The ubiquitin ligase Hyperplastic discs negatively regulateshedgehog and decapentaplegic expression by independent mechanisms

Communication is essential for multicellular development. Intercellular signals regulate the timing and pattern of cellular events such as growth,division, movement, and differentiation, allowing large groups of cells to behave in a coordinated manner during events such as organogenesis. Cell-cell signaling can regulate the activity of cells at many different levels,including gene expression, cell division and motility. Within the receiving cell, signal transduction cascades transform the reception of ligand at the cell surface into the proper cellular response. This may involve regulation of protein activity by post-translational modification or by destruction or proteolytic processing. Ubiquitination, the covalent addition of a multimeric chain of the 76-amino-acid Ubiquitin (Ub) protein, is the most common intracellular signal for proteolysis. Ubiquitination is a multi-step process that begins with the activation of a Ub molecule by an E1 or Ub-activating enzyme. The activated Ub is transferred to an E2 enzyme, which is then responsible, either directly or indirectly, for attaching the Ub to a substrate protein (reviewed by Ciechanover et al., 2000). Specificity of the ubiquitination reaction is achieved at the level of the E3 ubiquitin ligase, which is thought to directly bind the substrate. Many such ligases exist and have been classified into families based on the structure of the ubiquitination domain. HECT-domain E3 ligases directly attach Ub to a substrate, while RING domain E3s direct specific substrate ubiquitination by the E2(Jackson et al., 2000).

A small number of signaling pathways appear to direct most developmental processes. In Drosophila, the BMP family member Decapentaplegic (Dpp)and the founding member of the Hedgehog family (Hh) are used repeatedly throughout development. One function of Hh and Dpp is to direct the progressive differentiation of the eye imaginal disc (reviewed inLee and Treisman, 2002). In the second instar eye disc, hh is expressed in a complex pattern in both the disc proper and the peripodial membrane, before being refined to a small domain centered on the dorsoventral midline of the disc's posterior margin (Cho et al., 2000). Hh signals more anterior cells to express dpp and atonal (ato),which encodes the bHLH transcription factor required for the formation of the R8 `founder' photoreceptor in each cluster(Jarman et al., 1994). These cells then differentiate as photoreceptors and themselves express hh,allowing the cycle to propagate toward the anterior of the disc. dppis expressed in the morphogenetic furrow, an indentation at the front of differentiation, where it is responsible for coordinating the timing of differentiation through synchronization of the cell cycle(Penton et al., 1997). Loss of either Hh or Dpp blocks the initiation of differentiation, while loss of both blocks progression (Curtiss and Mlodzik,2000; Greenwood and Struhl,1999). While it is known that Hh activates dpp expression in the furrow (Heberlein et al.,1995; Heberlein et al.,1993; Ma et al.,1993), it is not known how hh expression is controlled,nor how dpp is turned off in cells leaving the furrow.

Much has been learned about the regulation of hh expression in tissues other than the eye. During embryonic development, maintenance ofhh expression in the posterior of each segment requires the homeobox gene engrailed (en)(Tabata et al., 1992). In the head segments, however, hh expression is not controlled by En, but by several different homeobox and segment polarity genes(Crozatier et al., 1999;Gallitano-Mendel and Finkelstein,1997). hh is also regulated by En later in development,in the posterior compartments of the leg and wing imaginal discs(Tabata et al., 1995;Zecca et al., 1995). En regulates hh expression indirectly, by repressing the expression ofcubitus interruptus (ci) in the posterior compartment(Dominguez et al., 1996;Eaton and Kornberg, 1990;Schwartz et al., 1995). In the absence of Hh signal, the full-length Ci protein (Ci₁₅₅) is proteolytically processed to a 75 kDa transcriptional repressor(Ci₇₅); when the Hh signal is received this processing event is inhibited, allowing the accumulation of Ci₁₅₅(Aza-Blanc et al., 1997). In the anterior compartment, Ci₇₅ represses hh expression(Dominguez et al., 1996;Methot and Basler, 1999). The Hh signal is transmitted to Ci through two transmembrane proteins, Patched(Ptc) and Smoothened (Smo) (reviewed byIngham and McMahon, 2001). Hh binds to Ptc and inhibits its activity; in the absence of Hh, Ptc promotes internalization, dephosphorylation and degradation of Smo(Alcedo et al., 2000;Denef et al., 2000;Ingham et al., 2000).

Unlike these other tissues, in the eye the source of hh expression changes over time; as cells are recruited to differentiate as photoreceptors,they begin expressing hh. Reflecting this difference in hhcontrol, en function is not required in the eye disc for hhexpression (Strutt and Mlodzik,1996). The relationship between hh and dppexpression is conserved in some tissues, but not in others. In the wing disc,as in the eye, dpp is transcribed in response to Hh(Basler and Struhl, 1994;Zecca et al., 1995); in the leg disc this response is limited, through the activity of Wg, to a dorsal domain in the anterior compartment (Basler and Struhl, 1994; Brook and Cohen, 1996; Jiang and Struhl,1996). In other tissues, such as the dorsal domain of the early embryo, dpp expression appears to be independent of Hh signaling(St Johnston and Gelbart,1987). Thus the mechanisms controlling the expression of these signals are likely to be at least partially context-dependent. In tissues where Hh regulates dpp expression, dpp is both activated by Ci₁₅₅ and repressed by Ci₇₅, allowing robust control ofdpp by Hh (Methot and Basler,1999).

To gain further insight into the control of photoreceptor development, we have conducted mosaic screens for mutations that perturb differentiation(Benlali et al., 2000;Lee and Treisman, 2001a;Lee and Treisman, 2001b;Treisman, 2001). Many components of the hh and dpp pathways were recovered in these screens. We report here our characterization of mutations in thehyperplastic discs (hyd) locus; hyd encodes a HECT domain E3 ubiquitin ligase (Callaghan et al., 1998; Mansfield et al.,1994). Mutations in hyd cause ectopic non autonomous differentiation in the eye. This effect is due to the independent activation of hh and dpp expression in hyd mutant tissue,reflecting independent effects of Hyd on hh expression and function.

The mosaic screen for eye differentiation mutants on 3R has been described previously (Lee and Treisman,2001a). Four alleles of hyd were recovered in this screen. Several previously isolated hyd alleles, including alleles in which Hyd protein is absent (Mansfield et al., 1994), gave indistinguishable clonal phenotypes in the eye disc. To make negatively labeled clones, FRT82, hyd or hyd,hh or slmb males were crossed to eyFLP1 orhsFLP122; FRT82, arm-lacZ or FRT82, Ubi-GFP females. To make clones in a Minute background, the females used were eyFLP1;FRT82, arm-lacZ, M(3)96C. To make positively labeled clones, FRT82,hyd males were crossed to UASGFP, eyFLP1 orhsFLP122; tub-GAL4, FRT82, tub-GAL80 females(Lee and Luo, 1999). The same females were used for clone-specific expression of transgenes carried by the FRT82, hyd males (Lee and Treisman, 2001a). Crosses in which hsFLP122 was used were heat-shocked at the first and second instar stage for 1 hour at 38.5°C. The alleles used were hh^rJ413(Heberlein et al., 1993),hh^AC (Lee et al.,1992), hh^ts2(Ma et al., 1993),slmb^B93 (Lee and Treisman, 2001a) and ato^IH3 (J. D. L.,unpublished). Transgenes used were tub-GAL4, tub-GAL80(Lee and Luo, 1999) and UAS-ci76 (Aza-Blanc et al.,1997). Reporters used were dpp-lacZ(Blackman et al., 1991) andhh^P30 (Lee et al.,1992). To examine the level of hh expression inhyd mutant clones, hyd clones carrying two copies of ahh-lacZ reporter in a heterozygous background were compared to wild-type clones with two copies of hh-lacZ in a heterozygous background. The clones shown have significantly stronger lacZexpression than would be expected from doubling the reporter dosage.

Imaginal discs were stained as described previously(Hazelett et al., 1998). Primary antibodies used were rat α-Elav (1:1)(Robinow and White, 1991),rabbit α-Atonal (1:10,000) (Jarman et al., 1994), rat α-Ci (1:1)(Motzny and Holmgren, 1995),rat anti-Smo (1:500) (Denef et al.,2000), mouse α-β-galactosidase (Promega; 1:200), rabbitα-β-galactosidase (Cappel; 1:5000), mouse anti-Neuroglian (1:1)(Hortsch et al., 1990), mouse anti-Ptc (1:150) (Capdevila et al.,1994) and rabbit α-Hh-N (1:2000)(Lee and Treisman, 2001b). Secondary antibodies were from Jackson Laboratories, conjugated to FITC, Texas Red or Cy5, used at 1:200. Fluorescent images were obtained using a Leica TCS NT confocal microscope. In situ hybridization was performed as described(Maurel-Zaffran and Treisman,2000).

To identify novel genes contributing to pattern formation in theDrosophila eye disc, we carried out a mosaic genetic screen in which we generated homozygous mutant clones of cells specifically in the eye disc(Benlali et al., 2000;Lee and Treisman, 2001a;Lee and Treisman, 2001b;Treisman, 2001). In this screen we recovered 4 alleles of hyd, which encodes a large protein containing a HECT family E3 ubiquitin ligase domain(Callaghan et al., 1998;Mansfield et al., 1994).hyd was initially isolated in a screen for mutations causing imaginal disc overgrowth (Martin et al.,1977). We found that adult eyes containing hyd mutant clones showed extensive overgrowth of the eye tissue surrounding ahyd clone, although the clones themselves did not persist to adulthood (Fig. 1B). In the third instar eye disc, we observed premature photoreceptor differentiation,visualized by expression of the markers Elav and Neuroglian, in hydmutant clones in a region just anterior to the morphogenetic furrow(Fig. 1E and data not shown). Very rarely, differentiation was seen in clones lying near the anterior margin(data not shown). The spatial restriction of this phenotype is consistent with work by others demonstrating the existence of a preproneural zone anterior to the furrow (Greenwood and Struhl,1999; Heberlein et al.,1995; Ma et al.,1993; Strutt et al.,1995). Ectopic differentiation spread beyond the borders of thehyd mutant clones into the surrounding wild-type tissue(Fig. 1E). The proneural transcription factor Atonal (Ato) was also misexpressed within and surroundinghyd clones anterior to the furrow(Fig. 1H). The overgrowth of wild-type tissue in adult eyes was visible in third-instar imaginal discs as folding and distortion of the disc epithelium surrounding hyd clones(Fig. 1E,Fig. 3A). This effect was more widespread than the ectopic differentiation phenotype, appearing in more anterior regions of the disc.

While hyd clones in third instar eye discs contained Elav-expressing photoreceptors, clones in the adult eye appeared to form either scars in the eye or head cuticle at its margin(Fig. 1B). We also noticed thathyd clones in the posterior of the eye disc were generally smaller than wild-type clones generated in a similar cross and contained reduced numbers of photoreceptors (Fig. 1D,E). The lack of mutant ommatidia in adult eyes could reflect either a loss of hyd mutant clones due to competition with the surrounding tissue, or a later requirement of hyd for viability. To further examine this question, we made hyd clones in aMinute genetic background (Morata and Ripoll, 1975), thereby reversing the growth disadvantage ofhyd mutant tissue (Fig. 1C). These animals survived to pharate adult stages but could not eclose; they had greatly reduced eyes that nevertheless appeared to contain properly formed hyd mutant ommatidia(Fig. 1C). This suggests that the lack of visible hyd clones in adults is due to their poor relative growth, leading to their elimination through competition with wild-type cells. We also examined third instar eye discs containinghyd/Minute clones, and found greatly reduced eye discs that appeared to contain full clusters of photoreceptors throughout the eye field(Fig. 1F). As cells arrest division when they differentiate, premature differentiation could explain the small size of hyd/Minute eye discs. Ato is normally expressed in the morphogenetic furrow and in the youngest R8 photoreceptors(Fig. 1G); in early third instar hyd/Minute mutant discs the only remaining Ato staining is in single (presumably R8) cells (Fig. 1I), implying that the morphogenetic furrow has already reached the anterior margin. From this we infer that differentiation inhyd/Minute discs still progresses from posterior to anterior, but at an accelerated rate.

Growth and differentiation of cells in the eye disc both depend on Hh and Dpp secreted by more posterior cells. The non autonomous differentiation and overgrowth caused by hyd clones suggested that the clones might be producing one or both of these molecules; a similar phenotype is produced by ectopic expression of hh or activation of the hh pathway ahead of the furrow (Heberlein et al.,1995; Pan and Rubin,1995; Strutt et al.,1995). We therefore looked at hh expression inhyd clones using both an enhancer trap line(Lee et al., 1992) and antibody staining. hyd mutant clones anterior to the furrow indeed expressed hh-lacZ and Hh protein earlier than surrounding wild-type cells (Fig. 2A-C and data not shown). When we made hyd clones in a Minute background, we observed hh-lacZ expression throughout the eye disc(Fig. 2E). The widespreadhh expression in hyd/Minute clones may explain the accelerated differentiation and small size of these eye discs.

The ectopic hh expression we observed in hyd mutant clones could be a consequence, rather than a cause, of their premature differentiation. To rule out this possibility we generated hydato double mutant clones. Clones mutant for ato cannot form the R8 photoreceptor (Jarman et al.,1995), which is itself required for the recruitment of photoreceptors R1-R7. Thus ato mutant clones do not differentiate,unless a cell at the margin of a clone is recruited by a neighboring wild-type R8. We confirmed that ato single mutant clones do not expresshh (data not shown). We found that while hyd ato mutant clones did not differentiate, they nevertheless expressed hh-lacZ and were capable of directing ectopic differentiation in surrounding wild-type tissue (Fig. 2G-I). This demonstrates that loss of hyd function has a direct effect onhh expression that is not simply due to differentiation.

As an E3 ubiquitin ligase, Hyd is likely to promote the degradation of one or more proteins. Based on the ectopic expression of hh inhyd mutant clones, we hypothesized that Hyd was acting in the anterior of the third instar eye disc to prevent premature expression ofhh. hyd is expressed in proliferating tissues in the embryo and larva, but its expression in the eye disc had not been described in detail(Mansfield et al., 1994). Using in situ hybridization, we found that hyd RNA was highly expressed in the anterior of the eye imaginal disc, especially around the dorsoventral midline (Fig. 2F).hyd was expressed at lower levels towards the dorsal and ventral margins but was still restricted to the anterior. This expression pattern is consistent with a role for hyd in preventing the premature expression of hh.

If the ectopic differentiation and overgrowth associated with hydclones is due to ectopic hh expression, it should be possible to rescue this phenotype by removing hh function from the clones. We therefore determined the phenotype of hyd hh double mutant clones. We obtained similar results with three hh alleles,hh^rJ413 (Heberlein et al., 1993), hh^ts2(Ma et al., 1993) grown at 29°C, and the null allele hh^AC(Lee et al., 1992). Whilehh mutant clones appear wild type unless located on the margin of the eye disc (Dominguez and Hafen,1997) (Fig. 3B),hyd hh double mutant clones showed a partial suppression of thehyd phenotype. Ectopic photoreceptors were no longer present in or around hyd hh clones (Fig. 3C), and hyd hh double mutant clones generated in aMinute background had only a few photoreceptors associated with the remaining wild-type tissue (Fig. 3D). Thus hyd mutant tissue requires Hh in order to differentiate. However, some hyd hh mutant clones were still able to stimulate proliferation of surrounding tissue, leading to overgrowth of the adult eye (Fig. 3C,E).

dpp has also been shown to stimulate proliferation in the eye disc(Pignoni and Zipursky, 1997). The remaining non autonomous overgrowth induced by hyd hh double mutant clones might therefore be due to ectopic expression of dpp. Using a dpp-lacZ reporter construct, we observed ectopic dppexpression in and around hyd mutant clones anterior to the furrow(Fig. 3F and data not shown). Ectopic dpp expression appeared more widespread than ectopichh expression and occurred in more anterior regions of the disc,raising the possibility that dpp misexpression is not merely induced by ectopic Hh in the clones but is an independent consequence of the loss ofhyd. Indeed, we found that dpp was still expressed in somehyd hh mutant clones (Fig. 3G,H), although its expression was limited to clones close to the morphogenetic furrow. Thus in the absence of hyd function,dpp expression is no longer strictly regulated by Hh signaling.

The persistence of dpp expression in hyd hh double mutant clones suggested that dpp regulation by Hyd was at least partially independent of hh (Fig. 3G,H). One molecule that is known to transcriptionally regulate both hh and dpp is Ci. In the wing imaginal disc it has been shown that Hh controls dpp expression both by suppressing the production of Ci₇₅, which inhibits dpp expression, and by activating Ci₁₅₅, which activates dpp expression(Methot and Basler, 1999). In addition, Ci₇₅ inhibits hh expression in the anterior compartment of the wing disc; ci transcription is repressed byen in the posterior compartment, allowing hh to be expressed there (Dominguez et al., 1996;Methot and Basler, 1999). To test whether hyd was acting through Ci to affect hh anddpp expression in the eye disc, we drove the expression of a truncated constitutive repressor form of Ci (Ci₇₆)(Aza-Blanc et al., 1997)specifically in hyd mutant clones using the MARCM system(Lee and Treisman, 2001a;Lee and Luo, 1999).dpp-lacZ was no longer expressed in hyd clones expressing UAS-ci₇₆ in or anterior to the furrow(Fig. 4A,B). However,Ci₇₆ did not prevent the ectopic expression of hh inhyd mutant clones (Fig. 4D-F). Continued hh expression in these clones sometimes led to ectopic differentiation in tissue surrounding the clone(Fig. 4D-F). Thus Ci₇₆ is sufficient to block dpp but not hhexpression in hyd clones, suggesting that Hyd regulates hhand dpp expression through at least partially independent mechanisms.

If the overgrowth phenotype induced by hyd hh double mutant clones is indeed due to their misexpression of dpp, it should be blocked by introducing Ci₇₆ into the mutant cells. Indeed, when we generatedhyd hh mutant clones expressing UAS-ci₇₆, the adult eyes no longer exhibited any overgrowth(Fig. 4C). Thus all the ectopic growth and differentiation caused by loss of hyd in the eye can be attributed to independent effects on hh expression and activation of the Hh pathway.

If hyd regulates dpp expression by altering Ci activity,loss of hyd should lead to upregulation of full-length, active Ci. We indeed observed increased levels of full-length Ci in hyd mutant clones in the anterior of the eye disc (data not shown). However, this could be due to misexpression of hh in the same clones. To determine whether hyd has a direct effect on Ci, we examined hyd hhdouble mutant clones anterior to the morphogenetic furrow. High levels of full-length Ci accumulated in these clones(Fig. 5A-C), confirming that Hyd normally reduces Ci levels independently of Hh activity.

Hyd might act directly on Ci to promote its proteolytic cleavage or degradation. An alternative possible target for Hyd activity is Smoothened(Smo). Smo is a transmembrane protein that acts positively in Hh signaling(Alcedo et al., 1996). Smo levels are kept low by the receptor protein Patched (Ptc) in the absence of Hh, but Smo is stabilized and localized to the membrane when Hh binds to Ptc(Alcedo et al., 2000;Denef et al., 2000;Ingham et al., 2000). To test whether Hyd normally contributes to Smo degradation, we stained eye discs containing hyd, hh clones with Smo antibody. No Smo accumulation was apparent in the clones (Fig. 5D-F). Thus loss of hyd leads to accumulation of full-length Ci without altering the level of Smo.

Since hyd is expressed in the wing disc and is required for its normal growth (Mansfield et al.,1994), we tested whether its effects on wing development might also be mediated by alterations in hh expression and Ci levels. In wild-type wing discs, hh is expressed uniformly throughout the posterior compartment of the wing pouch, while dpp is expressed in the anterior compartment in a stripe along the AP border(Basler and Struhl, 1994). Ci₁₅₅ is present at high levels in a similar stripe at the AP border and at lower levels elsewhere in the anterior compartment(Motzny and Holmgren, 1995). Expression of hh, dpp and Ci₁₅₅ in hyd clones remained restricted to the correct compartment. However, some hydmutant clones in the posterior compartment expressed elevated levels ofhh-lacZ (Fig. 6A,B). This misexpression of hh was correlated with a rounded shape and apparent overgrowth of the clones. The only known regulator of hhexpression in the wing disc is Ci, which is restricted to the anterior compartment by En-mediated repression; Ci₇₆ represses hhthere (Dominguez et al., 1996;Methot and Basler, 1999;Schwartz et al., 1995). These results suggest that a Ci-independent activator of hh expression must be present in the posterior compartment and kept in check by Hyd activity.

In addition, Ci₁₅₅ was upregulated in anterior hydmutant clones (Fig. 6E,F). In contrast to the eye disc, we did not observe any hh misexpression in anterior hyd clones (Fig. 6A,B); thus Ci upregulation in hyd clones must be independent of hh. This is consistent with our findings thathyd regulates Ci and hh independently in the eye disc. Smo levels were not significantly increased in hyd mutant clones in the anterior compartment of the wing disc (Fig. 6C,D), suggesting that as in the eye disc, hyd affects Ci independently of Smo.

The F-box protein Slmb has been shown to promote processing of Ci to Ci₇₅ as a component of an SCF ubiquitin ligase complex(Jiang and Struhl, 1998;Miletich and Limbourg-Bouchon,2000; Noureddine et al.,2002; Theodosiou et al.,1998). We therefore compared the effects of slmb andhyd mutations on Ci levels in the wing disc. Ci₁₅₅ was much more dramatically increased in slmb clones than in hydclones (Fig. 6I,J). We also observed an interesting difference between hyd and slmb in the regulation of dpp. dpp expression was increased in hydmutant clones close to the AP border, but was very rarely activated outside this domain (Fig. 6G,H). In contrast, slmb mutant clones activated dpp expression only when they lay outside the wing pouch (Fig. 6K,L) (Miletich and Limbourg-Bouchon, 2000), perhaps because of activation of Wg signaling, which represses dpp expression, within the wing pouch(Jiang and Struhl, 1998). Consistent with these third instar phenotypes, we have not observed anterior duplications like those resulting from loss of slmb in adult wings containing hyd mutant clones, although outgrowths did arise from internal regions of the wing (data not shown). Such duplications would requiredpp to be misexpressed at a distance from its normal domain of expression (Zecca et al.,1995). Ptc expression, which requires activation of the full-length form of Ci (Methot and Basler,1999), was not altered in either hyd or slmbmutant clones (Wang et al.,1999) (data not shown). Slmb and Hyd thus appear to have distinct effects on Ci protein accumulation and activity, suggesting that they have either different substrates or different effects on the same substrate.

We have shown that hyd acts as a negative regulator of bothhh and dpp expression in the anterior of theDrosophila eye disc. Loss of hyd function leads to the ectopic expression of both genes, resulting in non autonomous overgrowth of the disc and premature photoreceptor differentiation that propagates into the surrounding tissue. Our ability to suppress this overgrowth by preventing both expression of hh and activation of the Hh pathway indicates that the hyperplastic effects of hyd in the eye are entirely mediated by Hh signaling. This is probably not the case in the wing disc; in this tissuehh and dpp expression are also blocked by hyd, buthyd clones in the posterior compartment can autonomously overgrow without expressing Ci, the transcription factor required for the response to Hh (Fig. 6A-D). hydmay thus have an independent effect on the cell cycle in wing disc cells. The human homologue of hyd, EDD, is located in a chromosomal region that is disrupted in a variety of cancers(Callaghan et al., 1998). It will be of interest to determine whether loss of hyd is responsible for any of these syndromes, and if so, whether the tumorous growth can be attributed to misregulation of hh or dpp homologues.

Little is known about the control of hh expression in the eye disc. In the embryo and other imaginal discs hh expression is controlled by en, which defines the posterior compartment in a lineage-dependent manner. The eye disc has no anterior-posterior compartment boundary, and loss of en function in the eye has no effect(Strutt and Mlodzik, 1996). Since hh expression is repressed by Ci₇₆ in the anterior wing disc (Dominguez et al.,1996; Methot and Basler,1999), it seemed possible that this was also the case in the eye disc; Hh could then activate its own expression in more anterior cells by blocking Ci cleavage. However, we were not able to prevent hhexpression by providing Ci₇₆ to hyd mutant cells in the eye disc, although this did suffice to repress hh target genes such as dpp. In agreement with this result, ci mutant clones anterior to the furrow do not induce ectopic differentiation (N. Baker,personal communication), indicating that loss of the repressor form of Ci is not sufficient to allow hh transcription in the eye disc.hyd must therefore be a component of the Ci-independent mechanism that restricts hh expression.

Regulation of hh by hyd is also clearly independent of Ci in the wing disc, since loss of hyd leads to hh upregulation in the posterior compartment, where Ci is not present, and not in the anterior compartment. The Groucho (Gro) corepressor has been proposed to contribute to a Ci-independent mechanism of hh repression in cells close to the compartment boundary (Apidianakis et al.,2001). However, the effects of loss of hyd differ from those of loss of gro, which affects only the anterior compartment of the wing disc (Apidianakis et al.,2001) and promotes excessive photoreceptor differentiation only posterior to the furrow in the eye disc(Chanut et al., 2000),suggesting that a third mechanism of hh regulation may exist.hh expression may not be merely a default state resulting from the absence of the Ci repressor and Gro, but may require another activator, the levels of which are normally kept in check by Hyd.

Control of dpp expression by hyd, in contrast, appears to be mediated by Ci. Ci₁₅₅ is upregulated in hyd mutant cells in the eye disc in a hh-independent manner, and ectopicdpp expression in these cells can be blocked by Ci₇₆. In the wing disc, dpp misexpression is limited to theci-expressing anterior compartment, and is again associated with upregulation of Ci₁₅₅. Thus hyd acts on Hh signaling as well as hh expression, preventing full activation of the Hh pathway in anterior cells (Fig. 7).

Hyd is likely to act as an E3 ubiquitin ligase; its human homologue has been shown to ubiquitinate at least one substrate in vitro(Honda et al., 2002). The substrate specificity of HECT domain E3 ubiquitin ligases appears to reside within their unique N-terminal domains(Ciechanover et al., 2000),making it difficult to predict the sequence or structure recognized by Hyd. The only potential clue is that Hyd contains a peptide-binding domain homologous to the C-terminus of poly(A)-binding protein (PABP)(Callaghan et al., 1998;Kozlov et al., 2001); the human HYD protein can interact with Paip1, a binding partner of PABP(Craig et al., 1998;Deo et al., 2001).

Hyd independently regulates hh and dpp expression,suggesting either that Hyd has multiple substrates or that its substrate has multiple functions. dpp expression in hyd clones is blocked by Ci₇₆, placing the effect of Hyd on dpp upstream or at the level of Ci activity. Ci₁₅₅ but not Smo accumulates to high levels in hyd hh mutant cells in the eye disc and in anterior cells in the wing disc; thus Hyd may act on Ci itself, on a component of the Hh pathway between Smo and Ci, or on a nuclear cofactor that stabilizes Ci. Consistent with an effect on Ci or a cofactor, Hyd protein has been detected in both the cytoplasmic and nuclear compartments(Mansfield et al., 1994). The human Hyd homolog EDD appears to be predominantly localized in the nucleus,where it interacts with the progesterone receptor and DNA topoisomerase II-binding protein (Henderson et al.,2002; Honda et al.,2002). Ubiquitination can function to enhance the potency of transcriptional activation domains(Salghetti et al., 2001);however, the ectopic gene expression observed in hyd mutant clones would be difficult to explain by this mechanism.

The SCF ubiquitin ligase complex containing the F-box protein Slmb and the RING finger protein Roc1 has been implicated in the ubiquitination of Ci that mediates its processing to Ci₇₅(Jiang and Struhl, 1998;Noureddine et al., 2002;Theodosiou et al., 1998;Wang et al., 1999). There are several possible explanations for the apparent overlap between Slmb and Hyd functions. Slmb may directly ubiquitinate Ci, while Hyd acts on another substrate; the more dramatic effect of slmb than hyd clones on Ci accumulation argues for this possibility. However, it is also possible that Hyd, rather than Slmb, is the direct ubiquitin ligase for Ci. The consensus sequence for Slmb is not present in Ci, although it has been suggested that several weakly matching sequences might suffice for its recognition (Price and Kalderon,2002; Winston et al.,1999). In addition, two groups have obtained inconsistent genetic evidence as to whether slmb acts upstream or downstream ofsmo and protein kinase A (PKA)(Jiang and Struhl, 1998;Theodosiou et al., 1998;Wang et al., 1999). It is unlikely that Hyd and Slmb carry out the same process in different cells, ashyd is expressed throughout the wing pouch(Mansfield et al., 1994), andslmb is clearly active in the same region(Fig. 6I)(Jiang and Struhl, 1998),although its expression has not been examined. Finally, Hyd and Slmb could both act on Ci, either additively or with different outcomes. For example,ubiquitination by Slmb promotes processing of Ci to Ci₇₅(Jiang and Struhl, 1998),while ubiquitination by Hyd might promote complete degradation of the Ci protein. Unfortunately, we were not able to obtain large enough quantities ofhyd hh mutant tissue to test this possibility directly by western blotting. The lack of hh misexpression in hyd clones in the anterior compartment of the wing disc suggests that these cells still contain Ci₇₅ as well as Ci₁₅₅(Methot and Basler, 1999);loss of hyd may therefore stabilize both forms of the Ci protein rather than altering their ratio. However, this is not a definitive test ofhyd function, as loss of PKA does not lead to ectopichh-lacZ expression despite its effect on Ci processing(Jiang and Struhl, 1995;Jiang and Struhl, 1998;Li et al., 1995;Pan and Rubin, 1995). Hydmediated degradation of both forms of Ci, or its redundancy with Slmb for Ci cleavage, would explain the limited effect on dpp expression inhyd hh double mutant clones in the eye disc and in hydmutant clones in the wing disc; dpp misexpression in these cases is restricted to a region in which endogenous Hh may contribute to altering the ratio of the two forms of Ci.

Ubiquitination is a mechanism commonly used to regulate protein activity by targeting proteins for degradation or processing, during the cell cycle and in a number of signaling pathways (reviewed inCiechanover et al., 2000). The Slmb-containing SCF complex is also required for the degradation of Armadillo(Arm)/β-catenin, allowing Wnt signaling(Hart et al., 1999;Jiang and Struhl, 1998;Miletich and Limbourg-Bouchon,2000; Winston et al.,1999), as well as for the degradation of IkappaB(Spencer et al., 1999;Yaron et al., 1998). Hyd is unlikely to act on Arm in the eye disc, as Arm accumulation would prevent the ectopic photoreceptor differentiation seen in hyd mutant clones(Lee and Treisman, 2001a;Treisman and Rubin, 1995);this may indicate another difference in the substrate specificity of Hyd and Slmb. However, it is possible that hyd affects Wg signaling in the wing disc, as hyd mutant clones can induce ectopic expression of the Wg target gene scute (K. A., unpublished data). The HECT domain ligases Smurf1 and Smurf2 are important antagonists of BMP signaling,promoting downregulation of both Smads and receptors(Kavsak et al., 2000;Podos et al., 2001;Zhu et al., 1999). Itch/Suppressor of deltex, another HECT domain ligase, ubiquitinates Notch(Cornell et al., 1999;Fostier et al., 1998;Qiu et al., 2000). In addition, nuclear Notch is degraded by Sel-10, while the ligand Delta is the target of ubiquitination by Neuralized (reviewed byLai, 2002). Our placement ofhyd within the Hh pathway and upstream of hh expression expands this growing list of cases in which signaling pathways are regulated by ubiquitination.

We would like to thank Steve Cohen, Isabel Guerrero, Bob Holmgren, Yuh-Nung Jan, Tom Kornberg, Markus Noll, Tetsuya Tabata, Nicole Theodosiou, and the Bloomington Drosophila stock center for fly stocks and antibodies. We are grateful to Ruth Lehmann and Steve Burden for the use of their confocal microscopes. We thank the members of the Treisman lab for helpful discussions,and Florence Janody and Kerstin Hofmeyer for comments on the manuscript. This work was supported by NIH grants EY13777 to J. E. T. and GM53058 to A. S.