Introduction

The leech CNS has been used to study numerous topics in development and physiology (Kristan et al. 2005). The stereotyped cell lineage of the leech embryo is well characterized (Stent et al. 1992), and phenomena such as neurogenesis (Baptista et al. 1990), axonal pathfinding (Baker and Macagno 2000; Venkitaramani et al. 2004), synaptogenesis (Johnson et al. 2000; Marin-Burgin et al. 2005), and neuronal cell death (Macagno and Stewart 1987; Shankland and Martindale 1989) have been examined in detail. The morphogenesis and segmentation of the central nervous system (CNS) have also been investigated using cellular methodologies (Blair 1982; Stuart et al. 1989; Torrence et al. 1989; Ramírez et al. 1995; Shain et al. 1998, 2000), but little is known about the regulation of CNS morphogenesis at the molecular level.

In the leech, the mature CNS is composed of a segmented ventral nerve cord connected at its anterior end to an unsegmented supraesophageal ganglion (Stent et al. 1992). The segmented tissues arise from a defined set of large, embryonic stem cells called ‘teloblasts’ (Weisblat and Shankland 1985), each of which undergoes a series of highly asymmetric stem cell divisions to produce a linear column of primary blast cell daughters (Fig. 1). One bilateral pair of M teloblasts generates the mesodermal germ layer, and four bilateral pairs of N, O, P, and Q teloblasts generate the ectoderm. Blast cells are designated by the same letter as their parent teloblast in lower case and function as segmental founder cells for that teloblast lineage.

Fig. 1
figure 1

Schematic overview of the formation of the leech germinal band. The M, N, O, P, and Q teloblasts are a set of bilaterally symmetric embryonic stem cells, here shown for only one side. Each teloblast undergoes an iterated sequence of highly asymmetric cell divisions, producing a linear column of primary blast cell daughters. The five ipsilateral columns merge in parallel to form the germinal band. The individual teloblast lineages adopt specific positions in the band, with the ectodermal N, O, P, and Q lineages arrayed in the ventral-to-dorsal order over the surface of the mesodermal M lineage (inset). The right and left germinal bands fuse along the ventral midline to form the germinal plate. The midline is marked with an arrowhead

Full size image

Early in gastrulation, the five ipsilateral blast cell columns merge to form a germinal band in which the N, O, P, and Q lineages are arrayed in ventral-to-dorsal order over the surface of the M lineage (Fig. 1, inset). The oldest blast cells are the first to enter the germinal band and will eventually contribute their clones to the anteriormost body segments (Fig. 1). As the blastopore closes, the right and left germinal bands meet along the ventral midline and fuse to form the germinal plate (Fig. 1). The ganglia of the CNS develop along this midline.

The segmentation of the leech CNS is implicit in the division pattern of the teloblastic stem cells (Weisblat and Shankland 1985). For example, each N teloblast generates an alternating sequence of daughters designated ns and nf blast cells (Bissen and Weisblat 1987). Each ns cell generates a descendant clone of 70–80 neurons that populate the anterior quadrant of a segmental ganglion or neuromere (Shain et al. 1998). The nf cell produced by the next N teloblast division generates a descendant clone of similar size that primarily populates the posterior quadrant of that same hemiganglion. Together, the right and left N lineages account for roughly two thirds of the central neurons, with additional neurons migrating into the ganglion from the other teloblast lineages. Unlike the N and Q teloblasts, the M, O, and P teloblasts each generate only one blast cell per segment (Weisblat and Shankland 1985).

The developmental cues required for ganglion formation have been studied with embryonic cell ablation. Unilateral ablation of an M teloblast leads to irregular segmentation on the ipsilateral side of the CNS (Blair 1982; Torrence et al. 1989). This ablation eliminates only a few neurons and is thought to influence gangliogenesis by removing a positional framework of mesoderm-derived cues that guide the neural sublineages to appropriate locations (Torrence et al. 1989). A similar although milder disruption of CNS segmentation is produced by knocking down the expression of the Hau-Pax3/7A gene in the segmental mesoderm (Woodruff et al. 2007), suggesting that this transcription factor may be involved in the generation and/or deployment of such cues.

Other studies have suggested that the N lineage coordinates its own segmental morphogenesis autonomously. For instance, certain aspects of gangliogenesis still occur in the absence of the mesoderm (Shain et al. 2000). In addition, unilateral ablation of an N teloblast lineage disrupts ganglion formation (Blair and Weisblat 1982; Stuart et al. 1989), and the ablation of a single nf cell prevents the formation of an interganglionic fissure (Ramírez et al. 1995; Shain et al. 2000). Early reports suggested that the leech’s engrailed gene might play a critical role in the formation of this fissure (Ramírez et al. 1995), but subsequent work suggests that this is unlikely to be the case (Shain et al. 2000).

We here report the identification of two Pax6 genes from the leech Helobdella sp. (Austin) and show that Hau-Pax6A is primarily expressed in the N teloblast lineage. Genes of the Pax6 subfamily encode transcription factors with two discrete DNA-binding domains: an upstream paired domain and a downstream homeodomain (Balczarek et al. 1997). Pax6 genes are best known for their widely conserved role in eye development (Callaerts et al. 1997) and have also been implicated in the development of the CNS (Stoykova et al. 1997; Kammermeier et al. 2001; Callaerts et al. 2001). In leech embryos, Hau-Pax6A is expressed in a cell population associated with the developing eye primordium and also displays a segmentally iterated expression pattern that precedes and may contribute to the morphological segmentation of the CNS.

Materials and methods

Animals

Embryos were taken from a laboratory breeding colony established in 1997 with leeches collected in Austin, TX. This population was originally described as H. robusta on the basis of morphology (Seaver and Shankland 2000), but molecular phylogenetic analysis suggests they are a distinct and as yet unnamed species (Bely and Weisblat 2006), referred to here as Helobdella sp. (Austin).

Adult leeches were maintained in 1% artificial seawater and fed on pond snails collected at the University of Texas Brackenridge Field Laboratory. Embryos were raised at 24°C in a defined saline and staged according to the Stent et al. (1992) system for H. triserialis.

Identification and cloning of genes

Hau-Pax6A was isolated from the cDNA by the polymerase chain reaction (PCR) using degenerate primers targeted to conserved amino acid sequences (VSNGCV and FAWEIR) in the paired domain. We extended the mRNA sequence in the 5′ and 3′ directions by the rapid amplification of cDNA ends (RACE) technique, using the Gene Racer kit and gene-specific primers on cDNA synthesized at 63°C with Thermoscript reverse transcriptase (Invitrogen). Primer sequences are available upon request.

A partial transcript Hau-Pax6B was isolated from the cDNA using gene-specific primers predicated on a second Pax6 homolog identified from genome sequence traces of the leech H. robusta, available at http://www.ncbi.nlm.nih.gov/blast/tracemb.shtml.

The phylogenetic relationship of the leech Pax6 genes was examined with PAUP 4.0b10.

RNA analysis

The size of the Hau-Pax6A mRNA was determined by gel electrophoresis and Northern blotting. Total RNA from stage 8–10 embryos was isolated with Tri-Reagent (Molecular Research Center), run on a 1.5% agarose denaturing gel, transferred to nylon, and hybridized overnight with a 1.0-kb digoxigenin (DIG)-labeled riboprobe complementary to Nt 533–1,489. Hybridization was detected by alkaline phosphatase immunostaining using nitroblue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate substrates.

Accumulation of RNA during embryogenesis was measured with reverse transcription PCR (RT-PCR). Embryos were collected at various times after egg deposition, and maturing oocytes were dissected from the ovary of a gravid adult. For each developmental stage, total RNA was extracted with the RNAqueous kit (Ambion) and reverse transcribed using random primers. A region of the Hau-Pax6A transcript (Nt 533–772) was then amplified using gene-specific primers, and an 18 S rRNA sequence was amplified in parallel as a loading control (QuantumRNA, Ambion).

Cellular distribution of RNA was examined by whole mount in situ hybridization (Nardelli-Haefliger and Shankland 1992). DIG-labeled sense and antisense riboprobes were generated from the 1.0-kb clone used for northern blot analysis, and hybridization was visualized with the same reagents. Stained embryos were either cleared in glycerol or dehydrated through an ethanol series and cleared in a 3:2 mixture of benzyl benzoate/benzyl alcohol. Some specimens were embedded in POLYBed 812 plastic (Polysciences) and handcut by razor blade into 0.1-mm sections. All specimens were imaged with a Diagnostic Instruments Spot CCD camera.

Lineage tracing

To trace cell lineages, identified blastomeres were microinjected with a 10-kDa tetramethylrhodamine- or fluorescein-dextran-amine (Molecular Probes) at 50 mg/ml, with 2% Fast Green FCF added to monitor injection volume. Injected embryos were counterstained with 2.5 μg/ml Hoechst 33258 and viewed on a Nikon E800 fluorescence microscope.

Results

Gene isolation

A novel Pax6 gene was identified in Helobdella sp. (Austin) by degenerate RT-PCR, and the two ends of the gene product were amplified by RACE. The RACE products predicted a 3,520-bp mRNA with a polyA tail, and Northern blots of embryonic RNA revealed a single band of roughly this size (Fig. 2a). We designated this gene Hau-Pax6A and deposited a consensus cDNA sequence in GenBank under accession no. EF375876.

Fig. 2
figure 2

Expression of Hau-Pax6A RNA. a Northern blot analysis of total embryonic RNA detected a single prominent band. b RT-PCR detected a low level of Hau-Pax6A RNA in oocytes but not in zygotes or early cleavage stage embryos. Zygotic expression was first detected at embryonic stage 7 and increased at later stages. Sample loading was normalized by parallel amplification of 18 S rRNA

Full size image

After the isolation of Hau-Pax6A, we identified two Pax6 homologs in the online genome sequence of the closely related leech H. robusta. We used this sequence information to clone a second Helobdella sp. (Austin) Pax6 homolog, which we have designated Hau-Pax6B. A partial cDNA sequence of the latter gene has been deposited in GenBank under accession no. EF394359. Characterization and expression analysis will be deferred to a future manuscript.

Molecular characterization

Conceptual translation of the Hau-Pax6A cDNA revealed a paired domain at Nt 297–680 (Fig. 3a) and a homeodomain in the same reading frame at Nt 1443–1622 (Fig. 3b). Outside of these two domains, we did not detect any significant similarity to other known protein sequences. Some Pax6 genes encode a proline-, serine-, and threonine-rich transactivation domain in the C-terminal region (Czerny and Busslinger 1995), but the latter motif was not evident in Hau-Pax6A. The 3′ RACE products indicated that the RNA was polyadenylated, and a polyadenylation signal sequence was found at Nt 3499–3504.

Fig. 3
figure 3

Sequence alignment of the paired domain (a) and homeodomain (b) of the leech genes Hau-Pax6A and Hau-Pax6B with Pax6 homologs from other species. Amino acids identical to Hau-Pax6A are marked with a period. Sequences were taken from polychaete Platynereis dumerilii (AM14770), squid Euprymna scolopes (AF513712), nemertean Lineus sanguineus (X95594), planarian Dugesia japonica (AB017632 and AJ311310), fruitfly Drosophila melanogaster (NM_079889), and mouse Mus musculus (NM_013627)

Full size image

Splicing and polyadenylation suggested that the Hau-Pax6A transcript was translated mRNA, but none of our Hau-Pax6A cDNAs had an upstream ATG codon inframe with the paired domain coding sequence. The five longest 5′ RACE products had stop codons 0.2 kb upstream of the paired domain, with two intervening ATG sequences that were not in the same reading frame as the paired domain. There was an inframe CTG codon at Nt 282–284; if the latter sequence functioned as a noncanonical translation start site (Riechmann et al. 1999), then the consensus cDNA would encode a 933 amino acid protein.

There were sequence variations in our cloned Hau-Pax6A cDNAs suggestive of polymorphisms in the leech population. Most involved short indels in stretches of repetitive sequence, i.e., variable numbers of AAT/AAC repeats in three strings of asparagine codons and variable numbers of TCA repeats in a string of serine codons. These putative polymorphisms only affected the predicted protein sequence downstream of the homeodomain.

Phylogenetic analysis

Hau-Pax6A and Hau-Pax6B clustered with other known members of the Pax6 subfamily in phylogenetic analyses of paired domain and homeodomain sequences (Fig. 4). The Hau-Pax6A gene was most similar to Hau-Pax6B, with 87% amino acid identity in the paired domain and 90% in the homeodomain (Fig. 3). In contrast, the Hro-Pax6B gene product was more similar to Pax6 genes from other species, with 95% amino acid identity to the Pax6 paired domains from two squid species (Tomarev et al. 1997; Hartmann et al. 2003) and 93% identity to a Pax6 homeodomain from the sea urchin (Czerny and Busslinger 1995). Thus, it appeared that Hau-Pax6B has retained a protein sequence more characteristic of the Pax6 subfamily, while Hau-Pax6A has undergone a higher rate of sequence diversification.

Fig. 4
figure 4

The Helobdella genes Pax6A and Pax6B clustered with other Pax6 genes in phylogenetic analyses of paired domain (a) and homeodomain (b) protein sequences. Phylograms were generated by the neighbor-joining method with branch lengths reflecting levels of sequence divergence. Numbers represent bootstrap percentages from 1,000 replicates, and branch nodes incompatible with the 50% majority rule were collapsed. The paired domain tree was rooted with mouse Pax2 (P32114) and Drosophila Pax258 (NP_524633) as outgroups. The homeodomain tree was rooted with mouse Pax3 (AAH48699) as the outgroup. In addition to mouse Pax4 (NM_011038), other sequences used were the same as for Fig. 3

Full size image

Some evidence suggested that Hau-Pax6A and Hau-Pax6B arose from a gene duplication event that occurred at a relatively recent time in leech evolution. Only a single Pax6 gene has been reported in the only other annelid species examined, the polychaete Platynereis dumerilii (Arendt et al. 2002). In addition, the Hau-Pax6A and Hau-Pax6B homeodomains clustered with one another (54% of bootstrap replicates) in a neighbor-joining tree that included all known Pax6 genes from the superphylum Lophotrochozoa (Fig. 4b). On the other hand, the two leech Pax6 paired domains did not show as strong a phylogenetic affinity, with the Hau-Pax6A paired domain sorting to a more basal position in a neighbor-joining tree (Fig. 4a).

Developmental time course

The time course of Hau-Pax6A RNA expression was examined by RT-PCR using primer pairs flanking a known splice site. We detected a low level of Hau-Pax6A RNA in oocytes (Fig. 2b) but not in zygotes or early cleavage stages (stages 2–6). This finding suggested that maternal Hau-Pax6A RNA is present during oogenesis but degraded by the early stages of embryonic development. A second phase of de novo, presumably zygotic Hau-Pax6A expression began at embryonic stage 7 and increased during later stages (Fig. 2b).

In situ hybridization

The cellular distribution of Hau-Pax6A RNA was characterized by in situ hybridization. Reproducible staining patterns were observed from stage 7 through the end of embryonic development but not at earlier stages. Control hybridizations with sense probes gave little or no staining at any stage.

Germinal bands and plate. Beginning in stage 7, Hau-Pax6A RNA was detected in the anterior, i.e., most mature, portion of the germinal bands (Fig. 5a). Early expression was restricted to blast cell progeny of the N teloblast lineage, and expression persisted in this lineage as the bands fuse to form the germinal plate. A few cells in the O teloblast lineage also began to show a segmental pattern of expression during germinal plate formation (Fig. 5a). We did not detect any Hau-Pax6A expression in the M, P, or Q lineages at these stages.

Fig. 5
figure 5

Expression of Hau-Pax6A visualized by whole mount in situ hybridization. a Ventral view of stage 8 embryo showing the fusion of the right and left germinal bands (cf. Fig. 1) with anterior to the top. At this stage, there was a segmental pattern of Hau-Pax6A expression, including numerous cells in the N lineage (arrowheads) and a single segmentally repeated cell in the O lineage (arrows). b Anterior view of a stage 8 embryo with dorsal to the top. At this stage, there were bilateral clusters of stained cells (arrowheads) in the prostomial head domain. Hybridization was also visible more ventrally in the segmental nervous system. c, d Side views of a stage 9 embryo (c) and stage 10 embryo (d) with anterior to the left and dorsal to the top. Expression of Hau-Pax6A RNA disappeared in an anterior-to-posterior wave during the maturation of the segmental nerve cord but persisted in the supraesophageal ganglion (arrowheads) throughout embryogenesis. Arrows point to a single mature cell in each segmental ganglion that also retained Hau-Pax6A expression. e The head of a stage 11 embryo in the same orientation as parts c and d. The ventral surface of the rostral sucker is marked by a bracket. At this stage, there were clusters of Hau-Pax6A-expressing cells in the supraesophageal ganglion (outlined by dashes) and at the site of eye formation (arrow). Longitudinal rows of single Hau-Pax6A-expressing neurons appeared in the dorsal integument at this stage; the two most dorsal rows are in focus (arrowheads). f Transverse section of a stage 10 embryo in the plane of the supraesophogeal ganglion, which is outlined by dashes. The empty proboscis sheath (asterisk) passes through the center of the ring-shaped ganglion. Arrows mark Hau-Pax6A-expressing cells in the dorsal integument; arrowheads mark cells located in the anterior segments of the ventral nerve cord. Dorsal is to the top. Scale bars–50 μm

Full size image

We used fluorescent lineage tracers and nuclear staining to analyze the cellular pattern of Hau-Pax6A expression in more detail. The N teloblast lineage is comprised of alternating ns and nf blast cells, which undergo stereotyped sequences of asymmetric cell divisions within the germinal band. We found that the sequences of n and o blast cell divisions in Helobdella sp. (Austin) were identical to those reported for H. triserialis (Bissen and Weisblat 1989) and that Hau-Pax6A RNA was first expressed in cell ns.a, the anterior daughter of the ns blast cell (Fig. 6a,b).

Fig. 6
figure 6

Colocalization of the Hau-Pax6A RNA expression with injected cell lineage tracers. Anterior to the top. a, b RNA expression (arrowhead) first appeared in a segmentally repeated cell of the N teloblast lineage, here labeled with rhodamine dextran (a). Individual cells can be identified by nuclear size, seen by Hoechst staining (b). The ns and nf blast cells have completed their first divisions, and their paired daughter cells are marked. At this stage, the hybridization product was restricted to cell ns.a, the anterior daughter of an ns blast cell. An undivided ns blast cell can be seen at the bottom. Future medial to the right. c–e In this embryo, the O teloblast lineage was injected with fluorescein-dextran in such a way that a single primary blast cell clone showed a distinct pattern of faint labeling (asterisk; see Bissen and Weisblat 1987). The clone shown here contained nine progeny cell nuclei (arrowheads, d), including a single large cell that expressed Hau-Pax6A (outlined arrowhead) as seen by partial occlusion of Hoechst fluorescence (d) and brightfield illumination (e). Future medial to the right. f, g Brightfield view of the Hau-Pax6A expression along the ventral midline of the germinal plate (f) in a stage 8 embryo whose left N teloblast lineage had been labeled with rhodamine dextran (g). By this stage, the labeled cell lineage has adopted a segmental pattern of lateral bulges and was a mosaic of cells that do and do not express Hau-Pax6A. Four nonexpressing cells are marked with arrowheads. Segmental expression of Hau-Pax6A in the O lineage could be seen more laterally on both sides. Scale bars–10 μm (a, b), 20 μm (c–e), and 50 μm (f, g)

Full size image

As blast cell clones matured, the Hau-Pax6A expression pattern expanded to other cells of the N lineage including progeny of the nf blast cell. The dynamic character of the N lineage expression pattern prevented us from individually identifying cells at later stages. Nonetheless, we could see a clear segmental periodicity of Hau-Pax6A expression where the N lineages pass from the right and left germinal bands into the germinal plate (Fig. 5a, b). Expression persisted through the early stages of ganglion formation, at which time segmental repeats were composed of an iterated pattern of both Hau-Pax6A-expressing and nonexpressing cells (Fig. 6f,g).

In the O teloblast lineage, Hau-Pax6A expression began near the point of germinal band fusion and was associated with a single, segmentally repeated sublineage (Figs. 5a, 6d–g). To delimit individual clones of primary o blast cell descendants, we took advantage of the fact that a certain percentage of teloblast injections yield a distinctive faint labeling of the next blast cell produced (Bissen and Weisblat 1987). Focusing on these faintly labeled clones, we found that Hau-Pax6A expression first appeared after the primary o blast cell had divided to produce nine clonal progeny and was restricted to the largest cell in the clone (Fig. 6c–e). Based on size and position, we believe that this Hau-Pax6A-expressing cell corresponds to cell o.apapl described by Bissen and Weisblat 1989. The single large cell was later replaced by two small, adjacent Hau-Pax6A-expressing cells–possibly the result of cell o.apapl’s next division–which appeared to merge medially with the Hau-Pax6-expressing cells of the N lineage.

As the CNS matured, Hau-Pax6A expression faded in an anterior-to-posterior wave (Fig. 5c,d). Expression disappeared from the anteriormost segments during stage 8 and by stage 10 was only visible in the neuromeres associated with the caudal sucker. However, there was one bilateral pair of cells that continued to express Hau-Pax6A through the end of embryogenesis at the anterior edge of every segmental ganglion or neuromere (Fig. 5d).

Head development. Expression of Hau-Pax6A also appeared during late stage 7 in the unsegmented head domain, where it occurred in a bilateral array of cell clusters associated with the supraesophageal ganglion (Fig. 5b). In contrast to the segmented portion of the CNS, this head ganglion maintained robust Hau-Pax6A expression throughout embryonic life (Fig. 5c–f).

Some Hau-Pax6A-expressing cells were distributed in peripheral tissues closely associated with the supraesophageal ganglion (Fig. 5e,f). Helobdella develops a single bilateral pair of eyes on the dorsal lip of its rostral sucker, and a population of Hau-Pax6A-expressing cells was evident at this site during embryonic stages 9–11 (Fig. 5e). Eye pigmentation normally develops in stage 11 (Shankland and Martindale 1989), but our hybridization protocol bleached the eye pigment, and we were unable to use the pigment as an anatomical marker.

Dorsal integument. During the transition from embryonic stages 10 to 11, Hau-Pax6A expression appeared in an orthogonal array of isolated cells distributed in the dorsal integument. This integumental staining developed in an anterior-to-posterior wave and consisted of three longitudinal rows of segmentally repeating cells located symmetrically on either side of the dorsal midline (Fig. 5e).

Discussion

We have identified two Pax6 genes in the leech Helobdella sp. (Austin) and here characterize the embryonic expression of Hau-Pax6A. Like Pax6 genes from a wide range of other animal species (Callaerts et al. 1997, 2001; Stoykova et al. 1997; Arendt et al. 2002; Pineda et al. 2002), Hau-Pax6A is expressed during the development of the CNS and in a population of cells associated with the developing eye.

One noteworthy feature of the Hau-Pax6A mRNA is the absence of a canonical translation start site upstream of the paired domain coding sequence. The Hau-Pax6A mRNA may use a noncanonical start site (Riechmann et al. 1999), possibly a CTG codon (see “Results”). A second possibility is that an inframe AUG is generated by RNA editing before translation (Gott and Emeson 2000). If the latter is the case, we have only been successful to date in cloning the unedited sequence. It is worth noting that a second leech Pax gene, Hau-Pax3/7B, also lacks an inframe AUG codon upstream of its paired domain coding sequence (Shankland and Woodruff, unpublished results).

Hau-Pax6A in CNS development

Within the segmented body trunk, Hau-Pax6A expression precedes the appearance of overt morphological segmentation or ganglion formation and is largely restricted to the N teloblast lineage, which gives rise to roughly two thirds of the ganglionic neurons (Weisblat and Shankland 1985). Early embryonic expression was also observed in one segmentally iterated descendant of the O teloblast, which we tentatively identify as cell o.apapl. The latter cell is also part of a sublineage known to produce central neurons (Shankland 1987). However, there was no early Hau-Pax6A expression in the M, P, or Q teloblast lineages, although they also contribute neurons to the CNS (Weisblat and Shankland 1985).

The observed pattern of Hau-Pax6A expression in the leech CNS is suggestive of a developmental role in segmentation and/or ganglion formation. The N teloblast lineage is composed of an alternating sequence of ns and nf blast cells whose descendant clones form the structural framework of the ganglia (Shain et al. 2000). Moreover, ablation experiments indicate that the N lineage plays a critical role in ganglion morphogenesis (Blair and Weisblat 1982; Stuart et al. 1989). It has been suggested that proper ganglion formation depends on differences in the adhesivity of certain n blast cell sublineages (Shain et al. 2000), and it is interesting to note that Pax6 regulates cell adhesivity during development of the mammalian brain (Stoykova et al. 1997) and eye (Collinson et al. 2000). The early segmental pattern of Hau-Pax6A expression in the n blast cell clones of the leech embryo may lead to changes in adhesivity that facilitate ganglion formation.

Some authors have argued that a small population of N-derived engrailed-expressing cells may play a critical role in the formation of fissures between the segmental ganglia of the leech CNS (Lans et al. 1993; Ramírez et al. 1995). This hypothesis was later discounted by Shain et al. 2000, who showed that the formation of interganglionic fissures precedes en expression within the N lineage. This is in contrast to Hau-Pax6A, whose expression begins at an earlier stage in the N lineage development and clearly precedes the cell rearrangements that bring about ganglion formation.

Hau-Pax6A expression largely disappears from the segmental ganglia before the onset of neurotransmitter expression (Stuart et al. 1987; Shankland and Martindale 1989) or extensive axonal projections (Braun and Stent 1989). This timing suggests that Hau-Pax6A has little function in the differentiation of segmentally iterated neurons, although it is still expressed in one bilateral pair of segmentally repeated cells in the mature CNS. Hau-Pax6A expression also persists through late embryonic development in neurons of the unsegmented supraesophageal ganglion and could be playing distinct developmental functions in that region of the CNS.

Hau-Pax6A in eye primordia and other peripheral tissues

We also observed the expression of Hau-Pax6A outside the CNS. The most prominent peripheral expression surrounded the supraesophageal ganglion, including a bilateral cell population located at the site of eye formation. The expression of a Pax6 gene in the developing eye of the leech Helobdella is in no way surprising given the comparable results from many other bilaterian animals (Callaerts et al. 1997; Pineda et al. 2002).

Late Hau-Pax6A expression was also observed in an orthogonal array of integumental cells on the dorsal side of the body trunk. The distribution of these cells is similar to that of the sensillae described in adult leeches (Sawyer 1986), and they may represent one particular type of sensillar neuron or neural precursor. Leech sensillae are primarily mechanosensory but can also detect light (Kretz et al. 1976). Additional studies will be needed to ascertain whether the Hau-Pax6A expression seen in the dorsal integument of Helobdella embryos is associated with the development of extraocular photoreceptors.

Gene duplications in the Pax6 subfamily

The last common ancestor of protostomes and deuterostomes is thought to have had only a single Pax6 gene (Balczarek et al. 1997). However, duplications of Pax6 appear to have occurred in a number of evolutionary lineages derived from that ancestor. For instance, arthropods have two Pax6 genes, eyeless and twin-of-eyeless, that play distinct roles in eye development (Kammermeier et al. 2001).

Two lines of evidence suggest that the gene duplication which gave rise to Hau-Pax6A and Hau-Pax6B occurred after leeches had diverged from polychaete annelids. First, only a single Pax6 gene has been reported in the polychaete Platynereis dumerilii (Arendt et al. 2002) and in several other organisms from the superphylum Lophotrochozoa: two species of squid (Tomarev et al. 1997; Hartmann et al. 2003) and a nemertean (Loosli et al. 1996). Planaria have two Pax6 genes (Pineda et al. 2002); however, our phylogenetic sequence analysis failed to reveal any consistent alignment between the planarian and leech Pax6 gene pairs, suggesting that they arose from independent gene duplications.

A second line of evidence consistent with a recent gene duplication is the observation that the Hau-Pax6A and Hau-Pax6B homeodomains grouped together in our phylogenetic analysis of lophotrochozoan Pax6 genes. However, this grouping was not seen for their paired domain sequences, possibly because the more highly diverged Hau-Pax6A paired domain was attracted to a more basal position in that tree. Pax6 genes will need to be characterized in additional lophotrochozoan species before we can reconstruct the history of gene duplications with greater certainty.