Distinct roles for GFRa1 and GFRa2 signalling in different cranial parasympathetic ganglia in vivo Jari Rossi, 1 Andreas Tomac, 2 Mart Saarma 1 and Matti S. Airaksinen 1 1 Program in Molecular Neurobiology, Institute of Biotechnology, Viikki Biocentre, FIN-00014 University of Helsinki, Finland 2 Laboratory of Mammalian Genes and Development, National Institutes of Health, Bethesda, USA Keywords: GDNF, lacrimal, mouse, neurturin, otic ganglia, Ret, salivary, sphenopalatine, submandibular Abstract Neurturin (NRTN), signalling via the GDNF family receptor a2 (GFRa2) and Ret tyrosine kinase, has recently been identi®ed as an essential target-derived factor for many parasympathetic neurons. NRTN is expressed in salivary and lacrimal glands, while GFRa2 and Ret are expressed in the corresponding submandibular, otic and sphenopalatine ganglia. Here, we have characterized in more detail the role of GDNF and NRTN signalling in the development of cranial parasympathetic neurons and their target innervation. Gfra1 mRNA was expressed at E12 but not in newborn cranial parasympathetic ganglia, while Gfra2 mRNA and protein were strongly expressed in newborn and adult cranial parasympathetic neurons and their projections, respectively. In newborn GFRa1- or Ret- de®cient mice, where many submandibular ganglion neurons were still present, the otic and sphenopalatine ganglia were completely missing. In contrast, in newborn GFRa2-de®cient mice, most neurons in all these ganglia were present. In these mice, the loss and atrophy of the submandibular and otic neurons were ampli®ed postnatally, accompanied by complete loss of innervation in some target regions and preservation in others. Surprisingly, GFRa2-de®cient sphenopalatine neurons, whose targets were completely uninnervated, were not reduced in number and only slightly atrophied. Thus, GDNF signalling via GFRa1/Ret is essential in the early gangliogenesis of some, but not all, cranial parasympathetic neurons, whereas NRTN signalling through GFRa2/Ret is essential for the development and maintenance of parasympathetic target innervation. These results indicate that GDNF and NRTN have distinct functions in developing parasympathetic neurons, and suggest heterogeneity among and within different parasympathetic ganglia. Introduction In contrast with other peripheral neurons, studies on the development of parasympathetic neurons have been scarce. Classic studies with chick ciliary ganglion demonstrated the target-dependent survival of parasympathetic ciliary neurons (Landmesser & Pilar, 1978). However, studies on developing frog cardiac ganglia found no evidence of naturally occurring cell death: the number of para- sympathetic neurons always matched the growing target size with increased proliferation (Heathcote & Sargent, 1987). This apparent lack of programmed cell death, in contrast with other peripheral neurons that undergo cell death according to the neurotrophic factor hypothesis, was interpreted to re¯ect the proximity of parasympa- thetic neurons and their targets. Yet, organ culture and lesion experiments with rodent salivary glands showed that factors from the target organ are essential for the development and maintenance of parasympathetic ganglion neurons and their target innervation (Coughlin & Rathbone, 1977; Womble & Roper, 1987). GDNF and related neurotrophic factor neurturin (NRTN) signal via a receptor complex consisting of a ligand-binding GPI-anchored GDNF family receptor a (GFRa1 and GFRa2, respectively) and the transmembrane tyrosine kinase Ret (Airaksinen etal., 1999; Baloh etal., 2000). One earlier study has shown that Ret-de®cient mice completely lack some cranial parasympathetic ganglia already at birth (Marcos & Pachnis, 1996). However, it is not known which GDNF family ligands and co-receptors are involved. Chicken ciliary para- sympathetic neurons show differences and developmental changes in Gfra1 and Gfra2 expression and in responsiveness to GDNF and NRTN in vitro (Forgie etal., 1999; Hashino etal., 1999). Recent gene- targeting experiments have identi®ed NRTN signalling via the GFRa2 receptor as essential target-derived factor for many, but not all parasympathetic neurons (Heuckeroth etal., 1999; Rossi etal., 1999). NRTN is expressed in salivary and lacrimal glands, while GFRa2and Ret are expressed in the ganglia that innervate these glands. Mice lacking GFRa2 and NRTN show similar parasympathetic de®cits. However, the loss of neurons and target innervation among different cranialparasympatheticgangliaappearunevenandhavenotbeenfully characterized. Moreover, the in vivo role of GDNF and GFRa1onthe development of parasympathetic neurons has also not been studied. Here we have compared the development of the neuronal de®cit in cranial parasympathetic ganglia in mice lacking GFRa1, GFRa2 or the common signalling receptor Ret. We show that GFRa1/Ret signall- ing is essential for early gangliogenesis of otic and sphenopalatine neurons. In contrast, these neurons require NRTN/GFRa2 signalling later for target innervation. Finally, many submandibular neurons do not require GDNF family signalling during embryogenesis but do need GFRa2/Ret signalling for survival during target innervation. Materials and methods Histology Mice lacking Ret, GFRa1 or GFRa2, as well as their wild-type littermates in the original hybrid backgrounds, were genotyped as previously described (Durbec etal., 1996; Rossi etal., 1999; Tomac Correspondence: Dr Matti S. Airaksinen, as above. E-mail: matti.airaksinen@helsinki.® Received 29 June 2000, revised 15 August 2000, accepted 22 August 2000 European Journal of Neuroscience, Vol. 12, pp. 3944±3952, 2000 Ó Federation of European Neuroscience Societies