedgehog target genes in 6,7 12,14–16 Drosophila is mediated directly by the Cubitus interruptus protein, a member of the GLI family of zinc stream target genes including the ngn genes . Thus, high ngn finger DNA-binding proteins. Genes Dev. 10, 2003–2013 (1996). expression is indicative of cells differentiating to a primary fate. We 22. van den Heuvel, M. & Ingham, P. W. smoothened encodes a receptor-like serpentine protein required for hedgehog signalling. Nature 382, 547–551 (1996). therefore first investigated ngn3 expression in the developing 23. Chawengsaksophak, K., James, R., Hammond, V. E., Kontgen, F. & Beck, F. Homeosis and intestinal pancreas in detail. Low, uniform expression of ngn3 could be

© 1999 Macmillan Magazines Ltd NATURE | VOL 400 | 26 AUGUST 1999 | www.nature.com 877 letters to nature detected in the dorsal pancreatic epithelium at the ϳ14-somite stage primarily of differentiated endocrine cells, as indicated by the (embryonic day (E) 9) (Fig. 1a). By the 25-somite stage (E9.5), ngn3 expression of ISL1 (Fig. 1i, p), glucagon (Fig. 1j, q), peptide YY was expressed at high levels in a few cells of the dorsal pancreatic (PYY)19 (Fig. 1k, r) and NeuroD20–22 (data not shown). The endo- epithelium (Fig. 1b). ngn3+ cells in the ventral epithelium were first crine cells in the pancreatic buds of stage-matched control embryos detected at E10.5 (Fig. 1b; and data not shown), in keeping with the were relatively fewer and more scattered (Fig. 1i–j). No carboxy- delayed appearance of endocrine cells in the ventral compared with peptidaseA+ cells were detected in the pancreatic buds of E12.5 Ipf1/ the dorsal pancreas4,5,8,9. The relative number of ngn3+ cells Pdx1–ngn3 transgenic embryos (Fig. 1l, s). This block in pancreatic increased between E9.5 and E15.5 (Fig. 1a–e), but then decreased exocrine development also persisted at later stages, as indicated by such that, by E17.5 and in neonatal pancreas, there were only a few the lack of both amylase+ (Fig. 1n, u) and carboxypeptidaseA+ (data ngn3+ cells (Fig. 1f, g). Markers of differentiated pancreatic endo- not shown) cells in the bud of E15.5 transgenic embryos. Instead, crine cells, such as the islet-specific LIM-homeodomain protein endocrine cells expressing glucagon (Fig. 1m, t), PYYand ISL1 (data ISL12,17 (Fig. 1a–g), the ␤-cell-specific homeodomain protein IPF1/ not shown) were observed. Thus, ectopic expression of ngn3 in most PDX1 and the pancreatic hormones (not shown), were expressed in of the early pancreatic precursor cells results in a massive, pre- cells separate from ngn3+ cells, providing evidence that ngn3+ cells cocious differentiation of pancreatic progenitor cells into endocrine and post-mitotic endocrine cells represent distinct populations. cells. The accelerated, premature differentiation of endocrine cells This indicates that the ngn3+ cells may be endocrine precursor cells, depletes the pool of pancreatic precursor cells, resulting in a short- and may correspond to the primary fate in a lateral specification age of cells that can proliferate, branch and differentiate into model. exocrine cells (Fig. 1n–u). Collectively, these data show that ngn3 To test this idea, we used the Ipf1/Pdx1 promoter18 to broaden the acts as a pro-endocrine gene during pancreatic development, a expression of ngn3 to most pancreatic precursor cells in transgenic function analogous to that of ngn genes in neurogenesis14–16. mice. E12.5 Ipf1/Pdx1–ngn3 transgenic embryos exhibited small These findings implicate the Notch signalling pathway in pan- and poorly branched pancreatic buds (Fig. 1h, o), consisting creatic cell-type specification. We next altered Notch signalling at

Figure 1 ngn3 is expressed and functions as a pro-endocrine gene during p–u). Accelerated endocrine cell differentiation occurs at the expense of exocrine pancreatic development. a–g, ngn3 (dark blue) and ISL1 (brown) are expressed in cell differentiation in Ipf1/Pdx1–ngn3 transgenic pancreatic buds, as indicated by separate sets of cells in the pancreas from E9.0 (a), E9.5 (b), E11.5 (c), E13.5 (d), the presence of ISL1+ (i, p), glucagon+ (j, q, m, t) and PYY+ (k, r) pancreatic E15.5 (e) and E17.5 (f) embryos and neonates (g). h–u, Accelerated pancreatic endocrine cells but lack of carboxypeptidaseA+ (l, s) and amylase+ (red) (n, u) cells endocrine-cell differentiation in Ipf1/Pdx1–ngn3 transgenic mice. Whole-mount in E12.5 (h–l, o–s) and E15.5 (m, n, t, u) wild-type (h–n) and transgenic (o–u) immunohistochemistry of the pancreatic region of E12.5 wild-type (h) and Ipf1/ embryos. Epithelial specific anti-E-cadherin (E-cad) antibodies (green) (n, u) were Pdx1–ngn3 transgenic (o) embryos using anti-IPF1/PDX1 antibodies shows used to visualize the epithelium of E15.5 embryos and ngn3-expressing cells impaired development of the pancreatic epithelium in transgenic embryos (i–n, (blue) are shown by in situ hybridization and m and t. Scale bars, 0.05 mm.

© 1999 Macmillan Magazines Ltd 878 NATURE | VOL 400 | 26 AUGUST 1999 | www.nature.com letters to nature different steps in the signalling pathway: at the ligand, receptor and Jk11, in which Notch signalling would be impaired at the ligand and intracellular mediator levels. Notch1 (ref. 23) and Notch2 (ref. 23), intracellular mediator levels, respectively. RBP-Jk deficient mice die but not Notch3 (ref. 23), were uniformly expressed in the epithelium at around E8.5–9.5 (ref. 11). Nevertheless, analysis of E8.5–9.0 at the pancreatic anterioposterior level of E9.5 wild-type embryos RBP-Jk mutant embryos revealed an apparently increased number (Fig. 2a–c). Dll1 (ref. 24), but not Serrate1 (ref. 23), was also of ngn3+ cells in the dorsal pancreatic epithelium (Fig. 3a, h; and expressed at this stage, but only in the dorsal pancreatic epithelium data not shown). Dll1 mutant mice are developmentally arrested (Fig. 2d, e). The restriction of Dll1 expression to the dorsal between E10 and E12 (ref. 10) and thus only the early stages of epithelium of E9.5 embryos is consistent with the differentiation pancreatic development can be investigated in these mice. HES-1 − − of endocrine cells in the dorsal, but not ventral, pancreatic bud at expression was reduced in the dorsal epithelium of E9.5 Dll1 / this stage. The HES-1 gene25 was, however, uniformly expressed embryos (Fig. 3b, i). Moreover, the total number of ngn3+ cells (55, in the E9.5 pancreatic epithelium (Fig. 2f), implying that other 64 and 74 ngn3+ cells, respectively, in n ¼ 3 embryos; Fig. 3j) was − − ligands may be expressed in the ventral pancreas at this stage. By increased 3.5–4-fold in E9.5 Dll1 / embryos as compared with the E13.5, Notch1, Notch2, Dll1 and HES-1 were still expressed in the few, scattered ngn3+ cells (11, 17 and 20 ngn3+ cells, respectively, in pancreatic epithelium (Fig. 2g, h, j, l) and both Notch3 and n ¼ 3 embryos) in the dorsal pancreatic epithelium of Dll1+/+ Serrate1 were also detected (Fig. 2i, k). These results show that embryos (Fig. 3c). By E10.5, differentiating endocrine cells were components of the Notch pathway are expressed in the developing scattered among IPF1/PDX1+ progenitor cells within the protrud- pancreas. ing epithelial dorsal pancreatic bud (Fig. 3d, e, f, g). In E10.5 Dll1 To assess the role of Notch signalling during pancreatic cell-type mutant mice, the pancreatic bud was decreased in size and instead a specification, we analysed mice deficient in Dll1 (ref. 10) and RBP- cap of differentiated endocrine cells expressing NeuroD/BETA2

Figure 2 Notch signalling components are expressed in the developing Delta1 (d, j), Serrate1 (e, k) and HES-1 (f, l). a–f, Dorsal–ventral axis is as shown in pancreas. a–l, In situ hybridization analysis showing the expression in E9.5 f. Scale bars in a–f, 0.05 mm; g–l, 0.1mm. (a–f) and E13.5 (g–l) pancreas of Notch1 (a, g), Notch2 (b, h), Notch3 (c, i),

Figure 3 Lack of RBP-Jk and Dll1 leads to precocious pancreatic endocrine IPF1/PDX1 (brown) (d, k), ISL1 (a, e, h, l), glucagon (f, m) and PYY (g, n) differentiation. Analysis of E9.5 (a–c, h–j) and E10.5 (d–g, k–n) wild-type (a–g), shows accelerated, premature differentiation of pancreatic endocrine cells in RBP-Jk−/− (h) or Dll1−/− (i–n) mice using riboprobes corresponding to ngn3 the mutant embryos. Scale bars in a, h, 0.05 mm; b, c, i, j, d–gand k–n, (a, c. h, j) HES-1 (b, i) and NeuroD (blue) (d, k) and antibodies specific for 0.02 mm.

© 1999 Macmillan Magazines Ltd NATURE | VOL 400 | 26 AUGUST 1999 | www.nature.com 879 letters to nature

Figure 4 Repressed Notch signalling obstructs pancreatic epithelial branching Pdx1–N3IC transgenic (k–n) pancreatic buds using antibodies specific for E- and exocrine differentiation. a–c, h–j, The pancreatic epithelium is poorly cadherin (d, k), carboxypeptidaseA (e, l), amylase (f, m) and ISL1 (g, n) show that, branched, HES-1 expression (a, h) is reduced, p48 expression (b, i) is lost and in the transgenic embryos, the pancreating bud is decreased in size and poorly ngn3 (c, j) is expressed in cells throughout the pancreatic epithelium of E13.5 Ipf1/ branched, and that exocrine, but not endocrine, cell differentiation is impaired. Pdx1–N31C transgenic (h–j) as compared with stage-matched wild-type (a–c) Scale bars in a–c, h–j and d–g, k–n, 0.1mm. embryos. d–n, Immunohistochemical analysis of E15.5 wild-type (d–g) and Ipf1/

(Fig. 3d, k), ISL1 (Fig. 3e, l), glucagon (Fig. 3f, m) and PYY (Fig. 3g, Reduced Notch signalling leads to increased expression of the n) surrounded the IPF1/PDX1+ bud. The decrease in size of the pro-endocrine gene ngn3, which promotes the endocrine fate. − − pancreatic bud in Dll1 / embryos did not appear to be caused by Conversely, at normal levels of Notch signalling, cells express increased apoptosis, as judged from TUNEL assays (data not HES-1 and p48 and adopt the exocrine fate. Ⅺ shown). These results indicate that, in the absence of Dll1, epithelial, ...... + IPF1/PDX1 progenitor cells within the pancreatic bud differentiate Methods prematurely into endocrine cells, resulting in a depletion of cells . The generation of mice deficient in Dll1 (ref. 10) and RBP-Jk (ref. 11) within the pancreatic bud. These results provide evidence that lack has been described elsewhere and mutant mice were obtained from our local of Dll1 or RBP-Jk (that is, impaired Notch signalling) leads to an breeding colony. Mice and embryos were genotyped as described10,11. accelerated differentiation of pancreatic precursor cells into endocrine Preparation of construct for transgenic mice. A 4.5-kilobase (kb) NotI–NaeI cells. fragment located immediately upstream of the Ipf1/Pdx1 gene18 was sub-cloned N31C represses Notch 1-mediated up-regulation of HES-1 into a vector carrying a polyA site and a 1.6-kb BsgI–BamHI fragment of full- expression in vitro26. Thus, to experimentally perturb Notch signal- length ngn3 complementary DNA12 or a 2.3-kb cDNA fragment of the Notch3 ling at the receptor level, we generated Ipf1/Pdx1–N3IC transgenic intracellular domain (N31C)13. We generated transgenic mice by pronuclear − mice. In wild-type E13.5 embryos, the HES-1 (Figs 2l, 4a) and the injection of the purified expression cassettes (NotI–BamHI) (2.0 ng ml 1) into exocrine p48 (refs 27, 28) (Fig. 4b) genes were F2 B6/CBA hybrid oocytes as described29. The genotypes were determined by highly expressed in the forming acini of the branching pancreatic PCR analyses of genomic DNA extracted from the yolk sac. The primers used epithelia, whereas developing endocrine cells, and thus ngn3+ cells, were: 5Ј-TAGCGAGGGGGAAGAGGAGAT-3Ј (Ipf1/Pdx1 primer for 5Ј) and were found predominantly in the centre of the developing pancrea- 5Ј-TAGGTATGAGAGTGGGGCTAGG-3Ј (ngn3 primer for 3Ј) for the ngn3 tic bud (Fig. 4c). In contrast, the pancreatic epithelium of E13.5 transgenic mice, 5Ј-CAACCCAGGAGCCCCAGTTTC-3Ј (N31C primer for 5Ј) Ipf1/Pdx1–N3IC transgenic embryos was poorly branched and and 5Ј-TCGACCTGCAGGCATGCAAGC-3Ј (vector primer for 3Ј) for the reduced in size (Fig. 4g, h). Furthermore, p48 expression was N31C transgenic mice. absent (Fig. 4i), HES-1 was expressed at a low level (Fig. 4h), and In situ hybridization and immunohistochemistry. In situ hybridization, ngn3-expressing cells were found throughout the epithelium immunohistochemical localization of antigens and double-label immunohis- (Fig. 4j). The reduction in HES-1 and the relative increase in ngn3 tochemistry were carried out as described2. We used DIG-labelled RNA probes expression are consistent with the function of N3IC as a repressor26. of ngn3 (ref. 12), neuroD21 (provided by J. Lee), p48 (ref. 27) (provided by The concomitant loss of p48 expression indicates that high HES-1 O. Hagenbu¨chle), Notch1 (ref. 23), Notch2 (ref. 23), Notch3 (ref. 23), Dll1 expression may be required for p48 expression. At E15.5, the (ref. 24), Serrate1 (ref. 23) and HES-1 (ref. 25). The primary antibodies used were pancreatic epithelia are normally highly branched and lobulated rabbit anti-IPF1/PDX1 (ref. 30), rabbit anti-ISL1 (ref. 17), guinea pig anti-glucagon (Fig. 4d), and differentiated carboxypeptidaseA+ (Fig. 4e) and (Linco), rabbit anti-PYY (specific) (Euro Diagnostics), rabbit anti-amylase amylase+ (Fig. 4f) exocrine cells as well as ISL1+ (Fig. 4g) endocrine (Sigma), rabbit anti-carboxypeptidase A (ANAWA) and rat anti E-cadherin cells have appeared. In E15.5 Ipf1/Pdx1–N3IC transgenic embryos, (Zymed). The secondary antibodies used were fluorescein anti-guinea pig the pancreatic epithelium is poorly branched and the pancreatic (Jackson), fluorescein anti-rat (Jackson) and Cy3 anti-rabbit (Jackson). buds are decreased in size (Fig. 4k and data not shown). In addition, very few carboxypeptidaseA+ (Fig. 4l) and no amylase+ (Fig. 4m) Received 7 May; accepted 8 July 1999. cells are observed, whereas differentiated endocrine cells are 1. Pictet, R. L., Rall, L. B., Phelps, P. & Rutter, W. J. The neural crest and the origin of the insulin- producing and other gastrointestinal hormone producing cells. Science 191, 191–192 (1976). detected throughout the immature pancreatic epithelia, as judged 2. Fontaine, J. & Le Douarin, N. M. Analysis of endoderm formation in the avian blastoderm by the use by the expression of ISL1 (Fig. 4n) and hormones (data not shown). of quail-chick chimaeras. The problem of the neurectodermal origin of the cells of the APUD series. J. Embryol. Exp. Morphol. 41, 209–222 (1977). We conclude that N3IC expression in the developing pancreas 3. Le Douarin, N. M. On the origin of pancreatic endocrine cells. Cell 53, 169–171 (1988). impairs pancreatic epithelial proliferation, morphogenesis and 4. Slack, J. M. W. Developmental biology of the pancreas. Development 121, 1569–1580 (1995). exocrine, but not endocrine, cell differentiation. 5. Edlund, H. Transcribing pancreas. Diabetes 47, 1817–1823 (1998). 6. Lewis, J. Neurogenic genes and vertebrate neurogenesis. Curr. Op. Neurobiol. 6, 3–10 (1996). The results presented here provide new insight into cell lineage in 7. Beatus, P. & Lendahl, U. Notch and neurogenesis. J. Neurosci. Res. 54, 125–136 (1998). the pancreas and show that the Notch signalling pathway is critical 8. Pictet, R. & Rutter, W J. in Handbook of Physiology (eds Steiner, D. F. & Frenkel, N.) 25–66 (Williams and Wilkins, Washington, DC, 1972). in pancreas development. Pancreatic cell fate is altered by genetically 9. Ahlgren, U., Pfaff, S., Jessel, T. M., Edlund, T. & Edlund, H. Independent requirement for ISL1 in the perturbing Notch signalling at multiple steps in the pathway. formation of the pancreatic mesenchyme and islet cells. Nature 385, 257–260 (1997).

© 1999 Macmillan Magazines Ltd 880 NATURE | VOL 400 | 26 AUGUST 1999 | www.nature.com letters to nature

10. Hrabe de Angelis, M., McIntyrell, J. & Gossler, A. Maintenance of somite borders in mice requires the opment. Here we show that the Caenorhabditis elegans gene cam- Delta homologue Dll1. Nature 386, 717–721 (1997). 11. Oka, C. et al. Disruption of the mouse RBPJk gene results in early embryonic death. Development 121, 1 encodes a member of the Ror kinase family that guides migrat- 3291–3301 (1995). ing cells and orients the polarity of asymmetric cell divisions and 12. Sommer, L., Ma, Q. & Anderson, D. J. Neurogenins, a novel family of atonal-related bHLH transcription factors, are putative mammalian neuronal determination genes that reveal progenitor axon outgrowth. We find that tyrosine kinase activity is required cell heterogeneity in the developing CNS and PNS. Mol. Cell. Neurosci. 8, 221–241 (1996). for some of the functions of CAM-1, but not for its role in cell 13. Lardelli, M., Williams, R., Mitsiadis, T. & Lendahl, U. Expression of the Notch 3 intracellular domain migration. CAM-1 is expressed in cells that require its function, in mouse central nervous system progenitor cells is lethal and leads to disturbed neural tube development. Mech. Dev. 59, 177–190 (1996). and acts cell autonomously in migrating neurons. Overexpression 14. Ma, Q., Kintner, C. & Anderson, D. Identification of neurogenin, a vertebrate neuronal determination and loss of cam-1 function result in reciprocal cell-migration gene. Cell 87, 43–52 (1996). 15. Ma, Q., Chen, Z., del Barco Barrantes, I., de la Pompa, J. L. & Anderson, D. J. neurogenin1 is essential phenotypes, indicating that levels of CAM-1 influence the final for the determination of neuronal precursors for proximal cranialsensory ganglia. Neuron 20, 469– positions of migrating cells. Our results raise the possibility that 482 (1998). Ror kinases regulate cell motility and asymmetric cell division in 16. Fode, C. et al. The bHLH protein NEUROGENIN 2 is a determination factor for epibranchial placode-derived sensory neurons. Neuron 20, 483–494 (1998). organisms as diverse as nematodes and mammals. 17. Thor, S., Ericson, J., Brannstrom, T. & Edlund, T. The homeodomain LIM protein Isl-1 is expressed in in cam-1 disrupt the guidance of several cells that subsets of neurons and endocrine cells in the adult rat. Neuron 7, 881–889 (1991). 3 18. Apelqvist, A˚ ., Ahlgren, U. & Edlund, H. Sonic hedgehog directs specialised mesoderm differentiation migrate along the anteroposterior axis in C. elegans . We have found in the intestine and pancreas. Curr. Biol. 7, 801–804 (1997). that cam-1 mutants also display defects in asymmetric cell division 19. Upchurch, B. H., Aponte,, G. W. & Liter, A. B. Peptide YY expression is an early event in colonic and axon outgrowth. Normally, six V cells on each side of first-stage endocrine cell differentiation: evidence from normal and transgenic mice. Development 120, 245–252 (1994). larvae always divide asymmetrically, each producing an anterior 20. Naya, F. J., Stellrecht, C. M. & Tsai, M. J. Tissue-specific regulation of the insulin gene by a novel basic daughter that joins a growing epithelial syncytium and a posterior helix–loop–helix transcription factor. Genes Dev. 9, 1009–1019 (1995). 4,5 21. Lee, J. E. et al. Conversion of Xenopus into neurons by NeuroD, a basic helix–loop–helix protein. blast cell (Fig. 1a, b). In cam-1(gm122) mutant animals, 12% of V1 Science 268, 836–844 (1995). cell divisions are reversed, with the anterior daughter adopting the 22. Naya, F. J. et al. Diabetes, defective pancreatic morphogenesis, and abnormal enteroendocrine blast-cell fate and the posterior daughter joining the syncytium differentiation in BETA2/neuroD-deficient mice. Genes Dev. 11, 2323–2334 (1997). 23. Mitsiadis, T. A., Henrique, D., Thesleff, I. & Lendahl, U. Mouse Serrate-1 expression in the developing (Fig. 1c, d, Table 1). tooth is regulated by epithelial-mesenchymal interactions and fibroblast growth factor-4. Develop- Like V cells, the six Pn.aap neuroblasts, P3.aap through P8.aap, ment 124, 1473–1483 (1997). 24. Mitsiadis, T. A., Hirsinger, E., Lendahl, U. & Goridis, C. Delta-Notch signalling in odontogenesis: divide asymmetrically in first-larval-stage males to generate two 6 correlation with cytodifferentiation and evidence for feedback regulation. Dev. Biol. 204, 420–431 different daughter neurons . The posterior daughters, known as CP (1998). 25. Ishibashi, M. et al. Targeted disruption of mammalian hairy and of split homolog-1 (HES- 1) leads to up-regulation of neural helix–loop–helix factors, premature neurogenesis, and severe neural tube defects. Genes Dev. 15, 3136–3148 (1995). 26. Beatus, P., Lundkvist, J., O¨ berg, C. & Lendahl, U. The Notch 3 intracellular domain represses Notch- 1mediated activation through Hairy/Enhancer of split (HES) promoters. Development 126, 3925– 3935 (1999). 27. Krapp, A. et al. The p48 DNA-binding subunit of transcription factor PTF1 is a new exocrine pancreas-specific basic helix–loop–helix protein. EMBO J. 15, 4317–4329 (1996). 28. Krapp, A. et al. The bHLH protein PTF1-p48 is essential for the formation of the exocrine and the correct spatial organization of the endocrine pancreas. Genes Dev. 12, 3752–3763 (1998). 29. Hogan, B., Constantini, F. & Lacey, E. in Manipulating the Mouse Embryo: A Laboratory Manual (Cold Spring Harbour Lab. Press, New York, 1994). 30. Ohlsson, H., Karlsson, K. & Edlund, T. IPF1, a homeodomain-containing transactivator of the insulin gene. EMBO J. 12, 4251–4259 (1993).

Acknowledgements. We thank K. Falk, U. B. Backman, U. Valtersson and I. Berglund for technical assistance; T. Edlund for critical reading and comments; and members of our laboratories for helpful discussions. This work was supported by grants from the Swedish Medical Research Council and the Juvenile Diabetes Foundation, New York (to H.E.)

Correspondence and requests for materials should be addressed to H.E. (e-mail: Helena.Edlund@ micro. umu.se).

A C. elegans Ror receptor tyrosine kinase regulates cell motility and asymmetric cell division Figure 1 cam-1 defects. Anterior is left and dorsal is up. V-cell daughters were Wayne C. Forrester*, Megan Dell, Elliot Perens visualized by Nomarski optics (a, c). Anterior daughters of the V cells join the & Gian Garriga epithelial syncytium so that only the outline of the nucleus can be seen. The Department of Molecular and Cell Biology, University of California, Berkeley, outline of the posterior daughter cell distinguishes it from its sister. a, Wild-type L1 California 94720, USA larva showing V1 and V2 anterior (thick line) and posterior (thin line) progeny. b, ...... Tracing of in a. c, cam-1(gm122) L1 larva showing reversed V1 cell division. Ror kinases are a family of orphan receptors with tyrosine kinase d, Tracing of animal in c. CA and CP neurons were visualized by staining with anti- activity that are related to muscle specific kinase (MuSK), a serotonin antibodies (e, f). CA and CP neurons from P3.aap and P4.aap are receptor tyrosine kinase that assembles acetylcholine receptors marked. e, him-5(e1490) adult male showing wild-type CA and CP staining at the neuromuscular junction1,2. Although the functions of Ror pattern. The intensely stained neurons (large arrows) are CPs. Each CP axon kinases are unknown, similarities between Ror and MuSK kinases extends posteriorly (arrowheads). Faintly stained neurons (small arrows) just have led to speculation that Ror kinases regulate synaptic devel- anterior to each CP may be CA7. f, cam-1(gm122);him-5(e1490) male showing CP1 extending its axon anteriorly (arrowheads). The faintly staining presumptive CA neuron (arrowhead) is just posterior to CP1. Scale bars in a and e represent 10 ␮m * Present address: Department of Biology, Jordan Hall, Indiana University, Bloomington, Indiana 47405, USA. and 50 ␮m, respectively.

© 1999 Macmillan Magazines Ltd NATURE | VOL 400 | 26 AUGUST 1999 | www.nature.com 881