BIOLOGICAL OVERVIEW

In Drosophila each leg and wing primordium is subdivided into two cell populations: the anterior (A) and posterior (P) compartments. Cells in the P compartment are programmed by the selector gene engrailed (en) to secrete hedgehog (Hh). en is not active in A compartment cells, and as a consequence, these cells are competent to transduce the Hh signal. Dispatched (Disp), a protein with sequence similarity to Patched, is required for the release of Hedgehog from P cells. In the absence of Disp, Hh is retained in P cells, and Hh access to A cells is severely limited. These results indicate that Hedgehog is a tethered protein but that its retention is overcome by the activity of Disp, which is dedicated to the release of Hh (Burke, 1999).

Mature Hh is a modified protein, bearing a cholesterol moiety at its C-terminus. Hh precursor undergoes an autoproteolytic cleavage reaction to give rise to its active N-terminal portion (Lee, 1994 and Porter, 1995). This cleavage is accompanied by the covalent bonding of a cholesterol to the C terminus of this N-terminal portion, producing the active Hh, termed Hh-Np (where 'p' stands for 'processed'). Hh protein derived from a transgene that encodes only the N-terminal portion does not undergo cleavage and is consequently not linked to cholesterol. When this unmodified protein, Hh-Nu (where 'u' stands for 'unmodified'), is expressed in the Drosophila embryo, a broader than normal range of Hh action is observed (Porter, 1996a). One interpretation of these results is that Hh-Nu can move further than Hh-Np due to the absence of the cholesterol modification. Intriguingly, Ptc also contains a sterol-sensing domain (SSD), which has been shown in proteins such as HMG CoA reductase and SREBP cleavage-activating protein to be able to monitor sterol levels in membranes. One possibility is that Ptc interacts directly with the cholesterol moiety of Hh-Np via its SSD, thus sequestering Hh and restricting its motility (Burke, 1999 and references therein).

dispatched was identified in a genetic screen for components of the Drosophila Hh signaling pathway. A mutation on the third chromosome was identified, causing segment polarity phenotypes typical of those resulting from loss of hh or wg function. Animals zygotically homozygous for this mutation survive until early pupal stages. However, embryos that also lack the maternal component of this locus die during embryogenesis with a strong segment-polarity phenotype. Instead of the wild-type segmentally repeated pattern of denticle belts interspersed by naked cuticle, such embryos display a lawn of denticle belts and fail to secrete naked ventral cuticle. Germline clone-derived embryos are rescued by a wild-type paternal chromosome, indicating that the gene product is required only after the onset of zygotic transcription. Segment-polarity phenotypes are indicative of loss-of-function mutations in essential components of the Hh and Wg signal transduction pathways. Analysis of the disp mutant phenotype in the wing, in which Hh and Wg pathways function in different anatomical locations, indicates that disp mutation interfers with Hh and not Wg signaling (Burke, 1999).

Disp is necessary for Hh signaling in Hh-producing cells. The effect of disp mutant clones on Hh signaling was examined in the wing imaginal disc, where the A/P boundary can be precisely defined and where the transcription of the Hh target genes ptc and decapentaplegic (dpp) serve as immediate readouts of Hh signaling activity. Even large clones of disp^-/- A cells abutting the A/P boundary have no discernible effect on ptc-lacZ or dpp-lacZ expression. Thus, despite its Ptc-like structure, Disp plays no role in transducing the Hh signal in responding A cells. In contrast, large clones of disp^-/- P cells abutting the A/P boundary cause a dramatic reduction of both ptc-lacZ and dpp-lacZ expression. Since this requirement is very similar to that for hh itself, this result is interpreted as evidence that Disp is essential for the effective production of the Hh signal in P cells. Even a small patch of disp^+/- cells at the A/P boundary is sufficient to locally rescue Hh signaling, impressively demonstrating the potency of the Hh signal and its requirement for disp activity. One obvious explanation for the phenotypes associated with disp mutant cells would be an involvement of Disp in the expression of the hh gene itself. However, it has been found that hh-lacZ expression is unaffected in disp^-/- clones, which rules out a requirement for Disp in hh transcription (Burke, 1999).

These experiments show that Disp is necessary for Hh signaling in Hh-producing cells. They do not address, however, whether the ubiquitously expressed Disp protein plays a role in other signaling pathways or in physiological processes. To investigate this issue, animals were generated in which disp expression was restricted to Hh-secreting cells. This was achieved by introducing a UAS-disp transgene together with a P cell-specific en-Gal4 driver into a disp mutant background. The en-Gal4 driver is inactive in A compartment cells, which do not secrete Hh. en-Gal4 is also not active in eye imaginal disc cells, which do, however, secrete Hh. The en-Gal4 UAS-disp transgene combination rescues disp mutant animals to adulthood. The resulting flies display normal patterning in the wing, leg, notum, and abdomen, and give rise to viable offspring, which demonstrates that in larval and embryonic tissues, disp function is only required in Hh-producing cells. However, these rescued animals show a dramatic reduction in eye size, which indicates that Disp is also required for Hh signaling in the eye, and that the rescue observed in other tissues is due solely to disp⁺ transcripts provided by the en-Gal4 driver. Importantly, throughout all stages of development, A compartment cells develop and differentiate normally in the complete absence of functional Disp protein and become correctly patterned by numerous signaling molecules other than Hh. Thus, despite its ubiquitous expression, Disp is required exclusively for Hh signaling, and not for other known signaling pathways, nor for sterol homeostasis or membrane integrity (Burke, 1999).

Because disp is required in Hh-producing cells for Hh signaling, but not for hh transcription, an examination was made of whether Disp is required for the processing of Hh into the active signaling moiety, Hh-Np. This processing event involves the autocatalytic cleavage of full-length Hh precursor protein to the N-terminal portion Hh-N (Lee, 1994; Porter, 1995), with the concomitant covalent linkage of cholesterol to the C-terminal amino acid to form Hh-Np (Porter, 1996b). This cleavage event was assayed by Western blot analysis. Transgenes encoding either full-length hh cDNA (hh-F^HA) or only the N-terminal portion of Hh (hh-N^HA) were expressed under en-Gal4 control in imaginal discs. Each of these constructs was tagged with an HA epitope just N-terminal to the defined cleavage site to allow protein detection with an alpha-HA antibody (Burke, 1999).

In lysates of wild-type larvae expressing hh-F^HA, two prominent bands of ~50 kDa and ~30 kDa were observed that are absent in lysates from control animals. These two proteins correspond to unprocessed full-length Hh and processed Hh-Np, respectively. In lysates from animals expressing tagged hh-N^HA, only a single major protein species of ~30 kDa was detected that comigrates with the smaller protein seen from animals expressing hh-F^HA, confirming that this smaller band is the result of internal cleavage of the hh-F^HA product. When hh-F^HA was expressed in disp mutant animals, the same ratio of full-length Hh to Hh-Np was observed, indicating that Hh cleavage is occurring at the same efficiency in disp mutant cells. From this result it has been concluded that the defect in Hh signaling imposed by the lack of Disp is not due to faulty cleavage of the Hh precursor protein. Since the covalent addition of cholesterol is coupled to the cleavage reaction, which occurs normally in disp mutant cells, it is assumed that Hh-N is properly modified in the absence of Disp. In support of this assumption it should be noted that Hh lacking a C-terminal cholesterol moiety would produce an increased, rather than a decreased, spatial response to Hh. Hence, the possibility that Disp is required for cleavage and cholesterol modification of Hh is dismissed (Burke, 1999).

Two key findings are sited that pinpoint the role of Disp: the observation that disp mutant cells retain rather than secrete Hh; and the demonstration that Hh-Nu, the cholesterol-free variant of Hh-Np, bypasses the requirement for Disp. From these two results, it is concluded that the normal function of Disp is to liberate the cholesterol-tethered form of Hh from internal or surface membranes of producing cells. An investigation was carried out to see whether the distribution of Hh protein is altered in the absence of Disp. Wild-type Hh protein normally accumulates in intracellular punctate structures in A cells near the A/P border. These accumulations of Hh antigen colocalize with punctate Ptc staining, suggesting they might reflect vesicular signaling complexes. When disp minus discs are stained with Hh antisera, no Hh staining at all is observed in A cells, whereas staining in P cells is significantly higher than in wild-type discs. To confirm this increase in Hh levels, marked disp minus clones were generated and the distribution of Hh antigen in single discs was analyzed. Strong accumulation of Hh levels in disp minus P cells is observed in comparison to neighboring wild-type P cells. Together, these results indicate that in the absence of Disp, Hh is predominantly retained in producing cells and is thus unable to move in significant quantities to A cells. Since some weak Ptc expression is still observed in A cells of disp mutant discs, a small fraction of Hh protein must be escaping, but in vastly reduced quantities, below the limits of detection (Burke, 1999).

It was then asked whether the retention of Hh in disp mutant tissue might reflect defects in the intracellular trafficking of Hh protein. This possibility was raised by the observation that in embryonic epidermal cells, unprocessed Hh (Hh-Nu) is mainly apical while processed Hh (Hh-Hp) is predominantly basolateral; this suggests a role for the cholesterol modification in sorting. Using Hh antisera, no specific localization along the apical/basal axis of wing imaginal disc cells could be detected, and no alteration in Hh distribution in disp minus compared to wild-type tissue was observed. To examine whether the different isoforms might nevertheless be differently distributed in disp mutant cells, the surface distribution of Hh-F^HA and Hh-N^HA was examined in wild-type and disp mutant tissue. In these experiments, the antibody was applied prior to fixation and permeabilization in order to visualize only cell surface antigen. In both wild-type and disp mutant tissues, Hh-F^HA is detected on both the basal and apical surfaces, while Hh-N^HA is exclusively apical. Due to the difficulty in accurately quantitating levels of cell surface staining, it could not be determined if the accumulation of Hh seen within disp mutant cells also occurs at the cell surface. It is concluded, however, that Hh is still able to reach the surface of cells lacking Disp, and although the possibility that Disp is required to differentially sort some small, active fraction of total Hh protein, these results argue against a role for Disp in apical/basal sorting of Hh (Burke, 1999).

One candidate effector for the retention of Hh in disp mutant cells is the cholesterol moiety, which could conceivably tether Hh to the membranes of producing cells. This lipid modification has been proposed to restrict the range of Hh action, since expression of Hh-Nu results in a spatially extended Hh response in embryos (Porter, 1996a). Before assaying the relationship between the cholesterol modification of Hh and the function of Disp, the role of this modification was further clarified by establishing (1) that in the absence of modification, Hh-Nu possesses a vastly extended range of action in imaginal discs; (2) that this extended range of action is an intrinsic property of Hh-Nu, and (3) that cholesterol-free Hh-Nu is apparently not subject to sequestration by Ptc, yet retains the ability to form vesicular complexes with Ptc in receiving cells (Burke, 1999).

Expression of Hh-Nu in P cells of the wing imaginal disc results in dpp-lacZ expression in the entire A compartment of the disc and a consequent dramatic enlargement of the A compartment. Thus, Hh-Nu appears to have a range of action at least 5-fold larger than that of wild-type Hh. However, since this and previous experiments have been performed in the presence of endogenous Hh, it could not be ruled out that the observed extension of Hh activity depends on, or is even mediated by, endogenous Hh-Np whose range might be expanded in the presence of Hh-Nu. To address this, a situation was created in which Hh-Nu is the sole source of Hh in imaginal discs by expressing hh-N^HA in P cells under en-Gal4 in hh^ts2/ts2 animals that were shifted to the nonpermissive temperature during early larval stages. Even in the absence of endogenous Hh, Hh-Nu is still capable of inducing the same expanded anterior compartment morphology and shows normal punctate staining in anterior cells. Thus, Hh-Nu alone is able to associate with Ptc and signal in vivo. The cholesterol anchor of Hh appears to be required for the sequestration of Hh by Ptc, since untethered Hh is seemingly unrestricted in its range. The possibility cannot currently rule out that Hh-Nu is also, at least partially, sequestered by Ptc but that extracellular Hh-Nu levels are abnormally high and saturate the capacity of Ptc. Any sequestration of Hh-Nu must, however, be much less efficient than that of Hh-Np, since even discs containing endogenous Hh plus en-Gal4 driven Hh-Np do not show the dramatic effect caused by Hh-Nu alone (Burke, 1999).

Having confirmed that the cholesterol modification is needed for efficient Hh sequestration but not signaling, it was of interest to determine if the retention of Hh in posterior disp mutant cells is due to the lipid anchor. The hh^ts2/ts2 hh-N^HA experiments were repeated in a disp mutant background so that the Hh-Nu-secreting P cells were simultaneously lacking endogenous Hh and Disp. The patterning activity of cholesterol-free Hh-Nu is virtually unaffected by the lack of Disp. Hh-Nu causes the same 'extended anterior compartment phenotype' in hh^ts2/ts2 disp^-/- double mutant discs as it causes in the hh^ts2/ts2 single mutant background. Also, equivalent levels of ptc-lacZ expression are induced by Hh-Nu in hh^ts2/ts2 disp minus discs and in hh^ts2/ts2 disp plus discs, and punctate Hh staining is again observed in anterior cells. Thus, unlike Hh-Np, cholesterol-free Hh is neither retained nor compromised in its range of action if produced by disp mutant cells. Since the sole known structural difference between Hh-Np and Hh-Nu is the C-terminal cholesterol moiety, it has been concluded that it is this lipid anchor that is responsible for the retention of Hh-Np protein by disp mutant cells. From this it has been inferred that the function of Disp is to overcome this retention and thereby permit the release of lipid-modified Hh protein from Hh-producing cells (Burke, 1999).

Having established that the activity of Disp permits the release of tethered Hh protein, the specificity of this release mechanism was addressed by asking two questions: (1) is Hh activity also dependent on Disp if Hh is tethered by a nonlipid anchor? And (2), does Disp also liberate Hh protein if Hh is tethered by a lipid anchor other than cholesterol? To address the first question, a fusion protein (Hh-CD2) was used, in which the signaling domain of Hh is fused to the N terminus of the type I transmembrane protein CD2. This derivative of Hh is effectively tethered to expressing cells and it retains biological activity even in the absence of endogenous Hh. hh-CD2 was expressed under en-Gal4 control in a disp minus hh^ts2/ts2 mutant background and Hh-CD2's activity was found not to depend on the presence of disp. It is therefore concluded that Hh protein with a nonlipid tether, like Hh protein with no tether (Hh-Nu), functions independently of Disp (Burke, 1999).

Finally, it was asked if addition of lipids other than cholesterol would also tether Hh signaling activity, and whether such tethering could be overcome by Disp. A form of Hh-N (Hh-GPI) was generated that carries the C-terminal 54 residues of Drosophila Fasciclin I (Fas1), including the glycosyl-phosphatidylinositol (GPI)-anchoring signal of Fas1. As a control, a derivative of Hh-GPI (Hh-DeltaGPI) was used in which the GPI-anchoring signal was mutated. When Hh-DeltaGPI is expressed in marked clones of wing imaginal disc cells, ubiquitous expression of dpp-lacZ is observed in the entire A compartment, which is extended in size. This phenotype is the same as that of unprocessed Hedgehog (Hh-Nu), and indicates that the addition of heterologous sequences does not compromise the long-range signaling activity of Hh-Nu. In sharp contrast, expression of Hh-GPI induces ectopic dpp-lacZ expression only in Hh-GPI-expressing cells and in their immediate wild-type neighbors. Conversely, wild-type Hh in the same assay induces dpp-lacZ in wild-type cells up to five or more cell diameters away. Thus, the GPI moiety effectively tethers Hh to the surface of expressing cells. Disp, which is present and active in these cells, can not liberate Hh-GPI as it does Hh-Np, indicating that cholesterol is an important determinant of the Disp-dependent release mechanism of tethered Hh (Burke, 1999).

Hh is an extremely powerful signaling molecule. When unrestricted, it is able to move to and program cells far away from its origin of synthesis. Two mechanisms have evolved to restrict its range of action: (1) upregulation of ptc transcription with subsequent sequestration of Hh ligand by Ptc protein and (2) the employment of a lipid tether. Intriguingly, it appears that to function efficiently, the two mechanisms depend on each other. For example, if P cells secrete Hh protein that lacks the cholesterol tether, this Hh protein reaches the entire population of A compartment cells despite the presence of Ptc protein at the boundary. Conversely, in the absence of Ptc, the cholesterol moiety does not function as an anchor, since under such conditions even wild-type Hh protein can reach over impressively large distances. Why are these two mechanisms interdependent? The answer is not known, but it is intriguing that both the lipid tether of Hh as well as the SSD of Ptc point to the involvement of cholesterol as a common link (Burke, 1999).

Lipid modifications are employed to tether proteins to cellular membranes. In many instances these modifications fulfill the function of a posttranslationally added transmembrane domain. The use of a cholesterol tether for a signaling molecule that has to act at a distance is unprecedented. Since its discovery, it has posed a paradox: how can Hh be released from cells if it is covalently attached to a cholesterol moiety that is presumably inserted in lipid bilayers? Several different scenarios can be envisaged. Either Hh-N is liberated from its anchor by a cleavage event involving C-terminal proteolysis or anchor hydrolysis, or Hh-Np is released by displacing the cholesterol anchor from the lipid membrane or by the formation of extracellular microparticles through membrane vesiculation. Biochemical analyses in cultured cells and the properties of Hh in vivo argue against a cleavage event. The vast majority of Hh protein expressed in cultured cells is cell associated and not soluble. Moreover, two forms of Hh that carry a membrane anchor other than cholesterol—Hh-CD2 with a transmembrane domain and Hh-GPI with a glycosyl-phosphatidylinositol moiety—are not released from the cell surface, ruling out proteolysis as the mechanism of Hh release. The mechanism of release is likely to involve the function of Disp in the displacement of Hh's cholesterol tether (Burke, 1999).

The mechanism of Hh release does not operate effectively in the absence of Disp. These results reveal the existence of a specific pathway by which Hh overcomes its cholesterol impediment. Unexpectedly, the molecular analysis of the disp gene indicates that this pathway of Hh release—like that of Hh restriction—involves an sterol-sensing domain (SSD) protein. However, although Ptc and Disp are structurally similar, the two proteins play opposite roles. Ptc is required in A cells to sequester Hh-Np, and Disp is required in P cells to displace Hh-Np. In spite of these seemingly inverse roles, an intriguing parallel exists between Ptc and Disp: both SSD proteins appear to depend on the cholesterol modification of Hh to exert their function -- Ptc efficiently sequesters cholesterol-modified Hh but has little effect on Hh-Nu, and Disp enables the efficient release of cholesterol-modified Hh but has no apparent effect on the release of Hh-Nu. It is not known whether the SSDs of Ptc and Disp play a direct role in binding cholesterol-modified Hh. If this were the case, one could envisage an equilibrium between membrane cholesterol, lipid bilayer-associated Hh-Np, Disp-associated Hh-Np, and free Hh-Np, in which Disp functions as a catalyst for Hh release. Since the affinity of Hh-Np to Disp would have to be higher than to the lipid bilayer, the presence of a cofactor for the dissociation of Hh-Np from Disp must also be postulated. Once Hh-Np is released, reinsertion of it into membranes of nearby A cells might be hindered by an association with carrier proteins or proteoglycans, whose synthesis might depend on the function of Tout-velu, an EXD protein required for proper movement of Hh (Bellaiche, 1998). In a reverse sequence of events, Ptc could be involved in reinserting Hh-Np into the lipid bilayer of receiving cells (Burke, 1999).

PROTEIN STRUCTURE

Amino Acids - 1218

Structural Domains

Searches of genome databases reveal structural homologies of the Disp protein to the products of the vertebrate and Drosophila ptc and to NPC1, a patched-related protein involved in intracellular cholesterol trafficking (Carstea, 1997; Loftus, 1997). Disp contains 12 putative membrane-spanning domains. Like the Ptc and NPC1 proteins, Disp has a sterol-sensing domain (SSD), a domain first defined in HMG CoA reductase, an endoplasmic reticulum enzyme whose degradation is accelerated by sterols, and SREBP cleavage-activating protein (SCAP), which regulates cholesterol metabolism by stimulating cleavage of transcription factors SREBP-1 and -2, thereby releasing them from membranes. These two proteins are key regulators of intracellular cholesterol homeostasis, while NPC1 is thought to be involved in cholesterol trafficking, since defects in this protein cause an accumulation of cholesterol in lysosomes. Aside from the multitransmembrane domain structure and the SSD, no other homologies to Ptc or NPC1 proteins could be detected in Disp. The protein with the highest overall homology to Disp is the product of an as yet uncharacterized C. elegans gene (GenBank acc. no. AAC48001, here termed ceDisp). It is proposed that together Disp and ceDisp define a novel subfamily of SSD proteins (Burke, 1999 and references therein).

date revised: 22 February 2000

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.