Formation and subdivision of the head field in the centipede Strigamia maritima, as revealed by the expression of head gap gene orthologues and hedgehog dynamics

Abstract

Background

There have been few studies of head patterning in non-insect arthropods, and even in the insects, much is not yet understood. In the fly Drosophila three head gap genes, orthodenticle (otd), buttonhead (btd) and empty spiracles (ems) are essential for patterning the head. However, they do not act through the same pair-rule genes that pattern the trunk from the mandibular segment backwards. Instead they act through the downstream factors collier (col) and cap‘n’collar (cnc), and presumably other unknown factors. In the beetle Tribolium, these same gap and downstream genes are also expressed during early head development, but in more restricted domains, and some of them have been shown to be of minor functional importance. In the spider Parasteatoda tepidariorum, hedgehog (hh) and otd have been shown to play an important role in head segmentation.

Results

We have investigated the expression dynamics of otx (otd), SP5/btd, ems, and the downstream factors col, cnc and hh during early head development of the centipede Strigamia maritima. Our results reveal the process of head condensation and show that the anteroposterior sequence of specific gene expression is conserved with that in insects. SP5/btd and otx genes are expressed prior to and during head field formation, whereas ems is not expressed until after the initial formation of the head field, in an emerging gap between SP5/btd and otx expression. Furthermore, we observe an early domain of Strigamia hh expression in the head field that splits to produce segmental stripes in the ocular, antennal and intercalary segments.

Conclusions

The dynamics of early gene expression in the centipede show considerable similarity with that in the beetle, both showing more localised expression of head gap genes than occurs in the fly. This suggests that the broad overlapping domains of head gap genes observed in Drosophila are derived in this lineage. We also suggest that the splitting of the early hh segmental stripes may reflect an ancestral and conserved process in arthropod head patterning. A remarkably similar stripe splitting process has been described in a spider, and in the Drosophila head hh expression starts from a broad domain that transforms into three stripes.

Background

The mechanisms that specify the head segments of arthropods are fundamentally different from those that operate in the trunk [1,2,3,4]. Some of the genes that define segment boundaries in the trunk are also used in the head (engrailed, wingless, hedgehog), but how the expression of these segment polarity genes is linked to upstream axial patterning remains unclear.

Currently, most insights into arthropod head patterning are based on studies from the fly Drosophila melanogaster. A gradient of the maternally provided anterior morphogen bicoid (bcd) in the blastoderm stage embryo [5] leads to the activation of the so-called head gap genes orthodenticle (otd), empty spiracles (ems) and buttonhead (btd) in broad overlapping domains. They were termed gap genes because mutations in these genes delete a coherent block of specific head segments [6,7,8,9,10]. Sloppy paired has also been proposed as a head gap gene [11], but its role in head development remains unclear. It is not known whether these genes are direct targets of bcd, but cis-acting elements that contain bcd binding sites have been identified upstream of ems and btd [8, 9].

In Drosophila, the domains of expression of these head gap genes are out of phase by one segment [1, 2, 6]. Each factor is required for the correct establishment of a subset of segmental stripes of wingless (wg) and/or hedgehog (hh) expression in the head [12]. It has not yet been fully elucidated whether there is direct control of segment polarity gene expression by the head gap genes or whether their effects are transduced through second level mediators that would replace the pair rule patterning system that acts in the Drosophila trunk but not in the head (see [1]).

The whole head segmentation mechanism in Drosophila is likely to be in at least some respects specific to the higher Diptera: bcd as a major anterior determinant is only present in the higher flies [13,14,15]. Different maternal signals are used in other insects [16, 17]. In many arthropods, the embryo is patterned after cellularisation, and any link between maternal coordinates and embryonic patterning remains unknown. Since the earliest patterning mechanism is so specialised in Drosophila, it is also questionable whether the downstream mechanism of gap gene patterning is present in the same way in other arthropods.

More recently, the beetle Tribolium castaneum has been established as a second model for insect head patterning. Work in Tribolium has shown that orthologues of the Drosophila gap genes otd, ems and btd are expressed during early head development [18,19,20], but their expression domains are not as broadly overlapping as they are in Drosophila. RNAi mediated knockdown of ems and btd had no (for btd) or only rather subtle (for ems) effects when compared to the gap gene mutant phenotypes in Drosophila. Only the knockdown of otd produced a severe head phenotype [20]. RNAi knockdown of ems and otd orthologues in the milkweed bug Oncopeltus fasciatus also gave mild phenotypes. No btd orthologue could be isolated from this species [21]. In both Tribolium and Oncopeltus, these mild phenotypes may be due to limitations of the RNAi method and need to be confirmed by gene knockout. Regardless of this, the results suggest that orthologues of Drosophila head gap genes are expressed during early head development in other insects, but their role may be rather different.

Another transcription factor of interest for insect head patterning is collier (col), which acts downstream of the head gap genes btd and ems in the fly. Col was the first intermediately acting factor to be identified in Drosophila head segmentation [22]. It directly controls the expression of segment polarity genes in the intercalary segment [22, 23]. It is also required for correct expression of cap‘n’collar, a factor that specifies mandibular identity [22, 24]. Furthermore, col is repressed at its posterior limit by the trunk pair rule gene even-skipped, hence taking input from both the head and the trunk patterning system [22]. Expression of col in the pre-mandibular region is conserved in other insects and in the millipede Glomeris marginata and has been discussed as being responsible for the appendage-less phenotype of the intercalary segment in insects and myriapods [25, 26]. However, the lack of a homeotic phenotype after removal of col, and its function in head segmentation in Drosophila [22, 26, 27], rather speak in favour of a crucial role in head segmentation and integration of head and trunk patterning that might be conserved in myriapods. Therefore, we included col in our analysis of head patterning genes in the centipede.

Further insights into arthropod head patterning come from work on the spider Parasteatoda tepidariorum (formerly called Achaearanea tepidariorum). Here, the signalling factor hedgehog (hh) is dynamically expressed in a cephalic domain that splits into three segmental stripes [3, 4]. Subdivision of a cephalic hh domain into ocular, antennal and intercalary segmental stripes also happens in Drosophila head segmentation [28], though this process has been little studied. A study in the millipede G. marginata has shown that the anterior-most hh stripes, belonging to the ocular and the antennal segment, are also generated by a stripe splitting event [29]. These similar patterns suggest that a domain-splitting mechanism, rather than a sequential addition of segments [4], might generally characterise arthropod head segmentation. We ask whether subdivision of an early cephalic hh domain also occurs in this centipede, and how this domain and resulting subdomains relate to the expression of other early expressed head patterning genes.

We chose to study the centipede Strigamia maritima to provide information on head patterning from another distant outgroup to the insects and to facilitate an evolutionary comparison of head patterning across the arthropods. Our work complements the recent work on head patterning in another myriapod, the millipede G. marginata, where otd and a putative btd orthologue were found to be anteriorly expressed [30], and the work on head patterning in the spider P. tepidariorum discussed above. Only little data are available on gene expression during head patterning in crustaceans [31, 32].

The S. maritima genome has been sequenced and extensively annotated [33]. There is a comprehensive modern description of its embryology [34], and some developmental processes, including neurogenesis and trunk segmentation, have already been studied in considerable detail [35,36,37,38]. Research on Strigamia has so far been limited to descriptive work (as it has in all myriapods); nevertheless, the detailed analysis of gene expression, and especially the spatial and temporal relationship of expression in relation to the process of morphogenesis, has provided significant insight into patterning mechanisms and possible hierarchical interactions of the factors involved [37, 38].

We identified in the Strigamia genome candidates for orthologues of buttonhead (SP5), empty spiracles, orthodenticle, collier and cap‘n’collar, and characterised their genomic organisation. We visualised the spatial and temporal dynamics of the expression of these genes and the degree of overlap of their domains and analysed how their expression relates to the morphodynamics of head field formation and to the first expression of the segment polarity gene hedgehog. While we find that all of the investigated factors are anteriorly expressed during early centipede development, we find that the combinatorial pattern of the investigated genes in the centipede is more similar to that seen in the beetle Tribolium than it is to that of Drosophila. Hence, we conclude that some aspects of the head gap gene patterning system have evolved in the fly lineage. We additionally show two rounds of subdivision of an early cephalic hh domain, which is reminiscent of the hh dynamics described in the spider [4], although the precise temporal dynamic is different. We find that the resulting ocular, antennal and intercalary segmental stripes of hh expression co-localise with ocular otx and antennal/intercalary SP5/btd expression, respectively.

Results

The centipede orthologues of insect head patterning genes

The sequenced genome of S. maritima contains orthologues of most known insect head patterning genes. We found one likely buttonhead/SP5 orthologue (among three SP family genes), three orthodenticle/otx genes (otx-A, otx-B and otx-C), one empty spiracles (ems), one collier (col) and two cap‘n’collar (cnc1 and cnc2) genes (see also Additional file 1). In addition, we identified two orthologues of the segment polarity factor hedgehog (hh1 and hh2). Gene models of all these factors can be found at http://metazoa.ensembl.org/Strigamia_maritima/. A list of genes including accession numbers, used primer sequences and probe lengths is provided in Additional file 1: Table S1.

SP5/buttonhead as a marker for early head patterning

The buttonhead (btd) gene is a zinc finger domain transcription factor that was first characterised as a head patterning gene in Drosophila [6]. It is the orthologue of the vertebrate SP5 genes within the SP family of transcription factors (see [39]). The likely orthologous Strigamia gene is here referred to as SP5/btd (see Additional file 1 and Additional file 2: Fig. S1 and Additional file 3: Fig. S2 for the proposed classification of the three centipede SP genes). This gene is a valuable marker for early patterning of the head (see below). There are two other SP family genes in Strigamia, all localised in a single chromosomal cluster with SP5/btd. Orthologies within the SP family are not completely clear, but these other genes likely represent an SP6-9 orthologue and an SP1-4 orthologue (see Additional file 2: Fig. S1).

The Strigamia SP5/btd gene is expressed before the head field becomes apparent in a broad domain at the anterior edge of the forming embryo. Subsequently, as the head field develops, the expression resolves into three domains belonging to the ocular, antennal and intercalary segments (see below). The SP6-9 orthologue is expressed in a similar pattern to SP5/btd in the anterior head (compare Fig. 1, and Additional file 4: Fig. S3 B, C), but first turns on slightly later. Following early head patterning, both of these genes are also expressed in a segmentally re-iterated pattern throughout the trunk (see Fig. 1g and Additional file 4: Fig. S3 D). The third SP family gene, SP1-4 is widely expressed in a near-ubiquitous pattern throughout later embryonic development (Additional file 4: Fig. S3 E–H), a pattern similar to that observed for SP1-4 orthologues in other arthropods [39]. We have not studied the expression of these other family members in detail.

Fig. 1

Expression of SP5/btd in early Strigamia development. All panels show ventral views, except (c), which is a lateral view. a (stage 2.1) A broad ring of SP5/btd expression surrounds the uniform blastoderm. b (stage 2.2, early) SP5/btd expression lies at the transition between the single layered blastoderm in the anterior and denser (multi-layered) nuclei in the posterior. c (stage 2.2, early) Lateral view of similar staged embryo as in (b): the ring of SP5/btd expression has opened dorsally. d (stage 2.2, late) The initial ring of SP5/btd expression has condensed to a domain situated at the posterior of the emerging head field; this domain (green arrowhead) marks the future antennal and intercalary tissue. Bilateral (ocular) expression domains (circled, white arrowheads) emerge anteriorly adjacent to this domain. e (stage 2.3, early) Ocular domains detach from the more posterior domain. A dashed line marks the anterior margin of the condensed head field. f (mid-stage 2.3) The head primordium is now clearly defined; the anterior medial region (double headed arrow) occupies the anterior-most part of the head. Mandibular SP5/btd expression appears posterior to the antennal/intercalary SP5/btd domain, which is still continuous. g (stage 3.1) The antennal and intercalary regions of SP5/btd expression have separated. h–j Enlargements of the ventral parts of embryos stained for SP5/btd and engrailed (en). h (stage 2.2) Early SP5/btd expression lies directly anterior to the first (intercalary) stripe of en expression. i (stage 2.2, late) Intercalary en expression is directly posterior to antennal/intercalary SP5/btd expression; the mandibular en stripe emerges more posteriorly. j (stage 2.3 early) Antennal en expression appears within the main SP5/btd domain. oc ocular, int intercalary, ant antennal, md mandibular

SP5/btd expression and initial formation of the head field at the ventral side

The first visible sign of axial patterning in Strigamia is the transition from a uniform blastoderm to a blastoderm in which the posterior half is characterised by a higher density of nuclei. Expression of SP5/btd starts at this early blastoderm stage (2.1–2.2; all stage references according to [34]), while the cell distribution is still largely uniform. SP5/btd expression appears as a broad ring that spans the whole circumference of the egg (Fig. 1a). The ring is located directly anterior to the early expression of even-skipped-1 (eve1) (Fig. 2e), which is expressed throughout the posterior half of the blastoderm. eve1 expression subsequently resolves into stripes, the most anterior of which is assigned to the posterior mandibular segment (Fig. 2f; and see [37]). Hence, the SP5/btd expressing part of the blastoderm is likely to give rise to segments directly anterior of the mandibular, which are the intercalary and the antennal segment. Note that the mandibular domain of SP5/btd appears de novo (see Fig. 1f), posterior to this initial domain.

Fig. 2

SP5/btd expression in conjunction with otx-B (a–d) and eve1 (e, f). All panels show ventral views. a (stage 2.2, early) An anterior cap of otx-B expression is directly anteriorly adjacent to the early ring-domain of SP5/btd. b (stage 2.3, early) A narrow gap has emerged between the antennal/intercalary SP5/btd expression and the more anterior otx-B expression. c (early stage 2.3) Photograph of only the SP5/btd stain shows ocular patches of SP5/btd expression anteriorly to antennal/intercalary domain. c′ Same specimen as c after otx-B stain, the strong part of the otx expression overlaps with ocular SP5/btd expression. Otx-B expression more anteriorly in the condensing head field (compare Additional file 4: figure S3) is barely visible due to reduced sensitivity of double stainings. d (stage mid-2.3, SP5/btd only) d′ (same specimen as d) Ocular otx-B expression is subdivided into lateral patches and a medial domain, the distance between ocular otx-B and antennal/intercalary SP5/btd expression has increased. e (stage 2.2) Broad expression of eve1 lies posteriorly adjacent to SP5/btd expression. f (stage 2.3) The posterior dynamic expression of eve1 resolves into single stripes of which the first lies directly posterior to the mandibular segmental SP5/btd expression (red arrowhead)

The head rudiment forms as cells in the anterior half of the blastoderm condense ventrally and posteriorly to form a visible thickening [34]. As this condensation begins, the early ring of SP5/btd expression narrows in the A/P axis (Fig. 1b), and opens dorsally (Fig. 1c), becoming restricted to the ventral half of the egg. SP5/btd expression initially marks the transition between the denser nuclei of the forming head field, and the less condensed tissue that lies anteriorly to it (Fig. 1b).

Subsequently, more condensed tissue appears anterior to the primary SP5/btd domain (Fig. 1d). At the same time a bilateral pair of expression patches emerges directly anterior to, and initially connected with, the main stripe of SP5/btd expression (encircled areas in Fig. 1d). A little later these patches separate completely from the more posterior domain (Fig. 1e–g). The anterior-lateral position of these detached SP5/btd expression domains clearly assigns them to the ocular region, while the primary stripe of SP5/btd marks the antennal and intercalary segment (see below). Subsequently, further condensed material appears anterior to the ocular patches of SP5/btd expression. This will become the anterior medial region of the head (see Fig. 1f, double headed arrow). For a characterisation of the anterior medial region in the centipede see [40].

At face value, these observations suggest that the antennal and intercalary territory is the first region of the head to condense, followed by the ocular region and then the anterior medial region. However, this interpretation assumes that SP5/btd expression is a marker for stable populations of cells and does not dynamically shift across cell populations. This remains to be shown.

After the ocular patches have detached, the lateral extent of SP5/btd expression continues to narrow and the distance between the ocular patches and the more posterior stripe increases (Fig. 1d–f).

The major stripe of SP5/btd expression marks the antennal and intercalary segments

Based on its shape and position, and on comparison with expression of the segmental marker gene engrailed (en) (Fig. 1h–j), it is evident that the posterior SP5/btd stripe encompasses the future antennal and anterior intercalary segment. The earliest expression of en belongs to the intercalary segment and appears posteriorly adjacent to the early stripe of SP5/btd expression (compare single stain in Fig. 1b and double SP5/btd- en stain in Fig. 1h). Mandibular en expression appears posteriorly of the first en stripe well outside the early SP5/btd domain (Fig. 1i, j), whereas antennal en expression appears within the domain (Fig. 1j). In this antennal and intercalary region, SP5/btd expression appears before the onset of engrailed expression.

A later phase of SP5/btd and SP6-9 expression follows segment formation

From stage 3 onwards, the expression of both SP5 and SP6-9 appears more posteriorly, in a segmentally re-iterated pattern in the mandibular and more posterior segments (Fig. 1g, Additional file 4: Fig. S3 D). This more posterior expression follows the first appearance of en and other segmental markers in these segments, implying that it is not instructive for segment patterning in this region. We assume that this later expression marks the start of a second phase of segmental SP5/btd expression that might have a tissue specific patterning role in the more posterior segments after their formation.

SP5/btd in Strigamia is continuously expressed in the head rudiment, and its expression makes visible the dynamics of head field formation and subdivision. Therefore, we used SP5/btd as a reference marker to provide precise staging information, and against which to compare the expression of other head genes.

Simultaneous expression of two closely linked centipede otx genes

Analysis of the Strigamia genome revealed the presence of two closely linked orthodenticle (otd/otx) paralogues (see Additional file 3: Fig. S2 C). Expression of otx-A was previously characterised within the anterior head of Strigamia [41, 42]. We found that otx-B is expressed in a very similar pattern to otx-A throughout development, suggesting that the closely linked genes are co-regulated (see Additional file 5: Fig. S4 A, B for single expression of otx-B). We used a probe against otx-B in double stains with SP5/btd and hh. A third otx gene identified in the Strigamia genome, otx-C, did not show any detectable expression pattern during early development.

An emerging gap between ocular otx and antennal/intercalary SP5/btd expression

otx-B is expressed in the ventral anterior part of the early blastoderm, directly anterior to the ring of SP5/btd expression with no or minimal overlap (stage 2.2, Fig. 2a). But even at this early stage, we have seen no embryo in which it spans the whole egg circumference, as SP5/btd expression does. Expression is stronger in the most posterior part that is adjacent to SP5/btd expression than in the more anterior blastoderm, perhaps reflecting cell density.

Only a little later, by early stage 2.3, a gap emerges between the more posterior domain of SP5/btd domain and otx-B expression (Fig. 2b). The strong domain of otx-B expression is now found in the ocular segment where it overlaps with the ocular expression of SP5/btd (Fig. 2c, c′, d, d′). otx-B is also weakly expressed anterior to the ocular domain throughout the condensed head field. This weaker anterior expression is not clearly visible in the double stainings against SP5/btd and otx-B, but is seen in single in situ stainings (see Additional file 4: Fig. S4 A, B for otx-B, and [41, 42] for otx-A).

As the head condenses further the distance between otx expression and antennal/intercalary SP5/btd expression increases (Fig. 2c′, d′). The ocular otx expression domain becomes subdivided into a medial part and two slightly more anterior-lateral patches that overlap with ocular SP5/btd expression (Fig. 2d, d′).

In summary, the otx-A and otx-B genes are broadly expressed at the blastoderm stage before the head field condenses, in the anterior part of the embryo. Expression then becomes restricted to the condensing head field.

empty spiracles is not an early expressed gap-like gene

Expression of the Strigamia empty spiracles gene (ems) was not detected at early blastoderm stage prior to head condensation. First expression was found at early stage 2.3 (Fig. 3a, c) in two small bilateral domains that are located directly anterior to antennal SP5/btd expression (Fig. 3c, d). These first domains of ems expression localise in the emerging gap between ocular otx expression and antennal/intercalary SP5/btd expression. At its posterior edge, ems expression overlaps in a few cells with antennal SP5/btd expression (Fig. 3c, d). With further development, the region of overlap becomes a bit larger but is always restricted to the lateral part of the segment (Fig. 3c, d). ems is then expressed in a re-iterated fashion, marking a part of the ectoderm in every segment (Fig. 3b). The anterior antennal part of this segmentally repeated pattern is continuous with the earliest expression of ems. Notably ems is never expressed in the most anterior head, anterior to the antennal segment.

Fig. 3

Expression of ems (a–c), col (e–i) and cnc2 (j, k) singly (a, b, j, k) or in combination with SP5/btd (c–i). All panels show ventral views. a′, b′, j′, k′—nuclear stain of the same embryos a (stage 2.3, early) First expression of ems in a pair of bilateral domains. b (stage 4.2) Late segmental expression of ems. c (stage 2.3, early) ems expression lies anteriorly adjacent to the antennal part of the SP5/btd expression, hence located in the anterior antennal segment. d (stage 2.3, late) Overlap of ems and SP5/btd in the lateral part of the antennal domain (double headed arrow). e (stage 2.2, late), enlarged in f. First col expression (dark shading of nuclear stain by the BM-purple substrate) overlaps with the early expression of SP5/btd but is located a few cell widths more posterior. g (stage 2.3, early) col expression overlaps the posterior of SP5/btd expression, as SP5/btd becomes restricted to the condensing head field. h (stage mid-2.3) The col domain has narrowed down and is posteriorly included in the antennal/intercalary SP5/btd domain. i (stage 2.3, late) The col domain has split into an antennal and an intercalary stripe. j (stage 3.2) First appearance of cnc2 expression in a stomodaeal domain and within the mandibular segment. k (stage 4.2) Late segmental and stomodaeal expression of cnc2. st = developing stomodaeum, md = mandibular segment

In summary, ems is expressed in the anterior head, but its late appearance and the segmentally repeated pattern does not suggest a specific, instructive function in the early specification of the head.

collier is expressed early overlapping with SP5 and then becomes restricted to the anterior intercalary segment

Crozatier et al. [22] characterised collier (col) as a second level mediator between insect head gap genes and segment polarity genes, with a role in the intercalary and anterior mandibular segment of insects. In Strigamia expression of col is first detected at the earliest stage of head condensation (Fig. 3e, f), when the expression of SP5/btd covers about 50% of the egg circumference, and before the ocular patches of SP5/btd expression have emerged. It is expressed as a sharply defined stripe; overlapping with the SP5/btd expression domain but shifted a few cell widths posteriorly (Fig. 3e, f). In the same way as SP5/btd, its expression becomes restricted to the head field as head condensation progresses (Fig. 3g). At the earliest stages of head condensation, the posterior limit of col expression lies posterior to the limit of SP5/btd expression, but a little later the col expression domain lies completely within the SP5/btd domain (Fig. 3h). col expression only spans the posterior part of this SP5/btd expression, in a domain that becomes the anterior intercalary and most posterior part of the antennal segment (Fig. 3h, i).

Subdivision of the col domain

By late stage 2.3, the col domain becomes subdivided into two stripes, one anterior and one posterior, connected laterally but separated medially by non-expressing cells (Fig. 3i). The more anterior col stripe lies at the posterior edge of antennal SP5/btd expression, and the posterior col stripe is located within the intercalary segmental region (Fig. 3i). Comparing both col and en to SP5/btd expression suggests that the anterior col stripe is coincident with antennal en, which marks the posterior compartment of the segment (see Fig. 1i). Finding col expression within this antennal region is surprising because no antennal col expression has been reported in insects or in the millipede Glomeris (see [22, 25,26,27]).

Fading of the antennal/intercalary col expression after head field condensation

Expression of col in this antennal/intercalary domain is transient. It fades rapidly after head condensation and transcripts have disappeared completely by early segmentation stages (stage 3.1, 3.2). Just after disappearance from this region, it will re-appear de novo in the anterior medial head region where it is a marker for neural cells [40]. We consider these two expression domains of col to be completely independent from one another. Additionally, at late stages of development col is expressed in a segmentally iterated pattern in the neuroectoderm and in a subset of mesodermal cells (not shown). The specific transient expression prior to the formation of the intercalary and antennal segments suggests a role of col in the formation or delimitation of these segments in the centipede.

cap‘n’collar2 is expressed specifically in the mandibular segment

Strigamia cnc2 is expressed specifically around the stomodaeum and more posteriorly within the mandibular segment, whereas cnc1 is ubiquitously expressed in the embryo (see Additional file 1 and Additional file 4: Fig. S4 C, D and [33] for Strigamia cnc genes). The expression of cnc2 in the mandibular segment is only clearly detectable at stage 3.2 (Fig. 3j, an early trunk segmentation stage, [34]) and subsequent stages (Fig. 3k). In Drosophila the anterior mandibular compartment of cnc expression overlaps and is positively regulated by col expression [24]. In Strigamia col does not extend into the mandibular territory at all. Its posterior border lies within the intercalary segment (see above). Also, cnc2 is only expressed after col RNA has completely disappeared from its segmental expression domain. Hence, col expression and cnc2 expression never overlap in the same cells and a direct regulatory action of col on cnc2 (as found in the fly) can be excluded.

Emergence of the cephalic hh domain and first split

Two copies of the hedgehog (hh) gene are present in the Strigamia genome (hh1 and hh2) [33]. They are not linked but show a high degree of sequence similarity. RNA-probes generated against the two orthologues showed the same pattern. Data presented here was generated using a probe against hh1.

We have not been able to examine hh expression in the earliest blastoderm stage, but by early stage 2.2 hh1 is expressed in one broad stripe covering about 50% of the egg circumference (Fig. 4a, b, stage comparable to Fig. 1b). This domain is in approximately the same position as the early stripe of SP5/btd expression, which later splits into ocular, antennal and intercalary expression domains.

Fig. 4

Expression of hh1 (a–g) and hh1 expression combined with otx-B (h). a, c–g ventral view. b lateral view. a (stage 2.2, early) First expression of hh1 appears. b (same specimen as a) The early hh1 domain covers about 50% of the egg circumference. c (stage 2.2, late) The 1st split of the early hh domain has begun laterally; this produces the ocular hh stripe. d (stage 2.2, late) The splitting-off of the ocular stripe has reached the medial domain. The mandibular stripe is faintly visible; it appears de novo, the lateral ends connecting to the initial, more anterior domain. e (stage 2.3, early) The ocular stripe has completely separated. The more posterior domain has begun to split into antennal and intercalary stripes. f (stage 2.3, early) Antennal and intercalary domains have split centrally but stay attached laterally. g (stage 2.3, early) First appearance of the 1st maxillary stripe. h/h′ (stage 2.2, late, showing the lateral part of the head domain). h otx-B channel only. h′ otx-B expression overlaps with the ocular hh stripe. Dashed line is in same place in h and h′ and marks the posterior limit of both the otx-B and the ocular hh domain. oc = ocular domain, ant/ic = antennal/intercalary domain

From this single early hh domain, one narrow stripe splits off anteriorly, first laterally, and then medially (see Fig. 4c–e). It becomes the ocular hh stripe. Double stainings with otx show that the gap between the ocular hh stripe and the more posterior stripe of expression is coincident, or very nearly so, with the gap that appears between the early expression domains of SP5/btd and the otx genes (Fig. 2b, c′).

De novo appearance of mandibular hh expression

The more posterior part of the hh early expression domain initially remains as a single stripe in the antennal/intercalary region. Posterior to it mandibular expression appears (Fig. 4c). This mandibular expression most likely appears de novo. No ‘splitting-off’ event from the antennal/intercalary domain could be captured in the close series of developmental stages that were stained for hh expression.

Second split of the cephalic hh domain

Shortly after the appearance of mandibular hh expression, the antennal/intercalary hh domain splits into separate antennal and intercalary stripes (Fig. 4c, d). Antennal, intercalary and mandibular stripes of expression remain connected at their lateral ends (Fig. 4d–f), a pattern that is also seen for en expression (Fig. 1i). A similar split in the expression domains is also observed in the pattern of SP5/btd (see above), but this splitting is only completed at a slightly later stage in development (stage 3.1, Fig. 2f), possibly because of the greater A/P extent of the SP5/btd domains within each segment, which makes them directly adjacent or overlapping for longer.

Summary of results

Among the studied factors otx-B and SP5/btd are expressed first, followed by hh, col, en, ems and cnc2. All early expression patterns reflect the head condensation process, whereby expression domains condense from a broad region of the early blastoderm to the ventral head field. Much of this movement presumably reflects the ventral, and to a limited extent posterior migration of cells in the anterior half of the egg to form the multi-layered head primordium. Condensation of the head territory does not follow a strict anterior to posterior progression: the region of the intercalary and antennal segment appears to condense first, followed by the ocular and then the most anterior medial region.

As the head condenses, initially abutting or overlapping domains of gene expression become more resolved, and some expression domains split, reflecting the formation of defined segments. These changes are summarised in Fig. 5. The upper diagrams illustrate the dynamics of early expression of head gap gene orthologues and col, whereas the lower diagrams show hh expression at the same phases of development.

Fig. 5

Schematic drawing of combined head gap gene expression (upper panel) in comparison with the dynamics of hh1 expression (lower panel). The boundary between embryo rudiment and dorsal field, marked by a dashed line, is a gradual transition at the early stages, but becomes more defined during later development (compare Fig. 1). df = dorsal field (putatively ‘extra-embryonic’, characterised by lower cell density than the ventral part where the embryo forms) oc = ocular domain, ant/int = combined antennal and intercalary domain, ant = antennal domain, int = intercalary domain. Black arrows in b indicate split between the early otx-B and SP5 domains, red arrows in b and c indicate splits of the hh domain

Discussion

Head morphogenesis: shared characters in myriapods, spiders and sequentially segmenting insects

Symmetry breaking transitions from circumferential to ventrally restricted head gene expression occur in spiders and myriapods.

At the earliest stages of development in the myriapods Glomeris and Strigamia, early gene expression is often ring-like around the whole blastoderm, as we found for SP5/btd expression. As the embryonic primordium forms at the ventral side, these expression patterns become restricted to the embryo rudiment ([30, 42, 43] and this work).

Similarly, the early germ disc of the spider P. tepidariorum is radially symmetric. Anterior markers in the spider, such as otd-1 and hh, are expressed in a complete ring around the early, radially symmetric germ disc [3], which is similar to the early hemispherical and ring-like expression of SP5/btd in Strigamia. After the radial symmetry of the germ disc in the spider is broken, the rings of expression open dorsally and otd-1 and hh become restricted to stripes at the ventral side [3]. A similar process of ring opening is observed for SP5/btd in Strigamia. We are not sure whether otx and hh are also initially expressed in a complete ring—in the earliest embryos we have been able to stain, they cover only about 50% of the egg circumference, but we have successfully prepared very few embryos at these earliest stages, which are extremely difficult to handle.

In summary, a transition from circumferential gene expression to a ventral restriction is observed in both spiders and in myriapods. In spiders the migration of a cell cluster from the germ band centre to the rim, accompanied by decapentaplegic (dpp) signalling from these migrating cells, leads to a break of symmetry in the germ disc [44]. The trigger of the symmetry break in myriapods has so far not been identified, though a very early focus of dpp expressing cells has been observed in Strigamia [42].

The head rudiment only covers a ventral part of the anterior embryo in the centipede and in short-germ insects

The condensed head of Strigamia only covers a relatively small ventral part of the egg. Likewise in short-germ insects like the beetle Tribolium the head primordium does not cover the whole anterior part of the blastoderm but forms at the ventral side of the embryo, distant from the anterior pole of the egg, while dorsal and most anterior tissues are extra-embryonic (see [45, 46]). By contrast, in the long-germ insect Drosophila, the whole anterior cap of the embryo contributes to the head and only a small dorsal part of the anterior blastoderm forms amnioserosa [47]. Therefore, with respect to the ventral condensation process and the size of the head primordium relative to the egg-surface, head formation in Strigamia rather resembles the process in the sequentially segmenting insect Tribolium [20, 46] than head formation in the fly (see for example [1, 47]).

Long-germ development in insects is thought to have evolved multiple times independently from an ancestral short-germ type of development [48, 49], which is similar to the developmental mode in Strigamia where trunk segments are added sequentially from a posterior growth zone [34, 36, 50]. Hence, it is not surprising that also head formation in Strigamia is more similar to Tribolium than to Drosophila. Kittelmann et al. [46] suggested that the anterior displacement of the head anlagen, accompanied by the invention of the bcd dependent anterior patterning system, allowed for the evolution of long germ development in Drosophila.

The expression of head gap gene orthologues in Strigamia is more similar to that in the beetle Tribolium than either is to Drosophila

The head gap genes of Drosophila are characterised by staggered but extensively overlapping expression domains along the head anterior–posterior-axis: otd is expressed most anteriorly, followed by ems and then btd [6, 7, 10, 51].

In the beetle Tribolium, the relative sequence of otd, ems and btd expression is the same as in the fly, but the expression domains are more restricted and do not overlap to the large extent seen in Drosophila [20]. In the early Tribolium blastoderm, otd-1 and btd are expressed in broad domains that are directly adjacent to one another: otd-1 is expressed in two bilateral domains in the head lobes and btd is expressed directly posterior to it in a broad stripe. ems is not expressed at this early stage. Shortly thereafter, the otd and btd expression domains become separated by a gap, and within this gap the expression of ems emerges [20].

In the centipede Strigamia, the dynamics of early head gap gene expression are very similar to those in Tribolium: at the early blastoderm stage otx is expressed in the anterior head and SP5/btd is expressed directly posterior to it. At the interface, a few cells express both otx/otd and SP5/btd in Strigamia, as well as in the beetle [20]. With condensation of the head field, otx/otd and SP5/btd expression become separated in Strigamia as they do in Tribolium, and in both species ems expression emerges between them in two lateral domains (compare schematic drawing in Fig. 5 and [20]).

During subsequent development, the resolution of these domains to specific segments differs in Tribolium and Strigamia. Whereas in Strigamia the early SP5/btd domain divides into ocular patches and an antennal/intercalary domain, thus marking all cephalic segments, the early SP5/btd domain in Tribolium shrinks to become positioned in the mandibular segment [20]. In addition, early ems expression in Strigamia overlaps in its posterior part with antennal SP5/btd expression, but there is no overlap of SP5/btd and ems in Tribolium. Hence, the later expression of SP5/btd is positioned relatively more posterior in Tribolium than in Strigamia, although in both organisms the expression domain is directly posterior to otd at early development, and in both, otd/otx marks the ocular segment during later development.

In Strigamia, we never see the large areas of overlap that characterise early head gap gene expression in Drosophila. In Drosophila, otd is expressed in a domain that covers the ocular and the whole antennal segment (see [1]), whereas in Strigamia the posterior limit of otx expression is the ocular segment. The early head stripe of btd expression in Drosophila covers the antennal, intercalary and mandibular segment [1] but in Strigamia the posterior limit of early expression of SP5/btd is more anterior, in the intercalary segment. In Drosophila, ems is expressed in a large domain covering the antennal and intercalary segment [1]. In Strigamia, the earliest expression of ems only covers a small territory in the anterior antennal segment. This expression initiates late and seems to be part of a pattern of metameric expression that appears in every following segment in anterior to posterior progression.

In conclusion, there are similarities in head gap gene expression between Drosophila, Tribolium and Strigamia, in the AP-order of expression domains, in the early expression of otx and SP5/btd, and to some extent in the dynamics of SP5/btd pattern resolution. However, the overlap of early gap gene expression domains, and the extensive early expression of ems in Drosophila head patterning, are not found in the centipede, or in Tribolium.

Implications of gap gene expression in Strigamia

The Drosophila head gap genes are crucial for correct patterning of head segments: loss of any of these three genes results in the deletion of several contiguous segments in Tribolium, only otd has a severe head phenotype when knocked down via RNAi: extreme phenotypes delete all head and gnathal segments, and may also affect the thorax [19, 20]. In contrast, RNAi mediated knockdown of btd (SP5) did not lead to any head phenotype, and the knockdown of ems only affected the head cuticle of the antennal segment: in the strongest ems RNAi phenotypes, the ocular and antennal wingless (wg) stripes were merged. These limited phenotypes may reflect limitations of the RNAi gene knockdown, but that seems unlikely, given the general efficacy of RNAi in Tribolium.

The lack of an SP5/btd phenotype in Tribolium is an unexpected finding considering the deep conservation of early head specific SP5/btd expression in myriapods and insects. It is hard to believe that the highly conserved domain of anterior SP5/btd expression lacks function in most systems other than the fly. A re-investigation of SP5/btd function in other arthropod models, and possible functional redundancy with the SP6-9 orthologues, the expression of which overlaps with SP5/btd in the beetle [39], as it does in the centipede, might shed light on the evolution of the functional role of these SP genes.

Tribolium otd also has a characterised role in dorsal–ventral patterning, which by itself might explain the anterior defects after otd knockdown [45]. However, given the widely conserved early anterior expression of otx genes across animals [41, 52], it seems likely that otx genes in the centipede function in the correct establishment of the anterior head, and specifically in the ocular region.

Conclusions: shared patterns of head gap gene domains in beetle and centipede and their alterations in the fly

The striking similarities between Tribolium and Strigamia in the early otd-ems-btd pattern suggest that this may be an ancestral aspect of arthropod patterning. The observation that broad gap gene-like expression of ems is found neither in the centipede nor in the beetle suggests that ems has only been recruited to a major role in regional anterior patterning in the lineage leading to the fly. This view is supported by functional data from Tribolium, where ems knockdown only leads to subtle effects on antennal development [20].

The use of extensive, overlapping domains of head gap gene expression to pattern anterior segments can hence be interpreted as an invention of the Diptera, where it is associated with the long germ type of development, and might contribute to the fast development of the fly. The long-germ development by itself is an example where the fly has evolved a divergent mechanism that facilitates fast development: each trunk segment is specified by a unique combinatorial input of gap genes [53], whereas in other arthropods segments are added sequentially from a posterior growth zone (e.g. [36, 54,55,56]. A classical gap gene patterning system, where early expressed genes define broad regions of the embryo and directly or indirectly control segmentation genes has, however, not been described outside the insects [49].

col and cnc patterning

The gene network surrounding the intercalary segment, where col acts at an intermediate level between the early expressed head gap genes and segment polarity gene expression, is probably the best understood part of the Drosophila head segmentation mechanism [22, 27, 57]. We compared this gene network to the expression data from Strigamia.

An anterior shift of centipede col expression by half a segment in comparison with insects

Early segmental col expression in the insects, and also in the millipede Glomeris, is mainly localised within the intercalary segment [25,26,27]. In both Drosophila and Tribolium, the posterior limit of expression extends into the mandibular segment and the anterior limit in all species previously examined lies within the intercalary segment [22, 25, 26, 58]. In the fly col expression is also dynamic: it is first expressed more widely in the anterior mandibular and posterior intercalary segment and then becomes restricted to the posterior intercalary segment only. The late col domain in Drosophila includes the intercalary spots of en expression and also tissue posterior to that, but does not overlap with antennal en expression [22]. Hence, col expression is about half a segment more posterior in Drosophila than it is in Strigamia, where the anterior limit is in the antennal segment. Irrespective of the slightly different position of the expression domains, in both the fly and in the centipede the coherent domain later splits into two stripes that are connected at the lateral ends [22].

In the millipede, the anterior limit of col expression is also clearly distant from antennal en expression. The precise posterior extent of the expression remains unclear; however, as in the insects expression does not extend anteriorly into the antennal segment [25]. The work by Janssen et al. [25] also shows col expression in the centipede Lithobius forficatus. Unfortunately, these data are presented without a segmental marker as counterstain, so it is not possible to locate the col domain precisely, or to assess whether col expression is shifted anteriorly in this other centipede, as it is in Strigamia.

col as a putative mediator between head gap gene expression and segment polarity gene expression

In Drosophila col has been characterised as a mediator through which the head gap gene btd activates hh and en expression within the intercalary segment [22]. Here, the col domain lies completely within the btd domain, and in btd mutant embryos col expression is completely lost [27]. In Strigamia early expression of SP5/btd and col co-localise to a large degree and a regulatory input of SP5/btd on col seems possible. Strigamia col may also be involved in activating segment polarity gene expression within its intercalary expression domain, as does col in Drosophila [22]. A regulatory input of ems on col, as found in the fly [22] is, however, not possible in Strigamia as the two factors are not co-expressed.

In the posterior part of the Drosophila blastoderm col is repressed by eve; in an eve mutant embryo an additional col stripe appears posterior to the intercalary one [22]. A similar down-regulation of eve1 by col is possible in Strigamia: during early blastoderm eve1 characterises the tissue that will give rise to the mandibular and posteriorly following segments, an area where col is not expressed.

Separation of the cnc2 and col expression domains in Strigamia

cnc orthologues are expressed within the labrum and the mandibular segment of myriapods and pancrustaceans [30, 32, 58]. In Drosophila, cnc in the mandibular segment depends on the early head domain of btd [59]. By contrast, the early head expression of Strigamia SP5/btd does not co-localise with mandibular cnc2 expression. The later segmental expression of SP5/btd, which includes the mandibular segment, is unlikely to activate cnc2 because SP5/btd is restricted to the central part of the segment, whereas cnc2 fills the complete segmental area.

cnc expression in the fly also depends on col in the posterior intercalary segment [22]. By contrast, Strigamia col expression is not only spatially but also temporally separated from cnc2 expression. Hence, a direct activation of cnc2 by col can be excluded, and in contrast to Drosophila, cnc2 activation in Strigamia is independent of col and also of SP5/btd function. Interestingly, also in the beetle, cnc expression in the mandibular segment is not altered in col knockdown embryos [26].

In conclusion, some of the dynamics of the combined head gap gene, col, cnc2 and segmentation gene expressions in Drosophila are similar to those in Strigamia, which suggests a partial conservation of gene interactions between insects and myriapods. But there are also clear differences, as for example the anterior shift of col expression in Strigamia and the late expression of cnc2, separated from the SP5/btd and col expression domains.

hedgehog domain splitting

Gene expression patterns during early segmentation of the head field of Strigamia show that the ocular, antennal and intercalary segments derive from a shared domain, which is specified at the blastoderm stage. This is strikingly shown by the pattern of early hedgehog (hh) expression: one initial domain of expression splits twice to give rise to the ocular, antennal and intercalary hh stripes. At about the same time, or shortly afterwards, a single engrailed (en) stripe splits to give rise to the antennal, intercalary and mandibular en stripes, presumably as a downstream consequence of the earlier hh patterning. In the only other myriapod where head segmentation has been examined, the millipede Glomeris, the ocular and antennal stripe of hh expression also split from a shared domain. The intercalary stripe by contrast is described as appearing de novo after some delay [29].

The dynamic of hh stripe splitting in the spider P. tepidariorum is slightly different. An early broad domain of hh expression in the anterior part of the germ band undergoes two splitting events, as in Strigamia, but the first split gives rise to a posterior stripe, which becomes the pedipalpal hh stripe, and an anterior stripe that undergoes a second round of splitting to give rise to the most anterior (cephalic) stripe of hh expression and to the cheliceral stripe posterior to it [3, 4].

Current understanding of spider head segments, based on hox gene expression, homologises the spider cheliceral segment with the insect (and myriapod) antennal segment, and the pedipalpal with the intercalary segment [60, 61]. Assuming that the cephalic hh domain corresponds to ocular hh expression in myriapods, the early hh expression domain comprises the same three respective segments. However, the order of stripe splitting differs between the spider and the centipede: in the spider it is the anterior of the first two stripes that splits again [4], whereas in Strigamia it is the posterior stripe that undergoes a second split.

The subdivision of an early hh domain into three cephalic segmental stripes has also been described in Drosophila. In the fly, a wedge-shaped domain of hh expression is established at blastoderm stage in the cephalic region of the embryo. During gastrulation and germ band extension, this domain subdivides into an intercalary, an antennal and a procephalic (ocular) stripe [28]. Unfortunately, this early observation has not been followed up by more recent studies with current techniques. However, the fact that hh stripe splitting occurs in centipedes, millipedes, spiders and potentially also in insects suggests that this process may be a basic principle of head segmentation in arthropods.

In the spider, Kanayama et al. [4] have shown that the initial domain of hh expression is dynamic with respect to the cell population. It moves from the most anterior rim of the germ disc to a more posterior position, where it then splits into the cephalic segmental stripes. The initial domain overlaps with the posterior edge of otd expression, which is similarly dynamic. RNAi mediated knockdown of spider otd caused hh expression to remain static at the germ band rim. Cell-specific knockdown of otd after the hh travelling phase led to an interruption of the stripe splitting events through loss of hh in the otd negative cells. Hence, otd in the spider is required to maintain hh expression during both the travelling and the splitting phase of early hh expression [4].

We have no evidence for expression of hh that travels over cells before the onset of splitting during early Strigamia development, although such a process cannot be excluded based on the methods used here. As in the spider, at the onset of its subdivision, the initial hh domain is located some way from the anterior rim of the condensed head field. We showed that Strigamia otx gene expression is overlapping with the anterior part of this broad initial hh domain. The most posterior otx expression is then found in the ocular segment, where it overlaps with the most anterior stripe of hh. The more posterior, antennal/intercalary hh domain, which splits again, does not overlap with otd. Thus, in Strigamia, otx expression might be required for maintenance of hh in its ocular domain, but it is certainly not required for maintaining hh expression during the second stripe splitting event.

Conclusions

The head patterning system in the centipede, as inferred from our analysis of gene expression, reveals some remarkable features.

A set of gap gene orthologous genes is expressed during early head patterning in Strigamia, Tribolium and Drosophila. They are expressed in the same relative anterior/posterior order, but expression domains in Drosophila are broader, with larger areas of overlap [1], which we interpret as part of a patterning mechanism that is specific to the fly lineage. Greater similarities exist between the early dynamics of gene expression in Strigamia and the sequentially segmenting insect Tribolium [20].

The downstream network in the intercalary and the mandibular segment involves at least some of the same factors in insects and myriapods. Our analysis of the spatiotemporal dynamics of gene expression suggests that some but not all gene interactions are conserved. This raises a question as to head patterning in crustaceans, as current phylogenies place insects as an in-group of crustaceans [62,63,64]. Crustacean head segments differ from those of insects and myriapods as they have a second antennal segment instead of an appendage-less intercalary segment. col-dependent patterning seems to be absent from this second antennal segment [26], whereas expression of cnc in the mandibular segment of crustaceans is conserved [32].

Finally, a mode of head segmentation involving the splitting of hh domains, which appears quite different from segmentation in the trunk, may be a basic principle of arthropod head development. Future studies should address the molecular mechanism of this segmentation mode in different species, and its interdependence with tissue mechanics.

Methods

Embryo collection and fixation

Strigamia maritima embryos were collected from a wild population near Brora, Scotland [50]. The material was fixed for several days in 4% formaldehyde/0.5 × PBS and then transferred to methanol for storage at – 20 °C. Embryos were staged according to morphological features (according to [34]).

Genetic resources and gene classification

Genomic resources for S. maritima are available at http://www.ncbi.nlm.nih.gov/assembly/322118/ and a manually annotated version of the genome is available at http://metazoa.ensembl.org/Strigamia_maritima/Info/Index.

Gene orthologues were identified by Blast searches and reciprocal testing. A list of the identified genes with their Ensembl gene IDs is given in Additional file 1: Table S1. The phylogenetic tree for the classification of SP genes (Additional file 2: Fig. S1 and Additional file 3: Fig. S2) was created using Phylemon2 [65]. Multiple sequence alignments were performed on protein sequences using MUSCLE [66], and the gene tree was built based on maximum likelihood analysis in PhyML [67]. Tree calculation parameters are given in the legend of Additional file 2: Fig. S1.

Gene amplification and RNA in situ staining of gene expression

Specific primers were designed against the identified gene sequences. Products were amplified by standard PCR and subsequently cloned into the pGEM-T-Easy vector system (Promega). Inserts were verified by sequencing and then used as templates for in situ probe synthesis. Single and double colorimetric in situs using a BM-purple and a fast red staining reaction were performed as described in [35, 50]. All embryos were counterstained with the nuclear dye Hoechst (H34580 2 µg/ml) or Sytox green (1 µM).

Image acquisition

After the staining procedure, specimens were immersed in 90% Glycerol for microscopy. Whole mount in situ stained embryos were photographed using a Leica MZFLIII stereomicroscope with attached DFC500 camera or a Zeiss Axiophot compound microscope with attached Leica DFC300FX camera. Picture stacks with different focal planes were taken of the spherical shaped embryos and then reconstructed to images of whole embryos using the software Helicon Focus (Helicon soft Ltd.). In addition, photographs of the nuclear stain were always taken from the same angle as a reference for stage and morphology of the embryo. For the overlay of fluorescence pictures (Hoechst, Fast Red) with bright-field pictures of the BM-purple stain an RGB picture was created using Adobe Photoshop. The Fast Red stain was assigned to the red channel, the Hoechst stain to the blue channel and the BM-purple stain to the green channel. The latter was colour inverted to change from a dark stain on a white background to a light stain on dark background, which then allowed creating the overlay with the two other channels. Some of the embryos were imaged using a Leica Sp5 confocal microscope using the 543 mm laser line for fast red visualisation and the 405 mm diode for the Hoechst stain. Brightness and contrast of whole images were adjusted using Photoshop CS5.