Ks1, an epithelial cell-specific gene, responds to early signals of head formation in Hydra

Knowledge of the mechanisms of position-dependent differentiation of specific cell types is central to understanding how embryos are formed. 25 years ago, Lewis Wolpert (1969) introduced the concept of positional information and proposed that cells within a developing system are informed of their position with respect to one or more reference points. The cells then acquire a positional value and interpret their positional value by differentiating in a particular way. One example of pattern formation considered in detail by Wolpert was development and regeneration in Hydra (Wolpert, 1969; Wolpert et al., 1974).

Development in hydra consists mainly of head and foot formation from tissue of the body column (see Bode and Bode, 1984). When head and/or foot are removed, they regenerate within two to three days. A head always regenerates at the apical end and a foot regenerates at the basal end. These and other observations have produced strong evidence for the existence of positionally restricted ‘morphogenetic’ molecules in hydra (see Bode and Bode, 1984; Summerbell et al., 1991). The naturally occurring morphogens have not yet been identified. Considerable progress, however, has been made recently towards the molecular analysis of axial patterning in hydra.

Several homeodomain genes have been identified as candidates for specification of positional information (for review see Shenk and Steele, 1993). The expression pattern of one of them, cnox-2, is consistent with a role in axial pattern formation (Shenk et al., 1993). Furthermore, several genes have been isolated that become activated in hydra in spatial and temporal patterns (Kurz et al., 1991; Lopez de Haro et al., 1994).

The signal transduction system used in axial patterning in hydra has recently begun to be examined. Treatment with diacylglycerol (DAG) or 12-O-tetradecanoylphorbol-13-acetate (TPA) was found to induce ectopic heads in the body column (reviewed in Müller, 1993). Prolonged exposure of polyps to lithium causes appearance of ectopic feet along the body column (Hassel et al., 1993). Coadministration of DAG with arachidonic acid, a subordinate signaling molecule in the phosphatidylinositol-protein kinase C (PI-PKC) system, causes a strong increase in the head formation potential (Müller et al., 1993). DAG and TPA are known to activate PKC (Nishizuka, 1986) while lithium blocks several enzymes in the PI-PKC system (Berridge et al., 1989). Thus, it appears likely that a PKC-mediated intracellular pathway initiated by a ligand-receptor interaction on the cell surface mediates head specification in hydra.

Candidate genes responding to such a signaling system should be (i) expressed in head-specific cells, (ii) induced by TPA, and (iii) contain TPA-responsive elements in their promoter. Here we report the cloning and characterization of a gene, ks1, which fulfills these criteria and thus supports the view that a ligand-activated PI-PKC system is involved in the conversion of hydra gastric tissue to head tissue.

Polyps of Hydra magnipapillata (strain sf-1) and Hydra vulgaris were used in this study. Polyps were cultured according to standard procedure (Lenhoff and Brown, 1970) at 18°C.

TPA was dissolved in acetone at 10 mM and stored at —20°C. Polyps were incubated in TPA at a final concentration of 30-50 nM (in hydra medium) for 20 minutes at 18°C. After treatment the animals were transferred to hydra medium and cultured at 18°C until they were used.

1×10⁵ amplified recombinants of a oligo(dT)-primed epithelial cell-specific cDNA library (Lopez de Haro et al., 1994) were screened with 32P-labeled cDNA made from head and gastric tissue of epithelial polyps by the random primer labeling procedure. 19 clones were picked as head specific; 17 of these clones encoded a single gene, designated ks1. The remaining 2 clones encoded different genes and are described elsewhere (Lopez de Haro et al., 1994). Since none of the 17 ks1 cDNA clones was full length, one of the clones (cLK7) was used to rescreen a H. vulgaris cDNA library. The cDNA library in lambda ZAPII was prepared commercially (Stratagene) from Hydra vulgaris whole polyp poly(A)+ RNA and was generously provided by Drs Michael Sarras (University of Kansas Medical Center) and Hans Bode (University of California, Irvine). Rescreening resulted in the isolation of a number of additional clones, one of which was cLK7-41 Sequence analysis of cLK7-41 indicated the presence of an open reading frame beginning at the 5′ most ATG triplet. cLK7–41, however, did not encode 3′ sequences present in the initially isolated ks1 clones. To complete the sequence and to confirm the region of overlap between cLK7-41 and the initial cDNA clones, inverse PCR was carried out as described below.

To obtain the full-length sequence of ks1 as well as 5′-flanking sequences, inverse PCR was carried out as described (Gellner et al., 1992) using two 20-nucleotide primers oriented such that primer-extension proceeded outward from the known sequence. Cleavage of H. vulgaris genomic DNA with HindIII resulted in a 3 kb fragment, which in Southern blots strongly hybridizes with cLK7–41. For inverse PCR, therefore, HindIII-digested and ligated DNA was used as template DNA. Oligonucleotide primers were synthesized complementary to nucleotides 84–103 and identical to nucleotides 297–319 (see Fig. 2). The expected 1.7 kb inverse PCR product was obtained and cloned into pBS– for further analysis. The relationship of this PCR-derived genomic clone, gIPCR1, to the cDNA clones is shown in Fig. 1.

Nucleic acid isolation, sequence analysis and RNA blot analysis were carried out following standard procedures (Sambrook et al., 1989). DNA sequences were analysed using the Hibio DNASIS/PROSIS program (Hitachi). All polymerase chain reactions (PCR) were carried out using the buffer conditions and Taq polymerase supplied by the manufacturer (Amersham). The 5′ boundary of the ks1 transcript was determined by primer-extension following standard procedures. The 20-nucleotide IPCR5 primer was 5′ end-labeled, mixed with 9 μg of poly(A)+ RNA from head tissue, denatured and allowed to hybridize at 56°C for 16 hours as described previously (Gellner et al., 1992). After primer elongation and RNase A treatment, the primer extension products were separated by 8.3 M urea/6% PAGE and visualized by autoradiography. EMBL database accession number for ks1, X78596.

In situ hybridization on whole mounts and macerates was carried out with digoxigenin-labeled DNA probes as described by Kurz et al. (1991) with the following modifications. Postfixed whole mounts were washed for 15 minutes in PBT, then for 20 minutes in 1/1 hybridization solution (HS)/PBT at room temperature and for 20 minutes in HS at room temperature. Afterwards whole mounts were prehy-bridized for 20 minutes in HS at 45°C and hybridized as described.

We used a differential cDNA screening approach to isolate hydra genes expressed in head but not in gastric tissue. We focused our attention on epithelial cells, since this is the cell type responsible for morphogenesis in hydra. We isolated a number of clones encoding a tentacle-specific cDNA, which we have designated ks1. Since none of the cDNA clones was full length, we isolated additional clones from a cDNA library and a genomic clone by inverse PCR (see Materials and Methods) to complete the sequence.

The nucleotide sequence of the ks1 gene was obtained from the combined results of the H. vulgaris cDNA and genomic clones shown in Fig. 1. The ks1 sequence contains an open reading frame of 651 bp (Fig. 2). At the 5′ end, 392 nucleotides precede an ATG triplet. 15 bp upstream of the first in-frame methionine codon, there is an in-frame termination codon indicating that the first ATG triplet is indeed the initiator. The length of the transcript predicted from the combined length of the cDNA clones (906 bp) corresponds well to 0.9 kb mRNA detected on northern blots (see below) indicating that the full-length cDNA has been identified. Conceptional translation of the open reading frame predicts a 217 amino acid protein with a relative molecular mass of 25.4×10³.

The transcription initiation site was determined by primer extension using the IPCR5 primer complementary to nucleotide 84–103 of ks1. The labeled primer was hybridized to RNA from head tissue and extended by reverse transcriptase. The single extension product indicated that transcription initiates 68 bases upstream of the translational initiation codon (data not shown).

To identify sequences that potentially control ks1 expression, we examined the 5′ flanking sequence for potential transcription factor binding sites. Comparison of this region with promoters of other genes revealed several segments of sequence conservation (Fig. 3). The sequence TATTAA (single underline in Fig. 3A) 17 bp upstream of the transcription start site is similar to TATA-like sequences in many eukaryotic promoters. At positions —62 to —70 the ks1 gene contains the sequence 5′-AAGATTCAG-3′ (double underlined in Fig. 3A), which is similar to the TPA-responsive element (TRE, consensus sequence 5′-TGAGTCAG-3′, Angel et al., 1987). A sequence comparison between TREs in the control regions of TPA-inducible genes and the putative TRE in ks1 is shown in Fig. 3B. The short conserved TRE sequence serves as the binding site for the transcription factor AP1 and is the only element required for TPA induction in a number of genes (Angel et al., 1987). At position —175 to —182, there is a sequence related to the binding site for the transcription factor MLTF/USF in the TPA-inducible heme oxygenase gene (Muraosa and Shibahara, 1993). As shown below, hydra ks1 is responsive to TPA (see Fig. 9). Thus, it is the first TPA responsive gene known to contain both TRE and MLTF/USF sequences in the 5′ flanking region. At the 3′ terminus, a consensus polyadenylation sequence, AATAAA, is located at position 878–883 (Fig. 2).

A search of various protein data banks with the KS1 amino acid sequence revealed no homology between KS1 and any other protein. KS1, however, contains several interesting structural features (Fig. 4). Hydropathy analysis of KS1 predicted a strongly hydrophilic character (Fig. 4A). The N-terminal 15 residues are strongly hydrophobic, indicative of a signal sequence for translocation across the endoplasmic reticulum. An acceptable cleavage site for a signal peptidase is present between Ser16 and Met17 (von Heijne, 1983). It seems likely, therefore, that KS1 encodes a secreted protein. No potential N-linked glycosylation sites are present in the KS1 sequence.

One striking feature of the predicted KS1 protein is the large number of charged residues that are clustered into highly acidic and basic domains (Fig. 4B). Beginning at residue 32, 7 out of the next 15 amino acids are lysine and arginine. This basic domain (B1 in Fig. 4B) is followed by a acidic domain (A1 in Fig. 4B) consisting of 66% aspartic acid and glutamic acid residues. This acidic domain is immediately followed by a 16 amino acid basic region (B2), including a stretch of 6 consecutive lysine and arginine residues. The 35 amino acid domain A2 following this basic domain shows remarkable amino acid similarity to the first acidic domain A1. The carboxy terminal domain (B3) contains 42% lysine or arginine residues.

Analysis of the deduced amino acid sequence revealed two regions of internal sequence similarity comprising residues 31–96 and 97–146. The repeated sequences are aligned in Fig. 4C. Structural-prediction algorithms (Chou-Fasman) predict a helix-turn-helix structure for these repeats.

To determine the expression pattern of ks1, we analyzed northern blots of RNA from head, upper one third and lower two thirds of H. vulgaris polyps (Fig. 5). In head tissue a single, abundant 0.9 kb transcript was detected. The same transcript was also found in the upper third of the body column, although at a much lower level (lane 2 in Fig. 5). No ks1 transcripts were detected in the lower two thirds.

To determine the cell-type specificity of ks1 expression, we performed in situ hybridizations on macerated cells (Fig. 6). Fig. 6A shows that, in macerates hybridized with clone cLK12, tentacle-specific ectodermal epithelial cells (battery cells) were heavily labelled. The occurrence of ks1 transcripts in the upper third of the gastric region, which includes the ‘tentacle formation zone’ (Fig. 5, lane 2), suggests that there is a population of cells in this region not yet differentiated to tentacle-specific cells but already expressing ks1. To test this possibility, we performed in situ hybridization on macerates from the upper third of the gastric region. Fig. 6B shows that in this region ks1 hybridizes to a population of epithelial cells which are morphologically indistinguishable from gastric epithelial cells. Cell counts indicated that about 10% of the ectodermal epithelial cells in the ‘tentacle formation zone’ region are ks1 positive. These cells appear to represent body column cells determined to differentiate to tentacle epithelial cells. Thus, ks1 expression appears to be initiated early in head specification.

Activation of ks1 expression during head regeneration was studied by in situ hybridization on whole mounts (Fig. 7). Animals were decapitated directly below the tentacles and hybridized to ks1 1, 2 and 4 days later. ks1 expression could be detected in ectodermal epithelial cells by 2 days when tentacles were beginning to evaginate (Fig. 7C,D). When regenerating heads at this stage were viewed from above (Fig. 7D), staining could also be seen in the most apical region of the hypostome (arrow in Fig. 7D). In mature heads (Fig. 7E,F) staining was restricted to the tentacles: staining appeared to end abruptly at the base of the tentacle; no staining was found between the tentacles. In whole mounts, it was not possible to detect ks1-positive cells in the upper third of the gastric region. Thus, the whole-mount in situ procedure appears to be too insensitive to detect the ks1 expressing epithelial cells observed in macerates of the tentacle formation zone (Fig. 6B).

Treatment of hydra with the PKC activators DAG or TPA results in appearance of ectopic head structures in the body column (Müller, 1989, 1993). To investigate whether TPA caused changes in the pattern of ks1 expression, we exposed decapitated polyps to TPA for 20 minutes and allowed heads to regenerate for 1, 4 and 24 hours. Thereafter regenerating polyps were cut in the middle and the resulting upper and lower body column segments were assayed for ks1 expression using northern blotting. The results in Fig. 8 demonstrate that treatment with TPA alters ks1 expression significantly.

In the upper portions of TPA-treated polyps, the level of ks1 transcripts increased dramatically within 1 hour compared to untreated animals. In situ hybridization of macerated tissue from these animals indicated that there was no corresponding increase in the proportion of ks1-expressing cells compared to untreated animals. Thus, the increase in ks1 transcripts appears to represent an activation of transcription in ks1-expressing cells. In the lower portions of TPA-treated animals, there were no rapid changes in ks1 expression. There was, however, an increase in ks1 expression beginning 4 hours after TPA treatment. Cells in this region do not normally express ks1 (Fig. 5) and hence this expression represents ectopic activation of a positionally dependent gene.

The results in Fig. 8 also indicate that ks1 transcription is activated during normal head regeneration. The increase in ks1 transcripts is detectable beginning by 4 hours of regeneration and thus is delayed compared to the rapid effect of TPA (see Discussion).

Patterning in hydra has been studied extensively at the tissue level using grafting techniques and monoclonal antibodies (for reviews see Bode and Bode, 1984; Müller, 1993). No probes have been available, however, for examining the patterning process at the molecular level. In this report, we present the molecular cloning of such a probe.

The sequence of the ks1 gene product has no similarity to other proteins in various databases. KS1 is predicted to contain a leader sequence and several distinct domains of charged amino acid residues (Fig. 4C). The first cluster of basic and acidic amino acids (residues 32-96) shows significant sequence similarity to the second cluster of basic and acidic amino acids (residues 97-147) and may be due to sequence duplication. Several other proteins with highly acidic domains have been described and include nuclear-localized proteins such as nucleolin and cytosolic proteins such as the yeast sec7 protein. In all these proteins, the acidic domains are thought to serve as structural motifs for interaction with lipids or other proteins (Achstetter et al., 1988). Hydra KS1 appears to be a secreted protein and could be adsorbed on the cell surface by its highly charged amino acid domains. It may be a component of the carbohydrate-rich glycocalyx that covers the surface of hydra polyps.

A monoclonal antibody, CP8, which specifically recognizes ectodermal epithelial cells in the head (Javois et al., 1986), has been shown to detect a protein localized in granules near the apical surface of ectodermal epithelial cells. Since KS1 contains a leader sequence, it could encode a secretory protein and be located, at least transiently, in cytosolic granules. It is intriguing to speculate that the ks1 gene product is the CP8 antigen.

The upper gastric region subjacent to tentacles has previously been defined as the ‘tentacle formation zone’ (Hobmayer et al., 1990); epithelial cells in this region become committed to battery cell differentiation and are displaced into the tentacles. ks1 is expressed at high levels in battery cells as well as in a subpopulation of ectodermal epithelial cells in the tentacle formation zone. ks1-expressing cells in this region are morphologically indistinguishable from ks1-negative cells. Thus, ks1 is a very early marker for head-specific differentiation.

During head regeneration, at a stage when tentacles are just beginning to evaginate (C and D in Fig. 7), ks1 expression is observed at the most apical tip and in the tentacle buds. At later stages, no expression of ks1 is detected in the tip whereas the tentacles are strongly stained. This pattern of ks1 expression is strikingly similar to that found previously with the head-specific monoclonal antibody TS-19 (Bode et al., 1988). During head regeneration, TS-19 label first appears at the most apical tip. At later stages, it spreads to the evaginating tentacles and disappears from the apex.

The transient appearance of ks1 and TS19 in the most apical region of regenerating tips is consistent with the view (Bode et al., 1988) that the presumptive hypostome area passes through a stage of ‘tentacleness’ before final differentiation. This pattern of expression of ks1 and TS-19 can be explained by a model proposed recently by Hans Meinhardt (1993). According to Meinhardt’s model, tentacle formation is activated at regenerating tips; hypostome formation then locally inhibits tentacle formation and displaces it to the subhypostomal region. ks1 could be a suitable molecular probe to dissect the mechanism behind this patterning process.

The northern analysis in Fig. 8 indicates a rapid and a delayed effect of TPA on ks1 expression. The rapid effect appears to be due to direct activation of the second messenger system in cells already expressing ks1 since TPA treatment simply increases the amount of ks1 transcript but does not increase the number of positive cells. This suggests that ks1 expression is regulated by a ligand-activated PKC second messenger pathway. The observation that ks1 expression is also increased by 4 hours of regeneration in tissue untreated with TPA suggests that the endogenous ligand begins to be produced by this time.

By comparison, the delayed effect, which leads to generation of ectopic ks1-positive cells in the lower body column and requires at least 4 hours to become detectable, appears to be an indirect effect of TPA treatment. We imagine that TPA activates a signaling system for head formation in the body column and that this signal then induces ectopic formation of ks1-positive cells.

Two potential cis-regulatory elements are present in the 5′ flanking sequences of ks1 and might serve to transmit the TPA signal(s) to the transcriptional machinery. ks1 is the first gene in hydra found to be sensitive to TPA; its structure and expression pattern fulfill a prediction that followed from Müller’s results (Müller 1989, 1993). Whether other head-specific genes in hydra also have TPA-responsive sequence elements in their 5′ flanking sequences remains to be elucidated. It is interesting to note, however, that Cnox-2, a Hydra homeobox gene thought to be involved in axial patterning, also seems to be regulated by a signal transduction pathway involving PKC (Shenk et al., 1993).

Similar observations have been made in several other developing organisms. PKC has been suggested to participate in controlling differentiation along an embryonic axis in Xenopus (Otte et al., 1988), in determining cell fate in Dictyostelium (Ginsburg and Kimmel, 1989; Peters et al., 1989), and in establishing cell fate in sea urchin embryos (Livingston and Wilt, 1992).

Regardless of the mechanisms by which PKC modulates ks1 expression in Hydra, the results presented in this paper support the hypothesis that signaling mediated by ligand/receptor interactions at the plasma membrane is important in instructing hydra cells to differentiate according to their position in the embryo. Studying ks1 expression may now allow us to trace the mechanism necessary for transcriptional activation back to events at the cell surface.

We are grateful to Drs Klaus Gellner and Maria Lopez de Haro for critical discussion of the data, to Dr Michael Sarras and Dr Hans Bode for providing the H.vulgaris cDNA library, to Dr Robert Steele for comments on the manuscript, and to Gabriele Praetzel for excellent technical help. L. M. S. received a postdoctoral fellowship from the Alexander-von-Humboldt foundation, Bonn. Supported by the Deutsche Forschungsgemeinschaft (grants to C. N. D. and T. C. G. B.).