3.1 Hedgehog (hh) expression in the head of a harvestman and its bearing on the NHH

Early during head development, there is a clear two-fold splitting of an initially broad stripe of hh-expression in the spider Parasteatoda that leads to three stripes of expression in the three pGSs [18, 23]. In the light of the proposed NHH, this could be taken as evidence, since it may beautifully reflect the proposed splitting of an ancestral anterior segment into the three PGS in a process following Haeckel’s recapitulation theory [19]. Although the authors that support the NHH and discuss hh-splitting in this context never explicitly claimed this, but instead carefully suggest that any form of anterior hh-splitting could represent remnants of such a process [14], double-splitting, especially if identified as ancestral and unique for the three PGS would indeed provide strong evidence for the NHH.

As said, splitting of the most anterior stripe of hh is not restricted to Parasteatoda, but is also seen in other arthropods, including other chelicerates. Prior to this study, the only data on the early expression of hh during head development in another chelicerate came from the mite Archegozetes longisetosus [24]. The authors of this paper, however, never stated that there is any observed splitting, although they could also not exclude it. Instead they suggested that the anterior hh-expressing segments form rather independently from one another, and that the only possible evidence of stripe-splitting is the reported dorsal/lateral connection of the ocular and cheliceral stripes of hh at some point early during head development [24]. Although there may thus be a splitting of hh, it appears very unlikely judging from the available data that there are two splitting events corresponding to the three PGS in the mite [24].

In the harvestman Phalangium opilio, there is no expression of hh at the early blastoderm stage (stage 4.0) (Fig. 1A), but soon after at the late blastoderm stage (stage 4.1), hh appears in the form of a broad anterior stripe (Fig. 1B) (staging modified after [25]). Soon after that, at stage 4.2, intensity of this broad initial stripe increases (Fig. 1C). The stripe forms an almost continuous circle but is open at the dorsal side of the developing embryo (Fig. 1D/E). At stage 5.0–and I never observed any intermediate stages–several stripes of hh appear simultaneously that correspond to the pedipalpal and the four prosomal leg-bearing segments (L1-L4) (Fig. 1F). These stripes never reach the intensity and breadth of the initial most anterior stripe (Fig. 1F). At the following stages 5.1 and 5.2, the anterior stripe begins to split into the ocular and cheliceral stripes (Fig. 1G–I), and at stages 6.0 and 6.1 the distance between the two stripes further increases (Fig. 1J–M). A lateral view on corresponding embryos reveals that the two stripes that result from this splitting remain dorsally/laterally connected for a while (Fig. 1I,K,M) until they finally become disconnected at stage 7 (Fig. 1N–P). During this process, additional expression of hh appears in the anlagen of the stomodaeum, and additional stripes of hh appear that correspond to the developing opisthosomal segments (Fig. 1L–Q).

Fig. 1

Early expression of hh in the harvestman Phalangium opilio. In all panels, anterior is to the left, except panel D (on top view). Panels A−C, F−H, J, L, and O represent ventral views. Panel E represents a dorsal view of the embryo shown in D. Panels I, K, M, N, P, and Q represent lateral views (dorsal up). In all panels, blue asterisks mark the most anterior stripe of hh-expression that corresponds to the ocular and cheliceral segments; note that this stripe splits at later developmental stages. Panels A´–C´ represent SYBR green staining of corresponding embryos. Developmental stages are indicated (modified after [25]). Roman Numerals 1–4 leg-bearing segments 1–4; ch cheliceral segment, O1–O3 first to third opisthosomal segment, oc ocular segment, pp pedipalpal segment, s, anlage of the stomodaeum

The newly provided data from the harvestman thus show a splitting of the most anterior hh-stripe into two stripes corresponding to the ocular and cheliceral segments (as is also likely the case in the mite). Despite my best attempts, and more than 100 embryos that I investigated and that cover these particular developmental stages, I never observed any other splitting events suggesting that a single splitting into ocular and cheliceral hh-stripes may be the ground state for chelicerates. Under that assumption the splitting in Parasteatoda, or spiders in general, could represent a derived state. I address this question in the following section of the paper.

3.2 Is two-fold splitting of the anterior hh-stripe conserved in spiders?

Spiders possess at least two orthologs of hh of which only Group-1 hh-genes (hh1) are expressed in the form of SPGs [21]. In the following, I will simply refer to hh when referring to spider hh1 orthologs. Expression of hh2 orthologs is not relevant for the topic of this paper (cf. [21]).

In the common house spider Parasteatoda, the first splitting of hh corresponding to the separation of the pedipalpal from the ocular/cheliceral hh-stripe occurs approximately at the same time as the first opisthosomal hh-stripe forms; at this point, also the hh stripe corresponding to the fourth leg-bearing segment is present (but not the first to third) [18]. Appearance and splitting of this first stripe is fast and only takes about four hours. Approximately 4–6 h later, the ocular/cheliceral hh-stripe separates into two separate hh-stripes, representing the second splitting event [18]. At this time, all four hh-stripes in the leg-bearing segments are present, as well as the first two opisthosomal hh-stripes [18]. The developmental stages covered during this time are stage 6 and stage 7 (according to [26]). The same staging system is used for the other here investigated spiders.

In the mygalomorph spider Acanthoscurria, there is no detectable expression of hh at the germ disc stage (stage 5) (Fig. 2A). Shortly later, however, when the dorsal field forms (early stage 6), a strong anterior stripe appears (marked by a blue asterisk in Fig. 2B) together with a number of very weak stripes (Fig. 2B, black asterisks) and expression in the segment addition zone (Fig. 2B). Following the fate of the anterior stripe at subsequent developmental stages reveals that it corresponds to the ocular and cheliceral regions of the head (Fig. 2B-L). The faint stripes correspond to the pedipalpal, the first leg-bearing segment, the second leg-bearing segment, and the fourth leg-bearing segment; note, that expression of hh in the third leg-bearing segment is delayed (Fig. 2B-G). The most anterior stripe then broadens and begins to split (Fig. 2D-H). While splitting ventrally, the two resulting stripes remain dorsally/laterally connected until approximately stage 8 (Fig. 2G, J-L); note that this resembles expression of hh after splitting in the spider Parasteatoda [18], the harvestman (cf. Fig. 1I,K,M) and (likely also) the mite [24]. Apart from the observed splitting of the stripe corresponding to the ocular and cheliceral segments, I could not identify any further splitting of hh-stripes in Acanthoscurria.

Fig. 2

Early expression of hh1 in the tarantula Acanthoscurria geniculata. In all panels, anterior is to the left. Ventral views, except panels E, G, and J that represent lateral views. In all panels, blue asterisks mark the most anterior stripe of hh-expression that corresponds to the ocular and cheliceral segments; note that this stripe splits at later developmental stages. Black asterisks mark stripes of hh-expression posterior to the ocular/cheliceral stripe. The inlay in panel H shows a slightly different angle (as shown in panel H) of the connected ocular and cheliceral stripes of hh-expression. Panels A´–F´, I´, and J´ represent SYBR green staining of corresponding embryos. Developmental stages are indicated (after [26]). Roman Numerals 1–4 leg-bearing segments 1–4, O1 and O2, first and second opisthosomal segment; pp, pedipalpal segment saz, segment addition zone

The earliest expression of hh in Pholcus is located close to the anterior rim of the late germ disc (Fig. 3A). At this stage, the cumulus is on its way to the anterior rim of the disc to determine the future dorsal side of the embryo (cf. [27]). Approximately two hours later, the cumulus has reached the anterior rim of the disc, and a second stripe of hh appears in some distance to the aforementioned anterior stripe (Fig. 3B). Expression of hh is also present in the secondary thickening that marks the future posterior pole of the embryo, the centre of the disc (Fig. 3B/C). Shortly after, a third hh stripe appears between the two aforementioned stripes (Fig. 3D); note that this stripe may be dorsally/laterally connected with the other posterior stripe (Fig. 3D, yellow asterisk in brackets). A few hours later, three additional stripes appear which, based on their position in the early germ band, likely represent the pedipalpal and the first and fourth prosomal leg-bearing segments (Fig. 3E). A lateral view and a dorsal view on the same embryo as shown in panel E reveals that the stripes corresponding to the second and the third leg-bearing segment indeed are dorsally/ventrally connected (Fig. 3F/G, yellow asterisks). Still, at this point the first and most anteriorly located stripe is somewhat broader and stronger than the other stripes (Fig. 3F). At the next stage, and as the germ band extends, the distance between the stripes increases. Most importantly, however, the most anterior stripe appears to have split (Fig. 3H), which is most obvious from the fact that the two resulting stripes are dorsally/ventrally connected (Fig. 3I/J), and that the resulting stripes are thinner and less strong compared to the earlier broad/strong anterior stripe (Fig. 3; cf panels E/F with panels H–J). Failure to identify intermediate stages in which splitting should be clearer to see suggests that this is a fast event. Shortly later, the two most anterior stripes that represent the ocular and the cheliceral segments disconnect (Fig. 3K/L). Despite investigating approximately 30 embryos that correspond to this developmental stage, I never observed any further anterior splitting of hh-stripes in Pholcus that could correspond to the splitting of the pedipalpal segment as observed in Parasteatoda. It is, however, interesting to mention that there is a second splitting event, namely the splitting of a single initial hh-stripe that corresponds to the second and third leg-bearing segment. This represents a fundamental difference between Pholcus and Parasteatoda, suggesting that the splitting of hh-stripes is more variable in spiders than expected. In the context of the NHH, this represents an example of hh-splitting in segments that are not homologs of the PGS. This weakens the argument that hh-splitting may represent an indicative remnant of segment-splitting during the evolution of the PGS.

Fig. 3

Early expression of hh1 in the cellar spider Pholcus phalangioides. In all panels, anterior is to the left. Ventral views in panels B, E, H. Lateral views in panels A, D, F, and I−L. Dorsal views in panels C and G. Panels B and C, and panels E–G each show the same embryo from different directions. In all panels, blue asterisks mark the most anterior stripe of hh-expression that corresponds to the ocular and cheliceral segments; note that this stripe splits at later developmental stages. Note the splitting of the stripe corresponding to the second and third leg-bearing segment (panels D−G, yellow asterisks). Panels A´–L´ represent SYBR green staining of corresponding embryos. Developmental stages are indicated (after [26]). Roman Numerals 1–4 leg-bearing segments 1–4, ch cheliceral segment, cu cumulus, O1 and O2, first and second opisthosomal segment, oc ocular segment, pp pedipalpal segment, saz segment addition zone, st stage

A common theme in chelicerates, however, is the splitting of the most anterior stripe of hh into two corresponding to the ocular and the cheliceral segment as this appears to be conserved in mites, harvestmen and spiders. The remaining question is whether this represents a conserved feature of arthropods as a whole. I address this question in the chapter after next of this paper. Before, however, I would like to critically address the possibility of missing the splitting-off of the pedipalpal segment in the harvestman or any of the here investigated spiders.

3.3 Overlooking of the pedipalpal-split?

In Parasteatoda, formation of the most anterior hh-stripe, broadening of this stripe, and splitting-off of the pedipalpal hh-stripe is a quick process that only takes about four hours of development [18]. Therefore, it is conceivable that this short time window is not covered by my experiments. The second splitting event in Parasteatoda, however, is similarly quick and also only takes about four hours. Moreover this event is well documented for the harvestman and the two spider species investigated in my experiments. Taken together, this makes the repeated overlooking of a potential first splitting event (as seen in Parasteatoda) in all three species investigated here unlikely. Another point is that in Parasteatoda, the pedipalpal hh-stripe remains dorsally/ventrally connected for a while with the initial stripe, thus increasing the time window in which one would be able to detect signs of such a splitting event. This, however, I could never observe in any of the investigated chelicerate species. What I find instead is bending of the most anterior stripes of hh-expression towards a common lateral/dorsal fix point (best seen in the harvestman (Fig. 1K)). From some optical perspectives it may thus give the appearance that these stripes are connected, although in fact they are not (cf. discussion on Drosophila hh-splitting). Such bending of stripes is thus clearly not per se a sign of a previous splitting event. Yet another point to discuss is that the pedipalpal hh-stripe in Pholcus is relatively strong compared to the more posterior hh-stripes (Fig. 3E). This could indicate that this stripe, like in Parasteatoda, originates from the strong anterior stripe via splitting (note that also in Parasteatoda the pedipalpal stripe is strong compared to more posterior stripes [18]). But unlike in Parasteatoda, I could never detect any physical connection (not even dorsally/laterally) of these two stripes (Fig. 3F).

Summarized, there is no evidence for pedipalpal hh-stripe splitting in any of the three here investigated chelicerate species, although I cannot fully exclude this possibility.

3.4 Data on early hedgehog expression in other groups of arthropods reveals the ancestral state of stripe-splitting

In the dipteran fly Drosophila melanogaster, a wedge-shaped broad stripe of anterior hh expression forms early during development that has been described to split at later stages into three domains corresponding to the three PGS [2]. This, however, is not clear beyond doubt given the data provided in Lee et al. [2] (their Fig. 5g). Suspiciously, in the data they provide it looks as even the mandibular stripe is dorsally/laterally connected with the antennal and procephalic/ocular stripes/domains of hh-expression, although the mandibular stripe clearly formed earlier and separated from the anterior wedge-shaped domain ([2], their Fig. 5e and g). The short intercalary stripe, however, is not connected in any form with the more anterior stripes corresponding to the antennal and ocular stripes, or the posteriorly located mandibular stripe, and thus there is no evidence that this stripe resulted from stripe/domain splitting ([2], their Fig. 5g). In fact, a later study by Ntini and Wimmer [9] revealed that the intercalary hh-stripe in Drosophila indeed develops independently from the ocular and antennal stripes. Critical re-evaluation of the data thus shows that there is no two-fold splitting of hh-expressing anterior domains corresponding with the three PGS in Drosophila. This clarification is of importance because it could otherwise have implied an ancestral origin of the two-fold splitting and this would have lent support for the NHH.

In other insects like the beetle Tribolium castaneum [5], the cricket Gryllus bimaculatus [28], and the true bug Oncopeltus fasciatus [12] an early anterior stripe of hh may also split into two stripes, corresponding to the ocular and antennal segments. Interestingly, however, neither Farzana and Brown [5] nor Miyawaki et al. [28] claim any splitting of hh in Tribolium and Gryllus respectively, although it looks like splitting in all three species because the stripes are dorsally/laterally connected. Based on the available data, it appears that the most anterior stripe of hh splits into the ocular and antennal segments and that the intercalary stripe forms independently and with some delay in insects [5, 9, 12, 28].

To my knowledge, there are no data on early hh expression in any crustacean species; the only published data are on Parhyale hawaiensis hh and Artemia franciscana hh but in both papers the presented data do not include relevant early developmental stages [29, 30]. The situation in myriapods is complex. In the millipede Glomeris marginata, splitting of a broad anterior hh-stripe results in an ocular and an antennal hh-stripe [31], and thus shows the same pattern as in insects and (most) chelicerates. In the centipede Strigamia maritima, however, an initial broad domain appears to split into three stripes, the ocular, the antennal and the intercalary thus resembling the situation in the spider Parasteatoda [32]. Somewhat surprisingly, in Strigamia it appears that also the mandibular stripe is dorsally/laterally connected to the other three aforementioned stripes suggesting that this stripe too may be originated from the initial broad stripe, or, that like discussed for Drosophila, this represents an optical artefact [32]. If not the latter, this could represent splitting into four stripes corresponding to the three PGS plus the mandibular segment. In that case, the data would not support the NHH, but rather refute it, and would represent another unique pattern of hh-splitting (cf. Pholcus data, Fig. 3) (summarized in Fig. 4).

Fig. 4

Summary of reported hh-splitting events in arthropods. This figure shows a cladogram of the most relevant arthropod species discussed in this paper. 1: Chelicerata; 2: spiders sensu lato; 3 Mandibulata; 4 Myriapoda; 5 Pancrustacea. Note that there are no relevant early expression data in any crustacean species, indicated by a row of question marks. The pattern described for insects is based on conserved expression in several species. The question marks in the Strigamia line indicates that it is not clear beyond any doubt that this expression is indeed splitting-derived (see text for further information). The protocerebral (P), deuterocerebral (D) and tritocerebral (T) segments are indicated (these are the pre-gnathal segment (PGS)), as well as a number of posterior segments (PSs) that entail the gnathal segments (+ 1− + 3) and additional PSs (+ 4, + 5 …). Blue bars indicate expression of hh that was generated via stripe/domain-splitting. Yellow bars indicate expression of hh hat did not occur via a splitting process. Salmon-coloured bars indicate hh-expression that was generated via stripe/domain-splitting in segments posterior to the three PGS

Conclusively, the available data suggest that a single splitting event of hh in the arthropod head represents the ancestral state, and that the splitting into three stripes corresponding to the three PGS found in the spider Parasteatoda must be interpreted as derived, even within spiders. Likewise, the splitting into three or four stripes in Strigamia must under this scenario represent a derived state. Compared to the conserved pattern in insects, the situation in the spiders Parasteatoda and the centipede Strigamia thus reveal a high degree of variation concerning the splitting of the most anterior hh-domain in the arthropod head. The reason for this flexibility, however, remains unclear. In the final section of this paper, I would like to speculate on a possible reason for this variability.

3.5 On the patterns of “segment splitting” in the germ disc of spiders

In most arthropods, only the anterior segments develop almost simultaneously and from the blastoderm, and posterior segments are added sequentially, usually one by one, from the posterior located segment addition zone (e.g. [7]). The number of segments that form via the blastoderm varies between different groups of arthropods as for example in a myriapod [4] and an insect [33]. Splitting events as reported here and elsewhere for hh and other early expressed segmentation genes are likely only to occur in the segments that develop from the blastoderm, but not posteriorly added segments. In the latter, however, comparable morphogenetic processes may be represented by the dynamic expression of genes in the segment addition zone. In spiders, a relatively large number of segments, all prosomal segments, develop from the germ disc, a blastoderm-like structure (e.g. [34]). In other chelicerates including harvestmen and scorpions, for example, much fewer segments form from the germ disc (scorpions) [35] or the blastoderm (harvestmen) [25]. At the same time, the spider embryo undergoes dramatic morphogenetic changes as the radial germ disc transforms into a bilateral germ band (e.g. [36]). The area in which splitting events of segment-primordia (and associated gene expression patterns) can occur is thus larger and more likely (due to the drastic morphogenetic movements in spiders) than in other groups of chelicerates, and arthropods in general. I speculate that this circumstance may contribute to the variety of hh-stripe splitting patterns reported in this paper for Pholcus, and previously reported for Parasteatoda [18]. Differences of timing in development (heterochrony; either concerning onset of gene expression or timing of cell re-arrangement) and parallel transformation of the spider germ disc into a germ band may be the reason(s) why different splitting events of hh-stripes are observed. I speculate further that similar differences of splitting may be observed when investigating the early expression of pair-rule gene orthologs in spiders; note that Pechmann et al. [23] present data on hairy (h) expression in the germ disc showing splitting of multiple segment-wide domains in the germ disc. In order to investigate this further, I propose to systematically investigate the expression pattern of these genes early during spider development, and to also systematically perform cell-labelling experiments to reveal the early fate map of segment primordia (or otherwise contributing cells) in the germ disc of spiders.