FoxB, a new and highly conserved key factor in arthropod dorsal–ventral (DV) limb patterning
Forkhead box (Fox) transcription factors evolved early in animal evolution and represent important components of conserved gene regulatory networks (GRNs) during animal development. Most of the researches concerning Fox genes, however, are on vertebrates and only a relatively low number of studies investigate Fox gene function in invertebrates. In addition to this shortcoming, the focus of attention is often restricted to a few well-characterized Fox genes such as FoxA (forkhead), FoxC (crocodile) and FoxQ2. Although arthropods represent the largest and most diverse animal group, most other Fox genes have not been investigated in detail, not even in the arthropod model species Drosophila melanogaster. In a general gene expression pattern screen for panarthropod Fox genes including the red flour beetle Tribolium castaneum, the pill millipede Glomeris marginata, the common house spider Parasteatoda tepidariorum, and the velvet worm Euperipatoides kanangrensis, we identified a Fox gene with a highly conserved expression pattern along the ventral ectoderm of arthropod and onychophoran limbs. Functional investigation of FoxB in Parasteatoda reveals a hitherto unrecognized important function of FoxB upstream of wingless (wg) and decapentaplegic (dpp) in the GRN orchestrating dorsal–ventral limb patterning.
KeywordsAppendage patterning Forkhead domain Limb segmentation Development
Arthropod limbs develop along three different axes, the proximal–distal (PD) axis, the anterior–posterior (AP) axis, and the dorsal–ventral (DV) axis. In the model system Drosophila melanogaster, leg allocation and AP axis determination is under control of segment polarity genes such as wingless (wg) and hedgehog (hh) (e.g. [1, 2]). This is likely conserved in arthropods and onychophorans as indicated by gene expression and functional data (e.g., [3, 4, 5, 6, 7, 8, 9, 10]). The PD axis is established by the function of the so-called limb gap genes and the morphogens Wg and Decapentaplegic (Dpp) (e.g., [1, 11, 12, 13]), and gene expression data suggest that the function of limb gap genes is generally conserved among arthropods and onychophorans (e.g., [3, 10, 14, 15, 16, 17, 18, 19]). In Drosophila, the morphogens Dpp and Wg are also involved in the determination of the DV axis [20, 21, 22]. The wg gene is expressed in the ventral region of the leg imaginal discs and loss of Wg protein causes dorsalisation of these limbs [22, 23, 24]. Downstream of Wg act two T-box genes, the paralogs H15 and midline (mid), both of which are like wg expressed in ventral ectodermal cells of the limbs [4, 20, 25, 26, 27]. dpp is expressed along the dorsal side of the Drosophila leg imaginal disc and loss of Dpp causes ventralization of these limbs [28, 29]. Downstream of Dpp functions another T-box gene, optomotor-blind (omb), which is expressed along the dorsal side of the legs [20, 29]. Expression of omb can induce dorsal fate in ventral cells of the developing legs . Comparative gene expression data suggest that the role of ventral and dorsal leg patterning genes is conserved in arthropods, and partially also in onychophorans (e.g., [10, 26, 30, 31, 32]). However, functional evidence of a conserved DV patterning system in arthropods is sparse and exclusively based on data in insects [33, 7, 9, 25, 34]. While wg appears to be involved in DV limb development in holometabolous insects [7, 9], this does not appear to be the case for hemimetabolous insects .
Notably, many of the genes involved in DV limb patterning are duplicated in Drosophila, as well as in other studied arthropods and onychophorans. There are two H15-type genes in Drosophila and the millipede Glomeris marginata, and at least three H15 genes in spiders, and two copies of omb in onychophorans and spiders (summarized in ). Many of these genes have retained conserved expression patterns along the ventral and dorsal region of the developing limbs, respectively. Therefore, it is likely that their function(s) in DV patterning are at least partially conserved as well. This makes functional studies difficult because of likely redundant functions of such paralogs. Although wg is not duplicated in arthropods and onychophorans [6, 35, 36], it is a member of the Wnt class of genes (wg is Wnt1) of which arthropods ancestrally possess 12 classes (reviewed in [36, 37]). Many Wnt genes, although not paralogs of wg/Wnt1, are expressed in very similar patterns along the ventral side of the developing arthropod limbs (e.g., [38, 36]). It is therefore possible that other Wnt genes may substitute for wg function, and that functional data on the role of wg in arthropod limb development are inconclusive and potentially misleading (cf. [7, 9, 33]).
Here we report on the discovery of a hitherto unrecognized gene that is expressed along the ventral side of the investigated arthropods and an onychophoran, the forkhead transcription factor-encoding gene FoxB (Drosophila paralogs Dmfd4/Dmfd5 aka fd96Ca/fd96Cb [39, 40]. Although it exists in two copies (paralogs) in the model arthropods Drosophila and the flour beetle Tribolium castaneum, there is only one copy in the spider Parasteatoda tepidariorum. We therefore targeted the spider FoxB gene (Pt-FoxB) in our study and investigated its function in appendage development. Among other phenotypes, Pt-FoxB knockdown leads to altered leg morphologies, likely correlated with disturbed DV patterning during limb development. The expression of other known (or implied by conserved expression patterns) DV limb patterning genes such as omb, H15, and wg/Wnt1 is disturbed in Pt-FoxB knockdown appendages. This indicates a high-ranking function of FoxB in the gene regulatory network orchestrating DV limb patterning in spiders as well as Panarthropoda as a whole.
Research animals, embryo collection and developmental staging
Drosophila flies and embryos were obtained from the cultures in Göttingen (Oregon-R strain). Tribolium embryos were obtained from the cultures in Göttingen (San Bernardino strain). Glomeris embryos were collected and prepared as described in Janssen et al. . Parasteatoda spiders were obtained from the established Göttingen strain for RNA-interference experiments. They were kept separately in plastic vials at approximately 21 °C. They were supplied with water and fed with either sub-adult crickets (Acheta domesticus) or Drosophila. Euperipatoides kanangrensis embryos were obtained as described in Hogvall et al. . Developmental staging is after Janssen et al.  (Glomeris), Janssen and Budd  (Euperipatoides), Mittmann and Wolff  (Parasteatoda), and Strobl and Stelzer  (Tribolium).
Gene cloning, probe synthesis, whole mount in situ hybridization and nuclear staining
All gene fragments were isolated by means of RT-PCR with gene-specific primers based on sequence information from either sequenced genomes (Tribolium, Tribolium Genome Sequencing Consortium  and Parasteatoda, Schwager et al. ), or sequenced embryonic transcriptomes (Glomeris and Euperipatoides). Primer sequences are listed in Additional file 1: Table S1. Genes were cloned into either the PCRII vector (Invitrogen) or the PCR4-TOPO vector (Invitrogen). Sequences of the cloned fragments were verified by means of Big Dye chemistry on an ABI3730XL sequence analyser by a commercial sequencing service (Macrogen, Seoul, South Korea). Accession numbers of all gene fragments are listed in Additional file 2: Table S2.
Antisense RNA probes were in vitro transcribed with either Sp6, T7 or T3 RNA Polymerase (Roche) after amplification of the gene products from the plasmids using M13F and M13R oligonucleotides (Additional file 1: Table S1). Embryos of Tribolium were fixed as described in Schinko et al. . Embryos of Parasteatoda were fixed as described in Pechmann et al. , or, after RNAi treatment, were fixed in a formaldehyde–heptane mix (1:15) for 4 h. Thereafter, the vitelline membranes were removed with Dumont-5 forceps. There is no significant difference in the result of in situ hybridization with either of the spider fixation protocols. Embryos of Glomeris were fixed in the same way as spider embryos after RNAi treatment. Embryos of Euperipatoides were fixed in a 1:15 formaldehyde-PBS mix for 4–6 h; membranes were removed with Dumont-5 forceps prior to fixation. For all single-colour stainings in all embryos, we applied the whole mount in situ hybridization protocol as described in Janssen et al. . Double in situ hybridization was performed as described in Janssen et al.  with the exception that BM Purple (blue signal) (Roche) and SIGMAFAST Fast Red TR/Naphthol AS-MX (red signal) (SIGMA) were used for probe detection. Cell nuclei were visualized incubating embryos in 2 μg/ml of the fluorescent dye 4-6-diamidino-2-phenylindole (DAPI) in phosphate-buffered saline with 0.1% Tween-20 (PBST) for 20–30 min.
dsRNA synthesis and parental RNAi
After amplification from the plasmid using T7 and T7-Sp6 overhang primers (Additional file 1: Table S1), double-stranded RNA (dsRNA) for Pt-FoxB was synthesized using the MEGAscript T7 Kit (Life technologies). Sodium acetate was used for precipitation and RNA was dissolved in injection buffer (1.4 mM NaCl, 0.07 mM Na2HPO4, 0.03 mM KH2PO4, 4.0 mM KCL). Female spiders were injected laterally into the opisthosoma and either mated several hours prior to injection, or several hours after the first injection. We performed two independent rounds of injection that differed in the concentration of injected dsRNA. Each spider was injected three times (on 3 consecutive days) with each time 2.5 µl of 2.8 µg/µl dsRNA (first round of injections) or 4 µg/µl dsRNA (second round of injections) in injection buffer. Control spiders were injected with 2.5 μl of injection buffer. Altogether, 20 adult females were injected with Pt-FoxB dsRNA and 20 females with buffer only. Of all surviving spiders, cocoons of dsRNA-injected spiders were kept in glass vials separately. All dsRNA-injected and all control cocoons, respectively, were pooled and investigated each as one batch (see Additional file 3: Figure S1 for quantification). Embryos with morphologically distinguishable phenotypes were categorized into four distinct classes of which one, Class-I, is of particular interest for this study (see “Results”).
Amino acid sequences of the FoxA, FoxB and FoxC forkhead domains of Parasteatoda, Glomeris, Tribolium, Euperipatoides and Drosophila were aligned using ClustalX with default parameters in MacVector v12.6.0 (MacVector, Inc., Cary, NC). In our analysis, the forkhead domain of Drosophila FoxQ2 serves as an outgroup sequence.
Bayesian phylogenetic analysis was performed with MrBayes  using a fixed WAG amino acid substitution model with gamma-distributed rate variation across sites (with four rate categories). Unconstrained exponential prior probability distribution on branch lengths and an exponential prior for the gamma shape parameter for among-site rate variation were applied. Topologies were estimated using 0.5 million generations for the Metropolis-coupled Markov chain Monte Carlo analysis with four chains. The chain-heating temperature was set to 0.2, and the chains were sampled every 200 cycles. 25% of samples were used as burnin. Clade support values were computed with posterior probabilities in MrBayes. See Additional file 4: Figure S2.
Imaging and image processing
All pictures were taken with a Leica DFC490 digital camera mounted on a Leica dissection microscope (MZ-FLIII). The image processing software Adobe PHOTOSHOP CC 2015 (v. 1.0 for Apple Macintosh) was used for the application of linear corrections of contrast and brightness.
Expression of panarthropod FoxB genes in developing appendages
Expression of FoxB in the onychophoran Euperipatoides (as a panarthropod outgroup species) is restricted to ventral expression early during limb development (Fig. 1j–l). While the slime papillae, the jaws and the walking limbs express FoxB in this way, the most anterior appendages, the protocerebral frontal appendages (= the onychophoran antennae) do not express FoxB (Fig. 1j). Additionally, FoxB genes are expressed in the developing ventral nervous system in Drosophila , Tribolium, Glomeris, Parasteatoda, and Euperipatoides (Figs. 1 and 2, Additional file 7: Figure S5 and Additional file 8: Figure S6), and in Glomeris, FoxB is strongly expressed in the anal valves (Additional file 7: Figure S5). At early developmental stages, Parasteatoda FoxB is expressed in the centre of the germ disc where the cumulus forms (Fig. 2a), and slightly later when the germ field forms, FoxB is expressed ubiquitously (Fig. 2b). When the segments form, FoxB is expressed in the form of transverse segmental stripes that later refine to the position of the forming limb buds (Fig. 2c–i).
FoxB function in the spider Parasteatoda
Class-II embryos show an unusually slim germ band. Class-III embryos develop a partially duplicated germ band. In Class-IV embryos, the germ disc does not form correctly. In many of these embryos, the disc is either fully malformed or disintegrated. The yolk is exposed as if the overlying so-called extraembryonic ectoderm did either not form, or disintegrate after formation.
Completely undeveloped eggs likely represent unfertilized eggs that never developed a protective vitelline membrane, and therefore appear as hardened yolk masses after fixation. Our analysis shows that the number of dead embryos is similar in knockdown embryos and wild-type (or control) embryos (Additional file 3: Figure S1).
We also recognized a small percentage of Class-II and Class-IV phenotypes in control embryos, indicating that delayed and small embryos as well as disintegrating (or not-forming) germ discs occasionally occur under natural conditions as well (Additional file 3: Figure S1). This shows that the transition from radial to bilateral symmetry and the onset of germ band formation are critical steps in spider development. However, we did not find a single Class-I or Class-III embryo in control embryos (Additional file 3: Figure S1).
In FoxB knockdown cocoons, the number of embryos identified as wild type after fixation at different embryonic stages is higher than the hatching rate suggesting that FoxB knockdown may lead to additional effects that are difficult to recognize (e.g., defects of the nervous system).
Expression of DV appendage patterning genes in FoxB knockdown embryos
The conserved arthropod dorsal limb marker optomotor-blind (omb) (e.g. [10, 27, 31] is still strongly expressed in the dorsal tissue of the limbs in Pt-FoxB knockdown embryos (Fig. 6p–r, Additional file 11: Figure S9). Interestingly, however, we observed an extension of the omb expression pattern into ventral tissue, that normally does not express Pt-omb (Fig. 6p–r), indicating a partial dorsalisation of the appendages after loss of Pt-FoxB function. This is predominantly (if not exclusively) observed for the distal region of the pedipalps and legs.
FoxB genes represent evolutionary conserved markers of ventral limb tissue in ventral appendages
We could show that FoxB genes are expressed along the ventral side of all ventral appendages and that this expression is conserved in species of diverse panarthropod groups, namely the fly Drosophila, the beetle Tribolium, the millipede Glomeris, the spider Parasteatoda, and the onychophoran Euperipatoides. This suggests a conserved role for FoxB in DV appendage patterning in the entire clade Panarthropoda. Dorsal appendages, like the wings and halteres in Drosophila, in contrast, do not express FoxB, indicating that its function is restricted to ventral appendages. In all ventral appendage types including the highly modified spider opisthosomal appendages (i.e., the book lungs, the tracheal system, and the spinnerets), FoxB is expressed along the ventral ectoderm. This pattern is very similar to that of wg and H15, two other highly conserved ventral limb marker genes [5, 10, 27, 31].
Exceptions from this rule are the conserved FoxB expression domains in the dorsal tissue of the labrum in Tribolium, Glomeris and Parasteatoda (note that we did not investigate expression of FoxB in the labral discs of Drosophila). This apparent discrepancy, however, can be explained by the hypothesis that the labrum is the result of rotation and fusion of a pair of limbs. As a consequence, ventral and dorsal tissue is reversed in the labrum . The second exception concerns the frontal appendages of the onychophoran which do not express FoxB. These appendages are innervated from the protocerebrum and likely are homologous with the arthropod labrum ([58, 59]; discussed in e.g., ), although expression of FoxB does not support this notion.
Knockdown of FoxB in the spider Parasteatoda reveals a specific role in DV appendage development
The highly conserved expression of FoxB in the limbs of arthropods and the onychophoran strongly suggests an important and evolutionarily conserved function in panarthropod DV limb development. The functional analysis of FoxB in the spider Parasteatoda tepidariorum indeed revealed that FoxB is required for proper DV patterning during limb axis formation. The Class-I phenotype shows an abnormally crooked distal region of pedipalps and legs, most probably explained by a reduction of ventral tissue. Class-I appendages are also broader and softer than wild-type appendages, indicating that the overall integrity of the limbs is disturbed. This becomes even more evident in later stage Class-I appendages which are characterized by the occurrence of abnormal constrictions that finally lead to the complete budding off of limb parts, especially in the distal region (Fig. 5).
A very similar phenotype has been reported for wg/Wnt1 and its receptor-encoding gene frizzled-1 (fz1) in the beetle Tribolium [7, 61]. Here the phenotypes are called “candy cane” and “nonpareille”/“pearls on a chain” referring to the bending of the limbs (“candy cane”) and the budding off and fusion of distal limb segments (“nonpareille” and “pearls on a chain”) . Tribolium fz1 is expressed ubiquitously, but it fulfils a specific function in limb development as revealed by fz1 knockdown . Although the function of fz1 is not yet studied in other arthropods, it is also ubiquitously expressed in Parasteatoda, allowing for a conserved interaction of Wg and Fz1 in spider limb development .
The effect of knockdown of Pt-FoxB and Tc-wg, both of which are expressed in conserved patterns along the ventral side of appendages in all hitherto investigated arthropods, is strikingly similar (“Bandyklubba” phenotype and “candy cane” phenotype, respectively) suggesting that they might work in the same conserved gene regulatory network (GRN) in DV limb patterning.
FoxB acts as a key factor in the gene regulatory network (GRN) controlling DV appendage patterning
After FoxB knockdown in the spider, expression of both ventral marker genes, wg and H15.2, is missing (Fig. 9b, c). This indicates that FoxB acts upstream of wg in the GRN required for DV patterning. Since wg is acting upstream of H15 in Drosophila, the lack of H15.2 in FoxB knockdown appendages could be the result of the lack of wg, and thus a secondary effect of FoxB, or it could (as assumed for wg) be under direct control of FoxB. Our experimental setup cannot distinguish between these two possibilities, but it would be interesting to study in future experiments.
The expansion of dpp expression along the ventral region in limbs after FoxB knockdown indicates that FoxB normally acts as a repressor of dpp in ventral tissue, either directly, or via wg and/or H15.2 (Fig. 9d). We note, however, that no aspect of the topology model predicts our observation that the expression of Pt-dpp is progressively removed from the distal tip in Pt-FoxB knockdown embryos (Fig. 9d), and therefore this effect of Pt-FoxB RNAi cannot be explained by the model.
Also, the dorsal factor omb is intruding ventral and distal areas of appendages in FoxB knockdown embryos, which suggests that FoxB acts directly (or indirectly via Wg and H15.2) as a repressor of omb (Fig. 9e). The assumption that Dpp could act as a direct activator of omb  is not supported by our data, because the expansion of dpp along the ventral side of the limbs apparently does not cause ectopic expression of omb in this tissue (Fig. 9e). However, it is also possible that ventral tissue is not competent for omb expression, even in the presence of dpp.
In Drosophila, Hedgehog (Hh) activates dpp and wg in the leg disc due to an early asymmetry that allows ventral and dorsal cells to respond differently to Hh signalling (in dorsal tissue, dpp is activated, and in ventral tissue, wg is activated). Such asymmetry is provided by the relative earlier expression of Wg in ventral tissue [2, 20, 23]. Consequently, in the absence of Wg, Hh would activate dpp in ventral tissue, instead of wg .
This scenario is in line with our data. Since wg is absent from ventral tissue in FoxB knockdown embryos as a result of the missing function of FoxB, now dpp is dominantly expressed in this tissue. Once the asymmetry between wg and dpp expressing tissue is established, Dpp and Wg act as mutual antagonists in the Drosophila imaginal discs . If this mutual antagonistic function is conserved, or at least the repressive function of Wg on Dpp, this might explain why dpp expands into the now wg-free ventral limb tissue after FoxB knockdown. Again, we cannot distinguish between a possible direct or indirect repression of dpp via FoxB or/and Wg. Either way, our data suggest that FoxB is acting at a high level in the GRN orchestrating DV limb patterning.
Evidence for different regulatory mechanisms acting along the AP axis of developing limbs
It has been shown that different regions along the AP axis of the Drosophila leg are under control of different GRNs, or that given GRNs act differently in different regions of the leg. For example, the most proximal region of the Drosophila leg, the coxa, never expresses Distal-less (Dll), a gene that is otherwise involved in the formation of all other podomeres (leg segments) (reviewed in ). It has also been shown that wg plays a specific role in the development of the coxa . Similarly, it also appears that the proximal region (including the coxa) is patterned differently in the beetle Tribolium. Interestingly, here wg appears to have the opposite effect. While distal regions of the legs are affected in wg knockdown and Fz1 knockdown embryos, this is not the case for the coxal region [7, 61]. Therefore, it is possible that the proximal region and the distal region (defined as distal to the coxa) are generally regulated differently in arthropods [66, 67].
Our results on Parasteatoda FoxB function support this hypothesis and suggest that the differences between proximal and distal leg development may indeed date back to the last common ancestor of insects and spiders, i.e., the arthropod ancestor. Although Pt-FoxB is expressed all along the ventral side of pedipalps and legs, its knockdown affects only the distal and medial, but not the proximal Pt-wg expression (Fig. 9b). Similarly, Pt-FoxB knockdown leads to the misexpression of Pt-omb in the distal portions of pedipalps and legs, while medial and proximal portions are not affected (Fig. 9e). The most intriguing result, however, is the complete change of the Pt-dpp expression pattern after Pt-FoxB RNAi, especially in the distal tip. In this case, the loss of Pt-FoxB influences Pt-dpp expression even in portions of the limbs that never express Pt-FoxB. The reason for this is currently not clear.
Evidence for the coupling of DV patterning and joint formation
In Drosophila, the correct formation of joints depends on the PD patterning system and the so-called leg gap genes (e.g. [68, 69]; summarized in ). In a combinatorial mode, they activate Delta/Notch signalling (e.g., [71, 72, 73]) and downstream of Delta/Notch signalling act, e.g., the odd-skipped family genes, including odd-skipped (odd) itself [74, 75]. It has been shown that the involvement of Delta/Notch signalling and its downstream factors such as odd in arthropod joint formation is conserved in arthropods beyond Drosophila . In the spider Cupiennius salei, the odd ortholog odd-related-1 (one of three identified odd-related genes in this spider) is expressed in concentric rings in the limbs downstream of Delta/Notch signalling and its function is clearly correlated with that of joint formation . The same expression pattern is seen for odd in the limbs of Parasteatoda (Fig. 8).
Remarkably, we find that odd expression in concentric rings is disturbed after knocking down FoxB, but only in the ventral sector, while expression along the dorsal side of the limbs is not affected (except for the distal region where expression of odd is completely lost) (Fig. 8). Double in situ revealed that odd is indeed co-expressed with the patches of enhanced expression of FoxB (Fig. 3k). Together, this implies that odd expression in the limbs is likely under control of the DV patterning system downstream of FoxB function, at least ventrally. Since odd is one of the genes that is involved in joint formation in spiders, this finding is the first potential evidence that joint formation and DV patterning may be connected.
We thank the following colleagues for providing clones, cDNAs and embryos: Matthias Pechmann (University of Cologne, Germany), Alistair McGregor (Oxford Brookes University, England) and Gregor Bucher (University of Göttingen, Germany). We gratefully acknowledge the support of the New South Wales Government Department of Environment and Climate Change by provision of a permit SL100159 to collect onychophorans at Kanangra-Boyd National Park. We thank Glenn Brock, David Mathieson, Robyn Stutchbury and especially Noel Tait, for their help during onychophoran collection. Financial funding was provided by the Swedish Natural Science Council (VR), Grant No. 621-2011-4703 (to R.J.), and the Deutsche Forschungsgemeinschaft (Grant Nos. PR 1109/4-1 and PR 1109/6-1 to N.M.P.). In situ hybridization experiments were partially executed under the supervision of RJ by Katerina Günter and Johanna Sundqvist during the “Evolution and Development” course at Uppsala University in the spring of 2016, course no. 1BG397.
- 35.Janssen R, Le Gouar M, Pechmann M, Poulin F, Bolognesi R, Schwager EE, Hopfen C, Colbourne JK, Budd GE, Brown SJ, Prpic NM, Kosiol C, Vervoort M, Damen WG, Balavoine G, McGregor AP. Conservation, loss, and redeployment of Wnt ligands in protostomes: implications for understanding the evolution of segment formation. BMC Evol Biol. 2010;10:374.PubMedPubMedCentralCrossRefGoogle Scholar
- 38.Bolognesi R, Beermann A, Farzana L, Wittkopp N, Lutz R, Balavoine G, Brown SJ, Schröder R. Tribolium Wnts: evidence for a larger repertoire in insects with overlapping expression patterns that suggest multiple redundant functions in embryogenesis. Dev Genes Evol. 2008;218:193–202.PubMedPubMedCentralCrossRefGoogle Scholar
- 47.Schinko J, Posnien N, Kittelmann S, Koniszewski N, Bucher G. Single and double whole-mount in situ hybridization in red flour beetle (Tribolium) embryos. Cold Spring Harb Protoc. 2009;8:pdb.prot5258.Google Scholar
- 49.Janssen R, Andersson E, Betner E, Bijl S, Fowler W, Höök L, Lehr J, Mannelqvist A, Panara V, Smith K, Tieman S. Embryonic expression patterns and phylogenetic analysis of panarthropod Sox genes: insight into nervous system development, segmentation and gonadogenesis. BMC Evol Biol. 2018;18:88.PubMedPubMedCentralCrossRefGoogle Scholar
- 57.Zhang N. Developmental studies of appendage patterning and formation in spiders. Ph.D. Thesis. Georg-August-Universiät Göttingen, Göttingen, Germany. 2016. http://hdl.handle.net/11858/00-1735-0000-002B-7CB4-D.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.