Abstract
The resurgence of interest in the comparative developmental study of chelicerates has led to important insights, such as the discovery of a genome duplication shared by spiders and scorpions, inferred to have occurred in the newly proposed clade Arachnopulmonata. Nonetheless, several groups remain understudied in the context of development and genomics, such as the order Amblypygi (whip spiders). The phylogenetic position of Amblypygi in Arachnopulmonata poses them as an interesting group to test the extent of conservation of the proposed genome duplication in the history of arachnids. Moreover, whip spiders have their first pair of walking legs elongated and modified into sensory appendages (a convergence with the antenna of mandibulates), but the genetic patterning of these antenniform legs has never been investigated. Here, we break a 45-year hiatus of embryological studies in Amblypygi, by establishing genomic resources and protocols for cultivation of embryos and gene expression assays by in situ hybridization to study the development of the whip spider Phrynus marginemaculatus. We provide evidence that P. marginemaculatus retains duplicates of most Hox genes, and that paralogs of the leg gap genes dachshund and homothorax retain arachnopulmonate-specific expression patterns. We characterize the expression of leg gap genes Distal-less, dachshund1/2 and homothorax1/2 in the embryonic antenniform leg and other appendages, and provide evidence that allometry, and by extension the antenniform leg fate, is already specified very early in embryogenesis.
Introduction
In the past 25 years, comparative developmental study of chelicerates has resurged with the integration of molecular biology and genomics, spearheaded by developmental genetic investigations of the spiders Cupiennius salei and Parasteatoda tepidariorum (Akiyama-Oda & Oda, 2003; 2010; W. G. M. Damen, Hausdorf, Seyfarth, & Tautz, 1998; Schwager, Schoppmeier, Pechmann, & Damen, 2007; Schwager et al., 2017; Stollewerk, Schoppmeier, & Damen, 2003). Such works have revealed several important aspects of chelicerate development, such as the molecular mechanism of dorso-ventral axis patterning (Akiyama-Oda & Oda, 2003; 2006; Oda & Akiyama-Oda, 2008; Pechmann, Benton, Kenny, Posnien, & Roth, 2017), segmentation of the prosoma (Damen et al., 1998; Paese, Schoenauer, Leite, Russell, & McGregor, 2018; Pechmann et al., 2011; Schwager, Pechmann, Feitosa, McGregor, & Damen, 2009; Setton & Sharma, 2018) and opisthosoma (Paese et al., 2018; Stollewerk et al., 2003), and specification of prosomal versus opisthosomal fate (Khadjeh et al., 2012; Pechmann, Schwager, Turetzek, & Prpic, 2015). A few non-spider chelicerate models have also emerged in recent years and provided new perspectives on chelicerate development, such as the the horseshoe crab Limulus polyphemus (Xiphosura) (Blackburn et al., 2008; Mittmann & Scholtz, 2001), the mites Archegozetes longisetosus (Barnett & Thomas, 2012; 2013a; 2013b; Telford & Thomas, 1998) and Tetranychus urticae (Acariformes) (Dearden, 2002; Grbić et al., 2007; Khila & Grbić, 2007a), the tick Rhipicephalus microplus (Parasitiformes) (Santos et al., 2013), the Arizona bark scorpion Centruroides sculpturatus (Sharma, Schwager, Extavour, & Wheeler, 2014b) (Scorpiones) and the harvestmen Phalangium opilio (Opiliones) (Sharma, Schwager, Extavour, & Giribet, 2012a; 2012b; Sharma, Schwager, Giribet, Jockusch, & Extavour, 2013; Sharma et al., 2015b). For instance, studies in A. longisetosus and C. sculpturatus have suggested that changes in Hox gene number and expression domains are responsible for both reduced segmentation of mites and the unique body plan of scorpions (Barnett & Thomas, 2013a; Sharma, Schwager, Extavour, & Wheeler, 2014b).
In addition to the insights into chelicerate development, one of the major recent outcomes of increasing availability of genomic resources in in the group was the discovery of a whole or partial genome duplication in spiders and scorpions. This genome duplication is inferred to trace back to the most recent common ancestor spiders and scorpions, the newly proposed clade Arachnopulmonata that includes all arachnid orders with book lungs (Leite et al., 2018; Schwager et al., 2017; Sharma, Kaluziak, Pérez-Porro, González, Hormiga, et al., 2014a; Sharma, Santiago, González-Santillán, Monod, & Wheeler, 2015a; Sharma, Schwager, Extavour, & Wheeler, 2014b). This duplication event is also inferred to be independent from the multiple rounds of whole genome duplication undergone by horseshoe crabs (Xiphosura) (Kenny et al., 2015; Shingate et al., 2020). The duplication in spiders and scorpions encompasses several important developmental genes, such as the homeobox family (Leite et al., 2018; Schwager et al., 2017). Excitingly, there is increasing evidence of divergent expression and function of paralogs, such as in the case of Hox genes, leg gap genes and Retinal Determination Gene Network homologs (Gainett et al., 2020; Samadi, Schmid, & Eriksson, 2015; Schomburg, Turetzek, & Prpic, 2020; Schwager et al., 2017; Sharma, Schwager, Extavour, & Wheeler, 2014b; Turetzek, Khadjeh, Schomburg, & Prpic, 2017; Turetzek, Pechmann, Schomburg, Schneider, & Prpic, 2015).
Nevertheless, knowledge on the development of several arachnid orders still remains scarce or entirely unexplored in the contexts of embryological study and genomic architecture. One particularly interesting arachnid order whose genomic evolution and developmental biology remains largely unexplored is Amblypygi. Commonly known as whip spiders, Amblypygi comprise approximately 150 described species of nocturnal predators that are world-wide distributed primarily on the tropics and subtropics (Harvey, 2013; Weygoldt, 2000). Amblypygi is part of the clade Pedipalpi (together with the orders Thelyphonida and Schizomida), which in turn is sister group to spiders (Araneae) (Ballesteros & Sharma, 2019; Giribet, 2018; Lozano-Fernandez et al., 2019; Sharma, Kaluziak, Pérez-Porro, González, Hormiga, et al., 2014a). Considering the phylogenetic position of Amblypygi within Arachnopulmonata, a better understanding of Amblypygi genomics is anticipated to facilitate exploration of the evolutionary outcomes of genome duplication in Arachnopulmonata, and better characterizing the extent to which arachnopulmonate orders retained and specialized the ensuing paralogous genes. Unfortunately, the embryology of the order Amblypygi is known only from the works of Pereyaslawzewa (1901) a brief mention in Strubell (1892) (“Phrynidae”), Gough (1902), and the comprehensive study of Weygoldt (1975) in Phrynus marginemaculatus (formerly Tarantula marginemaculata).
Beyond the interest in their genomic architecture, whip spiders possess a fascinating biology and natural history. Whip spiders are emerging as model organisms in behavioral ecology, learning and neurophysiology in arachnids (Chapin & Hebets, 2016; Wiegmann et al., 2016). Their pedipalps, the second pair of appendages, are robust raptorial devices used for striking prey, including mostly arthropods, and even small vertebrates (Chapin & Hebets, 2016; Seiter, Lemell, Gredler, & Wolff, 2019; Weygoldt, 2000). However, their most conspicuous characteristic and namesake is their “whip”, a modified antenniform first walking leg that is not used for locomotion, but instead as a sensory appendage. The antenniform legs are extremely elongated relative to the other leg pairs, and the tibia and tarsus are pseudo-segmented into hundreds of articles that harbor an array of chemo-, thermo-, hygro- and mechanoreceptive sensilla (Igelmund, 1987; Foelix et al., 1975; Santer & Hebets 2011). The peripheral circuitry of the antenniform legs is complex, exhibiting peripheral synapses and giant neurons which are possibly involved in fast sensory responses (Foelix & Troyer, 1980; Santer & Hebets, 2011). The antenniform legs also have been shown to be important for foraging, mating, and intrasexual contests (Santer & Hebets, 2008; 2011). Notably, concentration of sensory structures, elongation and pseudo-segmentation of these legs constitute a striking convergence with the antenna of mandibulate arthropods (Pancrustacea and Myriapoda). Evidence from Hox gene expression and functional experiments support the view that the antenna of mandibulates is positionally homologous to the chelicera of chelicerates (deutocerebral appendage) (Sharma et al., 2015b; Telford & Thomas, 1998). Despite their different positions along the antero-posterior axis of the body, the serial homology of antennae (mandibulates) and antenniform legs (whip spiders) to walking legs constitutes a potentially useful comparison to address whether a striking morphological convergence is associated with convergence in mechanisms of genetic patterning.
Most of the knowledge about the genetics of antenna versus leg fate specification in arthropods was discovered in the fruit fly Drosophila melanogaster and involves the regulation by leg gap genes and Hox genes. Broadly overlapping domains of leg gap genes homothorax (hth), dachshund (dac) and Distal-less (Dll) in the antenna promote antennal fate through activation of downstream antennal determinants such as spineless (Casares & Mann, 1998; Dong, Dicks, & Panganiban, 2002; Dong, Chu, & Panganiban, 2001; D. M. Duncan, Burgess, & Duncan, 1998; reviewed by Setton et al., 2017); leg identity in thoracic appendages is mediated by mutually antagonistic hth (proximal), dac (median) and Dll (distal) domains, and repression of hth by the Hox gene Antennapedia (Antp) (Casares & Mann, 1998; Struhl, 1981; 1982). Ectopic expression of hth on the leg discs results in leg-to-antenna transformation, whereas ectopic Antp in the antennal disc results in antenna-to-leg transformation (Casares & Mann, 1998; Struhl, 1981; 1982). In chelicerates, similar to D. melanogaster, the chelicera (antennal positional homolog) has broadly overlapping domains of Dll and hth; legs and pedipalps lack a distal hth, and a proximal Dll expression domain, indicating a conservation of expression of positional homologs (Pechmann & Prpic, 2009; Prpic, Janssen, Wigand, Developmental, 2003, n.d.; Prpic & Damen, 2004; Sharma, Schwager, Extavour, & Giribet, 2012a). In line with this evidence, knockdown of hth in the harvestmen Phalangium opilio results in chelicera-to-leg transformation (and also partial pedipalp-to-leg transformation) (Sharma et al., 2015b). In contrast to D. melanogaster and other insects, arachnid Antp is not expressed on the leg-bearing segments, and knockdown of this gene in the spider Parasteatoda tepidariorum has revealed a role in repressing limbs in the opisthosoma (Khadjeh et al., 2012). Expression of dac in a mid-domain of chelicerate pedipalps and legs suggests a conserved role in mid segment patterning, and, accordingly, knockdown of dac in the harvestmen P. opilio and the spider P. tepidariorum results in defects on the mid-leg segments (Sharma et al., 2013; Turetzek et al., 2015). Additionally, loss of a dac domain of the chelicera of spiders in comparison to harvestman has been implicated in the transition from three-to two-segmented chelicera (Sharma et al., 2013; Sharma, Schwager, Extavour, & Giribet, 2012a).
Towards revitalizing embryological studies in Amblypygi, we generated herein protocols and comprehensive genomic resources for developmental genetic study of the whip spider Phrynus marginemaculatus, toward resuming the seminal studies begun by Peter Weygoldt with this promising model species. Leveraging a comprehensive developmental transcriptome for this species, and as a first step towards elucidating the patterning of antenniform legs, we characterized the expression pattern of the leg gap genes homothorax, dachshund and Distal-less in developing embryonic stages of P. marginemaculatus using newly established procedures for colorimetric whole mount in situ hybridization.
Methods
Establishment of amblypygid colony in the laboratory
Adult individuals of Phrynus marginemaculatus were purchased from a commercial supplier collects animals in the Florida Keys (USA). The maintenance of a colony of Phrynus marginemaculatus (Fig. 1A) in the laboratory followed guidelines provided is the original developmental works of Peter Weygoldt (Weygoldt, 1969; 1970; 1975) and notes on behavioral studies on this species (Fowler-Finn & Hebets, 2006; Rayor & Taylor, 2006; Santer & Hebets, 2009).While most species of Amblypygi have been reported as aggressive and better kept in separate containers (Weygoldt, 2000), individuals of P. marginemaculatus can be keep in a stable laboratory colony in communitarian large containers (Rayor & Taylor, 2006). A maximum of 20 animals can be allowed to interact and mate freely in plastic boxes (67.3 × 41.9 × 32.4 cm) with a 2 cm layer of damp coconut fiber saturated with distilled water. The substrate is continuously dampened with approximately 500 mL of dH2O on the substrate every week. All animals are keep in an acclimatized room at 26-28°C in a controlled reversed 16L:8D light cycle to simulate summer months in Florida. To prevent flies from reproducing in this humid environment, the lid of the container is provided with a 10 × 20 cm fine mesh cloth, which also allows for air circulation. Each container is furnished with pieces of tree bark, egg carton, and retreats made from polystyrene foam or florist’s foam, to provide enough space for animals to hide and suspend themselves during molting and egg-laying. After about two weeks of setting up a terrarium, the substrate hosted a stable colony of collembolans (possibly brought by the tree bark). The co-occurrence of collembolans does not appear to be disadvantageous in any way. Collembolans are invariably observed feasting on carcasses and likely contribute to preventing mite infestations (see also Weygoldt 2000). Animals are fed at least twice a week with nymphs of Acheta domestica ad libitum. Carcasses are removed as soon as detected to prevent the spread of mites. Freshly hatched praenymphs are kept with the female, and aggregate on the back of the brooding female (Fig. 1B). The next instar, the protonymph, disperse from the female upon molting. Protonymphs are transferred to communitarian terraria as described above. Protonymphs and subsequent instars until adulthood are fed pinheads of A. domestica and Gryllus bimaculatus ad libitum three times per week. The first adult individuals from our laboratory bred animals were detected 8 months after hatching, which accords to what has been described by Weygoldt (1975).
Terraria with adults are inspected daily for freshly molten individuals or ovigerous females, which are immediately separated from the main colony to avoid aggressive interactions. Ovigerous females carry the eggs outside the body in a brood sac attached to the ventral side of the opisthosoma (Weygoldt, 2000). Eggs are allowed to develop with the female until the desired stage for developmental work. Approximate stages may be inferred from days after egg laying (dAEL) at 26°C following the staging in Weygoldt (1975). For obtaining the eggs, females are anesthetized with CO2 and immobilized in customized plastic petri dishes (35 × 11 mm) with a central whole and foam (Fig. 1 C–D). With a blunt forceps pair, the pleura of the opisthosoma embracing the eggs may be gently displaced from the broodsac, and the brood sac removed as whole (Fig. 1E). The female may be returned to the colony unharmed. The brood sac easily dismantles when immersed in a phosphate buffer solution (PBS), and eggs may be cleaned with forceps under PBS. We obtained high mortality in keeping isolated eggs developing in petri dishes with humid (wet filter paper) or dry in 26°C incubators, as mold frequently develops. However, we found out that embryos will survive and develop immersed in PBS supplemented with kanamycin (1:5000) at 26°C incubator and even hatch into deutembryos (∼20 dAEL).
Phrynus marginemaculatus transcriptome assembly
We collected embryos of different embryonic stages as described above, and conducted total RNA extractions of each stage separately: hatched deutembryo (female #5, n=3; undetermined age, before eyespots), unhatched deutembryo (female #2 n=3; approx. 21 dAEL), and leg elongation stage (female #3, n=2; approx. 16-17 dAEL), late limb bud stage (female 6, n=1; 16 dAEL) and early limb bud stage (female 6, n=1; 14 dAEL) (voucher pictures are available upon request). Embryos were homogenized in Trizol TRIreagent (Invitrogen) and stored at –80°C until extraction. The combined total RNA of the five extractions was submitted to cDNA library preparation and sequencing in an Illumina High-Seq platform using paired-end short read sequencing strategy (2 × 100 bp) at the Center for Gene Expression, Biotechnology Center of University of Wisconsin-Madison (USA). The quality of the raw reads was accessed assessed with FastQC (Babraham Bioinformatics). De novo assembly of compiled paired-end reads was conducted with the software Trinity v. 2.6.6 (Grabherr et al., 2011), using Trimmomatic v. 0.36 (Bolger, Lohse, & Usadel, 2014) to removed low quality reads and adaptors. Completeness of the transcriptome was assessed with the Trinity script ‘TrinityStats.pl’, and BUSCO v. 3 (Waterhouse et al., 2017) on the longest isoforms using the database ‘Arthropoda’ and analysis of Hox genes. Whole transcriptome annotation was conducted following the Trinotate pipeline, by predicting proteins with TransDecoder v. 5.2.0 (Bryant et al., 2017) and searching best hits with blastx and blastp against Swiss-Prot and Pfam databases (The UniProt Consortium, 2018).
Gene identification
Hox genes (labial, proboscipedia, Hox3, Deformed, Sex combs reduced, fushi tarazu, Antennapedia, Ultrabithorax, abdominal A and Abdominal B), Distal-less, and dachshund orthologs were identified from the embryonic transcriptome of Phrynus marginemaculatus using BLAST searches (tblastn) with peptide queries of the homologs of Drosophila melanogaster and Parasteatoda tepidariorum. The best hits were reciprocally blasted to NCBI database and further assessed with a phylogenetic analysis.
For the phylogenetic analysis of Hox genes, we selected arthropod terminals with available genomes and well annotated references: the fruit fly Drosophila melanogaster (FlyBase), the beetle Tribolium castaneum (GCA_000002335.3) (Insecta), the centipede Strigamia maritima (GCA_000239455.1) (Myriapoda), the tick Ixodes scapularis (GCA_002892825.2), the mite Tetranychus urticae (GCA_000239435.1), the spider Parastetoda tepidarioum (GCA_000365465.2) and the scorpion Centruroides sculpturatus (GCA_000671375.2) (Chelicerata) (Chipman et al., 2014; Grbić et al., 2011; Gulia-Nuss et al., 2016; Herndon et al., 2020; Schwager et al., 2017) (Table S1). For the analysis of Distal-less, dachshund, and homothorax we selected terminals used in Nolan et al. (2020) (Nolan, Santibáñez López, & Sharma, 2020) and included the peptide sequence for P. marginemaculatus candidates (Table S1). Peptides were predicted with ORF Finder (NCBI) and TransDecoder v. 5.5.0 (Bryant et al., 2017) or obtained as CDS when possible from publicly available NCBI database. Peptide sequences were aligned using Clustal Omega (Sievers et al., 2011)in the software SeaView v. 4 (Gouy, Guindon, & Gascuel, 2010). Gene trees were inferred under maximum likelihood using IQTREE (Nguyen, Schmidt, Haeseler, & Minh, 2015)with automatic model selection (-m TEST) and 1000 ultrafast bootstrap resampling (-bb 1000).
Fixation and in situ hybridization
Collecting freshly laid eggs and keeping them in controlled temperature with the female or incubator (26°C) allows to select desirable stages for fixation. We fixed embryos for in situ hybridization with three methods: (1) intact eggs in a phase of 4% formaldehyde/PBS and heptane overnight, after an established spider fixation protocol (Akiyama-Oda & Oda, 2003); (2) piercing a hole in the vitelline membrane, and immersing eggs in 4% formaldehyde/PBS overnight; (3) dissecting the embryo out of the vitelline membrane and fixing in 4% formaldehyde/PBS for periods varying from 2–24h. All three procedures were conducted in a shaking platform at room temperature (∼22 °C). Following the fixation period, embryos were rinsed several times in 1x PBS/0.1% Tween-20 (Sigma Aldrich) (PBS-T), and gradually dehydrated into pure methanol. Embryos may then be stored in 20 °C freezer until next procedures.
cDNA was synthetized with SuperScriptIII kit (Thermo Fisher) following the manufacturer’s instructions, using the same total RNA used to generate the embryonic transcriptome. Riboprobe templates for the genes Pmar-dac1, Pmar-dac2, Pmar-hth1, and Pmar hth-2 in situ hybridizations were generated as follows: genes were PCR amplified from cDNA using gene-specific primers designed with Primer3 v. 4.1.0 (Koressaar & Remm, 2007) and complemented with T7 linker sequences (5′ -ggccgcgg-3′; 5′ - cccggggc-3′), and a second PCR using universal T7 primer to generate T7 polymerase templates specific for sense (control) and anti-sense (signal) probes (Table S2). Sense and antisense riboprobes for Pmar-Dll were generated with T7 and T3 polymerases using a plasmid template. For cloning of Pmar-Dll, gene-specific primers were used to amplify an 848 bp fragment (Table S2), which was incorporated to vectors and E. coli using TOPO TA cloning kit (Thermo Fisher) following the manufacturer’s instructions, and Sanger sequenced to confirm their identities.
For colorimetric in situ hybridization, we adapted a protocol for the spider Parasteatoda tepidariorum after Akiyama-Oda and Oda (2003). The staining reactions lasted between 2h and 8h at room temperature using nitro-blue tetrazolium (NBT) and 5-bromo-4-chloro-3′-indolyphosphate (BCIP) staining reactions (Sigma Aldrich). Embryos were then rinsed in PBS-T (0.05% Tween-20), counter-stained with 10µg/mL Hoechst 33342 (Sigma-Aldrich, St. Louis, MO, USA), post-fixed in 4% formaldehyde/PBS-T and stored at 4°C. Images were taken in a petri dish with agar under PBS-T, using a Nikon SMZ25 fluorescence stereomicroscope mounted with either a DS-Fi2 digital color camera (Nikon Elements software). Appendage mounts were imaged in 70% glycerol/PBS-T in a Zeiss LCM 710 confocal microscope.
Results
Overview of embryogenesis in Amblypygi
A characterization of the postembryonic development of P. marginemaculatus is available in the works of Weygoldt (Weygoldt, 1969; 1970). The following summary is after Weygoldt (1975), complemented with observations from the present study. We note that the timing of embryonic stages is approximate, as the temperature in Weygoldt (1975) was kept in a range of 24–27°C.
In our colony, a sample of 18 clutches from different females had an average of 19 eggs (min: 8 eggs; max: 39 eggs). Each egg is ovoid and measures 1.2–1.4 mm at the longest diameter. The yolk granules are light green, and fixation turns them light yellow. Upon secreting the brood sac and eggs into the ventral side of the opisthosoma, the brood sac is transparent, but gradually hardens and becomes dark brown in the course of two days. Around 6 dAEL, a blastoderm is formed. At 7–8 dAEL, the yolk is more condensed, a perivitelline space is formed, cells move into the yolk presumably in the region of the future blastopore. At 8–9 dAEL a blastopore is recognizable on the dorsal surface of the embryo, and a germ disc is formed. Towards 10 dAEL, the germ band starts to take form, the blastopore region is displaced posteriorly, and the first signs of segmentation of the prosoma are visible. A presumable cumulus, a group of migrating cells with dorso-ventral organizing properties, was reported by Weylgodt (1975) at this stage, but was not investigated by us. At 11-12 dAEL, the nascent prosomal segments elongate anteriorly and by 13-14 dAEL the segmented germ band occupies the ventral hemisphere of the egg. The limb buds of prosomal appendages begin to appear (Fig. 2A). The process of reversion begins, with a sagittal split of the germ band (Fig. 2A). From this point, until 19 dAEL, the ventral sulcus progressively enlarges with the dorsal displacement of both halves, and then begin to ventrally fuse again starting anteriorly (Fig. 2A–E). The head lobes on 15 dAEL are wing-shaped and uniform (Fig. 2 F). From 15–19 dAEL, ectodermal cells invaginate in the ventral midline and head lobes (neural precursor cells), which can be seen as evenly spaced dark dots (Fig. 2 F–H). Concomitantly, the anterior rim of the head lobes forms semi lunar groves (folds over itself) (Fig. 2 F–H). Opisthosomal segments 1–12 sequentially appear from the posterior growth zone (Fig. I–K). At 19 dAEL, the yolk moves into the caudal papilla (growth zone) and opisthosoma flexes ventrally (Fig. 2 K, N). The first pair of legs (future antenniform legs) is longer than the other appendages already on an early limb bud stage (15 dAEL), and more clearly on subsequent stages (Fig. L–N). A protuberance on the proximal end of the second walking leg pair (L2), the nascent lateral organ (osmoregulatory embryonic organ), is discernible at 16 dAEL and later stages, and becomes reniform (Fig. L–N). From 20–21 dAEL, the embryo begins to secrete a cuticle, which is darker on the fully formed lateral organ. The head lobes have completely folded onto themselves, and the distal segment of the chelicera is sclerotized. The embryo hatches into a deutembryo stage while still inside the brood sac (Fig. 1 F–H). Deutembryos remain in the brood sac for further 70 days without discernible external changes. The primordia of the eyes are formed towards the end of deutembryo development (70–106 dAEL), and eye spots of the median and lateral eyes become visible. From 90-106 dAEL the deutembryos molt into praenymphs, leave the brood sac and climb to their mom’s back. For a histological description of gastrulation, of the mesoderm, brain and nervous system, digestive system and gonad formation, we refer the reader to the exquisite original descriptions of Weygoldt (1975).
Phrynus marginemaculatus embryonic transcriptome
In order to provide a comprehensive resource for Amblypygi to facilitate developmental genetic study and comparative genomics, we assembled an embryonic transcriptome spanning different embryonic stages. The assembly of P. marginemaculatus transcriptome resulted in 544.372 transcripts composed of 277.432.101 bp and N50 of 440 bp. BUSCO scores, which reports the presence and number of copies of widely conserved single-copy genes (n=1066), indicated 94.2% completeness with 5.9% of genes duplicated. From the total number of transcripts (including isoforms), we predicted 37.540 peptide sequences with the Trinotate pipeline with a top blastp hit in Swiss-Prot database. A total of 801 of these peptides (2.1%) had a top hit to non-metazoan proteins, and presumed microbial contaminants.
To investigate the possibility of systemic gene duplication in Amblypygi and further test the transcriptome completeness, we searched for all ten Hox genes, emphasizing discovery of paralogs known from in the genomes of the spider Parasteatoda tepidariorum (except for fushi tarazu) and the scorpion Centruroides sculpturatus (except for Hox3) (Leite et al., 2018; Schwager et al., 2017; Sharma, Schwager, Extavour, & Wheeler, 2014b). One of the Hox3 copies of P. tepidariorum has no discernible expression pattern (Schwager et al., 2017) and had an unusually long branch in our analysis (Fig. S1). We therefore removed this copy from our analysis given its questionable annotation. We found orthologs of all ten Hox genes in Phrynus marginemcaulatus, namely labial, proboscipedia, Hox3, Deformed, Sex combs reduced, fushi tarazu, Antennapedia, Ultrabithorax, abdominal A and Abdominal B (Fig. 3A–B). We found evidence that all Hox genes, except proboscipedia are assembled in two Trinity genes and have amino acid differences (Fig. 3B). A Pmar-Dfd and Pmar-abdA were also found in a third Trinity gene, but their non-overlapping position on the alignment and nearly identic amino acid sequence to the respective paralog 2 indicate that they are fragmented assemblies of the same copy. These results suggest that Amblypygi share a duplicated Hox cluster with 19 Hox genes with spiders and scorpions, and that these copies are transcriptionally active during embryogenesis.
Gene expression analyses in Phrynus marginemaculatus
All three methods used here for fixing embryos of P. marginemaculatus yield embryos which are suitable for in situ hybridization (ISH). However, fixation of intact embryos in formaldehyde/heptane phase sometimes results in the vitelline membrane adhering to the embryos. Upon dissection of the vitelline membrane for ISH, parts of the embryo are commonly lost. Piercing a hole in the embryo for fixation in formaldehyde achieves a similar result, and is therefore not recommended. Dissecting vitelline membrane under PBS and immediately fixing the embryos in formaldehyde yielded the best results. We did not detect any difference in the signal-to-background ratio of colorimetric ISH between embryos fixed for 2 h or 24 h. To test gene expression assays and investigate the genetic patterning of antenniform legs, we selected the leg gap genes Distal-less, dachshund and homothorax, which constitute the subset of PD axis patterning genes associated with functional datasets in other arachnid model species (Schoppmeier & Damen, 2001; Sharma et al., 2013; 2015b).
Distal-less identification and expression
Similar to other arachnopulmonates and arthropods in general, a single Distal-less gene, Pmar-Dll, was found by similarity searches and its orthology to other metazoan Distal-less genes was confirmed with phylogenetic analysis (Fig. S2). Pmar-Dll is strongly expressed in all developing appendages, from early limb bud stage (15 dAEL) and is localized to the distal portion of the chelicera, pedipalp and legs throughout leg elongation (Fig. 4 A–I; Fig. 5A–L). In 17-18 dAEL embryos, the expression is heterogeneous and forms bands of stronger expression on the forming podomeres of the pedipalps and legs (Fig. 5 H–L). We did not detect expression on ventral midline (contra Sharma et al. 2012 and Mittmann and Scholtz 2001), or in the opisthosoma, and observed only diffuse expression on the head lobes of the stages surveyed. Expression on L1 of 16-17 dAEL embryos is subtly stronger on the proximal part of the expression domain (Fig. 5C–C’), but we did not detect clear differences in younger or older embryonic stages. The proximal and distal boundaries of strong Distal-less expression do not differ between L1 and the other legs, spanning two-thirds of the legs from the tip of the tarsus and excluding the most proximal third (Fig. 5C, 5I).
In contrast to other arachnid embryos (mygalomorph spiders (Pechmann & Prpic, 2009); araneomorph spiders (Prpic & Damen, 2004; Schoppmeier & Damen, 2001); Opiliones, Sharma et al. 2012; and mites (Barnett & Thomas, 2013b; Thomas & Telford, 1999)), Dll expression was not detected as a strong domain in the endite of pedipalp, which is very inconspicuous at these embryonic stages. We were not able to assess Dll expression in the pedipalpal endite in deutembryo stages due to deposition of cuticle.
dachshund identification and expression
dachshund is present as a single copy in Arthropoda, but is found independently duplicated in Arachnopulmonata and Xiphosura (Nolan et al., 2020; Turetzek et al., 2015). We found two Trinity genes with high similarity to Dmel-dac and confirmed their identity as dachshund paralogs with a phylogenetic analysis (Fig. S3). Pmar-dac1 and Pmar-dac2 are each nested in a clade with paralogs 1 and 2 of other arachnopulmonates, which appears as independent from horseshoe crab duplications (Fig. S3).
Pmar-dac1 is strongly expressed as broad mid band in the pedipalp and four pairs of legs (Fig. 6 A–F). This sharp mid band excludes the most proximal and the distal domains of these appendages (Fig. 7). No staining was detected on the chelicera (Fig. 6 A–F; 7 A, E). Additionally, expression occurs in three domains in the head lobes and strongly on the ventral midline of the prosoma. The continuous expression on the ventral midline stops before the opisthosoma (Fig. 6 A–F).
Pmar-dac2 is expressed as ring at the proximal region of chelicera, pedipalp and legs (Fig. 6 G–I). On 16-17 dAEL embryo a second thinner ring of expression is also present on the pedipalps and legs (Fig. 6 G–I; 7 I–L). Expression domains also occur on the neuromere of L1 and pedipalp, being larger on the pedipalp neuromere (Fig. 6 G). Three spot domains also occur on O1, O2 and O3 segments (Fig. 6 H–I).
homothorax identification and expression
Similar to the case with arachnid dachshund, homothorax is also found duplicated in arachnopulmonates (Leite et al., 2018; Turetzek et al., 2017). We annotated two Trinity genes in P. marginemaculatus transcriptome and the phylogenetic analysis shows that each copy is nested in two clades with the arachnopulmonate paralogs (Fig. S4).
Pmar-hth1 is conspicuously expressed on the developing appendages throughout the stages studied (15-18 dAEL; Fig. 8–9). On 16-17 dAEL embryos, Pmar-hth1 is expressed more strongly on the proximal parts of the chelicera, pedipalp, and legs, and is absent in a small distal domain (Fig. 9 A–D). On 17-18 dAEL, expression in all elongated appendages is similar, and spans most of the appendage length except for a small distal territory (Fig. 9 E-H). Additionally, three stronger stripe domains occur distal to the coxa of pedipalp and legs. Expression in L1 and other legs is similar in the studied stages, save for a slightly expanded distal domain on the proximal tarsus of L1 (Fig. 9 G–H). Expression on the ventral ectoderm and head lobes is weak and homogenous in early stages, and becomes stronger in 17-18 dAEL (not shown)
Pmar-hth2 is expressed in the developing appendages, but is reduced in intensity and different in pattern in comparison to Pmar-hth1 (Fig. 8 B–B’). Pmar-hth2 spans the whole chelicera (Fig. 9 I). Expression on the pedipalp spans most of the appendage, excluding only a distal territory (Fig. 9 J). In addition, there is a proximal and a sub terminal strong domain (Fig. 9 J). Expression on L1 and other legs is identical in all stages studies. Pmar-hth2 expression on the legs is consistently weaker in comparison to the pedipalp, but likewise possess a proximal and a distal sub-terminal domain (Fig. 9K–L). Expression on the ventral ectoderm, opisthosoma and head lobes is ubiquitous (Fig. 8B).
Discussion
Whip spiders retain arachnopulmonate gene duplications
Changes in Hox gene expression, regulation and targets are linked to changes in morphology, and Hox gene duplications have also been implicated in morphological innovations (Hughes & Kaufman, 2002; C. L. H. A. T. C. Kaufman, 2000; Ronshaugen, McGinnis, & McGinnis, 2002; Wagner, Amemiya, & Ruddle, 2003). Duplication in the Hox cluster is common in the vertebrates, which may have up to eight Hox clusters (teleost fishes) due to successive rounds of genome duplications (Wagner et al., 2003). On the other hand, genome and Hox cluster duplication is considered rare in Arthropoda (Pace, Grbić, & Nagy, 2016). Complete Hox cluster duplication in the phylum has only been reported in horseshoe crabs (Xiphosura) (Kenny et al., 2015; Nossa et al., 2014; Shingate et al., 2020; Zhou et al., 2020), and in spiders and scorpions (presumed ancestral in Arachnopulmonata). The phylogenetic placement of Amblypygi makes this group an opportune system to test the downstream prediction that P. marginemaculatus retains and expresses Hox paralogs shared by spiders and scorpions. In accordance with this hypothesis, we discovered that the whip spider P. marginemaculatus presents homologs of all ten Hox genes, and all excepting proboscipedia appear to be duplicated, as supported by our alignment and phylogenetic analysis.
A second line of evidence of a shared genome duplication event comes from conserved expression patterns of leg gap gene paralogs. It has recently been shown that duplicates of leg patterning genes extradenticle, optomotor blind, dachshund (dac) and homothorax (hth) present one copy with the arthropod plesiomorphic expression domain, and a second copy with a divergent expression shared between a spider and a scorpion (Nolan et al., 2020). For instance, dac1 of surveyed arachnopulmonates retains the plesiomorphic expression pattern of a broad mid band in the appendages, the condition observed in the single copy dac of all non-arachnopulmonate arachnids surveyed to date (Barnett & Thomas, 2013b; Sharma, Schwager, Extavour, & Giribet, 2012a), whereas dac2 is expressed as a proximal and a mid domain (Nolan et al., 2020; Pechmann & Prpic, 2009; Prpic & Damen, 2004; Turetzek et al., 2015). The same is true for homothorax: hth1 retains a plesiomorphic expression pattern on the proximal-mid appendage excluding the tarsus, whereas hth2 presents a strong short proximal domain and weakly striped pattern distally (Nolan et al., 2020; Pechmann & Prpic, 2009; Prpic & Damen, 2004; Turetzek et al., 2017). In the whip spider P. marginemaculatus, we indeed recovered two copies of dachshund and homothorax, with paralogs nested in two clades of arachnopulmonate copies with high nodal support (Fig. S3, S4). Moreover, Pmar-dac and Pmar-hth copies share with spider and scorpion the same pattern of expression of their paralogs: Pmar-dac1/2 and Pmar-hth1/2 conserve the plesiomorphic expression pattern of arachnids, while the copy 2 retain the arachnopulmonate-specific pattern. These data, together with the evidence from Hox gene duplications and the topologies of the dac and hth gene trees, provide compelling support for genome duplication as a synapomorphy of Arachnopulmonata.
A dachshund2 domain in the two-segmented clasp-knife chelicera of P. marginemaculatus
Expression of Pmar-Dll, Pmar-dac1/2 and Pmar-hth1/2 on the appendages is largely similar to spiders and scorpions (see above) (Nolan et al., 2020; Pechmann & Prpic, 2009; Prpic et al., n.d.; Prpic & Damen, 2004; Turetzek et al., 2015; 2017). A notable difference in the chelicera of the whip spider is the retention of proximal expression of one of the dachshund paralogs. In arachnids with three segmented chelicerae, such as daddy-long-legs, dac is expressed in the proximal segment of the appendage (Sharma, Schwager, Extavour, & Giribet, 2012a). The same is true for the three-segmented chelicera of scorpions, in which both dac1 and dac2 are expressed in the proximal segment (Nolan et al., 2020). In the two-segmented chelicera of spiders, dac1 is not expressed in the chelicera save for a non-ectodermal faint staining in late embryonic stage (Abzhanov & Kaufman, 2000; Pechmann & Prpic, 2009; Prpic et al., n.d.; Prpic & Damen, 2004; Turetzek et al., 2015). dac2 of spiders is also largely not expressed in the chelicera, save for small ventral domain at the base of the chelicera of Parasteatoda tepidariorum and Pholcus phalangioides (Turetzek et al., 2015). It has been proposed that the loss of a proximal dac domain is associated with the loss of the proximal segment in the chelicera of spider and other arachnids. This hypothesis is supported by functional data in the daddy-long-leg P. opilio: knockdown of the single-copy dac resulted in the loss of the proximal segment of the chelicera (as well as mid-segment truncations in pedipalps and legs) (Sharma et al., 2013). Similar to spiders, the two-segmented chelicera of the whip spider P. marginemaculatus does not express Pmar-dac1. Nonetheless, a clear complete ring of proximal expression was detected for Pmar-dac2.
In contrast to the classic leg gap phenotype incurred by knockdown of the ancestral dac copy in the harvestman, knockdown of dac2 in the spider P. tepidariorum resulted in the loss of the patella-tibia segment boundary in the pedipalp and walking legs; effects of this knockdown on the chelicera were not reported by the authors, nor is the function of dac1 known in P. tepidariorum (Turetzek et al. 2015). Given the interpretation that dac2 has a derived role in patterning a segment boundary in the spider, the incidence of the Pmar-dac2 stripe in the middle of the whip spider chelicera (and the absence of Pmar-dac1 in the chelicera) is tentatively hypothesized here to constitute evidence of a segment boundary patterning activity that separates the basal and distal cheliceral segment of Amblypygi (rather than evidence of a proximal segment identity in whip spiders that is homologous to the proximal cheliceral segment of groups like harvestmen and scorpions). Exploring this hypothesis in future will require the establishment of RNAinterference tools in P. marginemaculatus.
Antenniform leg distal allometry is specified early in development
To investigate a possible role of changes in leg gap genes expression in the patterning of the antenniform leg pair (L1) of whip spiders, we specifically searched for changes in expression domain between embryonic antenniform leg (L1) and the posterior legs (L2–4). The expression patterns of Pmar-Dll, Pmar-dac1/2 and Pmar-hth1/2 are mostly identical between the antenniform leg L1 and the posterior legs. We found only a slightly expanded Pmar-hth1 domain in the distal leg (compare Fig. G–H). It is unclear if this difference bears any functional significance in the differential fate of L1. Similar distal expression dynamics appear to distinguish the identity of the pedipalp and walking legs of the harvestman, as knockdown of the single copy of hth in P. opilio results in homeotic transformation of pedipalps to legs (Sharma et al., 2015b).
Besides the difference in sensory equipment, the antenniform leg pair of adults differ from the posterior walking legs in being much longer, and presenting numerous articles (pseudo-segments) on the tibia, metatarsus and especially the tarsus, whereas the walking legs have pseudo-segments only in the tarsus (Igelmund, 1987; Santer & Hebets, 2011; Weygoldt, 2000). Antenniform legs may be more than 2.5 times longer than the walking legs, and this allometry in adults is mostly accounted for by the length added by the distal leg segments, which may attain more than 100 distal articles (tibiomeres + tarsomeres) (Igelmund, 1987). Understanding the timing of the ontogenetic divergence of L1 relative to the walking legs is crucial for circumscribing target stages for investigation of the genetic patterning of this phenomenon. Current evidence indicates that this allometry on the antenniform legs takes place both embryonically and post embryonically.
The preaenymph, the stage that molts from the deutembryo and climbs to the mother’s back, has the full number of tibiomeres, but not of tarsomeres. The final number of tarsomeres is added in the next molt to the protonymph (Beck, Foelix, Gödeke, & Kaiser, 1977; Igelmund, 1987), and from this moment onwards the subsequent growth is isometric (Beck et al., 1977; Igelmund, 1987). In embryos of P. marginemaculatus, Weygoldt (1975) observed that the primordia of L1 are the first to appear in the germ band and are already longer than the remaining prosomal appendages, an observation paralleling the allometric growth of the L2 limb bud of the harvestman P. opilio (wherein L2 is the longest leg and also bears the most tarsomeres) (Sharma, Schwager, Extavour, & Giribet, 2012b). It is presently unclear at what point in development the distal podemeres of antenniform legs attain their allometry with respect to the proximal podomeres in the whip spider. In other words, are embryonic legs simply growing longer overall with respect to walking legs, or are they longer due to distal growth? The expression pattern dachshund in P. marginemaculatus early embryos reveals a further interesting aspect of this allometry. The expression of dac (including duplicates) in the legs of arachnids surveyed to date occurs on podomeres proximal to tibia, so that the tibia-metatarsus-tarsus territory may be identified as the a dac-free domain (Abzhanov & Kaufman, 2000; Barnett & Thomas, 2013b; Nolan et al., 2020; Pechmann & Prpic, 2009; Prpic & Damen, 2004; Sharma et al., 2013; Turetzek et al., 2015). The distal boundary of Pmar-dac1 an Pmar-dac2 in stages long before podomere formation reveals that the distal domain (Pmar-dac1 free cells) of L1 is already longer with respect to the proximal leg and the other legs (compare Fig. 7 C, D). These data support the inference that the distal allometry, and by extension the antenniform leg fate, are already specified from the onset of limb bud development.
Conclusion
The study of the embryology of Amblypygi has faced 45 years of hiatus since the seminal work describing the development of Phrynus marginemaculatus (Weygoldt, 1975). Here, we take the first steps in establishing P. marginemaculatus as a model for modern chelicerate evolutionary developmental study and for comparative genomics, with emphasis on discovering the genetic specification of the eponymous sensory legs. Future efforts to uncover the genetic underpinnings of this fascinating convergence with the mandibulate antenna should prioritize discovery of leg fate specification mechanisms in early whip spider development. Despite the establishment of a few promising arachnid models for studying chelicerate embryology, a pervasive challenge has been the establishment of toolkits for testing gene function (Akiyama-Oda & Oda, 2003; Khila & Grbić, 2007b; Schoppmeier & Damen, 2001; Sharma et al., 2013). Therefore, future efforts to study whip spider development must emphasize the development of functional for functional experiments in P. marginemaculatus.
Table S1: List of species, accession numbers and abbreviations of terminals used in the phylogenetic analysis of Hox genes, Distal-less, dachshund and homothorax homologs.
Table S2: List of primers for Pmar-Dll, Pmar-dac1/2, Pmar hth1/2, and cloned fragments for Pmar-Dll.
Acknowledgments
This work is dedicated to Peter Weygoldt, whose pioneer work in the development of Phrynus marginemaculatus and contributions to the study of Amblypygi biology continue to inspire the authors of this paper. Sequencing was performed at the University of Wisconsin-Madison BioTechnology Center. Microscopy was performed at the Newcomb Imaging Center, Department of Botany, University of Wisconsin-Madison. GG was supported by a Wisconsin Alumni Research Foundation Fall Research Competition award. Materials for study were supported by an Oscar Franke student competition award to GG from the International Society of Arachnology. This material is based on work supported by the National Science Foundation under grant IOS-1552610 to PPS.