ABSTRACT
ht-en protein, an annelid homolog of the Drosophila engrailed protein, is expressed during both early development and neurogenesis in embryos of the leech, Helobdella triserialis. In Helobdella as in Drosophila, early expression is in segmentally iterated stripes of cells within the posterior portion of the segment and later expression is in cells of the segmental ganglia. These findings suggest that dual expression of an en-class gene was present in a common ancestor of annelids and arthropods.
Introduction
Although the segmental body plans of such protos-tomes as annelids, arthropods and onychophorans are similar, there are wide differences in the cellular mechanisms by which segments form in embryos of these phyla (Anderson, 1973) and some authors have proposed that segmentation arose independently among them (Barnes et al. 1988). The process of segmentation has been analyzed extensively in arthropods, particularly by applying genetic and molecular techniques to Drosophila melanogaster. In this insect, regulatory networks of genes establish proper segment number, polarity and identity during early development. Many of these genes encode proteins containing homeodomains, highly conserved (approximately 60 aa) DNA-binding domains necessary for regulating gene transcription (Akam, 1987; Ingham, 1988 for reviews). Some of these genes exhibit what we shall refer to here as ‘dual expression’, meaning that, in addition to being expressed during early development, they are also expressed later, in overlapping subsets of cells within the nervous system (see Akam, 1987; Doe and Scott, 1988 for reviews).
One such homeodomain-containing protein that shows dual expression in Drosophila is encoded by the engrailed (en) gene (Fjose et al. 1985; Poole et al. 1985). In early development, en expression is detected in segmentally iterated circumferential stripes of cells where it is required to specify the identity of the posterior ‘compartment’ of each segment (Garcia-Bellido et al. 1973; Garcia-Bellido, 1975; Kornberg, 1981a,b;Lawrence and Struhl, 1982; Fjose et al. 1985; Poole et al. 1985; Kornberg et al. 1985; DiNardo et al. 1985; Lawrence and Morata, 1986; Brower, 1986). Later, during neurogenesis, en is expressed in a subset of segmentally iterated neurons (Brower, 1986; Patel et al. 1989a).
Homologs of en have been characterized in various arthropod, echinoderm and chordate species (Joyner and Martin, 1987; Dolecki and Humphries, 1988; Gardner et al. 1988; Fjose et al. 1988; Patel et al. 19896; Wedeen et al. 1991; Hemmati-Brivanlou et al. 1991). While expression of en during neurogenesis has been reported in all three of these phyla, the earlier embryonic expression of en during segmentation has been observed only in arthropods. Therefore, it has been proposed that the ancestral role for en was in neural specification and that only following the phylogenetic separation of the arthropods did it acquire a role in segmentation (Patel et al. 1989a).
To test this proposal, further study of en homologs in segmented protostomes other than arthropods is required. We have chosen to examine the expression of a homolog of en (ht-en) in embryos of Helobdella triserialis, a glossiphoniid leech, because embryogenesis has been described in some detail at the cellular level in this annelid (Weisblat et al. 1989; Shankland, 1991, for reviews). Here we present evidence showing that ht-en is expressed both in segmentally iterated stripes during early development and in a segmentally iterated subset of neurons during neurogenesis. We conclude from these data that, contrary to prior proposals, a dual expression pattern for en class genes was present in a common ancestor of annelids and arthropods. It thus seems likely that this common ancestor of annleids and arthropods was already segmented or at least had some serially iterated components in its body plan.
Materials and methods
Animals
Helobdella triserialis embryos were obtained from a laboratory breeding colony, periodically supplemented with leeches collected from lakes within Golden Gate Park, San Francisco. The colony was maintained and embryos were reared as described by Blair and Weisblat (1984) and Wedeen et al. (1990). Embryos were staged by the conventions of Fernandez (1980), as modified by Bissen and Weisblat (1989).
Recombinant clones
The subclones, pfhten1x and pfhten4x (fusions of one or four tandem copies, respectively, of the ht-en homeobox region with the E. coli lacZ gene) and phten4x-trpE (a fusion of four tandem copies of the ht-en homeobox region with the E. coli anthranilate synthase gene) were generated in the following way. First, a construct, phten1×0.24Sp+(KS), was made by cloning a 242 bp fragment (including the homeobox and the flanking 5 bp 5′and 57bp 3′) amplified from ht-en by polymerase chain reaction (PCR), into the Bluescript plus (KS) vector via the blunt ends of the SpeI site of the vector that had been 3′extended by Klenow polymerase. Using this subclone, two tandem repeats of the fragment were constructed in frame by digesting one aliquot of the plasmid with XbaI and the other aliquot with SpeI. (The XbaI and SpeI sites are present in the polylinker flanking the insert and are not present in the insert.) The two aliquots were then ligated together. This process was repeated on the resulting subclone to generate four in-frame copies of the ht-en homeobox region. Each copy coded for the following amino acid sequence from ht-en: - DEKRPRTAFTGDQLARLKREF-SENKYLTEQRRTCLAKELNLNESQIKIWFQNKRAKM-KKASGVKNQLALQLMAQGLYNHS (Wedeen et al. 1991), preceded by 4 additional amino acids, LELE, due to the contribution of the polylinker sequences from the Bluescript plasmid. The four tandem repeats were then removed from the Bluescript vector by digestion with XbaI and Hzndlll and ligated either into the Xbal and Wzndlll sites of pUR278 (Ruther and Muller-Hill, 1983) to generate a fusion with the lacZ gene (pfhten4x) or into the XbaI and Hindi 11 sites of pATH10 (Koerner et al. 1990) to generate a fusion with the bacterial anthranilate synthase gene (phten4x-trpE). The lacZ fusion construct, pfenlx was obtained by subcloning the XbaI-HindIII fragment from phtenlxO.24Sp+(KS) into the XbaI and HindIII sites of the pUR278 vector (Ruther and Muller-Hill, 1983).
Fusion proteins
Betagalactosidase fusions
LacZ fusion constructs.were transformed into strain, XL1 (Stratagene). They were grown as 500 ml cultures at 220 revs min-1 at 37°C to an O.D. 660 of 0.6, at which time, isopropyl β, D-thiogalactopyranoside (IPTG) was added to a final concentration of 1 mM. They were grown for an additional 2h and then harvested.
Fusion proteins were purified as follows. The 500ml culture was resuspended in 20 ml of 50 mM Tris-HCl, pH 7.5, 50 mM Tris Base, 25 mM EDTA, pH 8.0, 20% sucrose, 1 mg ml-1 lysozyme, and incubated on ice for 1h. 20 ml of 1% Triton X-100 was added rapidly and the cells were frozen at –70°C. Upon thawing, cells were sonicated to reduce the viscosity and insoluble proteins were pelleted at 4°C in an SS34 rotor at 10 000 revs min-1 for 20min. The pellet was resuspended by douncing in 6 ml 100 mM Tris-HCl 7.5, 0.1% Triton X-100. The insoluble fraction was again pelleted and resuspended in 6ml 100 mM Tris-HCl 7.5, 4M Urea. The process was repeated, resuspending the pellet in 50mM Tris-HCl, pH7.5, 8 M Urea. This final fraction was found to be highly enriched for the lacZ fusion protein.
Anthranilate synthase fusions
The trpE fusion construct was grown as recommended by Koerner et al. (1990). The trpE fusion protein generated from the phten4x-trpE construct was soluble in 50 mM Tris-HCl, pH 7.5, 50 mM Tris Base, 25 mM EDTA, pH 8.0. 20,% sucrose, 1 mg ml-1 lysozyme and all subsequent solutions. Therefore, the protein was purified by electroelution from SDS-polyacrylamide gels.
Antibodies
A dose of 0.5 mg of fusion protein generated from the pfhten4x construct was injected into a rabbit. After 2 weeks a booster injection of 0.25 mg was given. On a schedule of every two weeks the rabbit was alternately bled or boosted, until 4 bleeds were obtained.
Affinity purification of antibodies
A fusion protein containing 4 copies of the ht-en sequence fused in frame to one copy of the E. coli anthranilate synthase gene (phten4x–trpE) was used to affinity purify antibodies specific for the ht-en homeodomain region.
The en4x-trpE protein was purified by electroelution from an acrylamide gel slice and was coupled to an affigel-10 column (BioRad). Antibodies were bound to the column in PBS. Nonspecific antibodies were washed from the column with 10 mM Tris, pH 7.5 and specific antibodies were eluted in 100 mM glycine, pH4.0. The partially purified antiserum was then dialyzed against PBS overnight at 4 °C. Anti-βgal antibodies were further removed by passing the antisera over an affinity column which had coupled to it the βgal protein obtained from inducing the pUR278 vector and purified in the manner described above for βgal fusions. The IgG component of the purified antisera was concentrated to [30βgml-1] by binding to and eluting it from a protein A column (BioRad).
Antibody competition with bacterial proteins
A saturated culture of E. coli, strain XL-1, was harvested by centrifugation, resuspended in 1/10 original volume with water and lysed by sonication at 0°C. An aliquot of lysate was boiled 3 min and combined with an equal volume of non-heat treated lysate. An aliquot of this mixture was agitated for 48 h at 4 °C with 1/5 volume of affinity purified αht-en. Solid debris was precipitated by centrifugation and the supernatant was used at the appropriate dilution on western blots and embryos.
Western blot analysis
Westerns were blocked overnight at 4 °C in western blocking buffer (1% bovine serum albumin (Sigma no. A-7906), 2% nonfat dry milk (Carnation), 1 mM MgCl2, 25 mM Tris (pH 8), 137 mM NaCl, 2.7 mM KC1). Westerns were reacted at 4°C overnight with primary antibody (either a 50-fold dilution of αht-en or 2-fold dilution of mAb4D9 supernatant) and for one hour with 500-fold dilution of horseradish-peroxidase-conju-gated goat anti-rabbit secondary antibody (GAR-HRP; Cal Biochem) or 2000-fold dilution of horseradish-peroxidase-conjugated goat anti-mouse secondary antibody (GAM-HRP; Jackson Immunoresearch Lab), respectively. Nonspecifically bound antibody was removed by washing after each antibody treatment with six washes of blocking buffer for 10 min each. The HRP immunodetection was done in the presence of 0.03% cobalt and nickel ions (Adams, 1981).
Lineage tracing
Rhodamine dextran amine (RD A) lineage tracer (50 mg ml-1 in 0.2 % KC1) was pressure injected as described by Weisblat et al. (1980b) into individual teloblasts at stage 7.
α-hten antibody staining of embryos
All embryos were fixed for 1h in 4% formaldehyde, 50mM cacodylic acid (pH7.4) and immediately transferred to PBS (130 mM NaCl, 7mM Na2HPO4, 3mM NaH2PO4) where the vitelline membranes were dissected off unhatched embryos (Fig. 3). Stage 11 embryos (Fig. 4) were transferred from PBS to 4 units ml-1 chitinase (Sigma C1525 from Streptomyces griseus; 100 units ml-1 stock prepared in 75 mM Hepes, pH 7.5/130 ntM NaCl) diluted in 100 mM Hepes (pH 8), where they incubated at room temperature for 24 h. Chitinase-treated embryos were rinsed in PBS. All embryos were then incubated at least 12 h in block (2 % BSA, 2 % normal goat serum, 0.5% triton X-100, l×PBS) before adding the primary antibody.
Antibody dilutions and washes between antibody incubations were done in block. Antibody incubations were carried out for at least 24 h at room temperature. Washes were for at least 6h with hourly changes of the block solution. Embryos were incubated in a 50-fold dilution of the αht-en antibody, followed by a 500-fold dilution of GAR-HRP. Following the second wash, the embryos were preincubated in 5 mg ml-1 diaminobenzidine (DAB) in PBS for 15 min before treating with 0.03% hydrogen peroxide with 0.03% NiCl2 and CoCl2 (Adams, 1981). The color reaction was monitored under the dissecting microscope; embryos were transferred to PBS when the desired degree of staining had been achieved. The embryos were then rinsed in 100mM Tris, pH7.5, stained for 1h in lμgml-1 Hoechst in 100 mM Tris, pH7.5, and transferred to 90% glycerol in 100mM Tris, pH7.5. Embryos were transferred to 80% glycerol/10mM Tris, pH7.5/40 mg ml-1 A-propylgalate for dissection and mounting.
Microscopy and photography
Photomicrographs were taken on a Zeiss Axiophot microscope using DIC optics and Tech Pan 2415 film at ASA 100, except for Fig. 3A which was illuminated by fluorescence and obtained from a confocal microscope (Biorad); the image was photographed from a computer moniter using Plus-X pan at ASA 125.
Results
Summary of leech development
The development of glossiphoniid leeches from the newly laid egg to the juvenile leech has been divided into 11 stages (Fernandez, 1980). The process of segmentation occurs over an extended period of time lasting from stage 7–10; segments arise and mature as shown schematically in Fig. 1, from a posterior growth zone composed of 5 bilateral pairs of individually identified stem cells called teloblasts. One pair of teloblasts (M) generates mesoderm and four other pairs of teloblasts (N, O/P, O/P and Q) generate ectoderm. During stage 7 and 8, each teloblast undergoes a series of several dozen unequal divisions at the rate of about 1 per hour (Wordeman, 1982), generating a column (handle f) of primary blast cells. Ipsilateral bandiets come into parallel arrays called germinal bands, within which the bandiets occupy stereotyped positions and are designated m, n, o, p and q, as shown in Fig. 1. During stage 8, the two germinal bands coalesce from anterior to posterior along the ventral midline forming a structure called the germinal plate, from which segments arise. Ganglia first become evident in stage 9, again in a rostrocaudal progression. During stage 10, the edges of the expanding germinal plate meet along the dorsal midline, closing the body tube of the leech, and during stage 11 segmental tissues differentiate to a state approximating their mature form.
Summary of leech development. A schematic summary of the development of segmental tissues, as viewed from the ventral surface. In the lower half of the figure, paired teloblasts, M, N, O/P, O/P and Q at the posterior end each divide repeatedly at a common budding site, giving rise to columns (or bandiets) of blast cells shown in detail only on the right. Alternate blast cells in the n and q bandiets are shaded to denote the two classes of blast cells in these bandiets. Ipsilateral bandiets join to form germinal bands which coalesce along the ventral midline into the germinal plate. A cross section of the right germinal band shows the relative positions of the ectodermal (n,o,p and q) and mesodermal (m) bandiets. Above, a cross section through the germinal plate shows the segmental mesoderm (lighter lines) and ectoderm (heavier lines). Ganglia are shown developing along the ventral midline. Horizontal dotted lines in the germinal plate indicate segment borders.
Summary of leech development. A schematic summary of the development of segmental tissues, as viewed from the ventral surface. In the lower half of the figure, paired teloblasts, M, N, O/P, O/P and Q at the posterior end each divide repeatedly at a common budding site, giving rise to columns (or bandiets) of blast cells shown in detail only on the right. Alternate blast cells in the n and q bandiets are shaded to denote the two classes of blast cells in these bandiets. Ipsilateral bandiets join to form germinal bands which coalesce along the ventral midline into the germinal plate. A cross section of the right germinal band shows the relative positions of the ectodermal (n,o,p and q) and mesodermal (m) bandiets. Above, a cross section through the germinal plate shows the segmental mesoderm (lighter lines) and ectoderm (heavier lines). Ganglia are shown developing along the ventral midline. Horizontal dotted lines in the germinal plate indicate segment borders.
Segments arise in a highly determinate manner from the stereotyped divisions of the blast cells within the germinal bands and germinal plate; older blast cells in each bandiet contribute to more anterior segments, which accounts for the pronounced rostrocaudal temporal gradient seen throughout development (Braun and Stent, 1989). Each bandiet contributes a distinct subset of segmentally iterated neurons and nonneuronal progeny to the mature leech (Weisblat et al. 1984); each m, o and p blast cell generates a full segment’s worth (segmental complement) of definitive progeny for its cell line, although individual clones interdigitate and are distributed over more than one segment (Weisblat and Shankland, 1985). In the N (and Q) cell lines, two classes of blast cells, nf and ns (and qf and qs), arise in exact alternation, each contributing a specific subset of cells to a segmental complement of definitive N (or Q) progeny (Weisblat et al. 1984; Zackson, 1984; Bissen and Weisblat, 1987). Within the N cell fine in particular, the anterior portion of a segment is generated almost entirely from ns-derived progeny, whereas the posterior portion is generated almost entirely from the nf-derived progeny (Bissen and Weisblat, 1987). The nf and ns clones within a segment intermingle slightly, but do not extend beyond the segment borders (Braun and Stent, 1989).
An en homolog in Helobdella
We have previously reported the cloning of an en homolog from Helobdella and the sequencing of the homeobox and 3’ region of this gene (Wedeen et al. 1991). In several species in which en class genes have been studied, two en class genes have been identified (Poole et al. 1985; Joyner and Martin, 1987; Gardner et al. 1988; Fjose et al. 1988; Walldorf et al. 1989; Hemmati-Brivanlou et al. 1991). Exceptions are sea urchin and grasshopper, in each of which only one en homolog has been detected (Dolecki and Humphries, 1988; Patel et al. 1989a). In previous experiments, a single en class gene was detected in Helobdella using a Drosophila en homeobox fragment as a low stringency hybridization probe (Wedeen et al. 1990). Attempts to identify additional en class genes in the leech have been unfruitful: (1) screening 10 genome equivalents of a leech genomic library using low stringency conditions with an en homeobox probe yielded 10 recombinants, all with the same en homeobox cross-hybridizing region (Wedeen et al. 1991); (2) low stringency hybridization of a homeobox-containing fragment from ht-en to Southern blots of Helobdella and Drosophila DNAs identified 2 en class genes in Drosophila DNA and intense bands corresponding to ht-en in Helobdella. Additional faint bands in Helobdella DNA were detected by the ht-en probe under low, but not high, stringency conditions. These bands might correspond to other homeobox-containing genes or to additional en class genes. However, in the latter case, the Helobdella en class genes would be far more divergent from each other than are any known pairs of intraspecific en homologs (Poole et al. 1985; Joyner and Martin, 1987; Walldorf et al. 1989). Although we cannot definitively rule out the existence of a second en class gene in Helobdella on the basis of these negative results, there is no evidence for this possibility.
Polyclonal antibodies against an ht-en/fgal fusion protein
A region of ht-en that is highly conserved among en class genes (the homeobox plus the flanking 5′ 2 bp and 3′57 bp) was amplified by polymerase chain reaction and four tandem repeats of this sequence (ht-en4x) were cloned in frame with the E. coli β-galactosidase (βgal) gene (pUR278; Ruther and Muller-Hill, 1983) and with the E. coli anthranilate synthase (trpE) gene (pATH21; Koerner et al. 1990). The βgal fusion protein was induced, purified from the E. coli extract by the procedure of Ruther and Muller-Hill (1983) and injected into a rabbit. The iterated ht-en sequence was employed to bias the rabbit’s immune response toward the ht-en portion of the fusion protein (R. Kostriken, personal communication). Rabbit polyclonal antibodies directed against ht-en were affinity purified from the crude antiserum. Antibodies specific for ht-en were selected by affinity to the trpE fusion protein that had been purified by electroelution from an SDS-polyacrylamide gel (SDS-PAG) and bound to an affinity column (Affigel, Biorad). Anti-βgal antibodies were removed from the antiserum by running it over a second affinity column to which proteins from a crude bacterial lysate containing the βgal protein (induced pUR278; Ruther and Muller-Hill, 1983) had been bound. The purified antibody was designated αht-en.
In previous studies, a monoclonal antibody, mAb4D9, directed against a highly conserved portion of the Drosophila inv homeodomain had been used to determine the expression pattern of a putative en homolog in the leech (Weisblat et al. 1989; Patel et al. 1989b). We compared the specificities and affinities of mAb4D9 and αht-en on immunoblots of electrophoretic gels run with matched concentrations of crude extracts of E. coli strains bearing βgal alone or fusions of βgal with a single copy of the homeodomain and 3′ conserved region of either en (en/βgal) or ht-en (ht-en/βgal) (Fig. 3A). The Drosophila and Helobdella fusion proteins were of nearly equal size and serial 10-fold dilutions of each were electrophoresed in neighboring lanes. mAb4D9 detected ht-en/βgal with less than 1/1000 the sensitivity with which it detected en/βgal a degree of cross-reactivity equivalent to background. By contrast, αht-en identified both the fusion proteins, but demonstrated approximately 10-fold higher affinity for ht-en/βgal than for en/βgal (see lanes 7 and 10, Fig. 3B). Thus, we believe that the antigen detected by mAb4D9 in leech embryos is not ht-en, but rather some other antigen (possibly another homeodomain protein).
αht-en did not react with bacterial βgal either in the marker lane or in the induced crude extract (Fig. 2B, lane 6), but did detect a single low molecular weight bacterial protein (lanes 6 –10). Thus, to further test the specificity of this antibody and to insure that the antibacterial immunoreactivity would not generate a spurious pattern of staining in leech embryos, we performed the following competition experiments. Crude bacterial lysate from E. coli strains containing either (β gal or the ht-en4x/β gal fusion protein was incubated for 48 h with the purified antibody. After centrifugation, the supernatant was applied at 1:50 final dilution of the antibody to Helobdella embryos and to an immunoblot of a crude ht-en/βgal lysate. The α ht-en preincubated with ht-en4x/β gal-containing lysate failed to stain the immunoblot (Fig. 2C, lane 3) and also failed to stain the embryos (not illustrated). The α ht-en preincubated with lysate containing unfused β gal failed to stain the bacterial protein on the immunoblot, but did recognize ht-en/β gal (Fig. 2C, lane 2) and also revealed patterns of immunoreactive nuclei in embryos identical to those described below. Together, these results indicate that α ht-en specifically reveals the expression pattern of ht-en in Helobdella embryos. However, conventional western blot analysis using α ht-en on extracts of Helobdella embryos failed to show any bands (our unpublished observations) presumably due to the small relative amount of ht-en protein. Thus, we cannot formally exclude the possibility that α ht-en recognizes more than one protein in embryos.
Specificity of αht-en and mAb4D9 on immunoblots of E. coli extracts. E. coli strain XL1 was transformed with pUR278 (Ruther and Muller-Hill, 1983), enHD (Desplan et al. 1985), or pfht-enlx (see Experimental Procedures). Herein, these are referred to as βgal, en/βgal, and ht-en βgal, respectively. The lacZ promoter was induced as described in Experimental Procedures. Crude extracts of the cultures were electrophoresed on 8% (A and C) and 6% (B) SDS-PAGs and transferred to nitrocellulose. (A) Relative affinity of mAb4D9 for en/βgal and ht-en/βgal. Lane 1, molecular weight standards, as indicated in the left margin. Lane 2, en/βgal. Lane 3, ht-en/βgal.. Lane 4, a 10-fold dilution of the sample run in lane 2. Lanes 5–8, further serial 10-fold dilutions of en/βgal. Lane 9, identical to lane 3. Lanes 1–3 were stained for total protein using india ink stain. Lanes 4–9 were treated with mAb4D9 primary and horseradish-peroxidase-conjugated goat anti-mouse secondary antibody (GAM-HRP). The relative positions of the en/βgal and ht-en/βgal fusion proteins are shown by the upper and lower arrowheads, respectively, in the right margin. A 1000-fold dilution of en/βgal fusion protein (lane 6) is clearly identified by the mAb4D9 antibody. Staining specific to the ht-en/βgal fusion (lane 9) is not distinguishable from background. (B) Relative affinity of αht-en for ht-en/figal and en/βgal. Lanes 1 and 5, molecular weight standards as indicated in the left margin. Lanes 2 and 6, βgal. Lanes 3, 7 and 9, en/βgal); lane 9 is a 10-fold dilution of protein relative to lanes 3 and 7. Lanes 4, 8 and 10, ht-en/βgal; lane 10 is a 10-fold dilution of protein relative to lane 4 and 8. Lanes 1 –4 were stained for total protein using india ink stain. In lane 4, a doublet is observed, the lower band of which comigrates with)3gal in lane 2. The upper band represents the ht-en/fusion (large arrowhead in right margin). The slightly higher molecular weight band migrating in lane 3 is the en/βgal fusion protein (small arrowhead in right margin). Lanes 5 –10 were treated with αht-en and HRP-conjugated goat anti-rabbit secondary (GAR-HRP). In each of the crude extracts, an approximately 80 × 103Mr bacterial protein is recognized by α ht-en. In addition, the en/β gal fusion protein is recognized in lane 7 and, with about 10-fold higher efficiency, the ht-en/β gal. protein is recognized in lane 8. (C) Competition for α ht-en using bacterial extracts. Lanes 1-3, ht-en/β gal. Molecular weights are designated in the right margin. Lane 1 was treated with α ht-en. Lane 2 was treated with α ht-en that had been pre-incubated 48 h (4°C) with a 1:1 mix of native and denatured E. coli proteins from strain XL-1; lane 3 with β ht-en that had been preincubated 48 h with a 1:1 mix of native and denatured E. coli proteins from XL-1 that had been transformed with pfht-en4x and induced to produce the hr-en4x/β gal fusion protein. All lanes were treated with GAR-HRP. In the left margin the upper (large) arrowhead indicates the position of ht-en/β and the lower (small) arrowhead indicates the position of the approximately 80 × 103Mr bacterial protein to which α ht-en cross reacts.
Specificity of αht-en and mAb4D9 on immunoblots of E. coli extracts. E. coli strain XL1 was transformed with pUR278 (Ruther and Muller-Hill, 1983), enHD (Desplan et al. 1985), or pfht-enlx (see Experimental Procedures). Herein, these are referred to as βgal, en/βgal, and ht-en βgal, respectively. The lacZ promoter was induced as described in Experimental Procedures. Crude extracts of the cultures were electrophoresed on 8% (A and C) and 6% (B) SDS-PAGs and transferred to nitrocellulose. (A) Relative affinity of mAb4D9 for en/βgal and ht-en/βgal. Lane 1, molecular weight standards, as indicated in the left margin. Lane 2, en/βgal. Lane 3, ht-en/βgal.. Lane 4, a 10-fold dilution of the sample run in lane 2. Lanes 5–8, further serial 10-fold dilutions of en/βgal. Lane 9, identical to lane 3. Lanes 1–3 were stained for total protein using india ink stain. Lanes 4–9 were treated with mAb4D9 primary and horseradish-peroxidase-conjugated goat anti-mouse secondary antibody (GAM-HRP). The relative positions of the en/βgal and ht-en/βgal fusion proteins are shown by the upper and lower arrowheads, respectively, in the right margin. A 1000-fold dilution of en/βgal fusion protein (lane 6) is clearly identified by the mAb4D9 antibody. Staining specific to the ht-en/βgal fusion (lane 9) is not distinguishable from background. (B) Relative affinity of αht-en for ht-en/figal and en/βgal. Lanes 1 and 5, molecular weight standards as indicated in the left margin. Lanes 2 and 6, βgal. Lanes 3, 7 and 9, en/βgal); lane 9 is a 10-fold dilution of protein relative to lanes 3 and 7. Lanes 4, 8 and 10, ht-en/βgal; lane 10 is a 10-fold dilution of protein relative to lane 4 and 8. Lanes 1 –4 were stained for total protein using india ink stain. In lane 4, a doublet is observed, the lower band of which comigrates with)3gal in lane 2. The upper band represents the ht-en/fusion (large arrowhead in right margin). The slightly higher molecular weight band migrating in lane 3 is the en/βgal fusion protein (small arrowhead in right margin). Lanes 5 –10 were treated with αht-en and HRP-conjugated goat anti-rabbit secondary (GAR-HRP). In each of the crude extracts, an approximately 80 × 103Mr bacterial protein is recognized by α ht-en. In addition, the en/β gal fusion protein is recognized in lane 7 and, with about 10-fold higher efficiency, the ht-en/β gal. protein is recognized in lane 8. (C) Competition for α ht-en using bacterial extracts. Lanes 1-3, ht-en/β gal. Molecular weights are designated in the right margin. Lane 1 was treated with α ht-en. Lane 2 was treated with α ht-en that had been pre-incubated 48 h (4°C) with a 1:1 mix of native and denatured E. coli proteins from strain XL-1; lane 3 with β ht-en that had been preincubated 48 h with a 1:1 mix of native and denatured E. coli proteins from XL-1 that had been transformed with pfht-en4x and induced to produce the hr-en4x/β gal fusion protein. All lanes were treated with GAR-HRP. In the left margin the upper (large) arrowhead indicates the position of ht-en/β and the lower (small) arrowhead indicates the position of the approximately 80 × 103Mr bacterial protein to which α ht-en cross reacts.
Ht-en expression, stage 9. (A) Fluorescence photomicrograph of four half segments from a stage 9 embryo in which an N teloblast was microinjected (Weisblat et al. 1978) with lysinated rhodamine dextran (RDA) at early stage 7. Anterior is up, ventral midline at right. Tracer-labeled cells appear white; ht-en positive nuclei (black, indicated in one segment by black-on-white arrows) arise within lobes of cells derived from nf blast cells (short arrows) and not within those derived from ns blast cells (long arrows). (B,C) Bright-held photomontage and tracing of 22 segments in a dissected germinal plate. Boxes in C indicate fields shown at higher magnification in D and E. Anterior is up, ventral midline at center. Ganglia developing along ventral midline are designated by a solid outline; areas where ganglia will soon form are designated by a dotted outline. Posteriorly, development is least advanced; no ht-en expression is detected. Stripes first appear as 2 bilateral pairs of nuclei/segment (bottom arrow). More anteriorly, additional nuclei stain; when segmental ganglia appear, the boundaries between them are aligned with the stripes (see E). Concomitantly, expression disappears within the stripe except for two cells (nzl, nz2; left arrows and D) and new expression appears in ganglionic nuclei, first in two pairs of cells at ventral midline (upper right arrow). Still more anteriorly, additional ganglionic nuclei express ht-en. Slight deviations from bilateral symmetry in the staining pattern are not consistent from embryo to embryo and we attribute them to developmental noise. Scale bar, 26 μ in A, 50 μz in B, 23 μ in D and E.
Ht-en expression, stage 9. (A) Fluorescence photomicrograph of four half segments from a stage 9 embryo in which an N teloblast was microinjected (Weisblat et al. 1978) with lysinated rhodamine dextran (RDA) at early stage 7. Anterior is up, ventral midline at right. Tracer-labeled cells appear white; ht-en positive nuclei (black, indicated in one segment by black-on-white arrows) arise within lobes of cells derived from nf blast cells (short arrows) and not within those derived from ns blast cells (long arrows). (B,C) Bright-held photomontage and tracing of 22 segments in a dissected germinal plate. Boxes in C indicate fields shown at higher magnification in D and E. Anterior is up, ventral midline at center. Ganglia developing along ventral midline are designated by a solid outline; areas where ganglia will soon form are designated by a dotted outline. Posteriorly, development is least advanced; no ht-en expression is detected. Stripes first appear as 2 bilateral pairs of nuclei/segment (bottom arrow). More anteriorly, additional nuclei stain; when segmental ganglia appear, the boundaries between them are aligned with the stripes (see E). Concomitantly, expression disappears within the stripe except for two cells (nzl, nz2; left arrows and D) and new expression appears in ganglionic nuclei, first in two pairs of cells at ventral midline (upper right arrow). Still more anteriorly, additional ganglionic nuclei express ht-en. Slight deviations from bilateral symmetry in the staining pattern are not consistent from embryo to embryo and we attribute them to developmental noise. Scale bar, 26 μ in A, 50 μz in B, 23 μ in D and E.
Ht-en expression in stage 11 nervous system. (A) Photomontage (DIC optics) of three midbody ganglia and surrounding body wall from a dissected stage 11 embryo treated with aht-en and GAR-HRP to identify 6 pairs of ganglionic cells and 3 pairs of peripheral cells expressing ht-en. The N-derived peripheral neurons, nzl and nz2 are designated on the right side of the embryo with arrowheads. Another pair of presumptive peripheral neurons, located just lateral to each ganglion and not derived from the N cell line are designated on the left side by arrows; the righthand member of this pair is not visible in the anterior segment and is out of focus in the posterior segment (see B). Scale bar, 50 μ. (B) Tracing of A. Ganglia are outlined, /it-en-positive cells are indicated as black circles.
Ht-en expression in stage 11 nervous system. (A) Photomontage (DIC optics) of three midbody ganglia and surrounding body wall from a dissected stage 11 embryo treated with aht-en and GAR-HRP to identify 6 pairs of ganglionic cells and 3 pairs of peripheral cells expressing ht-en. The N-derived peripheral neurons, nzl and nz2 are designated on the right side of the embryo with arrowheads. Another pair of presumptive peripheral neurons, located just lateral to each ganglion and not derived from the N cell line are designated on the left side by arrows; the righthand member of this pair is not visible in the anterior segment and is out of focus in the posterior segment (see B). Scale bar, 50 μ. (B) Tracing of A. Ganglia are outlined, /it-en-positive cells are indicated as black circles.
ht-en expression during early development and neurogenesis
Throughout development, ht-en expression, as revealed by immunostaining with α ht-en, is bilaterally symmetric and localized to the nuclei of immunoreactive cells (Figs 3,4), in accord with the postulated role of en class proteins as transcription factors. The earliest detected expression of ht-en occurs in the germinal bands during late stage 7. Through mid-stage 8, expression occurs as a dynamic series of segmentally iterated patterns, the details of which remain to be determined.
By late stage 8, clones of cells derived from primary mesodermal (m) blast cells in the anterior portion of the germinal plate, which is developmentally more advanced, have formed hollow blocks of tissues corresponding to hemisomites (Zackson, 1982). Within the ectoderm, aggregations of cells are starting to become apparent, but ganglia are not yet evident (Fernandez, 1980; Torrence and Stuart, 1986). Within these anterior segments, ht-en immunoreactive nuclei appear as segmentally iterated transverse stripes. In each segment, the stripe extends across the central portion of the germinal plate, with a discontinuity at the ventral midline. By early stage 9, in accord with the rostrocaudal progression of development, the stripes are present in posterior segments. The subsequent temporal progression of ht-en expression can be inferred by observing more anterior, developmentally more advanced segments (Fig. 3). [Each segment is about 2h more advanced than the one posterior to it (Braun and Stent, 1989).] Several conclusions emerge from examining such preparations:
When it first becomes evident, the stripe consists of 2– 3 nuclei on either side of the midline (Fig. 3B and C). Over the next few hours, additional nuclei become immunoreactive, so that the stripe eventually contains 6– 7 nuclei on either side of the midline (Fig. 3E).
When segmental ganglia first become apparent, the boundaries between adjacent ganglia are aligned with the stripes (Fig. 3B and C).
As development progresses, ht-en disappears in all but two nuclei on each side, which move laterally as the germinal plate expands (Fig. 3D).
As the stripes fade, new ht-en expression appears in a subset of ganglionic cells, in a stereotyped spatiotemporal pattern, starting with a pair of anteromedial ht-en-positive nuclei within the developing ganglion (Figs 3B and C). Within a few hours, a second pair becomes detectable, just dorsal to the first. Soon thereafter, two more pairs of ventral nuclei begin expressing ht-en near the lateral margins of the ganglion; at about the same time, a fifth posteromedial pair appears. Finally, after several more hours, two pairs of anterodorsal nuclei begin to express ht-en. Except for the loss of expression by one of the anterodorsal pairs, the resultant pattern of ht-en expression in the midbody ganglia (Fig. 4) is maintained throughout advanced stages of gangliogenesis, as late as stage 11. A few extraganglionic cells also express ht-en in stage 11.
Lineage analysis of ht-en expressing cells
To ascertain the embryonic origin(s) of the stripes of nuclei expressing ht-en, fluoresecent lineage tracer was injected into one N teloblast of a series of embryos at early stage 7 (see Fig. 1). At early stage 9, the embryos were fixed and processed to reveal ht-en immunoreactivity. The clones of nf- and ns-derived progeny have not yet intermingled at this stage and form two transverse lobes of cells, extending laterally to greater and lesser extents, respectively (Fig. 3A; Weisblat et al. 1980a; Zackson, 1984). These embryos demonstrate that the stripe of ht-en-positive (black) nuclei is derived entirely from the nf blast cell, whose definitive progeny contribute to the posterior half of the segment.
Similar double label experiments on embryos at stages 10– 11 reveal that a variety of ganglionic and peripheral neurons derived from both nf and ns blast cells express ht-en. In particular, the ganglionic ht-en- positive cells at the anterior ventral midline arise from ns blast cells (data not shown). In addition, the 2 ht-en- positive peripheral nuclei that arise from the nf-derived stripe and persist in expression (Fig. 4) become cells nzl and nz2, two peripheral neurons already identified as being derived from the nf blast cells (Braun and Stent, 1989). The late expression of ht-en is not confined to the N cell fine, however, since additional ht-en- positive nuclei are observed by stage 11 near two epidermal specializations known as cell florets 3 and 6 (not shown), structures arising exclusively from the P and Q cell lines, respectively (Braun and Stent, 1989). Whether or not these cells are neurons remains to be determined.
Discussion
Patel et al. (1989a) described a monoclonal antibody, mAb4D9, which recognizes an 11 amino acid epitope within the homeodomain of the Drosophila invected (inv) gene product. This region of the inv homeodomain is identical to the corresponding region of the Drosophila en protein and is highly conserved in several other en class proteins. Therefore, mAb4D9 has been used to infer the expression pattern of en homologs in numerous species (Patel et al. 1989a; Hemmati-Brivanlou et al. 1989; Fleig, 1990; Whitington et al. 1991). In Helobdella embryos, mAb4D9 stains neurons in the second, third and fourth fused neuromeres of the subesophageal ganglion, a pattern more reminiscent of vertebrate than of arthropod staining patterns (Patel et al. 1989a; Hemmati-Brivanlou et al. 1989; Davis et al. 1991), and consistent with the notion of Field et al. (1988) that annelids and arthropods are not phyletically close (Patel et al. 1989a).
However, the ht-en sequence differs from that of en and inv by two amino acids in the mAb4D9 epitope. One of these differences (asparagine in Helobdella versus glycine in Drosophila) occurs at a residue that is critical for antibody binding, as evidenced by the finding that mAb4D9 does not cross react with the sea urchin or mouse en homologs, which have threonine and serine, respectively, rather than glycine at this position (Patel et al. 1989a). We have shown here that mAb4D9 recognizes the Drosophila en homeodomain with more than 1000-fold greater sensitivity than it recognizes the corresponding portion of Helobdella ht-en, as judged by immunoblots of fusion proteins. Based on genomic Southern blot analysis and library screening, it appears that ht-en is the only en class gene in Helobdella. These results suggest that mAb4D9 does not recognize an en class protein in Helobdella. The expression pattern observed with this monoclonal, may be that of a different homeodomain-containing protein. But in any case, it seems clear that the polyclonal antibody described here reveals the true pattern of ht-en expression in Helobdella.
It is hard to make meaningful comparisons between the timing of en expression in Drosophila and that of ht-en in Helobdella because the segmentation processes of Helobdella and Drosophila differ extensively at the cellular level. For example, identifiable founder cells for hemisomites in Helobdella arise sequentially beginning roughly 15 h after egg deposition, with the 8th cell cycle, as primary m blast cells which immediately begin generating morphologically distinct clumps of cells, before gastrulation has even begun (Zackson, 1982). In Drosophila by contrast, morphologically distinct segmental structures arise in the mesoderm about 3.5-4.5h after fertilization, but this is during the 15th cell cycle, by which time gastrulation is well under way (Campos-Ortega and Hartenstein, 1985). Thus, although embryogenesis proceeds at a much slower rate in Helobdella than in Drosophila, morphogenetic aspects of segmentation in leech are markedly advanced in terms of cell generations.
Because the segmentation processes are so different in Drosophila and Helobdella, it is difficult to identify homologies in these processes. However, in the early Drosophila embryo, the interface between the en-expressing cells and neighboring cells is necessary to establish the segmental and parasegmental borders (Martinez-Arias et al. 1988). Although en eventually becomes expressed in several rows of cells per segment, in theory such an interface could be achieved by a single row of en-expressing cells, as we observe for ht-en in Helobdella. Thus, although we do not yet understand ht-en function in the leech, it may be that the transient expression of ht-en in a segmentally iterated stripe is necessary to delineate the border between developing ganglia and/or other segmental structures.
The pattern of en class protein expression reported here for Helobdella, including both segmentally iterated stripes before ganglion formation and a segmentally iterated subset of presumptive neurons later in development, is strikingly reminiscent of the patterns of expression of en class proteins previously described for arthropods (DiNardo et al. 1985; Brower, 1986; Patel et al. 1989a; Fleig, 1990). We interpret these similarities as indicating that a pattern of dual expression of an en class protein was already present in a common ancestor of annelids and arthropods. From this, we predict that such dual expression patterns of en homologs will also be found in other segmented protestóme phyla and perhaps in molluscs as well. Moreover, the fact that ht-en is expressed during segmentation by cells descended from just the nf (posterior) class of n blast cells suggests that the anterior/posterior compartments of Drosophila and the nf/ns blast cells in Helobdella may both derive from a segmental subdivision that was present in the common ancestor, consistent with the previously proposed notion of homology between the ns and nf blast cells in leech and the anterior and posterior compartments of arthropods (Bissen and Weisblat, 1987).
ACKNOWLEDGEMENTS
The research reported here was supported by Damon Runyon Cancer Fund Postdoctoral Fellowship DRG 925 to CJW and NIH FIRST Grant HD 23328 to DAW. We are indebted to Rich Kostriken and Shirley Bissen for suggestions on the fusion protein production and antibody purification. We thank Rich Kostriken, Deborah Lans, Brad Nelson, Steve Torrence, Duncan Stuart and Nipam Patel for helpful discussions and suggestions.