A head-activator binding protein is present in hydra in a soluble and a membrane-anchored form

A very important step early in development is the decision of a cell to continue cycling or to become committed to a specific fate. In hydra the neuropeptide head activator (HA) functions as a signal for such a decision. If present at picomolar concentrations, stem cells continue cycling. At 100-fold higher HA concentrations stem cells become determined to undergo differentiation: interstitial cells to nerve cells, and epithelial cells to head-specific tentacle and hypostomal cells (reviewed in Schaller et al., 1996). How this differential effect of HA is achieved is unclear.

HA in hydra is produced by nerve cells and is present in a vesicle-stored form in a gradient from head to foot. When hydra are induced to regenerate a head, for example by cutting off the original head, HA is released from the cut surface. This release is necessary to initiate the processes which, at the morphological level, lead to restoration of a functioning head and, at the cellular level, to head-specific determination and terminal differentiation. Similar activation processes govern foot regeneration requiring foot-specific inductive signals (Schaller et al., 1996; Grens et al., 1999).

To understand the differential effects of HA on cell proliferation and differentiation we searched for proteins mediating HA actions. We had found that nerve-cell differentiation, induced by HA, involves activation of the cAMP pathway including CREB as a transcription factor (Fenger et al., 1994; Galliot et al., 1995). This implied that a signal-transducing HA receptor is a member of the G protein coupled seven transmembrane receptor family. HA, as released during head regeneration or isolated under isoosmolar conditions from hydra tissue, was found to be associated with a protein complex of molecular mass higher than 800 kDa (Schaller et al., 1986, 1989). We envisaged that the function of such a complex would be in prevention of HA degradation and limitation of diffusion thus ensuring local action. We describe in this paper the isolation and characterization of a HA-binding protein (HAB) which, in its membrane-anchored form may serve as receptor and/or in its secreted form as coreceptor, thus providing an excellent means of controlling the range of action and local concentration of HA.

HA (Glp-Pro-Pro-Gly-Gly-Ser-Lys-Val-Ile-Leu-Phe¹¹) and amino acid derivatives were purchased from Bachem (Heidelberg, Germany). [Tyr¹¹]HA in which Phe¹¹ is replaced by Tyr was prepared by solid phase peptide synthesis. [Tyr¹¹]HA was iodinated using the chloramine T method and purified by reverse-phase HPLC. HA bipeptide and ¹²⁵I-HA bipeptide were prepared as described by Neubauer et al. (1991). Synthesis of ¹²⁵I-Bpa-HA-HA bipeptide was performed according to the method of Hampe et al. (1996).

A multiheaded mutant of Chlorohydra viridissima (Lenhoff, 1965) and Hydra vulgaris were cultured as described by Scheurlen et al. (1996). Membranes and soluble fractions from 2-day starved hydra were prepared according to the methods of Neubauer et al. (1991). Briefly, hydra were homogenized in an isotonic buffer, large particles were removed by centrifugation at 1000 g, and membrane vesicles were sedimented at 50,000 g. Subsequently the content of the vesicles was released by hypotonic shock. A second high speed centrifugation at 50,000 g yielded the membrane fraction and a supernatant containing the soluble proteins from the vesicular fraction.

HAB was solubilized from the membrane fraction by gentle agitation in 50 mM Mes, pH 6 with 1% Triton X-100, 1 M NaCl and protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 1 mM Pefabloc SC and 1 µg/ml leupeptin) for 60 minutes at 4°C. Non-solubilized material was removed by centrifugation at 100,000 g for 30 minutes.

Binding studies and photoaffinity labeling were performed as described by Hampe et al. (1996). ¹²⁵I-HA bipeptide binding to soluble HAB was measured in a binding assay using GF/F glass fiber filters (Whatman) presoaked in 0.3% polyethyleneimine to separate free from bound HA ligand. All binding experiments were performed in triplicate and the standard deviations calculated. For determination of ligand-binding activity of the affinity purified protein, 50 µl dialyzed MgCl2 eluate was used for each tracer concentration.

Protein concentrations were determined with the bicinchoninic acid protein assay (Pierce) using bovine serum albumin as a standard.

For HA affinity chromatography HA was coupled to activated CH-Sepharose 4B (HA Sepharose) as described by Franke et al. (1997). The soluble vesicular hydra proteins (50 ml) were adjusted to pH 6.0 with 50 mM MES and, after addition of 0.5 M NaCl, 1% Triton X-100, 10 mM MgCl₂, 10 mM CaCl₂ and protease inhibitors, incubated with 2 ml HA Sepharose overnight at 4°C. After rigorous washing HAB was eluted from the matrix with 2 M acetic acid in 0.1% Triton X-100. If native HAB was wanted for ligand-binding experiments, elution was performed using 4 M MgCl₂ in 0.1% Triton X-100, and the eluate was dialyzed against 10 mM ammonium acetate pH 6.0, 1 mM CaCl₂, 0.1 mM KCl, 0.1 mM MgCl₂, 0.1% Triton X-100.

Oligonucleotide primers were constructed from partial peptide sequences of the 200 kDa hydra HAB. One pair, GGITTIGC(G/T)-AT(A/T)GATTATGT(A/C/G/T)GA(A/G)AA and GC(G/T)AT(G/T)-GC(G/T)ACIGCGAA(A/C/G/T)AC(A/G)TA, derived from the peptides YPENGLAIDYVENR and XYVFAVAIAHXNVE, resulted in PCR amplification of a 1600 bp fragment on an oligo(dT) primed cDNA from the multiheaded mutant of Chlorohydra viridissima (using 6 cycles with an annealing temperature of 45°C and 26 cycles at 55°C). This fragment served to screen a λZAP cDNA library from the multiheaded hydra mutant. The DNA was sequenced in both directions using internal oligonucleotide primers.

The 500 bp HAB fragment from Hydra vulgaris was amplified by PCR with the primers TATAGCTATGATGAAGGGAATACATGG and ATCACCAGGGACTTTGCGGTATCC at an annealing temperature of 55•C from an oligo(dT) primed cDNA.

Amino acid sequences corresponding to the deduced HAB fragment from Hydra vulgaris were aligned using the GeneBee server (http://www.genebee.msu.su). Phylogenetic analysis was performed using the program PUZZLE (Strimmer and von Haeseler, 1997). Irrespective of the parameters the program assigned a support value of 100% to each internal branch. Branch lengths correspond to the number of amino acid changes.

For northern blotting 2 µg of poly(A)⁺ RNA from Hydra vulgaris were subjected to electrophoresis and blotted onto nitrocellulose membrane. Hybridization with a ³²P-labeled HAB probe derived from Hydra vulgaris corresponding to aa 545-719 (Fig. 2A) was carried out using ExpressHyb Hybridization Solution (Clontech) at 50°C. The filter was washed with 2× SSC, 0.1% SDS at 37°C and autoradiographed using Kodak BIOMAX MS film.

HA and recombinant human RAP, and, as controls, bradykinin and ethanolamine were coupled to activated Sepharose 4B as described earlier (Jacobsen et al., 1996; Franke et al., 1997). Solubilized hydra membranes or HAB from the soluble vesicular proteins were incubated with the different Sepharoses at 4°C overnight. The amount of HAB in supernatant and retentate was assayed by western blotting and the HA-binding activity by filter-binding assay.

A protein containing the fibronectin type III domains of hydra HAB from amino acids 1357 to 1506, tagged amino-terminally with His6, was produced in E. coli using the vector pQE31 (Quiagen). The protein was solubilized with 6 M guanidine hydrochloride, purified on Ni-NTA resin (Quiagen), and used to immunize rabbits. Samples to be analyzed were subjected to electrophoresis under reducing conditions and transferred to Immobilon-P membranes (Millipore) by semi-dry blotting. HAB was detected using the HAB antiserum at a dilution of 1:3,000 and an alkaline phosphatase-conjugated goat anti-rabbit serum (Promega). Preabsorption was carried out by incubating the HAB antiserum with recombinant purified fibronectin domains blotted to Immobilon-P membranes.

Membranes from the multiheaded hydra mutant were solubilized in 1% Triton X-114, 0.15 M NaCl, 50 mM MES pH 6.0 and protease inhibitors for 60 minutes at 4°C. After centrifugation at 125,000 g CHAPS (0.1%) was added to the supernatant and heated to 37°C for 15 seconds. Centrifugation at 15,000 g resulted in separation of an aqueous phase containing most of the solubilized HAB and a detergent phase with the Triton X-114. The aqueous phase (500 µl) was incubated with the indicated antisera (50 µl) for 1.5 hours on ice. After precipitation with protein A Sepharose a ligand-binding assay was performed on the precipitate and the supernatant as described above.

Hydra membranes were incubated in the presence of protease inhibitors either with 50 mM MES pH 6, 1 M NaCl with or without 1% Triton X-100 or with 100 mM Na2CO3 pH 11.5 for 2.5 hours at 4°C. Soluble and membrane-bound proteins were separated by centrifugation at 125,000 g for 60 minutes. HAB content in the supernatants was assayed by western blotting, and HA-binding activity by ligand-binding analysis.

HA affinity purifed HAB from hydra membranes and from the soluble vesicular hydra proteins were dried in vacuo and incubated with 2 M NH2OH-HCl, 6 M guanidinium-HCl, adjusted to pH 9 with LiOH for 4 hours at 45°C (Bornstein and Balian, 1977). After cleavage the samples were desalted on a NAP5 column (Pharmacia), concentrated in vacuo and subjected to reducing 12% SDS-PAGE and western blotting using the HAB antiserum.

The expression pattern of HAB in whole animals was determined using the in situ hybridization procedure described by Grens et al. (1996). The HAB probes used for digoxigenin labeling were antisense RNAs from the multiheaded mutant of Chlorohydra viridissima (corresponding to the first 230 aa of the open reading frame; further probes yielded identical results) and from Hydra vulgaris (corresponding to aa 545-719 in Fig. 2A). Sections of the multiheaded mutant of Chlorohydra viridissima were hybridized with a ³⁵S-UTP labeled antisense RNA probe (corresponding to aa 1215-1356) and autoradiographed according to the method of Hermans-Borgmeyer et al. (1996).

For single cell preparations hydra were macerated in 7% acetic acid in 7% glycerol for 30 minutes at 25°C. Paraformaldehyde was added to a final concentration of 4% before spreading cells on slides. After drying, the slides were washed twice with 0.1% Triton X-100 in PBS, treated with 0.1 M glycine, pH 7.2, for 25 minutes, and blocked with 1% bovine serum albumine, 0.1% casein hydrolysate for 45 minutes. Incubation with primary antiserum was carried out overnight and with secondary alkaline phosphatase conjugated antibody for 3-4 hours at room temperature.

Gastric regions from daily fed multiheaded mutants of Chlorohydra viridissima were sliced in hydra medium (cutting medium), transferred to fresh medium and collected 4 hours later. Hydra and pieces were homogenized by ultrasonication, an aliquot was removed for determination of protein content, and the rest used for western blot analysis. Conditioned media were concentrated using Strataclean resin (Stratagene) before analysis by western blotting.

In membrane fractions of hydra two types of HA-binding sites were found, a ‘low-affinity’ site with a K_d in the nanomolar range and a 50-100 times less abundant ‘high-affinity’ site with a K_d in the picomolar range. The low-affinity HA-binding site was 30-fold more abundant in the multiheaded mutant of Chlorohydra viridissima than in wild-type animals (Neubauer et al., 1991). The low-affinity HA binding site was solubilized from membrane fractions of the mutant and purified by HA-affinity chromatography. A 200 kDa protein was isolated, which we designate HA-binding protein, HAB, because it bound to and could be photoaffinity labeled with HA ligands (Hampe et al., 1996; Franke et al., 1997).

HA-binding activity was also present in a vesicular fraction which contained HA. HA and the binding activity could be released by osmotic shock. The HA-binding activity was not sedimentable at 100,000 g, and was therefore soluble. Scatchard analysis revealed a nanomolar K_d identical to that found for HAB isolated from the membrane fraction. Photoaffinity labeling marked a soluble 200 kDa protein (Fig. 1A). It was labeled by the HA photoligand in the absence, but not in the presence, of excess unlabeled HA. An unidentified protein of about 160 kDa was also specifically labeled. Purification on HA Sepharose allowed isolation of the soluble 200 kDa HAB (Fig. 1B), which bound HA ligands with an affinity very similar to HAB from hydra membranes (Fig. 1C). To clone the gene coding for hydra HAB, we determined the amino-terminal peptide sequence of the purified membrane protein and sequenced eight peptides after a tryptic digest. Based on these peptide sequences oligonucleotides were deduced and used as primers for an initial polymerase chain reaction (PCR) on cDNA from the multiheaded mutant of Chlorohydra viridissima. The entire open reading frame was derived by several rounds of screening a λZAP-cDNA library of the mutant. The cDNA (GenBank accession number AF092920) encoded a protein of 1661 amino acids with a predicted molecular mass of 189 kDa. It contained all the sequenced peptides (underlined in Fig. 2A), proving identity of the cloned cDNA with purified HAB. Hydrophobicity analysis revealed that HAB, in addition to an amino-terminal signal peptide, contains a single transmembrane segment located near the carboxy terminus (Fig. 2B). Therefore, HAB is a type Ι membrane protein with a short 55 amino acid long carboxy-terminal tail in the cytoplasm.

HAB is made as a proprotein, a furin-cleavage site with the consensus motif RXXR (Molloy et al., 1999) is located directly in front of the sequenced amino terminus of the mature protein at amino acid 84 (Fig. 2A). Cleavage at this site separates a short propeptide from the residual membrane-anchored protein. The large extracellular part of HAB contains several possible glycosylation sites explaining the difference between calculated (179 kDa) and apparent (200 kDa) molecular mass. It is composed of several domains never found before in this combination and alignment in other proteins. The amino-terminal half of hydra HAB shows homology to the yeast sorting protein VPS10, including the spacing of 12 conserved cysteine residues (highlighted in Fig. 2A by oval shadowing). This VPS10 domain is followed by a module typical for the family of low density lipoprotein (LDL) receptors. It consists of six LDL receptor class B repeats with a YWTD motif, an EGF-like domain, and seven cysteine-rich class A repeats. Two fibronectin type III domains precede the transmembrane region (Fig. 3A). In the 3’ non-coding region of the HAB cDNA five copies of the cytoplasmic polyadenylation signal (U/A)UUUUA(U/A) and one AUUUA signal for fast mRNA degradation hint at a highly regulated mRNA expression (Shaw and Kamen, 1986; Sheets et al., 1994; Scheurlen et al., 1996).

HA is present in identical sequence in mammals (Bodenmüller and Schaller, 1981; Schaller and Bodenmüller, 1981) where it stimulates cell proliferation of neuronal precursor cells and enhances neurite outgrowth (Quach et al., 1992; Ulrich et al., 1996; Kayser et al., 1998). While this work was in progress homologues of hydra HAB with the same combination and alignment of modules were isolated from human, rabbit and chicken by searching for new LDL-receptor homologues (Jacobsen et al., 1996; Yamazaki et al., 1996; Mörwald et al., 1997). To study the function of HAB in mammals, we cloned a mouse homologue. Its specific expression in neurons of the developing and adult brain is compatible with a morphogenetic function of HA in mammals (Hermans-Borgmeyer et al., 1998). HABs from hydra and from the higher organisms differ only in the number of class A and fibronectin type III domains (Fig. 3B). The most conserved part of HAB between mammals and hydra is the carboxy-terminal tail (Fig. 3C), which contains highly conserved motifs typical for protein-protein interactions required for internalization, phosphorylation, and trans-Golgi sorting. A putative motif for G-protein coupling is also present (Okamoto et al., 1990; Anand-Srivastava et al., 1996). The homology between domains of hydra HAB and its vertebrate counterparts is higher than to any other protein. Intensive screening of the EST data base revealed no further mammalian HABs, and PCR experiments supported this notion. Taken together we conclude that hydra HAB and its mammalian homologues are orthologues.

The only other known protein with a VPS10 domain in higher organisms, sortilin (Petersen et al., 1997), was recently shown to be a third receptor for the neuropeptide neurotensin (Mazella et al., 1998). Like HAB, sortilin is a type I transmembrane receptor, but it neither contains LDL receptor nor fibronectin type III domains. A phylogenetic tree based on the highly conserved carboxy-terminal part of the VPS10 domain (Fig. 3D) emphasizes the close relationship between hydra HAB and its vertebrate homologues (48% amino acid identity). The VPS10 domain of HAB is less related to sortilin (31%) and to the yeast protein VPS10 (26%). HAB from a different hydra species, Hydra vulgaris (see below), differs much more from HAB of the multiheaded mutant of Chlorohydra viridissima (74% identity) than HABs from mammals and birds (92%), or from mouse and human (97%), in line with the old evolutionary origin of hydra and an early split of hydra species.

To study HAB function and identity with the soluble 200 kDa HA-binding protein, a polyclonal rabbit antiserum was raised against the hydra HAB fibronectin domains, produced in E. coli with a hexahistidine tag. The α-HAB antiserum recognized similar amounts of both the membrane-bound and the soluble HAB, both migrating in SDS gels as bands of approximately 200 kDa (Fig. 4A). Pre-immune serum or serum preabsorbed to recombinant protein were used as control and did not recognize HAB. HA Sepharose, but no control Sepharose bound the two immunopositive HAB forms.Likewise only HA Sepharose was able to precipitate the HA-binding activities (Fig. 4B). The HA-binding activity was also precipitable with Sepharose, to which the human receptor associated protein (RAP) was coupled (Fig. 4B). Since RAP interacts with many LDL-receptor like proteins, this is additional evidence that the low affinity HA-binding site from hydra belongs to this family. The HA-binding activity was immunoprecipitated from solubilized membranes and from a hydra fraction containing soluble vesicular proteins with α-HAB, but not with pre-immune serum (Fig. 4C). These experiments confirm the identity between soluble and membrane-bound HA-binding activities and HAB.

We investigated the transition from the membrane-bound to the soluble form by analyzing both extractability from the membrane and differences in the carboxy-terminal part of the proteins. For affinity purification, membrane-bound HAB was solubilized with 1% Triton X-100 in the presence of 1 M NaCl. Treatment of hydra membranes with this combination released HAB, as did incubation with Na₂CO₃, which is used to detach peripherally attached membrane proteins (Thomas and McNamee, 1990). Treatment with 1 M NaCl also converted some membrane-bound HAB into the soluble form indicating that at least part of membrane HAB is not an integral membrane protein (Fig. 5A). After incubation with 50 mM MES, pH 6, neither HAB nor HA binding was found in the supernatant, whereas after treatment with NaCl or Na₂CO₃ up to 20% of the HA-binding sites were released from the membranes.

To determine whether soluble HAB lacks the putative transmembrane domain, we studied HAB expression in COS7 cells. We found that a 640 amino acid carboxy-terminal HAB construct containing part of the extracellular domain, the putative transmembrane domain, and the intracellular tail, was present exclusively in cell membranes. Conditioned medium was devoid of the HAB fragment, and membrane-bound HAB could not be released from COS7 membranes by Na₂CO₃ treatment. This is taken as evidence that soluble HAB in hydra lacks the putative transmembrane domain.

The existence of membrane-bound and soluble forms of HAB could be due to alternative splicing. We exclude this, since only one dominant mRNA size was detected on northern blots, and since in four independently isolated cDNA clones the transmembrane domain was always encoded. Alternatively, the soluble form could be generated by proteolytic cleavage between the fibronectin and the transmembrane domain, as found in other proteins (Sadoul et al., 1988; Probstmeier et al., 1992). The calculated molecular mass of the transmembrane and the intracellular domain is 8 kDa. Since this is not easily detectable in a 200 kDa glycosylated protein, we treated affinity purified HAB from hydra membranes and from the soluble fraction with hydroxylamine. This reagent should cleave HAB between Asn¹²¹⁴ and Gly¹²¹⁵ (Fig. 2A) thus generating a 447 amino-acid long carboxy-terminal fragment. From the membrane fraction two cleavage products in similar amounts were detected with αHAB, one migrating with an apparent molecular mass of 73 and one of 58 kDa (Fig. 5B). The soluble HAB yielded only one cleavage product of 58 kDa. This supports the notion that HAB is processed to the soluble form by a protease cleaving outside the transmembrane region. No obvious protease cleavage site is recognizable at this position.

In situ RNA hybridizations with probes from different parts of HAB show especially strong signals in the ectoderm of the subhypostomal region of the multiheaded mutant of Chlorohydra viridissima (Fig. 6A). This mutant is characterized by the production of many heads, no buds and very few feet (Lenhoff, 1965). Since in the subhypostomal region proliferating cells become determined for their final differentiation fates, interstitial cells for nerve-cell and epithelial cells for head-specific determination, HAB may play a role in this process. Hydra has only two cell layers, ectoderm and endoderm. HAB is present in both with stronger expression in the ectoderm (Fig. 6B-D). In northern blots a mRNA of about 7 kb was labeled. Southern blotting, even under low-stringency hybridization conditions, showed that the HAB gene is a single copy gene (data not shown).

The nucleotide probes from Chlorohydra viridissima did not cross-react in northern blots nor in in situ hybridizations with mRNA from Hydra vulgaris. To study distribution of HAB in a hydra with a non-mutant morphology, we amplified a highly conserved part of the VPS10 domain from Hydra vulgaris cDNA by PCR (nucleotides 448-914 of GenBank accession number AF159157). In situ RNA hybridizations on this hydra species showed expression of HAB both in head and foot regions and in developing buds (Fig. 6E). HAB expression demarcated areas where determination of cells for head- or foot-specific fates occurs. The midgastric region, which contains mostly proliferating cells, showed lower levels. This finding is strongly supported by northern analysis where HAB expression was minute in the mid-gastric region, but strong in head and foot regions (Fig. 7). No strong expression of HAB mRNA was observed in very early buds, but signal intensity increased in later bud stages (Fig. 6E). The future head region of buds showed strong signals shortly before tentacle outgrowth occurred. Similarly, HAB mRNA levels increased in future foot regions. The basal disc of buds and of mature hydra appeared free of HAB mRNA. Sense probes were used for all constructs and showed no reactivity (Fig. 6B,F).

To analyze HAB in cells, hydra were dissociated by maceration and immunoreacted with α-HAB. Immunoreactivity was found in almost all cell types of hydra, but in very varying degrees (Fig. 8). Both ecto- and endodermal epithelial cells stained strongly with α-HAB (Fig. 8A) and staining was punctate (Fig. 8D). Interstitial cells of all nest-size classes were labeled (Fig. 8B-D,F), as were nerve cells (Fig. 8C,F). Staining of cells was prevented if the α-HAB antiserum was preabsorbed with the recombinant HAB fibronectin domains used as antigen (Fig. 8G,H).

For HA we had found that release from the head-regenerating tip is necessary during the first 2-4 hours after cutting to initiate head regeneration. To study HAB levels during head regeneration, gastric regions of the multiheaded mutant were salami-sliced and collected 4 hours later. Gastric regions without cutting, head regions, and total hydra were also collected, and variable amounts of protein subjected to SDS-gel analysis prior to immunostaining (Fig. 9A). HAB content in the 4-hour regenerates was almost undetectable indicating a drastic decrease. Concentration of HAB in gastric regions was lower than in whole hydra, whereas it was highest in heads. HAB decrease was partly due to release, as evidenced by its appearance in the medium, in which the animals were cut (Fig. 9B). Control hydra without cuts released no HAB into the medium. The time period of cutting extended over a period of 0-30 minutes indicating very rapid and immediate release of HAB within hydra and into the medium. HAB content in the medium did not increase with regeneration time indicating that HAB release was an immediate event and/or that degradation of HAB also occurred rapidly.

We found that HA in hydra interacts with the HA-binding protein HAB. HAB is produced as a type I transmembrane receptor and is processed in hydra to a secreted product. The amino-terminal sequence of mature HAB starts at amino acid 85 of the open reading frame, indicating that HAB is made as a proprotein. Mature HAB starts with a VPS10 domain which was first discovered in a yeast sorting protein. HAB, like all other VPS10 domain-containing receptors, contains a cleavage motif for a furin-like protease. For the neurotensin receptor sortilin (Mazella et al., 1998) cleavage at this site is required before ligands can bind (Petersen et al., 1999). Since HAB purified by HA-affinity chromatography is devoid of the propeptide, we assume that only processed HAB is able to bind HA. The sortilin propeptide, both before or after cleavage, inhibits ligand binding. It would therefore be interesting to find out whether the HAB propeptide also has such a function, which would provide a very subtle means of regulating HA binding to HAB and HA action.

Mature secreted and membrane HAB display the same, 200 kDa, apparent molecular weight in SDS-PAGE. For better resolution we proteolytically cleaved HAB from the two sources. Cleavage at the extracellular face of the transmembrane domain. Presence of HAB as a transmembrane protein, as a protein peripherally attached to membranes, and as a secreted product provides further means to regulate the range of action of HA in time and space.

During head regeneration in hydra HAB, like HA (Schaller, 1976), is rapidly secreted into the medium. We interpret this to mean that conversion of the transmembrane to the secreted form of HAB and/or its release is highly regulated and occurs very fast after stimulation. Secreted binding proteins have been found for a variety of other hormones, cytokines, and growth factors. Some of them are derived from transmembrane proteins and are also thought to modulate the local actions of their ligands (Hanneken et al., 1994).

HAB is a very large protein consisting of several domains found in identical alignment in only one protein in higher of membrane HAB yielded two carboxy-terminal fragments differing in size by 15 kDa. From secreted HAB only the smaller fragment was derived. This indicates that secreted HAB is released by proteolytic cleavage organisms, designated SorLA by one group (Jacobsen et al., 1996) and LR11 by another (Yamazaki et al., 1996). Taking into account the high homology between these two proteins and the absence of other homologues, we assume that hydra HAB and SorLA are orthologues. By containing class A and B repeats and an EGF-like domain HAB belongs to the LDL-receptor family (Gliemann, 1998). Members of this family, especially the VLDL receptor, the ApoE receptor 2, and the multiligand receptors LRP and megalin, are very important for brain development as evidenced by knock-out of the respective genes in mice. These proteins are important for internalization of extracellular ligands, but some evidence suggests that they also play a direct or indirect role in signal transduction (Willnow et al., 1996a; Gliemann, 1998; Trommsdorff et al., 1998, 1999; Hilpert et al., 1999). One member of the family, LRP, is posttranslationally modified by a protease and is found in a secreted form in human plasma (Willnow et al., 1996b; Quinn et al., 1997). To understand the function of SorLA in mammals, we cloned a mouse homologue (Hermans-Borgmeyer et al., 1998). We found very prominent expression in the adult and developing brain in very specific neuronal cell populations. During neuronal development in the telencephalon SorLA mRNA is abundant in cortical, but absent from basal zones. This suggests that SorLA is involved in establishing borders. Expression is strong in areas containing differentiating neurons, but is absent from regions where proliferation of stem cells occurs (Hermans-Borgmeyer et al., 1998). HAB mRNA expression in hydra also demarcates areas where determination to head- or foot-specific fates is observed.

This implies that HAB in hydra may have similar functions as in mouse. Its presence close to the foot region indicates that HAB interacts not only with HA as a factor important for head-specific determination, but also with foot-specific signals. For both signals we had found that they are bound to carrier complexes (Schaller et al., 1986; Schaller, 1996). HAB might be part of these complexes thus ensuring local action of inducing signals. A similar mode of action has been proposed for secreted or membrane-bound binding proteins of other morphogens like the Wnt proteins, TGFβ1, or hedgehog (Chuang and McMahon, 1999; Hsieh et al., 1999; Munger et al., 1999). Alternatively HAB may be necessary to accrue enough HA locally to achieve the high concentration necessary for determination as opposed to cell cycle progression, in analogy to the function of glycosaminoglycans for the Wnt proteins or the fibroblast growth factors (Schlessinger et al., 1995; Cumberledge and Reichsman, 1997). It remains to be shown whether HAB presents HA to a signaling receptor or whether the transmembrane form of HAB can itself transduce the HA signal to G proteins like some other single membrane-spanning receptors with short cytoplasmic tails (Okamoto et al., 1990; Anand-Srivastava et al., 1996).

We are grateful to F. Buck for peptide sequencing, to I. Hermans-Borgmeyer for discussion, and to O. Sperl for the artwork. This work was supported by the Deutsche Forschungsgemeinschaft (SFB 444).