The evolutionary origins and diversity of the neuromuscular system of paired appendages in batoids

Abstract

Appendage patterning and evolution have been active areas of inquiry for the past two centuries. While most work has centred on the skeleton, particularly that of amniotes, the evolutionary origins and molecular underpinnings of the neuromuscular diversity of fish appendages have remained enigmatic. The fundamental pattern of segmentation in amniotes, for example, is that all muscle precursors and spinal nerves enter either the paired appendages or body wall at the same spinal level. The condition in finned vertebrates is not understood. To address this gap in knowledge, we investigated the development of muscles and nerves in unpaired and paired fins of skates and compared them to those of chain catsharks. During skate and shark embryogenesis, cell populations of muscle precursors and associated spinal nerves at the same axial level contribute to both appendages and body wall, perhaps representing an ancestral condition of gnathostome appendicular neuromuscular systems. Remarkably in skates, this neuromuscular bifurcation as well as colinear Hox expression extend posteriorly to pattern a broad paired fin domain. In addition, we identified migratory muscle precursors (MMPs), which are known to develop into paired appendage muscles with Pax3 and Lbx1 gene expression, in the dorsal fins of skates. Our results suggest that muscles of paired fins have evolved via redeployment of the genetic programme of MMPs that were already involved in dorsal fin development. Appendicular neuromuscular systems most likely have emerged as side branches of body wall neuromusculature and have been modified to adapt to distinct aquatic and terrestrial habitats.

1. Background

Emergence and diversification of paired appendages are central to vertebrate evolution [1]. During evolution of paired appendages, skeletons, muscles and nerves have been assembled to support appendage movement, but the evolutionary trajectories and underlying genetic mechanisms of diverse appendicular neuromuscular patterning remain largely unknown [2]. Cartilaginous fishes, consisting of chimaeras, sharks, skates and rays, hold prominent phylogenetic positions in vertebrate evolution, representing primitive conditions of paired appendages [3,4]. In addition to their significance in evolutionary studies, cartilaginous fishes exhibit remarkably diverse paired fins. For example, skates, rays (batoids) and angel sharks have evolved extraordinarily broad paired fins [5]. To generate power for forward propulsion, batoids primarily rely on undulatory movement of wide pectoral fins [6]. This motion is achieved by a unique arrangement of the skeleton, muscles and nerves [7]. Despite their phylogenetically significant position and functional and evolutionary diversities, the knowledge about the developmental processes and mechanisms of appendicular neuromusculature in diverse cartilaginous fishes is limited [8].

In amniotes, hypaxial muscle precursors emerge from the ventrolateral dermomyotome of somites and develop ventral and appendicular musculature during embryogenesis [9,10]. These muscle precursors are categorized into subpopulations based on their positions along the anteroposterior axis [11]. In limb segments, hypaxial muscle precursors delaminate from the dermomyotome and migrate into the limb bud as migratory muscle precursors (MMPs). At interlimb levels, the ventral lip of the dermomyotome directly extends into the body wall forming body wall muscles [11]. Intriguingly, analysis of developmental processes of appendage and body wall muscles in chondrichthyans and other fish have identified differences to the condition in amniotes [12–15]. Whereas MMPs migrate and differentiate into appendage muscles in amniotes, appendage muscles of chondrichthyans have been argued to be derived from a direct extension of the ventrolateral dermomyotome into the pectoral fin [16,17]. It has also recently been shown that catsharks have delaminated MMPs entering into the pectoral fin, a seemingly comparable mechanism with that of amniotes [12]. Furthermore, in amniotes, hypaxial muscle precursors migrate exclusively either into the body wall or paired appendages [11]. In catsharks, hypaxial muscle precursors emigrate into both [2,12]. Lacking comparative data, particularly from cartilaginous fish, we currently do not know the phylogenetic polarity or significance of these changes to developmental patterning.

Genetic underpinnings of appendage and body wall muscle development have been revealed in mouse and chick embryos. At limb segments, MMPs express the homeodomain transcription factor Lbx1 (ladybird homeobox 1). The loss of Lbx1 results in a lack of appendage muscles and MMP migration [18–20], indicating that Lbx1 function is critical for the migratory ability of MMPs. By contrast, direct extension of the ventrolateral dermomyotome at the interlimb level takes place without expression of Lbx1 [11,21]. A previous study suggested that position-dependent expression of Lbx1 in hypaxial muscle precursors along the anteroposterior axis is determined by combinations of Hox genes in chicken embryos [22].

In vertebrates, concomitant with musculature development, motoneurons (MNs) originate from the ventral neural tube and extend to and innervate their target muscles. Coincident with Hox expression along the anteroposterior axis, MNs differentiate into lateral motor column neurons (LMCs) that innervate paired appendage muscles at the brachial and lumbar levels, hypaxial motor column neurons (HMCs) that innervate hypaxial muscles at the thoracic and sacral levels or medial motor column neurons (MMCs) that regulate epaxial muscles [23]. At brachial and lumbar levels, Hox6 and Hox10 genes induce high expression level of FoxP1, a transcription factor that promotes LMC differentiation via direct binding to its regulatory regions [24]. By contrast, at the thoracic level, Hoxc9 represses FoxP1 expression and induces development of the HMCs [25]. Intriguingly, little skates (Leucoraja erinacea) and chain catsharks (Scyliorhinus canicular) lost the HoxC cluster during evolution [26,27]. The loss of the HoxC cluster results in high FoxP1 expression, and the development of a continuous LMC between brachial and lumbar domains in skates [28]. Unexpectedly, a previous study identified a minor population of MNs that is similar to the HMC at pectoral and pelvic domains with LMC in the neural tube of skates [28], implying that skates possess unique patterns of innervation at the brachial and lumbar level compared with other finned vertebrates. However, developmental patterns of peripheral nerves in skates and other rays have not yet been sufficiently investigated.

To understand how batoid neuromuscular systems arose, we investigated their developmental patterns in skates (L. erinacea) and compared them with chain catsharks (Scyliorhinus retifer) using whole-mount in situ hybridization and antibody staining. These data provide molecular insights into the evolutionary mechanisms behind the generation of diverse neuromuscular systems as well as their evolutionary origins.

2. Material and methods

(a). Animal husbandry

All animal experiments were approved by the animal committees of Rutgers University (protocol no. 201702646) and the University of Chicago (protocol no. 71033). Leucoraja erinacea and S. retifer embryos were purchased from the Marine Resource Center of The Marine Biological Laboratory. Embryos were fixed by 4% paraformaldehyde (PFA) or Bouin's solution and subjected to immunostaining, histology or in situ hybridization. Stages were determined by referring to previously published studies [29–31].

(b). Characterization of orthologue genes

Cloning and characterization of Wnt3 and Fgf8 were described in the previous study [32]. Lbx1, Pax3, Cyp26a1 and Hox genes used in this study were cloned from cDNA of stage 23 and 30 skate embryos by reverse-transcriptase PCR and integrated into pCRII-TOPO vector (Invitrogen). PCR cloning primers are listed in electronic supplementary material, table S1. After cloning, each sequence was compared with the previously annotated skate transcriptome [32].

(c). Whole-mount immunostaining and in situ hybridization in chondrichthyans

Whole-mount in situ hybridization was performed as previously described [32]. The details of the protocol and replicate numbers are given as electronic supplementary material.

(d). Paraffin sectioning and haematoxylin and eosin staining

Two shark and two skate embryos were fixed in Bouin's solution overnight at room temperature and then washed with 70% ethanol, followed by 100% ethanol. Paraffin sectioning of fixed embryos (8 µm) and haematoxylin and eosin staining were performed by the Human Tissue Resource Center at the University of Chicago (https://pathcore.bsd.uchicago.edu/index.php).

(e). Cryosectioning of skate embryos

After whole-mount in situ hybridization, embryos were re-fixed in 4% PFA and immersed in a graded series of sucrose/PBS solutions (10%, 15%, 20%). Embryos were placed in OCT compound (Tissue-Tek) overnight. The following day, embryos were embedded and sectioned (8 µm) by Leica CM3050S.

(f). Phalloidin staining

Skate embryos were recovered from egg cases and fixed in 4% PFA overnight. The following day, embryos were treated in a sucrose/PBS solution series, embedded and cryo-sectioned as described above. Cryosections of four skate embryos were rinsed in PBTriton three times and incubated with 10% sheep serum/PBTriton for 30 min at room temperature. Sections were washed by PBTriton three times and incubated with 1 : 1000 phalloidin–Alexa488 (Invitrogen) and 1 : 4000 DAPI in PBTriton for 1 h. After sections were washed three times in PBTriton, fluorescent images were captured on a Zeiss LSM510.

(g). Reconstruction of innervation in shark and skate embryos

Haematoxylin and eosin-stained sections were photographed with a Leica M205 FCA. Images were imported into Amira software (ThermoFisher). Image direction was aligned, and maximum-intensity projection images were created. Nerves innervating the pectoral fin and body wall muscles were segmented manually and pseudo-coloured. For creating maximum-intensity projection images of nerves in retinoic acid (RA)-treated embryos (figure 3u,v), embryos stained by 3A10 immunostaining were cryo-sectioned and photographed without haematoxylin and eosin staining.

(h). Investigation of Hox expression patterns in NT, PAM and LPM

After whole-mount in situ hybridization, expression domains of Hox genes were investigated under a stereomicroscope and photographed (Leica M205 FCA and MC170 HD). To determine the anterior limit of expression of Hox genes in NT and LPM, the number of somites lateral to these tissues was counted. Expression at ventral somites was used to identify anterior limits of Hox expression in PAM.

(i). Culture of skate embryos with retinoic acid

Five skate embryos were cultured from stage 23 to 30 in 900 ml of salt water (Instant Ocean) with/without all-trans RA (Sigma; final concentration of 2 × 10⁻⁶ M). After the culture, embryos were fixed with 4% PFA and immunostained with 3A10 antibody. We repeated this experiment twice, investigating 10 embryos for each negative control and RA treatment. Furthermore, we counted observable somites number adjacent to the pectoral and pelvic fins in negative control (2 embryos) and RA-treated embryos (5 embryos). The data are summarized in electronic supplementary material, figure S3.

3. Results

(a). Dual contribution of muscles into the body wall and pectoral fin

To understand the developmental mechanisms of cephalic and appendicular muscles in batoids, we performed whole-mount in situ hybridization for Pax3 and Lbx1 in skate embryos. Pax3 was expressed in myoblasts of the pectoral fin and body wall (figure 1a,b). Subsequent sectioning of stained embryos confirmed Pax3 expression in the dermomyotome, myoblasts of the pectoral fin and ventrally extended body wall myoblasts (electronic supplementary material, figure S1). This dual contribution pattern of Pax3 expression in the pectoral fins and body wall is similar to that of shark embryos [12], although it was extended further posteriorly in skate fins compared to shark fins. Similarly, Lbx1 expression was observed in myoblasts of the pectoral fin, but not in the body wall, unlike Pax3 expression (figure 1c). This suggests that the myoblast population derived from the dermomyotome separates into fin and body wall muscles with and without Lbx1 expression, as MMPs and non-MMPs, respectively.

Figure 1.

Musculature development in paired fins of skate. (a–c) Whole-mount in situ hybridization of Pax3 and Lbx1. (a,b) Dorsal and ventral views of the expression patterns of Pax3 in the pectoral fin. Pax3 is expressed in both muscles of the pectoral fin (arrow) and dermomyotome (arrowhead) (a), or the pectoral fin (arrow) and body wall muscles (arrowhead) (b). White arrow points to the umbilical cord (b). (c) Ventral view of Lbx1 expression pattern in the pectoral fin. Lbx1 is expressed only in the pectoral fin (arrow) and not in the body wall muscles (arrowhead). (d–o) Immunostaining of skate embryos by myosin heavy chain antibody at stage 29 (d–i) or stage 30 (j–o) in lateral (d–f, j–l) or ventral (g–i, m–o) view. At stage 29, the constrictor branchialis, cucullaris and other cephalic muscles develop (d,g). Abductor and adductor muscles start to develop in the pectoral (e,h) and pelvic fins (f,i), yet they do not fully develop distally. Note that the abductor and adductor muscles of the pectoral fin and hypaxial muscles of the body wall develop at the same axial level (e,h). At stage 30, cephalic muscles are more developed compared to stage 29. Cucullaris extends dorsal to branchial arches (j). Interhyoideus and coracomandibularis are clearly identified (m). Abductor and adductor muscles in the pectoral fins (k,n) and pelvic fins (l,o) develop towards the distal direction. abd.m., abductor muscles; add.m., adductor muscles; a.m., adductor mandibulae; c.a. + c.b., a complex of coraco arcualis and coracobranchialis; c.h.d., constrictor hyoideus dorsalis; c.m., coracomandibularis; e.m., epaxial muscles; h.m., hypaxial muscles; i.h. + i.m., a complex of interhyoideus and intermandibularis; l.h., levator hyomandibulae. All scale bars are 0.5 mm. (Online version in colour.)

We next performed whole-mount immunostaining of skate embryos using myosin heavy chain antibodies. Basic components of cephalic and appendicular muscles were observed at stages 29 and 30 of embryonic development (figure 1d–o). The constrictor branchialis, which is necessary for gill movement, developed in the pharyngeal arches (figure 1d). The cucullaris, which articulates the skull and pectoral girdle, was formed dorsal to the pharyngeal arches and lateral to epaxial muscles (figure 1d,e). Hypobranchial muscles developed at the ventral gill region (figure 1g). In the pectoral fin, adductor and abductor muscles were observed at the dorsal and ventral sides, respectively (figure 1e,h). Hypaxial body wall muscles developed posteriorly from the base of the umbilical cord and backward, consistent with Pax3 expression patterns (figure 1e,h,k,n). This embryonic pattern is distinct from tetrapod musculature, in which the early development of appendage and body wall muscles occurs exclusively along the A–P axis. Adductor and abductor muscles were formed in the pelvic fin as well, but staining of myosin heavy chain was weaker than those of the pectoral fin at stage 29 (figure 1f,i). At stage 30, abductor and adductor muscle staining was stronger and extended to the distal tip of pectoral and pelvic fins compared to stage 29 (figure 1k,n,l,o).

(b). Dual innervation of spinal nerves into the body wall and pectoral fins

A previous study reported that the localization of LMC neurons expands posteriorly in the neural tube of skates [28]. However, the peripheral innervation pattern of batoid spinal nerves during embryogenesis, particularly a topological relationship of LMC and HMC along the A–P axis, has not been sufficiently described. To investigate developmental patterns of skate nerves, we used a 3A10 antibody that recognizes neurofilament-associated proteins. In contrast with the contribution of Spinal (Sp) nerves 5–8 into the pectoral appendage in mice [33], Sp nerves 1–32 innervated into the pectoral fin muscles in skate (figure 2a). Particularly, Sp nerves 1–10 developed the brachial plexus and innervated anterior pectoral fin muscles (figure 2a–c). In the posterior part of the pectoral fin, Sp nerves bifurcated at the base of the pectoral fin; one branch innervated body wall muscles (ventral branch of Sp nerves) and the other branch entered pectoral fin muscles at the same axial level (pectoral nerve) (figure 2d). After entering into the pectoral fin, pectoral nerves branched into dorsal and ventral rami for adductor and abductor muscles, respectively. In contrast with the previous study, however, we did not confirm a contribution of Occipital (Oc) nerves into pectoral fin muscles. Similar to pectoral nerves, Sp nerves 33–48 branched and innervated into the pelvic fins and body wall (figure 2a,e).

Figure 2.

Developmental pattern of nerves in chondrichthyan paired fins. Immunostaining of nerves in developing embryos of skate (L. erinacea) and shark (S. retifer) with 3A10 antibody. (a–e) Skate embryos at stage 29. (a) Dorsal view shows the innervation patterns for the pectoral and pelvic fins. The brachial plexus consists of Sp nerves. (b) Dorsal (sensory) and ventral (motor) root of spinal nerves for paired appendages. (c) Lateral view of the branchial arch domain. Occipital nerves do not contribute to the brachial plexus; only spinal nerves form the brachial plexus. The hypoglossal nerve and the pectoral fin nerves were observed. (d) Dorsolateral view of the pectoral fin region. Spinal nerves first branch into the pectoral fin laterally (arrowhead) and the body wall muscle ventrally (arrow) (shown in inset). Then, a second branching event occurred and the pectoral fin nerves innervate the dorsal and ventral muscles of the pectoral fin. (e) Dorsolateral view of the pelvic fin region. Inset shows the branching event where the spinal nerves split and innervate the pelvic fin (arrowhead) and body wall (arrow) muscles. (f–g) Skate embryos at stage 31. (f) Dorsal view of the pectoral fin region. Sp nerves directly innervate the posterior pectoral fin without forming the brachial plexus. (g) Ventral view of the pectoral fin region showing the brachial plexus and a second plexus-like structure (*) innervating the pectoral fin. (h) Three-dimensional reconstruction of neuromuscular systems in skate embryos at stage 29. The ventral roots of Sp nerves (green) exit from the ventral neural tube and innervate into dorsal and ventral pectoral fin (orange) as well as body wall muscles. (i,j) Shark embryos at stage 28. (i) Dorsal view of the pectoral region. Inset shows the branching event as the Sp nerves split and innervate the pectoral fin (arrowhead) and the body wall (arrow) muscles. Distally, the pectoral fin nerves branch dorsoventrally. (j) Ventral view of the pectoral fin. The v.b.s. is observed in the body wall. (k) Reconstruction of neuromuscular systems in shark embryos at stage 29. The ventral roots of Sp nerves (green) innervate into dorsal and ventral pectoral fin (red) as well as body wall muscles (yellow). a.b., abductor muscle of the pectoral fin; a.d., adductor muscle of the pectoral fin; b.m., body wall muscle; b.p., brachial plexus; d.b.p., dorsal branch of the pectoral nerve; d.g., dorsal root ganglion; d.r., dorsal root; h.g.n., hypoglossal nerve; v.b., ventral branch of spinal nerves; n.t., neural tube; p.f.n., pectoral fin nerves; sp., spinal nerves; v.b.p., ventral branch of the pectoral nerve; v.b.s., ventral branch of the spinal nerve; v.r., ventral root. All scale bars are 1 mm. (Online version in colour.)

At stage 31, Sp nerves established a brachial plexus-like structure in addition to the genuine brachial plexus of Sp nerves 1–10 at the middle of the pectoral fin. Dorsal and ventral fin muscles were also innervated by Sp nerves at this stage (figure 2f,g). The plexus-like nerve bundle may be indispensable to innervate into the pectoral fin through the pectoral girdle. The posterior pectoral fin was innervated by Sp nerves that branched at the base of the fins and extended into the body wall and pectoral fin without forming plexus structures.

(c). Conserved developmental pattern of muscles and nerves in chondrichthyans

Our results showed that the myoblast population and spinal nerves contribute into both the pectoral fin and body wall at the same vertebral level during skate embryogenesis (figures 1 and 2). To further compare developmental patterns of pectoral fin muscles and nerves of skates with those of sharks, we used myosin heavy chain antibodies to stain developing muscles and nerves in chain catshark embryos (S. retifer). At stage 30, staining confirmed abductor and adductor muscles in the pectoral fin (electronic supplementary material, figure S2). In the trunk, hypaxial body wall muscles developed at the same axial level as the pectoral fin muscles (electronic supplementary material, figure S2). In addition, Sp nerves 7–16 bifurcated into the body wall and pectoral fin without forming a plexus structure at stage 28 (figure 2i,j).

To confirm peripheral structures of Sp nerves, we stained serial sections of skate and shark embryos from stages 29 to 32 with haematoxylin and eosin solutions. After photographing serial sections, we produced maximum-intensity projections of stained sections and segmented muscles and nerves using Amira software (Material and methods). The segmentations of muscles and nerves in skate and shark embryos showed that Sp nerves that originated from the ventral part of the neural tube innervated both pectoral fin and body wall muscles peripherally, indicating that they contain motor nerves that regulate movement of body wall and pectoral fin muscles (figure 2h,k). Collectively, these results suggest that skates and sharks retain comparable developmental patterns of neuromuscular systems, in which muscles and nerves contribute into both the paired appendages and body wall at the same axial level, and batoids have posteriorly extended this dual contribution compared with sharks, supporting movement of their exceptionally wide fins.

(d). Dynamic rearrangements of Hox expression patterns in skates

Previous studies have revealed evolutionary diversity of MNs for paired appendages of elephant sharks, skates and zebrafish [28,34]. However, the molecular mechanisms that coordinate evolution of appendage muscles, nerves and skeletons remains unknown. Particularly in skates, development of the pectoral fin initiates from a strikingly wide fin bud that derives from the lateral plate mesoderm (LPM) [29,32]. Subsequently, muscles and nerves enter the enlarged pectoral fin bud from wider domains of the paraxial mesoderm (PAM) and neural tube along the anteroposterior axis compared with shark embryos. This observation led us to hypothesize that vertebrates have mechanisms that coordinate development of muscles, nerves and skeletons to support appendage movement.

To understand the genetic mechanisms involved in coevolution of muscles, nerves and skeletons in skate, we investigated gene expression patterns of Fgf8 (fibroblast growth factor 8), Wnt3, Cyp26a1 (cytochrome P450 26a1) and Cdx2, which provide anteroposterior positional information during gastrulation and specify prospective limb regions [35–37]. Expression patterns of these genes at stage 23 did not differ remarkably from those of other vertebrates at the comparable stage (figure 3a–d). To further explore the genetic mechanisms for coevolution, we tested expression patterns of Hox genes that provide positional information along the anteroposterior axis during gastrulation [39,40] as the downstream targets of Fgf8, Wnt3, Cyp26a1 and Cdx2 in skates. Consistent with Hox expression patterns of catsharks [38], Hoxa2, Hoxa3, Hoxa4, Hoxd1, Hoxd3 and Hoxd4 were expressed from the rhombomeres to the caudal tip of neural tube (figure 3e,f,g,m,n,o,q). However, comparison of the anterior limit of Hoxa9, Hoxa10, Hoxa11 and Hoxd8 expression showed that their expression domains shifted posteriorly in the neural tube of skates compared to catsharks [38] (figure 3i,j,k,p,q, Material and methods). In the PAM of skate embryos, expression domains of Hoxa9, Hoxa10 and Hoxa11 also shifted posteriorly from that of sharks (figure 3i,j,k,q). Furthermore, in skate LPM, Hoxa4, Hoxd4 and Hoxa5, which are capable of inducing Tbx5 expression via direct binding to regulatory regions in mice [41], exhibited broader expression domains than in sharks and other tetrapods (figure 3g,h,o,q). These results show that dynamic rearrangements of Hox expression patterns have occurred in the neural tube, PAM and LPM of skates.

Figure 3.

Expression pattern of Hox genes during skate gastrulation. (a–p) Whole-mount in situ hybridization of Fgf8, Wnt3, Cyp26a1, Cdx2 and Hox groups A and D in L. erinacea at stage 23. Note that Hox genes show colinear expression along the anteroposterior axis. In inset h, the black arrow, the white arrowhead and the black arrowhead point to the anterior limits of the expression in the neural tube, PAM and LPM, respectively. All scale bars are 1 mm. The photos are scaled except for E, G, N and P, which are scaled with each other. (q) Schematic summary of Hox expression patterns in L. erinacea in the neural tube (red), pharyngeal arches (yellow), PAM (green) and LPM (blue). Expression levels of Hox genes are indicated by colour darkness. The darkest bars show the anterior limit of expression of each Hox gene in S. retifer embryos [38], indicating that expression patterns of Hoxa9, a10, a11 and d8 have shifted posteriorly in L. erinacea. The posterior limits of Hox expressions in S. retifer embryos are not available in previous studies. Anterior limits of each gene in the neural tube (NT), LPM and PAM were determined by extending the anterior border of the adjacent somite (Material and methods). (r) Innervation staining of an RA-L. erinacea embryo. Note the overlap of the pectoral and pelvic fins (arrow). (s,t) Innervation staining of L. erinacea embryos cultured with retinoic acid, resulting in narrower pectoral fin size than RA embryos and creating a thoracic region between the pectoral and pelvic fins (bracket) (t). (u,v) Maximum-intensity projections of sections of innervation staining after RA treatment. At the pectoral fin level, spinal nerves exclusively innervate the pectoral fin in embryos treated by retinoic acid (p.f.n.), whereas they exclusively enter the body wall at the newly created thoracic domain (b.w.n). b.w.n., body wall nerve; p.f.n., pectoral fin nerve; v.r., ventral nerve. All scale bars are 1 mm. (Online version in colour.)

To test the effects of rearrangements of Hox expression patterns on the development of pectoral fins in skates, we cultured skate embryos with RA, which alters Hox expression patterns [42], from stage 23 (before the limb bud induction) to stage 30 (the stage at which the pectoral fin expands along the A–P axis) (figure 3r–v). The pectoral fin develops in the LPM adjacent to somites 1–36 (or 37) and the pelvic fin from somites 37–50 in skate embryos (electronic supplementary material, figure S3). Pectoral fins of embryos cultured with RA were narrower than control embryos along the anteroposterior axis (the narrowest is from somites 1–28, 5/5 embryos; figure 3r,s) with the expression shift of Hoxa9 (electronic supplementary material, figure S4), creating a thoracic domain between the pectoral and pelvic fins. RA-treated embryos lost the brachial plexus (electronic supplementary material, figure S5) and the branching pattern of pectoral nerves–Sp nerves directly innervated the pectoral fin from the neural tube (figure 3t,u). Furthermore, Sp nerves innervated the body wall at the newly created thoracic domain (figure 3v). These results suggest that Hox expression is responsible for regulating the width of skate fins, plexus formation and innervation patterns.

(e). Migratory muscle precursors in dorsal fin development

While developmental processes of pectoral muscles and nerves have been investigated in multiple fish taxa [2,43], evolutionary origins of the developmental programmes for neuromuscular systems in paired appendages remain unexplored. Therefore, we investigated Pax3 expression that represents muscle precursor cells by whole-mount in situ hybridization. Pax3 is expressed weakly at the proximal base of first and second dorsal fins at stage 29 (figure 4a), indicating that muscle precursor cells are present in developing dorsal fins. Then, to test whether these Pax3-positive cells are MMPs, we investigated Lbx1 expression in skate dorsal fins. At stage 29, Lbx1 begins to be expressed in the proximal domains of first and second dorsal fins, which is similar to Pax3 expression (figure 4b). At stage 30, Lbx1 expression extended to the distal part of dorsal fins and was undetectable at stage 31 (figure 4c). Sectioning of embryos after whole-mount in situ hybridization showed that Lbx1-expressing cells resided beneath the epidermis of dorsal fins (figure 4d). Furthermore, actin staining by phalloidin showed that MMPs accumulate actin internally, which is a signature of differentiating muscle precursor cells (figure 4e,f). These results demonstrate that the dorsal fin muscles of skates originate from MMPs.

Figure 4.

MMP cells in dorsal fins. (a) Pax3 expression in the first and second dorsal fins at stage 29. The expression is confirmed at the base of the dorsal fins (arrows). (b,c) Lbx1 expression in the dorsal fins at stage 29 (b) and 30 (c). Lbx1 is highly expressed in the muscle precursor cells in the first and second dorsal fins at stage 29 and the expression extends in the distal direction at stage 30. (d) Section of the embryo stained by whole-mount in situ hybridization of Lbx1. Arrowheads indicate expression of Lbx1. (e,f) Section staining of a dorsal fin by phalloidin (actin; green) and DAPI (nucleus; blue). MMP cells are observed right under the epidermal tissues. f is a magnified image of e without overlay of DAPI. All scale bars are 1 mm. (Online version in colour.)

4. Discussion

Comparison of skates and chain catsharks highlights a conserved developmental pattern of neuromuscular systems—the dual contribution into the body wall and paired appendages at the same axial level. Intriguingly, the dual contribution of hypaxial muscle precursors into the body wall and pectoral fins has been previously reported in species of catsharks [12]. Furthermore, the branching pattern of spinal nerves into the pectoral fin and body wall was also described in the anatomical study of dogfish (Squalus acanthias), lungfish (Propterus dolloi) and teleosts [34,44–46]. The cumulative knowledge of appendage neuromuscular systems in diverse taxa implies that their extension into both the body wall and paired appendages at the same axial level, which is conserved in chondrichthyans, actinopterygians and sarcopterygians, while absent in agnathans, is a synapomorphy for gnathostome appendages (figure 5).

Figure 5.

Summary and hypothesis of neuromuscular evolution in appendages. Summary of neuromuscular evolution in appendages. In skates, sharks, zebrafish and lungfish, muscles and nerves branch into the body wall and paired appendages at the same axial level (blue in the tree represents animals that possess the dual contribution of neuromuscular systems). Amniotes lost this branching pattern, and neuromuscular components exclusively contribute to either the body wall or the paired appendages during embryonic development. In skates, the dual contribution of neuromuscular components is extended posteriorly to support undulatory swimming. Note that skate dorsal fins also show Lbx1 expression in their MMP cells. Lbx1 expression in unpaired fins raises the possibility that genetic networks of MMPs may have been deployed from unpaired to paired fins during evolution. Phylogenetic tests of Lbx1 expression in other species are critical to conclude the evolutionary history of Lbx1. (Online version in colour.)

In contrast with ancestral dual contribution of the neuromuscular systems in fish, tetrapods do not possess similar patterns at the brachial segments (figure 5). This loss may be related to appendage narrowing; the narrowing width of pectoral appendages during the fish-to-tetrapod transition [47,48] may have been associated with the loss of this feature. In mammals, rostral intercostal nerves (the second and third in humans) have the intercostobrachial branches, and the subclavius nerve branches from the brachial plexus, both of which extend into the pectoral and body wall domains [49]. These nerves may be a remnant of the branching nervous system from the primitive condition of vertebrates, although comparative analysis of their developmental patterns as well as underlying molecular mechanisms across diverse taxa is critical to conclude.

How has the pattern of dual contribution of neuromuscular systems been modified or lost during the evolution of limbed vertebrates? In chicken embryos, Hoxa4 is expressed in the LPM lateral to somites 3–7 and Hoxa5 is expressed in the LPM adjacent to somites 4–10, both of which are capable of inducing Tbx5 expression [41]. In skates, these Hox genes are broadly expressed in the LPM lateral to somites 7–50, which is significantly wider than other vertebrates (figure 3q). While LPM expression domains of Hox paralogous groups 4 and 5 in sharks are not precisely described, these dynamic changes of Hox expression patterns in skate LPM are most likely involved in evolving wide pectoral fins. The size alteration of paired fins with the expression shift of Hox gene in RA-treated embryos supports this idea. Intriguingly, RA-treated skate embryos showed a similar innervation pattern to tetrapods; spinal nerves innervated appendages exclusively at the pectoral fin level, whereas they contribute into the body wall at the newly created thoracic domain (figure 3u,v). Our data are consistent with the previous study, which showed inhibition of RA activity by disulphiram in chicken embryos before the limb bud development shifted the position of the limb bud posteriorly [50]. However, besides the regulation of appendage width, RA plays multiple roles in limb development, such as induction of the limb bud from the body trunk [35] and differentiation of LMC neurons in the spinal cord [51]. While our result supports the scenario that RA treatment changes Hox expression patterns which, in turn, regulate formation of the brachial plexus and innervation of the pectoral fin and body wall, the differentiation process of LMC and HMC neurons in RA-treated embryos would be tested in future studies. Hox expression is also indispensable for hindlimb development. Particularly, Hoxa9–11 genes likely determine hindlimb position in the LPM and vertebrae identity in the PAM [52,53]. In skate, Hoxa9, Hoxa10, Hoxa11 and Hoxd8 are shifted caudally compared with sharks in the neural tube, LPM and PAM (figure 3q). The posterior shift of these Hox genes probably caused extensive and concomitant remodelling of muscles, nerves and skeletons in the pelvic fin of skates. In the tetrapod lineage, the opposite trend, restricting expression domains of a certain set of Hox genes such as the Hox4 and Hox5 paralogous groups, might transform the neuromuscular pattern of dual contribution into an exclusive one.

The newly defined expression of Lbx1 in skate dorsal fins sheds light on the evolutionary origin of appendicular neuromuscular systems (figure 5). If genetic networks of MMPs were deployed from unpaired fins to paired fins [54], then MMPs with Lbx1 expression in dorsal fins are likely to reflect the ancestral condition of paired appendage muscles. After paired appendages had obtained muscles with basic genetic networks including Pax3 and Lbx1, they might have been elaborated by recruiting other signalling pathways such as HGF/MET [55]. Alternatively, genetic networks of paired appendage muscles may have been assembled de novo, by recruiting Pax3, Lbx1 and HGF step by step during their evolution. Further analysis of regulatory regions of Lbx1 in multiple species including agnathans, which have only unpaired fins, could pursue this hypothesis. Our current analysis of neuromuscular development in chondrichthyans illuminates the evolutionary origins and diversification mechanisms of neuromuscular systems of paired appendages.

Data accessibility

Gene sequences of little skate used for cloning in situ hybridization probes in this study are previously published and available at the National Center for Biotechnology Information Sequence Read Archives (NCBI SRA), www.ncbi.nlm.nih.gov/sra (BioProject accession code no. PRJNA288370).

Authors' contributions

N.T., D.M., N.A., N.H.S. and T.N. designed the research; N.T., D.M., K.B., K.F. and T.N. performed the research; N.T., D.M., K.B., N.A. and T.N. analysed data; and N.T., D.M., K.B., G.S., K.F., N.A., N.H.S. and T.N. wrote the paper.

Competing interests

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

This work was supported by institutional support provided by the Rutgers University School of Arts and Sciences and the Human Genetics Institute of New Jersey; a Marine Biological Laboratory research grant (to T.N.); and the Brinson Foundation and University of Chicago Biological Sciences Division (to N.H.S.).