The Full-length Unprocessed Hedgehog Protein Is an Active Signaling Molecule

Abstract

The hedgehog (HH) family of ligands plays an important instructional role in metazoan development. HH proteins are initially produced as ∼45-kDa full-length proteins, which undergo an intramolecular cleavage to generate an amino-terminal product that subsequently becomes cholesterol-modified (HH-Np). It is well accepted that this cholesterol-modified amino-terminal cleavage product is responsible for all HH-dependent signaling events. Contrary to this model we show here that full-length forms of HH proteins are able to traffic to the plasma membrane and participate directly in cell-cell signaling, both in vitro and in vivo. We were also able to rescue a Drosophila eye-specific hh loss of function phenotype by expressing a full-length form of hh that cannot be processed into HH-Np. These results suggest that in some physiological contexts full-length HH proteins may participate directly in HH signaling and that this novel activity of full-length HH may be evolutionarily conserved.

Introduction

Underscoring the importance of sonic hedgehog (SHH)³ in developmental patterning, inherited and sporadic mutations in SHH are frequently found in patients diagnosed with the human developmental disorder holoprosencephaly (1–5). A number of SHH point mutations have been identified from holoprosencephaly patients, providing a unique opportunity to test the hypothesis that these mutations will reveal novel mechanistic insights into SHH biogenesis.

EXPERIMENTAL PROCEDURES

SHH Activity Assays

The full-length cDNA of human SHH (a gift from Dr. Cliff Tabin) was cloned into pcDNA3.1. All of the SHH mutants were made by site-directed mutagenesis using the QuikChange II kit (Stratagene) and expressed as described (6). Cell-to-cell signaling was determined by co-culturing SHH mutant-expressing cells with cells expressing SHH-dependent luciferase reporter gene (7, 8). Total cellular SHH activity was determined by incubating whole cell lysate from SHH-expressing cells with cells expressing SHH-dependent luciferase reporter gene (9) (lysate assay) as described (10). All of the activity assays were done in triplicate, and each experiment was repeated at least three times. The activity data are presented as mean ± standard deviation (error bars) in the various figures.

SHH Carboxyl-terminal Immunoprecipitation

Bosc cells expressing SHH-U or wt SHH were lysed in 1% Triton X-100 lysis buffer (150 mm NaCl, 10 mm Na₂HPO₄, 1% Triton X-100, pH 6.5) supplemented with a protease inhibitor mixture for 10 min on ice. Following nuclei removal by centrifugation, the samples were immunoprecipitated for 3 h at 4 °C using either antibodies raised against a SHH carboxyl-terminal domain peptide (R & D Systems) or control IgG and protein G beads (Sigma). The various immunoprecipitates were washed twice with PBS and incubated with confluent cells expressing SHH-dependent luciferase reporter genes overnight to assay SHH activity as described above or analyzed by immunoblotting.

Immunofluorescence

To examine PTC-dependent internalization of SHH-U, Bosc cells transiently expressing GFP, SHH-U, the SHH receptor PATCHED (PTC) (FLAG-tagged), or a control plasmid were co-cultured on microscope slides for 2 days in various combinations. These slides were washed with cold PBS, fixed in 4% paraformaldehyde for 10 min at room temperature, and permeabilized/blocked with PBS containing 5% donkey serum and 0.1% Tween 20. The various samples were incubated with goat anti-SHH carboxyl-terminal domain peptide (R & D Systems) or mouse anti-FLAG peptide (Sigma) antibodies followed by incubation with donkey anti-goat IgG-Alexa fluor 594 or donkey anti-mouse IgG-Alexa fluor 488 (Molecular Probes), respectively. All of the images were acquired with a Leica DMIRE2 confocal laser-scanning microscope, using 1 airy unit intensity of collected light.

Immunoblotting Tissues

Stage 22 chicken limb buds or rabbit gastric glands were isolated (11) and then lysed in 3 ml of radioimmune precipitation assay buffer (150 mm NaCl, 10 mm Na₂HPO₄, 1% sodium deoxycolate, 0.1% SDS, 2 mm EDTA, 50 mm NaF, 1% Triton X-100, pH 7.2, and freshly added 200 μm Na₃VO₄) supplemented with protease inhibitor mixture. The samples were immunoprecipitated using anti-SHH 5E1 monoclonal antibody or control IgG (Sigma) and True-blot beads. The immunoprecipitates were analyzed by immunoblotting with anti-SHH antibodies H-160 (Santa Cruz Biotechnology). Oregon-R (wt), hh^ac (12), and UAS-hh-M4 lines of Drosophila were also used to examine the expression of various forms of HH in different tissues. hh^ac is a null allele. The UAS-hh-M4 line (13) specifically expressed HH in larval tissues. Wild type embryos, homozygous hh^ac embryos, and heads of wt third instar hh-Gal4>UAS-hh-M4 larvae enriched in imaginal discs were dissected in ice-cold PBS and homogenized in lysis buffer (150 mm NaCl, 50 mm Tris, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, pH 7.5) supplemented with a protease inhibitor mixture for 30 min on ice. The proteins were separated by SDS-PAGE and analyzed by immunoblotting with rabbit “Calvados” polyclonal anti-HH antibody (14) or mouse anti-α-tubulin monoclonal antibody (Sigma).

Drosophila Analyses

For eye analysis three transgenic constructs UAS-hh-U (15), UAS-hh (16), and UAS-hh-N (16) were specifically expressed in the developing retina of hh¹ animals with the eye-specific promoter GMR-GAL4, and rescue of the phenotype was evaluated. For leg disc analysis, animals with the following genotypes were generated: y, hsflp/+; tub≪Gal4, UAS-CD8-GFP/+; ptc/lacZ. They were then heat-shocked for 22 min at 37 °C between the L1 and L3 stages. The discs were fixed and analyzed as previously described (15). The images were collected on a Zeiss LSM510 confocal microscope.

RESULTS

We have examined a set of SHH holoprosencephaly point mutations that encode a mutated carboxyl-terminal domain of SHH (2–4). During the processing of SHH into SHH-Np, this carboxyl-terminal domain functions to regulate the covalent addition of cholesterol to its amino-terminal domain and subsequent cleavage of full-length SHH (Fig. 1A for a schematic of the various forms of HH described here). We engineered these SHH holoprosencephaly mutations, SHH-D222N, SHH-T267I, SHH-G290D, SHH-A373T, and SHH-P424A, into a human SHH cDNA construct. Plasmids expressing these SHH holoprosencephaly mutations, wt SHH, or a control plasmid, were transfected into Bosc cells, a human embryonic kidney cell derivative (17), and the abundance and processing of the mutant proteins were examined from cell lysates. The processing pattern of three SHH holoprosencephaly mutants, SHH-D222N, SHH-T267I, and SHH-A373T, was significantly different from that of wt SHH, showing an increased abundance of their full-length form (SHH-F) and a corresponding decrease in their processed SHH-Np form (Fig. 1B). Thus, three of the five SHH holoprosencephaly mutations examined showed accumulation of full-length SHH and decreased SHH-Np. Next we estimated the activity of these SHH mutants by co-culturing cells expressing these mutant proteins with NIH-3T3 cells stably expressing a SHH-dependent luciferase reporter gene (8) (Fig. 1C). Interestingly, two of the processing-deficient mutants, SHH-D222N and SHH-A373T, exhibited an activation of the SHH reporter gene ∼65% of wt SHH. Together these results confirm the importance of the SHH carboxyl-terminal domain to the processing and activity of SHH.

FIGURE 1.

SHH holoprosencephaly mutations that encode carboxyl-terminal domain point mutants affect the processing and activity of the SHH protein. A, schematic diagram showing protein encoded by different SHH constructs. SHH is produced as a full-length protein (SHH-F) comprised of an amino-terminal domain (open box) and a carboxyl-terminal domain (hatched box) that regulates auto-processing of the two domains. During the processing of full-length SHH, cholesterol (Chol) is covalently added to the newly generated carboxyl terminus of SHH (SHH-Np). SHH-U refers to a mutant form of full-length SHH in which the critical processing site amino acid Cys¹⁹⁸ has been changed to Ala (C198-A) to prevent autoprocessing. SHH-N refers to a form of amino-terminal SHH peptide that is not cholesterol-modified but used in various assays as a substitute for SHH-Np. B, lysate from cells transfected with plasmids expressing wt SHH or the various SHH holoprosencephaly mutations, SHH-D222N, SHH-T267I, SHH-G290D, SHH-A373T, SHH-P424A, or a control (Ctrl) plasmid was analyzed for the expression of the unprocessed full-length SHH (SHH-F), the processed SHH-Np, or tubulin (TUB.), by immunoblotting. The SHH-N arrowhead shown indicates where the non-cholesterol-modified protein migrates on this immunoblot. On average, we see a small amount of SHH-Np in the cell lysate of SHH-D222N-expressing cells, unlike the individual experiment shown here. C, cells expressing the indicated SHH mutants were assayed for their SHH activity by co-culturing them with cells expressing SHH dependent luciferase reporter gene. The activities of the various SHH mutants are shown as a percentages of wt SHH activity.

The observed levels of SHH-D222N and SHH-A373T activity were unexpected. The simplest explanation for this result was that the full-length form of these proteins, which appears to be the predominant form, is active. However, this surprising outcome might also result from some low abundance processed form of SHH-Np inefficiently produced from the mutant full-length SHH, as is observed for SHH-A373T. To distinguish between these possibilities, we engineered a SHH mutant that is unable to process into SHH-Np, SHH-U, by mutating the highly conserved Cys¹⁹⁸ that is absolutely essential for full-length SHH processing (18, 19) (Fig. 1A). We expressed SHH-U, SHH-D222N, SHH-T267I, and wt SHH in Bosc cells and examined the ability of these mutant proteins to process into their SHH-Np forms (Fig. 2A). Consistent with what has been previously described for SHH-U (18), the ability of this mutant to process into its SHH-Np form is completely attenuated, whereas the SHH-D222N mutant yields some small amount of its processed form. We next examined the activity of SHH-U by co-culturing cells expressing various SHH mutants with cells expressing a SHH-dependent luciferase reporter gene (supplemental Fig. S1). Cells expressing SHH-U were able to activate the SHH pathway ∼44% as well as wt SHH. Thus, although the SHH-U mutant is unable to convert into its processed SHH-Np form, it still exhibits significant levels of SHH activity when measured in co-culture assay. Consistent with this activity in a cell-cell contact assay, SHH-U localized to the plasma membrane of these cells (supplemental Fig. S2).

FIGURE 2.

Full-length unprocessed forms of SHH are active. A, the processing efficiency of wt SHH, SHH-D222N, SHH-T267I, and SHH-U was analyzed by immunoblotting the lysate of cells expressing the indicated constructs. The expression of unprocessed full-length SHH (SHH-F), the processed form of SHH (SHH-Np), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) are indicated. B, cell lysate, normalized to total cell protein (as described in Fig. S3), containing the indicated SHH mutants was added to cells expressing SHH-dependent luciferase reporter gene to determine their relative SHH activity. This comparison was done under conditions where the correlation between wt SHH activity and amount of cellular lysate assayed was linear. The activities of the various SHH mutants are shown as percentages of wt SHH activity. C, SHH was isolated from the cell lysate of SHH-U or wt SHH-expressing cells using antibodies specific to the SHH carboxyl-terminal domain and then immunoblotted with antibodies specific for the SHH amino-terminal domain. Lanes 1–3 show aliquots of the indicated cellular lysate, whereas lanes 4–9 show various specific or control immunoprecipitates (IP). For the wt SHH (WT) immunoprecipitation (lanes 8 and 9), the amount of lysate used is four times that used for SHH-U, because full-length SHH levels are lower in wt SHH-expressing cells (compare lane 2 with lane 3). D, full-length SHH was immunoprecipitated from indicated amounts of SHH-U or wt SHH-expressing cell lysate (depicted as μg of protein) using antibodies specific to the SHH carboxyl-terminal domain (dark gray bars, unstriped or striped) or control antibodies (light gray bars, unstriped or striped). Antibodies to the SHH carboxyl-terminal domain were also used to immunoprecipitate from 330 μg of cell lysate from cells transfected with a control plasmid (white bar). The various immunoprecipitates were then incubated with cells expressing SHH-dependent luciferase reporter gene to assay SHH activity. Ctrl, control.

To further validate the ability of SHH-U to activate the SHH pathway, we compared its activity to wt SHH using a cellular lysate assay (9), which allows us to compare wt SHH and SHH-U under conditions where we could easily determine the linear range of our assay. The ability of different amounts of cell lysate from wt SHH transfected Bosc cells to activate cells expressing a SHH-dependent luciferase reporter gene was first used to determine the linearity of this assay (supplemental Fig. S3A). Activation of the SHH reporter gene by wt SHH-containing cell lysate was inhibited by the SHH antagonist cyclopamine (20, 21) (supplemental Fig. S3B), consistent with this assay being specific for SHH. Lysates from Bosc cells transfected with plasmids expressing SHH-U, SHH-D222N, SHH-T267I, wt SHH, or a plasmid control were normalized to total protein content (Fig. 2A) and assayed for SHH activity (Fig. 2B) under conditions where this assay was linear (supplemental Fig. S3A). SHH-U and SHH-D222N activated the SHH pathway to ∼70% of wt SHH, whereas SHH-T267I exhibited only minimal levels of activity.

Our results suggest that the full-length form of SHH is capable of signaling. However, one could still argue that a small amount of full-length SHH is being proteolytically clipped in a nonspecific manner to reveal an active SHH fragment. To further confirm that full-length SHH can itself activate the SHH pathway, we immunoprecipitated SHH-U using two unrelated antibodies targeted to its carboxyl-terminal domain and determined its activity. The specificity of the antibodies used in this analysis was validated by evaluating their ability to immunoprecipitate full-length SHH, but not SHH-Np, from the lysate of Bosc cells expressing wt SHH or SHH-U (Fig. 2C and supplemental Fig. S4 show results for different antibodies). The various immunoprecipitates were resolved by SDS-PAGE and immunoblotted using antibodies to the amino-terminal domain of SHH. This result verified that the antibodies used here can specifically immunoprecipitate full-length SHH but not SHH-Np. We next determined the activity of these various immunoprecipitates by assaying their ability to activate the SHH- dependent luciferase reporter gene. The SHH-U immunoprecipitate activated SHH signaling in a dose-dependent manner (Fig. 2D and supplemental Fig. S4B), giving up to 3–5-fold more activity than a similar immunoprecipitate from control cell lysate or from immunoprecipitates using nonspecific antibodies. Moreover, the immunoprecipitate from lysates of wt SHH-expressing cells, in which the full-length SHH species is less abundant, resulted in less activity. The decreased activity observed in this latter control is consistent with the activity we describe here being derived from full-length SHH and not from some SHH fragment associating nonspecifically with the SHH carboxyl-terminal antibodies. Taken together these results confirm that the full-length form of SHH has the ability to activate SHH signaling.

If the activity of SHH-U we observe is derived from the full-length protein, we would expect to observe SHH-U interact with the SHH receptor PATCHED (PTC). We could then monitor this direct interaction as PTC-dependent endocytosis using antibodies to the SHH carboxyl-terminal domain and PTC (22). We first expressed a control plasmid, GFP, SHH-U, or the SHH receptor PTC in individual populations of Bosc cells and then verified that these distinct populations of cells only showed reactivity to antibodies directed against the proteins they were engineered to express (Fig. 3A, top three rows). We then co-cultured these different cell populations expressing control plasmid, GFP, SHH-U, or PTC (Fig. 3A, bottom two rows). Our results show that SHH-U is specifically taken up by PTC-expressing cells when SHH-U-expressing cells are physically associated with PTC-expressing cells ∼75% of the time (Fig. 3A, fourth row from top, and 3B) but not by GFP-expressing cells (Fig. 3A, bottom row, and 3B). This result suggests that SHH-U can interact with PTC in a functional manner. Consistent with this suggestion, the SHH inhibitory 5E1 monoclonal antibody was also able to attenuate the PTC-dependent endocytosis of SHH-U (supplemental Fig. S5).

FIGURE 3.

SHH-U interacts with the hedgehog receptor Patched. A, cells expressing either control plasmid (Ctrl cells), PTC (PTC cells), SHH-U (SHH-U cells), or GFP (GFP cells) were generated. Each population of cells was stained with 4′,6′-diamino-2-phenylindole (DAPI, blue) to mark the cells and antibodies specific to the SHH carboxyl-terminal domain (red) and PTC (green). Representative images are shown. First row, control cells; second row, PTC cells; third row, SHH-U cells; fourth row, PTC cells co-cultured with SHH-U cells; fifth row, GFP cells co-cultured with SHH-U cells. SHH-U is endocytosed by PTC when these two cell types are physically associated (fourth row, arrow). The co-culture figure (fourth and fifth rows) is displayed at 4× enhanced magnification, using Photoshop, relative to the various control panels. B, the ability of PTC cells to endocytose SHH-U from adjacent SHH-U cells was quantified in three separate experiments. On average, we observed SHH-U being endocytosed by 134 PTC-expressing cells out of 180 PTC-expressing cells in physical association with SHH-U-expressing cells counted. Conversely, 0 of 180 GFP-expressing cells in physical association with SHH-U-expressing cells internalized SHH-U.

SHH plays an important role in the dorsal-ventral patterning of the neural tube, regulating cell fate through the induction or repression of homeodomain transcription factors such as Nkx2.2 and Pax6 (23). Thus, the extent to which these transcription factors are regulated by ectopically expressed SHH can serve as an indicator of SHH activity in vivo. Therefore, we electroporated SHH-U, wt SHH, or a control plasmid into one side of the developing chick neural tube to determine whether SHH-U exhibited activity in this in vivo system (supplemental Fig. S6). Consistent with our in vitro data, SHH-U expression resulted in ectopic repression of the normal expression of Pax6 when compared with the nonelectroporated contralateral side of the neural tube (supplemental Fig. S6, O and T). SHH-U was also able to induce the ectopic expression of Nkx2.2 in more dorsal regions of the neural tube than observed in the contralateral control half of the neural tube (supplemental Fig. S6, N and S). The ability of SHH-U to regulate the expression of Pax6 and Nkx2.2 was ∼50% of that of wt SHH, when approximately equal amounts of expression plasmid were electroporated into a chick neural tube (Table 1). However, when the amount of wt SHH or SHH-U expressed was compared with the level of endogenous chicken SHH expression in the floor plate and neural tube, using the anti-SHH 5E1 monoclonal antibody, we noted that wt SHH appears to be more highly expressed than SHH-U (supplemental Fig. S6, compare C and H with M and R). This observation was similar to our in vitro observations (Fig. 2C), which also shows that wt SHH is more highly expressed than SHH-U. Based on this differential expression of wt SHH and SHH-U, we assume that the regulation of Nkx2.2 and Pax6 expression by SHH-U represents an underestimate of its inherent activity relative to wt SHH. These results demonstrate that SHH-U also has substantial activity in an in vivo setting, ectopically regulating the expression of two homeodomain genes normally regulated by SHH in the neural tube.

TABLE 1.

SHH-U induces different cell fates in the developing neural tube

SHH-U induces different cell fates in the developing neural tube. Plasmids expressing SHH-U or wt SHH were electroporated into one side of a Hamburger and Hamilton stage 10–12 chick neural tube. The ectopic induction or repression of the SHH-regulated homeodomain transcription factors Nkx2.2 and Pax6 was then determined by in situ analyses, relative to the nonelectroporated contralateral side of the neural tube in three separate experiments.

SHH constructs	Nkx2.2 induction	Pax6 repression	Total embryos
	%	%
wt SHH	79	68	28
SHH-U	48	32	25

If the activity of full-length HH protein is physiologically relevant, one might expect to observe its differential accumulation in various tissues. To test this prediction we examined distinct tissues for their relative amounts of full-length HH proteins. First we decided to probe adult rabbit gastric glands for the presence of full-length SHH, in which full-length SHH has been suggested to play a role in the maintenance of gastric function (11, 24), as well as stage 22 chick limb buds, in which SHH-Np has been shown to be involved in limb patterning (25). Cell lysates were generated from these primary tissues, followed by immunoblot analyses. Our results show that under conditions in which the levels of SHH-Np are normalized, increased levels of full-length SHH were found in gastric gland tissue relative to limb bud tissue (Fig. 4A). These results suggest that the ratio of different SHH species may be differentially regulated. To explore the evolutionary conservation of such differential processing, we performed a similar analysis on Drosophila tissues. Thus, we collected Drosophila embryos and imaginal disc tissue from third instar larvae and compared the ratio of full-length HH protein to HH-Np by immunoblot analyses (Fig. 4B). Our results show that full-length HH proteins are the dominant species of HH during fly embryonic development, in which multiple species of HH co-migrate at the molecular size of full-length HH. Consistent with these immunoreactive proteins being full-length HH, these bands were absent in lysate from embryos deficient in hh (hh^ac). In contrast, HH-Np was the dominant species of HH in imaginal disc tissue, although full-length HH was still observed in this tissue. Collectively, these results show that the accumulation of full-length HH protein is consistent with it playing a physiologically relevant role, albeit in a spatially and temporally defined manner.

FIGURE 4.

Full-length hedgehog proteins enrich differentially in tissues. A, SHH was isolated from the lysate of cells expressing SHH-U or wt SHH, limb buds of stage 22 chicken embryos, or rabbit gastric glands using antibodies specific to the SHH amino-terminal domain and then immunoblotted with another antibody specific for the SHH amino-terminal domain. Lanes 1-6 show various specific or control immunoprecipitates for SHH from control (Ctrl), SHH-U, and wt SHH overexpressing cells, whereas lanes 7–10 show endogenous SHH from specific tissues. The lower panel is a shorter exposure of the same blot to visualize SHH-Np normalization. B, immunoblot analysis of lysates from Drosophila expressing UAS-hh, hh^ac, or wt Drosophila reveals the differential expression of full-length HH (HH-F) to HH-Np in wt embryos relative to wt imaginal discs or UAS-hh-expressing heads (compare lane 1 with lanes 3 and 4). hh^ac is a null allele, so it acts as a control for the specificity of the anti-HH antibody used. All of the panels are part of the same blot, shown at different exposure levels (lanes 3 and 4 showing full-length HH and HH-Np are longer exposure than lanes 1 and 2, whereas all lanes showing tubulin are same exposure). The size standards are in kDa. A portion of the membrane was probed for tubulin (TUB.) as a loading control.

To determine whether the inherent activity of full-length HH proteins was evolutionarily conserved, we examined the activity of analogous Drosophila HH mutants during eye development, where HH is thought to act predominantly in a localized manner (26). Using an eye-specific GAL4 driver, we expressed UAS constructs hh-U (15), wt hh (16), and hh-N (16) (analogous to SHH-U, wt SHH, and SHH-N (Fig. 1A)) in the eye imaginal disc and examined their ability to rescue the ommatidia development defect found in hh¹ mutant Drosophila (27, 28). Because the summation of individual ommatida development within each adult eye gives a quantitative determinant of HH activity, we were able to quantify the relative activity of these various HH proteins by determining the number of developed ommatidia per retina in transgenic flies (Fig. 5A). From these analyses we observed that the ommatida and lamina developmental defects of these mutant flies (Fig. 5A, second column) were substantially rescued by the expression of hh-U (Fig. 5A, fourth column), wt hh (Fig. 5A, third column), and hh-N (Fig. 5A, last column). Notably, the ability of HH-U and wt HH to rescue ommatidia development was quite similar. Moreover, hh-N expression resulted in more correctly developed ommatida (Fig. 5A, last column), consistent with this in vivo activity assay not being saturated for HH. The inability of the HH-U protein to be processed into HH-Np in vivo was verified by immunoblotting imaginal discs removed from the various transgenic flies (Fig. 5B). These results also clearly demonstrate the ability of full-length HH proteins to activate some aspects of HH signaling in vivo, without the requirement to process full-length HH to its cholesterol-modified HH-Np form. In addition, these results show that the inherent activity of full-length HH proteins is evolutionarily conserved, implicating the importance of this novel HH activity to its function.

FIGURE 5.

The activity of full-length, unprocessed hedgehog proteins is evolutionarily conserved. A, the extent of rescue of the eye hh¹ phenotype was established by quantification of ommatidial development in the wild type, hh¹, and hh¹ animals bearing one of the transgenes: wt hh, hh-U or hh-N driven by eye-specific GAL4 driver GMR-GAL4. Between 4 and 10 retinas were counted for each genotype. B, immunoblot analysis reveals the predicted protein products from the three transgenes wt hh, hh-U, and hh-N expressed in eye imaginal discs. Notably, the product from hh-U appears to migrate at a similar molecular weight to the full-length unprocessed wt HH polypeptide (HH-F) product prior to its auto-cleavage. The size standards are in kDa. A portion of the membrane was probed for actin expression, as a loading control.

DISCUSSION

The well accepted model of HH biogenesis suggests that the cholesterol-modified, processed form of HH, HH-Np, is the only known active HH species. Contrary to this model, our results suggest that full-length unprocessed HH proteins also exhibit substantial levels of activity. We speculated that the two most likely reasons for our results are that the full-length unprocessed HH proteins retain some substantial level of activity or that we were observing the activity of some small undetectable amount of inefficiently processed HH-Np. We favor the former model because of our analyses of HH-U proteins, which are unable to process into their cholesterol-modified forms. However, it remains possible that these full-length HH-U proteins are subject to some small amount of nonspecific proteolytic cleavage that liberates an active amino-terminal product, which would then be responsible for the bulk of activity we observe here. Our results are not consistent with this latter hypothesis for three main reasons: 1) the quantitative nature of the assays used here, both in vitro and in vivo, to determine the activity of the full-length HH proteins are inconsistent with the majority of the activity observed being contributed by a small undetectable fraction of the protein, 2) we were able to specifically immunoprecipitate active SHH-U using two distinct antibodies to the SHH carboxyl-terminal domain, and 3) the holoprosencephaly mutant SHH-T267I, which should be subject to the same putative nonspecific proteolytic clipping as SHH-U and SHH-D222N, exhibited negligible activity.

Contrary to the results presented here, it has been previously suggested that full-length HH proteins are not active (18, 19, 25). We do not know the reasons for the apparent discrepancy between the results of the various groups involved but speculate that the level of expression and type of assay used may be important to visualize the activity of full-length HH proteins.

Our ability to assay the activity of HH-U proteins was limited to assays that involved cell-cell contact, with the activity contributed by any SHH-U in the conditioned media being negligible (data not shown). Consistent with this observation, SHH-U was only internalized by PTC when cells expressing each construct were in direct contact. Our ability to rescue the ommatidia defect of hh¹ flies might also be due to more localized signaling, because HH is only thought to act over small distances in the fly eye (26). This suggests that the biological functions of HH-U proteins might be limited to those that do not require HH to act far from its site of synthesis. Although it is generally accepted that long and short range HH signaling is differentially regulated, this research has focused on how other distinct proteins direct the processed, cholesterol-modified form of HH proteins through different mechanisms (7, 14, 16, 29–33). Our results broaden this discussion, because we suggest that unprocessed full-length HH proteins might be directly responsible for a subset of localized HH signaling. In this model, long and short range HH signaling would be controlled by distinct forms of HH proteins, with long range signaling regulated solely by the cholesterol-modified processed protein and short range signaling controlled, at least in part, by full-length unprocessed HH proteins. Consistent with this model, we did not observe significant accumulations of full-length HH proteins in tissues thought to depend primarily on ability of HH to signal over extended distances, such as chicken limb buds (34, 35). In contrast we detected large amounts of full-length HH protein in Drosophila embryos, in which HH is thought to act in a more localized manner (34, 36). In conclusion, our results suggest that the relationship between HH processing and activity is more complex than previously thought and that in some biological contexts the full-length forms of HH family members may play a significant physiological role.