A 25.7×10³M_r hydra metalloproteinase (HMP1), a member of the astacin family, localizes to the extracellular matrix of Hydra vulgaris in a head-specific manner and has a developmental function

Because of its high regenerative capacity and its simplified histological organization (an epithelial bilayer with an intervening extracellular matrix; ECM) hydra is an excellent developmental model for analysis of cell/ECM interactions (Sarras et al., 1991a,b, 1993, 1994). Previous studies have shown that ECM plays a critcal role in a broad spectrum of developmental processes in hydra. In this regard, it has been established that in general, ECM functions at a number of levels to modulate development. At one level, ECM components have been shown to affect cellular processes through signal trans-duction pathways involving a number of cell surface receptor systems. These receptor systems include (1) ECM-specific receptors such as integrins, which interact with particular sequences within ECM components (e.g. RGD sequences; see review by Hynes, 1992) and (2) growth factor receptors, which interact with growth factors bound to the ECM (e.g. bFGF or TGF-β; see review by Schubert, 1992). At another level, ECM is thought to affect morphogenesis of tissue by functioning as an extracellular barrier that restricts the evagination or invagination of epithelial sheets. During histogenesis, the restriction imposed by the ECM is thought to be selectively eliminated at particular regions along an epithelial sheet by the action of matrix metalloproteinases that degrade ECM components and thereby weaken the ECM as a structural barrier. The selective degradation of ECM in regions of increased cellular proliferation rates and/or cell shape changes would result in an out-pocketing in the epithelial layer and lead to a transition from a flat epithelial sheet to a branched tubular structure as seen in glandular tissues (see review by Hay, 1991). With this in mind, the current study was designed to build on previous studies and investigate the potential role of ECM-associated metalloproteinases during hydra morphogenesis, especially as related to head regeneration.

The main species used in this study was Hydra vulgaris (formerly Hydra attenuata). Cultures were maintained at 18°C in hydra medium (HM) as described by Sarras et al. (1991a, 1993).

Zymograms for detecting gelatin-degrading enzymes contained 0.4% copolymerized gelatin and were prepared as described previously (MacKay et al., 1992) and regular SDS-PAGE was performed as described by Maizel (1971).

All steps of the protein purification procedure, except reverse phase HPLC, were carried out at 0°-5°C. The purification methodology was developed and followed using gelatin zymography to monitor the purification steps. Hydra vulgaris that had been fasted for 2 days were collected, rinsed several times in hydra medium, and collected after each rinse by pelleting at 200 g. Washed hydra were either extracted immediately or frozen at –70°C. For initial protein extraction, approximately 5.0 ml of freshly harvested or thawed hydra were suspended in five volumes of TNC buffer (0.05 M Tris, 0.5 M NaCl, 5 mM CaCl₂, 0.02% NaN₃adjusted to pH 7.50 at 5°C with HCl) containing 1 mM PMSF. Following homogenization and centrifugation (30 minutes at 100,000 g) protein of the supernatant was precipitated with 60% ammonium sulfate. The resuspended pellet was dialyzed against 0.01 M Tris-HCl buffer, pH 7.50 at 5°C. As will be discussed, five hydra metal-requiring gelatinase bands were identified in Hydra vulgaris including a band with a mass of 25-29×10₃. These five gelatinases could be separated using a Sephacryl-300 gel filtration column (1 cm × 100 cm). To simplify discussion of these gelatinases, the 25-29×10³M_r hydra gelatinase, which is the focus of the current study, will be termed hydra metalloproteinase 1 (HMP1).

HMP1 could be purified using Reactive Yellow 86 Agarose (Y86) equilibrated with 0.01 M Tris, pH 7.5 buffer. Hydra metalloproteinases including HMP1 were eluted using increasing concentrations of NaCl. Isolated HMP1 was further purified using an analytical C₁₈ column. About 1.0 μg of native HPLC purified HMP1 was subjected to Edman degradation using a gas phase sequencer. To obtain partial interior sequences, 2.0 μg of HPLC purified HMP1 was reduced and carboxymethylated and digested with 1.0 μg of TPCK-treated bovine trypsin for 3 hours at 37°C. The digest was lyophilized and separated by HPLC using a C₁₈ column.

Cloning of HMP1 was accomplished via a strategy involving initial PCR screening of a Hydra vulgaris λZAPII cDNA library (Sarras et al., 1994) using degenerate primers designed from the microsequence data obtained from the purified enzyme. Using the PCR product as a probe, standard cloning and sequencing procedures were followed (Sambrook et al., 1989).

Antiserum to HMP1 was elicited in female New Zealand White rabbits. Preimmune bleedings were made and the animals were immunized with 0.5-1.0 μg of Y86 purified HMP1 or with a multiple antigenic peptide (MAP; see Fig. 8A) following the procedures of Vaitukatis (1981). For some experiments IgG fractions were purified from serum.

Immunofluorescence and ultrastructural localization of HMP1 was carried out as previously described (Sarras et al., 1991a, 1994). The primary antibodies raised to Y86 purified HMP1 and the anti-MAP antibody were diluted from 1:25 to 1:100 for these studies.

Techniques have been developed that allow one to introduce macro-molecules into the basolateral inter-epithelial compartment of hydra (Zhang and Sarras, 1994) so that their effect on cell proliferation, cell migration, cell differentiation and morphogenesis can be monitored under in vivo conditions. In the present study the technique was utilized to analyze the effect of (1) purified HMP1, (2) anti-HMP1 IgG, or (3) mammalian growth factors thought to be related to the action of astacin-like metalloproteinases on hydra morphogenesis, cell proliferation, and cell differentiation. In this technique brief exposure to dimethyl sulfoxide (DMSO) was used to allow introduction of macromolecules into the inter-epithelial compartment of hydra (Fraser et al., 1987; Zhang and Sarras, 1994). DMSO (2.0% v/v) loading of reagents was carried out as previously described (Zhang and Sarras, 1994). At various time points following DMSO loading,1.0 mM 5-bromo-2′-deoxyuridine (BrdU) (Sigma) was injected into the gastric cavity to label cells in S-phase. As previously shown (see review by Bode, 1986), such cell proliferation is restricted to the body column of hydra since no cell division occurs in the tentacles or foot process. Dissociated hydra cells (maceration technique) were immunocytologically stained for BrdU uptake and analyzed as previously described (David, 1983). All cell counts were normalized to the total number of epithelial cells based on hemocytometer counts of the macerate solution.

Head regeneration was studied to determine the effect of anti-HMP1 IgG on hydra head morphogenesis. For all head regeneration experiments, hydra were starved for 1 day and decapitated as described by Sarras et al. (1991b). One to four hours after decapitation, anti-HMP1 IgG, non-immune IgG, or pre-immune IgG was DMSO loaded into regenerating hydra at a concentration of from 0.5 to 1.0 mg/ml in HM (concentrations indicated in figure legends). A second IgG DMSO loading was performed at 24 hours. Head regeneration was monitored over 48 hours at which time controls had begun to develop hypos-tomes and tentacles. Hydra were considered ‘blocked’ if they were observed with a sealed apical pole with no appearance of tentacle buds or a hypostome. To determine if any observed blockage was reversible, hydra were monitored for an additional 48 hours. Recovery from blockage was determined if previously blocked hydra went on to develop tentacles and a hypostome. In some experiments, 10 mM hydroxyurea was used to inhibit DNA synthesis (Cummings and Bode, 1984; Zhang et al., 1994). Hydra morphogenesis and cell differentiation has been reported to continue under conditions in which DNA synthesis and cell proliferation is inhibited (Cummings and Bode, 1984; Zhang et al., 1994).

Epithelial cells along the longitudinal axis of hydra are in a constant state of turnover. This turnover, in part, results from division of cells in the body column and the subsequent displacement of daughter cells into the tentacles and the foot process where they are eventually lost from the epithelial sheet. When cells along the body column are displaced into the tentacles and the foot process, they cease to divide and transdifferentiate into cell types that are specific for these body regions. For example, differentiated ectodermal cells of the body column which function in fluid transport and mucin secretion, trans-differentiate into battery cells (cells which function to ensheathe nematocytes) once they enter the base of the tentacles (Bode, 1986). We have utilized this transdifferentiation process to ascertain the potential role of HMP1 in hydra pattern formation. For these experiments, annexin XII (antibody to annexin XII provided by Dr Haigler, UCI, Irvine, CA) was used as a cell differentiation marker for battery cells (Schlaepfer et al., 1992) and DMSO loading was used to introduce anti-HMP1 IgG (or control IgG) into hydra for 5-8 consecutive days. Following DMSO loading, hydra were fixed for whole-mount immunofluorescence and stained with antibody to annexin XII.

The DMSO loading procedure was also employed to determine the ability of purified HMP1 to activate latent form TGF-β ₁ using an in vivo hydra cell proliferation assay. All procedures used in these experiments were performed as described above.

The extract of Hydra vulgaris exhibited a number of gelatinolytic bands when monitored by SDS-PAGE gelatin-zymography. These bands included a species at 67×10³M_r, a triplet at 51-58×10³M_r and a species at about 25-29×10³M_r (Fig. 1A). The 67×10³ and 25-29×10³M_r species appeared as a doublet in some gelatin-zymograms. These three groups were consistently observed in every zymogram analyzed with the species Hydra vulgaris. Other weak gelatinase bands could sometimes be observed, but were not consistent in all zymograms. These gelatinase activities were observed in all hydra species and strains analyzed: H. oligactis, H. utahensis, H. magnipapillata (Reg 16 strain), H. magnipapillata (strain 105), and H. vulgaris (strain CA7). All gelatinase activities were inhibited by chelators such as EDTA and 1,10 phenanthroline, but were not affected by serine or cysteine proteinase inhibitors, indicating that they were metalloproteinases (Fig. 1A). Treatment with 4-aminophenylmercuric acetate (APMA, an activator of matrix metalloproteinases), did not alter the gel mobility of any of the metallogelatinase activities identified in hydra (data not shown). The 25-29k×10³M_r gelatinolytic metalloproteinase has been termed hydra metalloproteinase 1 (HMP1) because it was the first one to be purified.

In order to ascertain the regional distribution of hydra gelatinases along the longitudinal axis of the organism, adult polyps were decapitated and head and body segments separately pooled. As indicated in Fig. 1B, HMP1 and the 51-58×10³M_r gelatinases appeared more concentrated in the head segments, while the 67×10³M_r gelatinase appeared more concentrated in the body column.

Using a 100-cm Sephacryl-300 column it was possible to separate the three size classes of hydra gelatinases from one another for use in some initial studies designed to compare the relative activity of these gelatinases in terms of pH optima and substrate analysis. While gelatin-zymography indicated that the gelatinase activities had been separated from one another, silver stain analysis of SDS-PAGE preparations indicated that these separated fractions did contain other non-gelatinase proteins (data not shown).

Attempts were then made to obtain a purified preparation of each size class of hydra gelatinases. We were initially successful in developing a procedure for the purification of HMP1 using a Y86 dye affinity column in combination with stepped salt elutions. As shown in Fig. 1C, the 67×10³M_r gelatinase was present in the initial flow through and also appeared immediately after the 0.15M NaCl elution step. The 51-58×10³M_r gelatinases appeared in early fractions of the 0.15M NaCl elution step, and in early fractions of the 1.5M NaCl elution step (Fig. 1C). HMP1 appeared in late fractions (approximately fractions 51-56) eluted with 1.5M NaCl. The final HMP1 product showed a single band of approximately 25.7×10³M_r on SDS-PAGE, and it exhibited corresponding gelatinolytic activity in zymography (Fig. 2A).

The purified HMP1 was subjected to N-terminal sequence analysis as well as internal sequencing after tryptic digestion. As shown in Fig. 2B, the sequence matched (approx. 35%) with members of the astacin metalloproteinase family.

While the microsequence analysis indicated that HMP1 belongs to the astacin family, it was essential that additional sequence data be obtained that spanned the metal binding region of HMP1 to prove that it contained the ‘astacin signature’ which is conserved among all members of the family. To obtain this, degenerate oligonucleotide PCR primers were designed from the following sequences: (1) FDNNFDN (starting with amino acid 18 of the N-terminal sequence) and (2) FWNNIQQ (starting with amino acid 9 of the tryptic fragment; see Fig. 2B). The corresponding oligonucleotides were a forward primer with 128-fold degeneracy (5′-TTYGAYAAYAAYTTYGAYAAY-3′? and a reverse primer with 96-fold degeneracy (5′-YTGYTGDATRTTRTTC-CARAA-3′? both of which had no restriction endonuclease sites at the 5′ end. These primers were used with the Hydra vulgaris λZAPII cDNA expression library and a 309 bp PCR product was obtained. Following blunt end ligation of this PCR product into pGEM7, sequence analysis of subsequent subclones indicated that at least 9 subclones contained the ‘astacin signature’ (Fig. 2C) and showed structural homology with the astacin family. Using the subcloned PCR product as a screening probe, a 1.4 kb cDNA clone encoding HMP1 was obtained from the hydra cDNA library. The deduced amino acid sequence indicated that the proteinase domain of HMP1 had about 42% identity in sequence with other members of the astacin family (Fig. 3). This degree of overall homology (approx. 40%) is characteristic of members of the astacin family (Dumermuth et al., 1991). HMP1 had a non-proteinase C-terminal domain (Figs 3 and 4) which did not match with either the EGF or CUB domains reported to be in other members of the astacin family (Childs and O’Connor, 1994; Finelli et al., 1994); however, this domain did show homology with the EGF-like and Cys-rich domains of a number of other proteinases to include the sperm/egg fusion metalloproteinase, PH30, and subtilisin/kexin-like proteinases. Based on the deduced amino acid sequence, the mature form of the protein (minus pre/pro domain) has a mass of 27×10³ vs the SDS-PAGE apparent mobility mass of 25.710³.

While substrate specificity studies have been conducted with two members of the astacin family, the in vivo substrate(s) for this metalloproteinase family are not known. Based on available data from previous studies with astacin metalloproteinases and matrix metalloproteinases (MMPs) and the fact that HMP1 is localized in the ECM of hydra, a number of protein substrates were tested for their susceptibility to hydrolysis by HMP1. Initial gelatinase assays utilizing ¹⁴C-acetylated type I gelatin, indicated that the activity of HMP1 was readily detected without treatment with APMA which is an activator of MMPs. Of all ECM components tested (Type I and Type IV collagen, fibronectin, and laminin), HMP1 was only effective in degrading fibronectin. Like meprin A, HMP1 was capable of degrading the insulin B chain. In contrast to astacin, HMP1 was very ineffective in hydrolyzing a synthetic ‘elastase’ peptide, N-succinyl-(L-Ala)₃-p-nitroanalide.

Zymography also indicated that HMP1 was unable to digest BSA or casein, but was very effective against type I gelatin.

Rabbit polyclonal antibodies were generated to purified HMP1 (anti-HMP1). As shown in Fig. 5A, these antibodies reacted with a protein of similar mobility to HMP1 while no binding was observed with pre-immune or non-immune serum. The effect of anti-HMP1 IgG on HMP1 activity was determined using 14C-acetylated type I gelatin digestion and gelatin-zymography. In these experiments, preincubation of HMP1 with anti-HMP1 IgG (6-30 μg/ml) resulted in a 75% reduction in HMP1 activity while the control IgG, at an equivalent protein concentration, had no effect. As compared to other hydra metalloproteinase activities (67×10³ and 51-58×10³M_r HMPs), anti-HMP1 IgG had a specific inhibitory effect on HMP1 activity as monitored by gelatin zymography (Fig. 5B). The antibody raised to the multiple antigenic peptide (anti-MAP-HMP1) designed from a region of the catalytic domain of HMP1 also reacted with a band of similar mobility to HMP1 but it was found to be less effective in blocking HMP1 gelatinolytic activity (data not shown).

The expression of HMP1 was examined using the polyclonal antibodies raised to purified HMP1. As shown in Fig. 6A, anti-HMP1 localized to the tentacles of hydra. Higher magnification indicated that antibody binding appeared to be concentrated in the ECM of the tentacles (Fig. 6B), but was lost once the ECM entered the body column (Fig. 6C). The localization of HMP1 in the tentacle ECM was further confirmed with 1.0 μm frozen sections and at the ultrastructural level as shown in Fig. 7A. Using TEM, immunogold labelling was observed throughout the entire width of the tentacle ECM. The localization of HMP1 to hydra tentacles was also observed with antibody raised against a synthetic multiple antigenic peptide (Fig. 8A,B) corresponding to an amino acid sequence within the catalytic domain of HMP1.

To determine whether HMP1 has a role in hydra morphogenesis and cell transdifferentiation, blocking experiments were performed using rabbit anti-HMP1 IgG. As shown in Fig. 9A, DMSO loading of anti-HMP1 IgG resulted in the reversible blockage of head regeneration in hydra. As shown in Fig. 9B, blockage of head regeneration by anti-HMP1 IgG was also observed under conditions in which cellular DNA synthesis in hydra was inhibited by 10 mM hydroxyurea. Studies by others (Cummings and Bode, 1984) and controls shown in Fig. 9B indicated that inhibition of cellular DNA synthesis by 10 mM hydroxyurea does not prevent head regeneration. In these experiments, 10 mM hydroxyurea was found to delay development as compared to untreated controls and therefore an observation period of 72 hours was utilized for monitoring head regeneration and recovery from blockage was followed through 120 hours following the initial time of decapitation. In addition, DMSO loading of anti-HMP1 IgG for 5-8 consecutive days resulted in a loss of annexin XII staining (a battery cell differentiation marker) in the tentacle ectodermal cells (Fig. 10C,D) as compared to DMSO loading of control IgG which had no effect on annexin XII staining patterns (Fig. 10A,B).

Because members of the astacin family (e.g. Drosophila Tolloid and mammalian BMP1) have been tied to the activation of latent form TGF-β family molecules, we directly tested the ability of HMP1 to activate latent form TGF-β₁. As shown in Table 1, we first established, using the DMSO loading procedure, that members of the TGF-β superfamily as well as a number of other mammalian growth factors (e.g. insulin, EGF and bFGF), are biologically active in stimulating epithelial cell proliferation in hydra when used at low concentrations (e.g nanogram levels). The members of the TGF-β super family tested were activin A (mammalian recombinant) and TGF-β₁ (porcine theca cells). While activin A was obtained in its active form, porcine theca cell TGF-β₁ was isolated in its latent form and had to be activated by heating at 80°C for 5-10 minutes. Having shown that mammalian growth factors are biologically active in hydra, we next tested the ability of purified HMP1 to activate latent form TGF-β₁. As shown in Table 1, while heating could activate latent form mammalian TGF-β₁, preincubation of TGF-β₁ with purified HMP1 for up to 24 hours did not result in an activated form of this growth factor. In running controls for this experiment, we found that DMSO loading of purified HMP1 (≤1 μg/ml) alone could stimulate cell proliferation along the body column of hydra (Table 1). This proliferative response was neutralized by preincubation of HMP1 with anti-HMP1 IgG (Table 1). Because of this observation, it was necessary to inactivate the enzyme after it was preincubated with latent form TGF-β₁. This was accomplished by adding anti-HMP1 IgG to the HMP1-TGF-β₁ solution immediately following the preincubation period. After HMP1 activity was blocked by the antibody, the TGF-β₁-HMP1 solution was then DMSO-loaded into hydra to determine if the latent form of the growth factor had been enzymatically activated by preincubation with HMP1. Subsequent heating of the preincubation mixture resulted in activation of latent form TGF-β₁, indicating that HMP1 had not simply degraded the growth factor thus preventing any effect on cell proliferation.

The elucidation of the primary structure of HMP1 indicates it is a member of the astacin family of metalloproteinases. Characteristic of this family was the presence of an ‘astacin signature’ (HEXXHXXGFXHEXXRXDRD) in HMP1, which contains the zinc binding site (Dumermuth et al., 1991). In addition, the position of all cysteines in the astacin domain of HMP1 aligns with the position of cysteines in all other members of the family, suggesting that the secondary and tertiary structure of the proteinase domain may be similar among highly divergent animal groups such as Cnidaria and Chordata. Further structural studies are now underway to clarify this point.

The mature form of HMP1 also contains a C-terminal domain which is not part of the proteinase domain. The function of this C-terminal domain is unknown and while it shows no homology with the EGF and CUB domains observed in other members of the astacin family, it does show homology with the EGF/Cys-rich domains of a number of metalloproteinases such as PH-30 (Wolfsberg et al., 1993) and subtilisin/kexin-like proteinases (Lusson et al., 1993). The HMP1-PH30 relationship is interesting because PH-30 is a sperm-egg fusion protein that contains a snake venom metalloproteinase domain with some similarity to the astacin family and a disin-tegrin domain (Wolfsberg et al., 1993). Studies are currently underway to determine if this domain is involved in protein-protein interactions as reported for the C-terminal domains of Drosophila tolloid.

Although some members of the astacin family, such as crayfish astacin, are expressed in differentiated tissues and likely function as endopeptidases during such processes as digestion (Pfleiderer et al., 1967), many members are expressed during early stages of development and appear to have some role in axis specification and cell differentiation (Dumermuth et al., 1991). The localization of HMP1 to the apical pole of hydra is consistent with observations made in Drosophila and sea urchin, which both show an axis-specific expression pattern for astacin-class metalloproteinases. In the case of Drosophila, the expression pattern for tolloid mRNA changes during embryonic development (Shimell et al., 1991) and eventually parallels the expression pattern seen for decapentaplegic (dpp) which is a member of the TGF-β family. Genetic studies to be discussed later, have indicated some functional relationship between tolloid and dpp. Similarly, expression of SpAN and BP10 has also been reported to occur in an axis-specific manner during sea urchin development (Lepage et al., 1992; Reynolds et al., 1992). SpAN and BP10 are both astacin family members with strong homology with human BMP-1 and Drosophila tolloid. As monitored by in situ hybridization, SpAN mRNA is distributed in a gradient along the animal-vegetal pole of 9-20 hours blastulas with its highest concentration being at the animal pole (Reynolds et al., 1992). Like SpAN, the localization of BP10 mRNA is concentrated at one pole of the blastula (Lepage et al., 1992). Immunolocalization studies using an antibody to BP10 were able to detect a similar distribution pattern for BP10 protein and morphological analysis indicated that as with SpAN, BP10 was more concentrated at the animal pole (Reynolds et al., 1992). The localization of Xenopus BMP1 has not been reported, but northern blot analysis indicates that its mRNA is present during early embryonic stages (Maeno et al., 1993). The localization pattern of HMP1 and other astacin-like proteins is therefore consistent with gene products that are involved in the regulation of developmental processes.

A functional role for members of the astacin family in developmental processes has been suggested for a wide variety of species, e.g. Drosophila (Shimell et al., 1991), sea urchin (Lepage et al., 1992; Reynolds et al., 1992; Hwang et al., 1994), Xenopus (Sato and Sargent, 1990; Maeno et al., 1993), fish (Yasumasu et al., 1992) and developing mammalian tissues, e.g. bone (Rosen and Thies, 1992). Besides the morphological findings discussed above, a number of functional studies suggest that HMP1 is somehow involved in developmental processes in hydra. Similar to that reported for sea urchin development (Lepage et al., 1992), anti-HMP1 IgG blocked hydra morphogenesis as monitored by analysis of head regeneration. This blockage appeared specific in that (1) it was reversible and (2) control IgG at equivalent concentrations was not inhibitory as compared to anti-HMP1 IgG. As shown by Lepage et al. (1992), incubation of sea urchin embryos (16-32 cell stage) with purified anti-BP10 IgG resulted in abnormally shaped larvae with an axially positioned archenteron. This same altered phenotype was also observed in embryos incubated with synthetic peptides corresponding to a portion of an internal EGF domain of BP10. While the underlying mechanisms of action are not clear from these studies, the data suggest that BP10 and HMP1 may be affecting signalling events related to morphogenetic processes. More direct data regarding the relationship of astacin homologues to cell signalling events comes from detailed genetic studies with Drosophila tolloid. These studies indicate that tolloid is involved in a cascade of interactions between zygotic gene products necessary for cell fate specification in the dorsal axis of the embryo (see review by Hecht and Anderson, 1992). In this cascade of events, the TGF-β -like family member, dpp, is expressed dorsally and plays a critical role in pattern formation of the dorsal half of the embryo (Ferguson and Anderson, 1991). In this signalling process, tolloid acts to increase or potentiate dpp activity and serves to help establish the dorsal-ventral gradient of this TGF-β class molecule (Shimell et al., 1991). More recent genetic studies have indicated that mutations toward the C terminus of tolloid involving EGF and CUB repeats are critical for tolloid-dpp interactions (Childs and O’Connor, 1994; Finelli et al., 1994). It has been proposed that once the tolloid-dpp complex is formed, the astacin-like proteinase domain of the tolloid activates dpp so that the growth factor can function in cell fate signalling processes along the dorsal-ventral axis of the embryo. The experiments described in the current hydra study are consistent with some aspects of what has been shown for Drosophila. HMP1 does have a C-terminal domain distinct from the proteinase domain. While this C-terminal domain does not share homology with the EGF or CUB domains seen in other members of the astacin family, it does show homology with the EGF and Cys-rich domains seen in a number of metalloproteinases such as PH-30 and developmentally related receptors like Notch. Whether this C-terminal domain functions in protein-protein interactions as reported for Drosophila tolloid remains to be deter-mined. The ability of purified HMP1 to stimulate hydra cell proliferation in the body column when artificially introduced into this region using the DMSO loading technique is consistent with the activation of some endogenous signalling factor(s). It is important to note however, that HMP1 is not expressed along the body column where cell proliferation occurs in hydra and therefore this effect doesn’t reflect the normal action of HMP1 in the tentacles. HMP1 would of course also enter the tentacle region of hydra during DMSO loading, but whole-mount BrdU-labelling experiments show that it does not stimulate cell proliferation in tentacles (data not shown) as predicted from previous work (Bode, 1986). Although not reflective of the normal in situ localization of HMP1, the stimulatory effect of HMP1 along the body column suggests that the enzyme can potentially affect cell signalling pathways as reported for tolloid. Data from a number of experiments indicate that cell signalling events by HMP1 does not involve the regulation of cell proliferation during head regeneration as might at first be deduced from its effect in the body column. For example, previous studies have clearly shown that head regeneration and battery cell transdifferentiation (Cummings and Bode, 1984) as well as morphogenesis of hydra cell aggregates (Zhang et al., 1994) can occur when cellular DNA synthesis is blocked. Even when hydra are exposure to 10 mM hydroxyurea for prolonged periods and cells are prevented from reaching mitosis, head regeneration and battery cell transdifferentiation can occur (Cummings and Bode, 1984). These facts coupled with the observation that the inhibitory effect of anti-HMP1 IgG on head regeneration continues even under conditions in which cellular DNA synthesis was blocked with 10 mM hydroxyurea, makes it unlikely that HMP1 functions to stimulate cell proliferation in hydra during the process of head regeneration. Demonstration that repeated introduction of anti-HMP1 IgG into the inter-epithelial compartment of hydra results in a loss of expression of annexin XII, a cell differentiation marker for battery cells of the tentacles, does provide evidence for a relationship between HMP1 and cell signalling processes in hydra that may be related to differentiation pathways as opposed to pathways associated with modulation of cell proliferation and division.

The underlying mechanism of action of HMP1 is not clear from these studies. If processes in hydra parallel what has been reported for Drosophila, then one could speculate that HMP1 is functioning to activate a latent cell signalling factor. This activation pathway could involve a direct hydrolysis of the latent cell signalling factor by HMP1 or could occur through an intermediate enzyme that must be first activated by HMP1 before it can activate the latent cell signalling factor. Because HMP1 localizes to the tentacle ECM, it is reasonable to propose that the substrate for this pathway may be matrix-associated. Whether this matrix-associated cell signalling factor has any homology with a true growth factor remains to be deter-mined, but it is certainly well established that growth factors such as TGF-β can be ECM-associated in a number of developing systems (Schubert, 1992). In addition, recent studies by Taipale et al. (1994) have also shown that proteolytic activation of TGF-β ₁ can occur following the interaction of this growth factor with a binding protein and the ECM. Nevertheless, while a number of mammalian growth factors such as insulin, EGF, bFGF, Activin A, and TGF-β ₁, are biologically active in hydra, we were unable to show that purified HMP1 could activate latent form mammalian TGF-β ₁ as monitored in a hydra cell proliferation assay. Given the fact HMP1 appears to be involved in cell signalling processes, we therefore propose that while the endogenous substrate for HMP1 may not be structurally equivalent to mammalian TGF-β ₁, there may likely exist some endogenous matrix-associated growth factor-like molecule upon which HMP1 acts. It is our current working hypothesis that activation of such a matrix-associated latent factor in the tentacles by localized expression of HMP1 would be one of a number of mechanisms to explain the trans-differentiation of body column cells to tentacle battery cells.

The authors wish to acknowledge the excellent technical support provided by Ms. Jacquelyn K. Huff during the course of these studies. We also wish to thank Dr J. May of the University of Kansas Medical Center, Wichita Campus for providing us with latent form TGF-β₁, and Dr H. Haigler of the University of California, Irvine for providing us with antibody to annexin XII. Lastly, the authors wish to thank Dr A. Fowler, Dept. of Biol. Chemistry, UCLA, for her expert N-terminal protein sequencing of HMP1. This work was supported by funds provided by NIH grants RR06500 and AR39189.