Introduction

Hox and ParaHox genes play dominant roles in the development of metazoan animals (cf. Brooke et al. 1998). In particular, these genes have well characterized roles in patterning the anteroposterior axis of bilaterally symmetric triploblasts, and are implicated in patterning the oral-aboral axis of diploblastic cnidarians (Ferrier and Holland 2002; Finnerty et al. 2003). It is widely thought that mutations affecting Hox/ParaHox genes, most particularly their regulation, provide a powerful and parsimonious means for macroevolutionary change, while duplication of these genes opens new avenues for anagenetic innovation (e.g. Plaza et al. 2001; Ronshaugen et al. 2002; Yanze et al. 2001).

Hox and ParaHox gene clusters are thought to have descended by duplication of a (thus far hypothetical) ProtoHox gene cluster (Brooke et al. 1998), although some uncertainties remain about the precise structure and function of the ancestral ProtoHox gene cluster. It is also unclear when the hypothesized gene cluster duplication occurred, or indeed when a single ancestral ProtoHox gene tandemly duplicated to form this ProtoHox gene cluster. These events most probably precede the divergence of the lineages leading to cnidarians and to triploblast bilaterians (Finnerty and Martindale 1999; Ferrier and Holland 2002), thus focusing attention on more basal lineages of metazoan life. Unfortunately, uncertainty persists concerning phylogenetic relationships at the base of the Metazoa. Many researchers think that sponges (phylum Porifera) represent the most basal lineage of metazoans, while others consider the phylum Placozoa to be basal (for the different views see Rokas et al. 2003, and Ender and Schierwater 2003). From published data we cannot decide between the two conflicting hypotheses, but at least it is clear that Trichoplax is not a derived cnidarian (Ender and Schierwater 2003). Regardless of which scenario is correct, the phylum Placozoa is likely to offer valuable insights into Hox and ParaHox gene origins. One reason for stressing the importance of Placozoa is that analyses of sponges have yielded no valid genes closely related to Hox or ParaHox genes (and thus candidates for ProtoHox genes). In contrast, the partial sequence of a single Hox/ParaHox-related gene, named Trox-2, has been reported from the placozoan Trichoplax adhaerens (Schierwater and Kuhn 1998). This gene can be considered a valid candidate for a ProtoHox gene.

Comparison of the 60 amino acid homeodomain sequences reveals close similarity between Trox-2 and the cnidarian Cnox-2 gene, the best studied homeobox gene from the phylum Cnidaria (Schierwater et al. 2002). In turn, Cnox-2 has been proposed to be orthologous to the Gsx family of ParaHox genes characterized in bilaterians (Finnerty and Martindale 1999; Ferrier and Holland 2002). However, it is also possible that Trox-2 is a ProtoHox gene, and that following gene duplication one of the daughter genes (Cnox-2, Gsx) retained closer sequence similarity to the ancestral ProtoHox gene, because of functional constraints. In support of this, no other Hox/ParaHox-type gene has been detected in Placozoa, in contrast to the multiple Hox/ParaHox genes detected in all cnidarians analysed. In both Hydrozoa and Anthozoa, the Cnox-2 gene is involved in axis formation, that is, the development of the foot versus head region, which may or may not be homologous to the anteroposterior axis of triploblastic bilaterian animals. The temporal and spatial patterns of expression vary with respect to cell layer, developmental stage, and species (see Schierwater et al. 2002; Finnerty et al. 2003), but all studies on cnidarian Cnox-2 genes report some kind of axial expression (e.g. Cartwright et al. 1999; Shenk et al. 1993; Yanze et al. 2001).

Placozoans possess neither a main body axis nor an entoderm in a traditional sense (cf. Syed and Schierwater 2002). This raises a highly intriguing question. What could be the expression pattern, and function, of a candidate ProtoHox gene in a non-symmetric animal lacking a defined body axis? The placozoan body appears as an irregular “hairy plate” (“tricho plax”) consisting of four recognized somatic cell types, which are arranged in an upper epithelium and a lower (or feeding) epithelium with dispersed fibre cells in between these epithelia (see Syed and Schierwater 2002 for morphological details; Martinelli and Spring 2003 for suggestion of additional cell types). Here we report the cDNA sequence of Trox-2 from T. adhaerens and show that Trox-2 expression marks the separation boundary between the lower and upper epithelium, and thus follows the irregular body shape of the animal. In addition, Trox-2 functional inhibition studies suggest an indispensable role in growth and reproduction.

Materials and methods

Animal material

For all parts of this study we used a single clone, “Grell-Bochum”, which we have been culturing in the lab since 1995 (Schierwater and Kuhn 1998). Under laboratory conditions T. adhaerens reproduces vegetatively by fission. For laboratory culture, we use only the products of binary fission. Other products of vegetative reproduction are small spheric swarmers which, unfortunately, have never been found to develop under laboratory conditions. Under certain conditions we find sexual organisms and putative blastula stages which do not develop further (cf. Grell 1972, 1983). Thus all animals studied are the product of vegetative reproduction and thus represent a single clone, which we call TriGre-3. This type of clonal propagation appears to be a shared primitive character for metazoans (Blackstone and Jasker 2003). Prior to DNA or RNA extraction, animals are cultured for 24 h in seawater without algae to avoid contamination with the food, the alga Pyrenomonas (Cryptophyta). Individuals subjected to whole-mount in situ analyses were treated either the same way or used unstarved. Expression patterns were the same in both cases. Gene inhibition studies by means of morpholino oligos were performed under standard culturing conditions except that morpholino oligos were dissolved in the seawater (see later).

Molecular cloning

Whole genomic DNA was extracted from live T. adhaerens according to Ender and Schierwater (2003). Total RNA was extracted with the aid of Trizol Reagent (Invitrogen) and used for cDNA synthesis (GeneRacer, Invitrogen) following the manufacturer’s protocol. To obtain full length cDNA sequences of Trox-2 5′ and 3′ RACE was used (e.g. Kuhn et al. 1999 for protocol). The gene-specific primers used were: 5′ RACE reverse primer ATCGGCTACTGTTGAACTCC; 3′ RACE forward primer TCATGATGAAAACGAAAG. Desired fragments were recovered by agarose gel electrophoresis and cloned into the pCRTOPO-II plasmid vector (Topo-TA, Invitrogen). Several independent clones were sequenced, and the full length transcript of the formerly partially isolated Trox-2 gene (Schierwater and Kuhn 1998) was identified (GenBank acc. AY531880).

In situ hybridization

Whole-mount in situ hybridization experiments were carried out according to standard protocols (Gröger et al. 1999; Müller et al. 1999) which were modified for fluorescence detection. Briefly, specimens of T. adhaerens were fixed overnight in ice-cold Lavdowsky fixative (ethanol/formaldehyde/glacial acetic acid/aqua bidest: 50/10/4/36), washed with PBST (PBS buffer containing 0.1% Tween 20), permeabilized by proteinase K treatment for 2 min at 37°C, and post-fixed for 1 h in 4% paraformaldehyde/0.2% glutaraldehyde. Optionally and for long-term storage, samples were dehydrated in ascending PBST/methanol dilution series. Digoxigenin-labeled RNA sense and antisense probes were synthesized from a 384-bp PCR fragment (Fig. 1) flanked by T7 and Sp6 promoter sites according to the manufacturer’s protocol (Roche). Prior to hybridization, specimens were rehydrated in PBST, washed in 50% hybridization buffer and prehybridized in hybridization buffer (50% formamide, 5× SSC, 1% SDS, 50 µg/ml heparin, 100 µg/ml yeast tRNA) for 1 h at 55°C. Hybridizations were carried out overnight in the presence of 100 ng/ml RNA probe at 55°C. Samples were washed two times, (1) 20 min at 55°C in 50% formamide/5× SSC/1% SDS and, (2) 20 min at 37°C in 50% formamide/2× SSC/1%SDS, equilibrated in 2× SSC/0.1% Tween 20, and subsequently treated with 20 µg/ml RNAse A in 2× SSC/0.1% Tween 20 for 30 min at 37°C. Animals were washed first in 2× and subsequently in 0.2× SSC/0.1% for 20 min at 55°C. Finally, specimens were equilibrated and stored prior to detection in PBST. Immunological detection of bound RNA probe was done either by alkaline phosphatase linked anti-digoxigenin (Roche) or by FITC-conjugated anti-digoxigenin (Sigma) fluorescence detection. Unspecific binding of antibodies was blocked by incubating samples in blocking buffer containing 100 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.5% blocking reagent (Roche) for 1 h. Incubation in antibody solution (blocking buffer plus 1/1,000 diluted anti-Dig-AP or 1/200 diluted anti-Dig-FITC, respectively) was done for 40 min at RT. After three washing steps in PBST (20 min each), colorimetric detection was carried out in reaction buffer (100 mM Tris-HCl, pH 9.5, 100 mM NaCl, 50 mM MgCl2, 0.1% Tween 20) and substrate (4.5 µl NBT/ml, 3.5 µl/ml BCIP; Roche). For fluorescence signal detection, specimens were counterstained in propidium iodide and mounted on microscope slides with a 9:1 glycerol/PBST medium containing DABCO as anti-fade reagent (Sigma). Visual inspection and documentation were achieved using an inverse Zeiss fluorescence microscope (Axiovert 200 M) equipped with green and red spectral filter sets (490 nm and 545 nm excitation, respectively) and connected to a CCD camera.

Fig. 1
figure 1

Trox-2 cDNA sequence from Trichoplax adhaerens. The homeodomain as well as the sequence of the double-stranded RNA used in RNAi experiments (position 160 to 544) are underlined; the complement of the morpholino antisense oligonucleotide (nucleotide position 45–69) and the TSFKIESLI motif are highlighted in bold

Full size image

Gene inhibition studies

For the gene inhibition studies, replicates of ten animals were maintained in a single culture dish in a volume of 5 ml artificial seawater. Food was supplied ad lib. and water was changed at 3-day intervals. Trox-2 gene inhibitions were obtained by (a) morpholino antisense oligonucleotides and (b) double-stranded RNA.

Trox-2 inhibition by morpholino antisense oligonucleotides

A chemically modified morpholino antisense oligonucleotide (Gene Tools, LLC; Partridge et al. 1996; Summerton et al. 1997; Summerton and Weller 1997) 25 nucleotides long was designed complementary to the Trox-2 cDNA, in a 5′ region close to the start methionine codon (sequence shown in Fig. 1). Transfection with morpholino antisense oligonucleotides was done with the aid of EPEI solution (ethoxilated polyethylenimine; Gene Tools, LLC) in a total volume of 10 ml (cf. Morcos 2001). The modified oligonucleotide was added to the seawater to a final concentration of 1 µM. Control oligos comprised either a random sequence of the same length, or the sense sequence of the experimental Trox-2 morpholino.

Trox-2 inhibition by RNAi

Double-stranded RNA was synthesized by means of SP6/T7-RNA polymerase from a 384 bp cDNA template (Fig. 1). This template was cloned into pGEM-T and is referred to as Trox-2-ss3–9.00. Single RNA strands (total length 545 bp; 384 bp Trox-2 cDNAs plus 161-bp vector sequence) were synthesized at 37°C for 2 h in 20 µl transcription buffer (Boehringer Mannheim) in the presence of 250 ng template, 40 U RNase, 2 U T7 or SP6 RNA polymerase, and 10 mM each NTP, including fluorescein UTP. After digestion of the DNA template, ssRNA was precipitated and resuspended in ddH2O, containing 1 µl RNase inhibitor, before complementary strands were annealed. Transfection of animals with dsRNA was done in a total volume of 200 µl with the aid of FuGENETM-6 Transfection Reagent (Roche Applied Science) following the FuGENETM-6 protocol. Transfection success was verified by means of semi-quantitative RT-PCR. Ten Trox-2 transfected and 50 Cnox-1 transfected control animals were collected 2 days after transfection, total RNA was extracted and RT-PCR was carried out according to Li et al. (2000). Controls for RNAi comprised (1) untreated animals in normal seawater, and (2) animals transfected with dsRNA synthesized from the Cnox-1 gene of the hydroid Eleutheria dichotoma.

Sequence similarity between Trox-2 and Cnox-2

The Trox-2 gene is a comparatively short gene: the cDNA including the full open reading frame consists of only 645 bp (Fig. 1). This represents the first complete homeodomain protein sequence deduced for the phylum Placozoa. The homeobox and homeodomain sequences of Trox-2 have high sequence similarity to the Cnox-2 genes of Cnidaria, although as we have noted previously sequence similarity outside the homeobox declines to the level of randomness (Schierwater and Kuhn 1998) except for the partially conserved “TSFKIESLI” motif, which seems to be characteristic of the Cnox-2/Gsx orthology group (Finnerty et al. 2003). Using homeodomain sequences to build a Neighbour-Joining tree places Trox-2 at the base of Cnox-2 proteins, defining a group of genes that we refer to as the Diplox-2 family (Diploblast Hox Gene 2; Fig. 2). We would like to note that this result is based on similarity and does not necessarily reflect phylogenetic relationships. Some recent analyses have concluded that cnidarian Cnox-2 genes are orthologous to the Gsx ParaHox gene of bilaterians (Finnerty et al. 2003; Ferrier and Holland 2001), although this conclusion is still controversial (Schierwater and Desalle 2001). Taken at face value these phylogenetic analyses suggest that Trox-2 is a ParaHox gene, although if this inference is correct it is strange that no additional Hox/ParaHox genes have been reported from Trichoplax. We therefore raise the possibility that Trox-2 is a ProtoHox gene, and that the Cnox-2 and Gsx gene families retain plesiomorphic sequence characters that distort phylogenetic tree reconstruction. Resolution of this issue awaits genomic characterization of Trox-2 and extensive screening for ANTP class homeobox genes in Trichoplax. At present, the hypothesis that Trox-2 is ancestral to all Hox/ParaHox genes, in the sense that it is most similar and most closely related to the proposed ProtoHox gene (cf. Brooke et al. 1998; Gauchat et al. 2000), is congruent with existing data and with a proposed basal position for Placozoa (Schierwater et al., in preparation).

Fig. 2
figure 2

Neighbour-Joining tree of all known full-length Diplox gene homeodomains. Trox-2 groups at the base of known Cnox-2 genes from Cnidaria (Diplox-2 group, highlighted), while all other Diplox genes are separated on different clades. The tree is intended to reflect basic similarity and not necessarily phylogenetic relationships, and thus no support values are given. For the cnidarian Diplox-2 homologs, only the name of the genus from which the Cnox-2 gene was isolated is given in the figure. Note that other Diplox genes may carry the same cnox gene number but are not homologous (for family suggestions see Kuhn et al. 1999). GenBank accession numbers are: Hydractinia AF031953, Podocoryne AF268446, Sarsia AF285145, Chlorohydra X64626, Hydra AJ277388, Cassiopeia AF124592, Nematostella AAD39349, Acropora AF245689, Eleutheria cx4 AAB48009, Eleutheria cx3, Podocoryne cx4 AY036893_2, Cassiopeia cx3 AF124593, Acropora AntpC S36771, Podocoryne cx1 X81455, Eleutheria cx5 AAB48010, Hydra cx3 L22787, Hydra cx4 AJ252181, Hydra cx3 AJ252182, Hydra cx1 S39066, Cassiopeia cx1 AF124591, Cassiopeia cx5 AAD32579, Cassiopeia cx4 AAD32578, Eleutheria (Sagasser and Schierwater, unpublished), Carybdea (Schroth and Schierwater, unpublished), Eleutheria cx1 (Kuhn and Schierwater, unpublished)

Full size image

Although we have used the generally accepted method of sequence identity and phylogenetics to address how Trox-2 relates to other genes, we accept that this might not reveal the complete picture. For example, a recent approach for classifying homeobox genes advocates the use of complex sequence attributes (Sarkar et al. 2002). Whether this would provide greater resolution, or indeed detect alternative homologies, is not yet known, and awaits assembly of larger data sets.

Trox-2 expression marks the epithelial separation boundary

Whole-mount in situ hybridization analyses in growing and fissioning placozoans revealed a clear and reproducible signal just inside the outer margin of the animal in almost all specimens (Fig. 3). We can detect this signal both with NBT/X-phosphate staining and fluorescent labeling (Fig. 3). A few animals (<5%) did not show any expression signal. These animals are probably not in a phase of growth and/or fission. By focusing through the animal, expression signals can be localized between the upper and lower epithelium, close to the peripheral margin. This boundary area, where the two epithelia connect, follows the outer animal shape homogeneously around the entire body, and thus appears as a closed irregular circle. Interestingly this spatial expression pattern is reminiscent of formerly described peripheral features, including a marginal rim or zone (Schwartz 1984; Rassat and Ruthmann 1979), peripheral birefringent granules (Pearse et al. 1994), and Rfamide staining (Schuchert 1993). In cross sections of in situ stained animals, small, dispersed, Trox-2-expressing cells are identified in this location (Fig. 3). These cells are much smaller than either the epithelial cells or the inner fibre cells. This report is the first recognition of the existence of this cell type in Trichoplax.

Fig. 3
figure 3

Trox-2 expression in Trichoplax adhaerens. Growing and fissioning placozoans express Trox-2 homogeneously close to the outer edge of the body (NBT-X-phosphate staining; pink, upper main panel), immunofluorescence assay (green, lower main panel). Trox-2-expressing cells are small cells between the upper und lower epithelium and are not fibre cells (see arrows in small cross-section picture). Homogeneous actin gene expression throughout the entire animal and both actin and Trox-2 sense control are also shown. Further explanations are given in the text

Full size image

Although we cannot deduce the precise function or fate of these cells, we propose (from their position and gene expression properties) that they could be a population of omni- or even totipotent cells, which can differentiate into epithelial or other cells during development and growth. Comparable cell types include the I-cells of Cnidaria and the archaeocytes of sponges. In this context, it is interesting to note that two other regulatory genes in Trichoplax, PaxB and Brachyury, are expressed in the same general region but with a less regular spatial pattern (showing concentration to certain areas close to the outer edge of the animal; Hadrys et al., in preparation; Martinelli and Spring 2003). We speculate that the latter two genes may be expressed in specific differentiating cells derived from the more widespread omni- or totipotent cells. In the absence of embryonic developmental data and a comprehensive characterization of these inner non-fibre cells, speculations about a potential proto-mesoderm would be tempting but unsupported.

Trox-2 is indispensable for growth and fission

We inhibited Trox-2 function by means of both RNAi and morpholino antisense oligonucleotides. The large double-stranded RNA molecules used for RNAi were successfully transferred into the animal by means of the FuGENETM-6 protocol. This is clearly demonstrated by the homogeneous dispersal of fluorescent signal throughout the body. Furthermore, this transfection causes a dramatic reduction in the amount of endogenous Trox-2 mRNA, as revealed by RT-PCR (Fig. 4, compare lane 6 with lane 2). Transfection of a control double-stranded RNA does not affect Trox-2 mRNA levels (Fig. 4, lane 4), and neither treatment affects levels of actin expression, used as a control reference (Fig. 4, lanes 1, 3, 5).

Fig. 4
figure 4

Transfection of Trichoplax adhaerens with double-stranded Trox-2 RNA. The FITC-labeled dsRNA homogenously enters the entire body (left) and dramatically reduces Trox-2 expression (right). RT-PCR products from untreated control animals are shown in lanes 1 and 2, from control dsRNA animals in lanes 3 and 4, and from Trox-2 dsRNA-infected animals in lanes 5 and 6. The strong product in lanes 1, 3, and 5 is actin expression (control reference). Note the strong decline of Trox-2 product in lane 6 compared to 2 and 4 (a small amount of product in lane 6 is visible on the original gel)

Full size image

The smaller morpholino oligonucleotides were dissolved in seawater and also successfully entered living placozoans, presumably taken up by pino- or phagocytosis. Evidence that the oligonucleotides were successfully taken up is the striking effect on fission and growth of Trichoplax cultures (Fig. 5a). We found that animals stopped growing and fissioning cum grano salis immediately. In contrast, control cultures treated with a control morpholino continued to increase in density at a rapid rate. To our knowledge this is the first report of infecting a marine animal with morpholino oligonucleotides by means of pino- or phagocytosis. Transfection with double-stranded Trox-2 RNA (RNAi) affects placozoans in the same manner (Fig. 5b). Animals stopped growing and fissioning, in sharp contrast to untransfected animals or controls transfected with dsRNA matching a non-Trichoplax Hox-like gene. Following both methods of Trox-2 inhibition, animals started dying after 3–4 weeks without any obvious change in morphology and without leaving any kind of reproductive product. It was clear that cultures of experimental (inhibitor-treated) animals showed cessation of fission, rather than an increase in death rate to a level compensating for vegetative reproduction. Since culture dishes contained no more than 10–15 individuals and were checked at least once every day, any remains of dead bodies would have been noticed. We did not observe any obvious change in body size in experimental animals over 3 weeks, suggesting that the cease in fission is a result of cessation of growth (because binary fission—and vegetative reproduction in general—depends on growth in between fission acts). It would be desirable to find out how growth cessation relates to epithelial cell division rates (cf. Bosch and David 1984) and the spatial distribution of the latter within animals.

Fig. 5a, b
figure 5

Trox-2 gene inhibition leads to cessation of growth and fission in T. adhaerens. Shown are the time course of increase in numbers as a result of (a) inhibition by double-stranded RNA and (b) inhibition by morpholino antisense oligonucleotides. In both cases, control 1 is untreated animals cultured in normal seawater. Control 2 used a sense-control morpholino oligonucleotide (a) or dsRNA targeted to a Hox-like gene, Cnox-1, from a different species, the hydrozoan Eleutheria dichotoma (b)

Full size image

Our findings clearly indicate that Trox-2 is indispensable for growth and fission in T. adhaerens. Combining this observation with the gene expression data suggests a possible mechanism underlying the effect on growth and fission. We suggest that inhibition of Trox-2 function is blocking the supply of differentiated somatic cells derived from Trox-2-expressing multi- or omnipotent precursor cells at the periphery of the animal. As a result, growth and consequent fission would be impossible. Thus, the gene inhibition studies are congruent with our hypothesis that Trox-2 is crucially involved in cell differentiation processes at the boundary between the lower and upper epithelium.

Implications for ancestral gene functions

Hox and ParaHox genes are involved in axial patterning in both Cnidaria and Bilateria, suggesting that these genes evolved axial roles before these two major metazoan lineages diverged. It is likely that the lineage leading to the Placozoa diverged even earlier (Ender and Schierwater 2003; Schierwater et al., in preparation). If so, analysis of Hox/ParaHox genes in Trichoplax should give insight into the roles of these genes at an earlier point in metazoan evolution. T. adhaerens is also interesting in this context because it does not have a defined body axis. Our gene expression and gene function studies suggest that Trox-2 plays a key role in defining the boundary between the two principal cell layers in the disk-shaped body. It is conceivable that this reflects an ancestral role for Hox/ParaHox genes prior to the evolution of true body axes in Metazoa.

Finally, the successful application of the morpholino and RNAi strategies indicate that is possible to inhibit gene function in Placozoa. The protocols used here will open new opportunities to study gene function in the simplest—and possibly also the most primitive—of metazoan animals.