bb10e03.qxd 03/13/2000 01:44 Page R180

R180 Dispatch

Hedgehog signalling: How modulates the signal Philip W. Ingham

By coupling to cholesterol, Hedgehog can be anchored of this tight association with the membrane seem to fit well to the cells where it is made, yet to act as a morphogen, with some of the properties of the signal. In the Drosophila it must be able to move away from its source. Novel embryo, for instance, target such as wingless (wg) and genes have now been identified that control the release patched (ptc) are activated only in cells immediately adjacent and dispersal of Hedgehog, shedding new light on the to those expressing hh [6], consistent with the signalling part played by cholesterol in these processes. activity being tethered to the surface of the secreting cell.

Address: MRC Intercellular Signalling Group, Developmental Genetics Programme, The Krebs Institute, University of Sheffield, Firth Court, But a growing body of evidence suggests that Hh Western Bank, Sheffield S10 2TN, UK. must also be capable of acting at some distance from their E-mail: [email protected] source. Studies in the chick have shown that Sonic hedgehog (Shh) acts in a concentration-dependent manner Current Biology 2000, 10:R180–R183 to regulate cell fate in the neural tube [7], implying that 0960-9822/00/$ – see front matter the spreads over many cell diameters, while in © 2000 Elsevier Science Ltd. All rights reserved. Drosophila similar conclusions have been drawn from studies of the wing [8,9] and abdomen [10,11] of the adult Developing embryos are replete with a variety of potent fly. In the Drosophila imaginal disc, the precursor of the signalling proteins, the activities of which must be adult wing — where the mechanism of Hh signalling has precisely regulated for development to follow its proper been analysed in greatest detail — hh expression is course. In principle, such signals may take two distinct restricted to cells of the posterior lineage compartment. Hh forms. Proteins tethered to the membrane of a signalling protein secreted by these cells passes across the cell — for instance via a transmembrane domain or lipid compartment boundary and modulates the expression of modification — might be expected to act as short-range various genes in anterior compartment cells in a concentra- switches, influencing the behaviour of their nearest neigh- tion-dependent manner [8,9] (Figure 1). Although Hh View metadata, citationbours and insimilar an essentially papers at core.ac.uk binary fashion. By contrast, proteins protein is only clearly visible in cells closest to the bound- brought to you by CORE

that are secreted by expressing cells could act over rela- ary, functional data imply that the HhN signalling form provided by Elsevier - Publisher Connector tively long distances, with the potential to influence cell must travel over at least six to eight cell diameters. fate in a dose-dependent manner. In practice, however, some signalling proteins have been found to possess But if HhN is tethered to lipid rafts in hh expressing cells, characteristics of both of these idealised forms. how can it escape these cells to generate the inferred graded distribution? To address this fundamentally cell bio- Members of the Hedgehog (Hh) family of signalling logical problem, Basler and colleagues [1] adopted a genetic proteins are a case in point: their activity can be detected approach. They reasoned that mutations in genes required several cell diameters away from the cells in which they are for the release of HhN should result in phenotypes similar expressed, yet the proteins are covalently coupled to cho- to those caused by a reduction in hh activity; and as the lesterol, a modification that results in their tight association Drosophila genome has effectively been saturated for purely with the cell surface, at least when expressed in tissue zygotically-acting genes, they exploited a strategy previ- culture cells. How these apparently conflicting properties ously developed by Perrimon et al. [12] to identify genes can be reconciled has been the subject of some debate that are expressed and required both zygotically and mater- over the past three years. Now, studies of two newly nally. In this way, they isolated the disp mutant, which identified Drosophila mutants, dispatched (disp) [1] and when homozygous in both the zygote and in the germline tout velu (ttv) [2,3], have provided novel insights into the of its female parent, results in an embryonic lethal pheno- control of the release and movement of lipid-modified Hh. type identical to that of an hh mutant.

Coupling to cholesterol occurs during the autoproteolytic Using clonal analysis to remove disp selectively from cells cleavage of full-length Hh, a process that yields a mature, in the wing imaginal disc, Burke et al. [1] found that expres- amino-terminally-derived signalling form of the protein, sion of disp, like that of hh, is completely dispensable in designated HhN [4]. Recent studies by Eaton and cells of the anterior compartment, but that it is critically colleagues [5] revealed that, in Drosophila cells, this choles- required in the Hh-secreting cells of the posterior compart- terol-coupled HhN accumulates in lipid microdomains ment. Strikingly, when such cells lack disp function they similar to the cholesterol-rich ‘rafts’ previously described in accumulate high levels of normally processed HhN but fail mammalian cells. At first sight, the expected consequences to release any of this protein (Figure 1). As a consequence, bb10e03.qxd 03/13/2000 01:44 Page R181

Dispatch R181

the up-regulation of ptc and other Hh target genes in neigh- Figure 1 bouring anterior compartment cells is compromised, and the normal growth and patterning of the wing is disrupted.

As processing of the Hh protein occurs normally in disp mutant cells, the simplest explanation of these findings is that disp is required to release cholesterol-coupled HhN from the membrane of secreting cells. In line with this sug- gestion, Burke et al. [1] found by expressing a truncated cDNA that secretion of an unprocessed form of HhN, HhNu, that consists of just the amino-terminal portion of the full-length Hh protein and is therefore not coupled to cholesterol, is completely independent of disp activity. Moreover, the function of disp seems to be specific for cho- lesterol modification, as HhN tethered to the membrane by a glycosylphosphatidylinositol (GPI) linkage is not released in the presence of Disp activity.

How, then, might Disp accomplish this highly specific release of cholesterol-linked HhN from the lipid rafts of secreting cells? One simple mechanism might be by cleav- ing the cholesterol moiety from the protein. But there are Distribution of Hh protein and target expression in imaginal disc various reasons for believing that such a cleavage event cells. Hh protein (blue) is made and processed in posterior compartment cells, from where HhN moves across the compartment boundary does not occur. Principal amongst these is the fact that (vertical line) into anterior compartment cells. In wild-type (+) discs, the endogenous HhN and transgenically-produced HhNu protein forms a gradient of activity which is presumed to reflect the behave in fundamentally different ways once they are distribution of the protein (progressively lighter blue shading) and which released from the same posterior compartment cells. activates the expression of different target genes; for simplicity, only one Whereas HhNu spreads unhindered across the anterior such gene, ptc, is illustrated (red shading). In contrast to the cholesterol-coupled form, HhNu passes freely throughout the anterior compartment, activating Hh target genes as it goes [1], compartment, activating target genes in all cells. Inactivation of disp in endogenous HhN induces an ordered pattern of target the posterior compartment has a dramatic effect on the distribution of , reflecting its graded distribution away HhN: the protein accumulates to higher than normal levels in expressing form the compartment boundary. cells (darker shading), but no HhN is observed in anterior compartment cells. There is weak activation of ptc in cells closest to the boundary, but whether this is mediated by membrane-anchored HhN or via disp- An indication as to how this distribution is controlled came independent ‘leakage’ of HhN from the secreting cells is unclear. But with the discovery [9] that the Hh receptor subunit loss of disp activity has no effect on HhNu distribution, indicating that it encoded by ptc actively sequesters endogenous HhN, but is specifically involved in release of the cholesterol-coupled form. Absence of ptc from anterior cells results in the free movement of not HhNu. This finding strongly suggests that the choles- cholesterol-coupled HhN across anterior cells; it also results in the terol moiety is essential for the interaction between HhN independent activation of target genes, as shown. Removal of ttv from and Ptc (though it should be noted that unmodified HhN anterior compartment cells, by contrast, blocks the movement of HhN can bind to Ptc in vitro and indeed must interact in some across anterior cells. HhN is undetectable even in the cells abutting the compartment boundary, though, as when disp activity is absent from way with Ptc in vivo to activate the Hh pathway); and if posterior cells, these cells do express ptc at reduced levels. this is the case, then HhN must remain coupled to choles- terol even when released from secreting cells by Disp. This property fits well with the finding that Ptc contains a proteins are predicted to have multiple membrane-span- so-called sterol-sensing domain, a region of the protein that ning domains, five of which show striking similarity to the may directly mediate its interaction with cholesterol [13]. sterol-sensing domains of HMGCoA reductase and SCAP.

The sterol-sensing domain was originally identified in two Intriguingly, when Burke et al. [1] cloned and sequenced proteins involved in cholesterol homeostasis: HMGCoA disp, they found that it too encodes a multipass trans- reductase and SCAP, for ‘SREBP cleavage activating membrane protein with significant sequence similarity to protein’, where SREBP stands for ‘sterol regulatory both Ptc and NPCI. Like these latter proteins, Disp has a element binding protein’. The presence of the sterol- sterol-sensing domain, suggesting that it also functions by sensing domain in Ptc was revealed by the discovery that binding the cholesterol moiety of HhN [1]. In this view, Ptc is homologous to the NPC1 protein [13], mutant forms Disp would act to displace HhN from lipid rafts in secret- of which are associated with defective cholesterol transport ing cells, facilitating release of the protein from the mem- in patients suffering from Niemann–Pick syndrome. Both brane (Figure 2). Interaction with Ptc on the membrane bb10e03.qxd 03/13/2000 01:44 Page R182

R182 Current Biology Vol 10 No 5

Figure 2 embryos phenotypically indistinguishable from those lacking Hh activity [3]. By contrast to disp (and hh), however, Bellaiche et al. [2] found that ttv activity is required within cells of the anterior compartment of the wing imaginal disc, both for the efficient activation of Hh target genes and for the movement of cholesterol-coupled HhN — but not HhNu [3] — between them (Figure 1).

One way of interpreting these findings, as suggested by Burke et al. [1], is to postulate that the proteoglycan synthesised by Ttv [3] competes with Ptc for binding of cholesterol-coupled HhN. In this view, after HhN is released from Disp its binding to the proteoglycan would prevent its sequestration by Ptc, and hence facilitate its transfer to more distantly located cells. Such a model, however, cannot account for the finding that ttv is still required for HhN movement in the absence of ptc expres- sion [2], nor the fact that, in the absence of Ttv, HhN fails to accumulate even in anterior cells immediately adjacent to the compartment boundary [2]. Taken together, these two findings suggest that Ttv may also be required to Possible scheme for the release and uptake of cholesterol-coupled release HhN from secreting cells. As Burke et al. [1] point HhN mediated by Disp, Ptc and Ttv. A section of the membrane of a out, release of HhN from secreting cells is likely to require HhN-secreting cell (right) is shown containing cholesterol-modified HhN in domains. HhN is postulated to be displaced from rafts a cofactor that can displace it from Disp: a proteoglycan by binding to the sterol-sensing domain (blue) in the Disp multipass produced by Ttv that is membrane tethered and acts in transmembrane protein (green). According to this scheme, HhN is trans but not in cis would seem to have all the desired released from Disp by interacting with a ttv-dependent membrane properties of such a cofactor (Figure 2) . Clarification of tethered proteoglycan (orange) on adjacent receiving cells. Once this and other aspects of this fascinating process will released from the membrane of secreting cells, HhN may remain associated with the proteoglycan, allowing it to pass to more distantly doubtless emerge in the coming months! located cells, or it may interact with Ptc (red). Interaction of the cholesterol moiety with the sterol-sensing domain (blue) of Ptc may Acknowledgements result in its reinsertion into lipid rafts, in a reverse of the Anita Taylor and Helen Skaer made helpful comments on the manuscript. The author is supported by the Wellcome Trust and the MRC. Disp-dependent process. Alternatively, Ptc may sequester HhN by endocytosis of the Ptc–HhN complex (not shown). References 1. Burke R, Nellen D, Bellotto M, Hafen E, Senti KA, Dickson BJ, Basler K: Dispatched, a novel sterol-sensing domain protein dedicated to of neighbouring cells could reverse this process, with Ptc the release of cholesterol-modified hedgehog from signaling cells. Cell 1999, 99:803-815. playing the opposite role to Disp, promoting the incorpo- 2. Bellaiche Y, The I, Perrimon N: tout-velu is a Drosophila homologue ration of HhN into microdomains of receiving cells. In of the putative tumour suppressor EXT-1 and is needed for Hh diffusion. Nature 1998, 394:85-88. line with this scenario, immunohistochemical analysis 3. The I, Bellaiche Y, Perrimon N: Hedgehog movement is regulated has previously identified distinct basolateral accumula- through tout velu-dependent synthesis of a heparan sulfate tions of HhN that appear to form in both secreting and proteoglycan. Mol Cell 1999, 4:633-639. 4. Porter J, Young K, Beachy P: Cholesterol modification of receiving cells of the polarised epithelia of Drosophila hedgehog signaling proteins in animal development. Science embryos [14,15]. While it is currently unclear whether 1996, 274:255-259. these accumulations correspond to lipid rafts, it is strik- 5. Rietveld A, Neutz S, Simons K, Eaton S: Association of sterol- and glycosylphosphatidylinositol-linked proteins with Drosophila raft ing that HhNu is not similarly localised. lipid microdomains. J Biol Chem 1999, 274:12049-12054. 6. Ingham PW: Localised hedgehog activity controls spatially restricted transcription of wingless in the Drosophila embryo. Nature 1993, The existence of such transfer of HhN between adjacent 366:560-562. cells raises the question of how HhN moves beyond these 7. Briscoe J, Sussel L, Serup P, Hartigan-O’Connor D, Jessell TM, cells to form the gradient of activity observed in imaginal Rubenstein JL, Ericson J: Homeobox gene Nkx2.2 and specification of neuronal identity by graded Sonic hedgehog signalling. Nature discs. Important insights into this process have come from 1999, 398:622-627. the recent functional analyses [2,3] of tout velu (ttv), a 8. Strigini M, Cohen S: A hedgehog activity gradient contributes to Drosophila homologue of a family of human genes coding ap axial patterning of the Drosophila wing. Development 1997, 124:4697-4705. for glycosylaminoglycan (GAG) transferases, enzymes of 9. Chen Y, Struhl G: Dual roles for patched in sequestering and proteoglycan synthesis. Mutations in ttv, like those in disp, transducing hedgehog. Cell 1996, 87:553-563. 10. Struhl G, Barbash D, Lawrence P: Hedgehog acts by distinct gradient were discovered because elimination of its activity from and signal relay mechanisms to organize cell-type and cell polarity both the female germline and the zygote results in in the Drosophila abdomen. Development 1997, 124:2155-2165. bb10e03.qxd 03/13/2000 01:44 Page R183

Dispatch R183

11. Lawrence P, Casal J, Struhl G: The hedgehog morphogen and gradients of cell affinity in the abdomen of Drosophila. Development 1999, 126:2441-2449. 12. Perrimon N, Lanjuin A, Arnold C, Noll E: Zygotic lethal mutations If you found this dispatch interesting, you might also want with maternal effect phenotypes in Drosophila melanogaster. II. to read the August 1999 issue of Loci on the second and third identified by P-element-induced mutations. Genetics 1996, 144:1681-1692. 13. Beachy PA, Cooper MK, Young KE, von Kessler DP, Park WJ, Hall Current Opinion in TM, Leahy DJ, Porter JA: Multiple roles of cholesterol in hedgehog protein biogenesis and signaling. Cold Spring Harb Symp Quant Genetics & Development Biol 1997, 62:191-204. 14. Taylor AM, Nakano Y, Mohler J, Ingham PW: Contrasting distributions which included the following reviews, edited of patched and hedgehog proteins in the Drosophila embryo. Mech Dev 1993, 43:89-96. by Norbert Perrimon and Claudio Stern, on 15. Tabata T, Kornberg TB: Hh is a signalling protein with a key role in Pattern formation and developmental patterning Drosophila imaginal discs. Cell 1994, 76:89-102. mechanisms:

Cell polarity in the early Caenorhabditis elegans embryo Bruce Bowerman and Christopher A Shelton The polarisation of the anterior–posterior and dorsal–ventral axes during Drosophila oogenesis Fredericus van Eeden and Daniel St Johnston Wnt signaling and dorso-ventral axis specification in vertebrates Sergei Y Sokol Establishment of anterior–posterior polarity in avian embryos Rosemary F Bachvarova Polarity in early mammalian development Richard L Gardner Diverse initiation in a conserved left–right pathway? H Joseph Yost Extracellular modulation of the Hedgehog, Wnt and TGF-β signalling pathways during embryonic development Javier Capdevila and Juan Carlos Izpisúa Belmonte Fringe, Notch, and making developmental boundaries Kenneth D Irvine Polarity determination in the Drosophila eye Helen Strutt and David Strutt Wnt signalling: pathway or network? Alfonso Martinez Arias, Anthony MC Brown and Keith Brennan Epithelial cell movements and interactions in limb, neural crest and vasculature Cheryll Tickle and Muriel Altabef Cell movements in the sea urchin embryo Charles A Ettensohn Roles of the JNK signaling pathway in Drosophila morphogenesis Stéphane Noselli and François Agnès Cell migration in Drosophila Alexandria Forbes and Ruth Lehmann Cell migration and axon growth cone guidance in Caenorhabditis elegans Catherine S Branda and Michael J Stern The full text of Current Opinion in Genetics & Development is in the BioMedNet library at http://BioMedNet.com/cbiology/gen