Reorganization of sea urchin gene regulatory networks at least 268 million years ago as revealed by oldest fossil cidaroid echinoid

Echinoids, or sea urchins, are rare in the Palaeozoic fossil record, and thus the details regarding the early diversification of crown group echinoids are unclear. Here we report on the earliest probable crown group echinoid from the fossil record, recovered from Permian (Roadian-Capitanian) rocks of west Texas, which has important implications for the timing of the divergence of crown group echinoids. The presence of apophyses and rigidly sutured interambulacral areas with two columns of plates indicates this species is a cidaroid echinoid. The species, Eotiaris guadalupensis, n. sp. is therefore the earliest stem group cidaroid. The occurrence of this species in Roadian strata pushes back the divergence of cidaroids and euechinoids, the clades that comprise all living echinoids, to at least 268.8 Ma, ten million years older than the previously oldest known cidaroid. Furthermore, the genomic regulation of development in echinoids is amongst the best known, and this new species informs the timing of large-scale reorganization in echinoid gene regulatory networks that occurred at the cidaroid-euechinoid divergence, indicating that these changes took place by the Roadian stage of the Permian.

gate is a regulatory circuit wiring design that is key to the specification of skeletogenic mesenchyme in euechinoids and the use of which in skeletogenesis is probably peculiar to this clade.
Echinoids are important and common constituents of modern ecosystems [13][14][15] . Though they have a diverse and storied history ranging back more than 400 myr to the Ordovician 16 , echinoids do not become abundant in the fossil record until 200 myr later in the Mesozoic 1,2 . Echinoids radiated in the Mesozoic after undergoing a bottleneck at the Permo-Triassic mass extinction (252 Ma) where they experienced a severe reduction in diversity 17,18 . The euechinoidea and cidaroidea clearly diverged before this mass extinction at the end of the Permian 19 , though the details of the timing of this divergence are not well constrained due to the rarity of echinoids in Palaeozoic strata. Apart from disarticulated spines, echinoids in the Palaeozoic are exceedingly rare. Most Palaeozoic echinoids had poor preservation potential compared to post-Palaeozoic forms, with many clades displaying imbricate, overlapping, plating which presumably lacked stereomic interlocking 20,21 . Because of this non-rigid test plating, Palaeozoic echinoids presumably disarticulated rapidly following their death, and thus well-preserved specimens in the Palaeozoic are usually limited to Lagerstätte deposits 22 . The stem-group cidaroid herein described from the Guadalupian of Texas, Eotiaris guadalupensis n. sp., is the earliest putative crown group echinoid known in the fossil record, and as such, provides new insight to the timing of the divergence of the euechinoids and cidaroids, which must have preceded it, and the associated morphologic and developmental gene regulatory changes that are the basis for this divergence.

Stratigraphy and Geologic Setting
All new specimens of Eotiaris guadalupensis n. sp. are known from the Lamar Member of the Bell Canyon Formation in the Guadalupe Mountains of west Texas ( Supplementary Fig. S1). Specimens described by Kier 23,24 are from the Word and Road Canyon Formations of the Glass Mountains of west Texas ( Supplementary Fig. S1). The Lamar Limestone is Lower Capitanian, about 264-263 Ma and the Road Canyon Formation is, at its youngest, 268.8 Ma. Stratigraphy and geologic setting is herein treated for only newly described material and detailed stratigraphic and locality information are in supplementary information. Diagnosis-Eotiaris with straight, clavate and bulbous spines covered in numerous spinules arranged helically around the shaft.

Systematic Palaeontology
Derivation of name-guadalupensis from the Guadalupe Mountains of west Texas, from where the type material was collected.
Description-Test regular and small, known only from disarticulated interambulacral columns. Columns range in width from 4.2 mm to 9.3 mm (Fig. 1A,B,D  about 9.4 mm to 20.6 mm. Apical system unknown, and adapical interambulacral plates are not preserved articulated to the interambulacral columns of the test. Adapical interambulacral plates likely imbricate whereas ambital and adoral interambulacral plates rigidly sutured (Fig. 1E). Peristomial plates unknown, however apophyses are present on most oral interambulacral plates (Fig. 1G,H). No buccal notches present.
Lantern and teeth unknown. Ambulacra unknown, although likely beveling under interambulacral plates as interior adradial interambulacral plate edges are denticulate.
Interambulacral plating arranged into two rows. First four to six plates usually rigidly sutured with more adapical plates disarticulated (Fig. 1D,E). Plates pentagonal, about 1.3 to 1.6 times as wide as high. Primary tubercles large, sunken, and confluent below ambitus (Fig. 1A,E). Areoles at ambitus on specimen USNM 610601 about 2.6 mm wide and 2.6 mm high. Boss crenulate with mamellons undercut and perforate. At ambitus, one row of secondary tubercles on each plate separates tubercles. Above ambitus, multiple rows of secondary tubercles separate ambitus on large specimens. On large specimens, about four rows of secondary tubercles between the edge of each tubercle and the perradial suture at ambitus (Fig. 1E). About three rows of secondary tubercles between primary tubercles and adradial suture at ambitus. Adorally, this is reduced to two rows and eventually one row on the most adoral plates. On smaller specimens, the number of secondary tubercles arranged laterally to the primary tubercles are reduced to one. Interior of interambulacral plates slightly concave with seven or eight denticles per plate at ambitus.
Spines ranging in morphology from straight ( Fig. 1F) to clavate to bulbous (Fig. 1C). Proximal fourth to third of spine shaft smooth, ending in diagonally oriented ridge, which contains the first row of spinules. Spinules oriented diagonally, along this raised ridge with more distal rows parallel to first row. Spine morphology variable, with some maintaining constant width and others tapering distally. Others ending in large clavate bulb covered in spinules. It is likely that spines varied aborally to orally, as is present in some archaeocidarids 22 and recent cidaroids such as Eucidaris clavata 33 . Although this variability exists, all spines of differing morphologies contain diagonally oriented ridge bearing first row of spinules. Acetabulum of spine bearing perforation and faint crenulations. A single non-clavate spine is found associated with an interambulacral fragment which is 5.0 mm in length (Fig. 1B). The interambulacral fragment is 7.6 mm wide indicating a probable test diameter of 16.8 mm. This would indicate that the spines were likely less wide than the diameter of the test. Spines have a prominent milled ring proximally. Bulbous spines hollow distally in bulb and non-bulbous spines hollow distally. Secondary spines and pedicellariae unknown.
Remarks-This taxon has been mentioned previously by Kier 23,24 from the Roadian and Wordian of west Texas, albeit as a single disarticulated interambulacral area and as misidentified cidarid secondary spines respectively. The inclusion of more material, and the association of the spines with the test of this species allow for a more thorough description herein. All new specimens of this taxon are known from the Lamar Member of the Bell Canyon Formation from the Guadalupe Mountains of west Texas, however, previously described specimens, now assigned to this taxon, indicate its stratigraphic range expands into the Roadian. The spines of this taxon are known from the Word Formation 23 of the Glass Mountains, however, they were originally incorrectly described as secondary spines of a larger cidarid. These spines were collected from in between the Willis Ranch and Appel Ranch members of the Word Formation, which are lower Wordian in age 34 . Furthermore, Kier 24 attributed a specimen from the Road Canyon Formation of the Glass Mountains to Miocidaris sp. This specimen (Figs 1A and 2D) is herein assigned to Eotiaris guadalupensis. This extends the stratigraphic range of this taxon into the Roadian, as the Road Canyon Formation is Roadian to Kungurian in age 34,36,37 . All of the material described herein has been silicified.
Morphologically, E. guadalupensis is very similar to Eotiaris keyserlingi from the Zechstein of the UK and Germany, differing significantly only in the morphology of its spines. Both bear rigidly sutured tests with plate imbrication adapically, sunken tubercles with multiple rows of scrobicular tubercles and crenulate and perforate tubercles. The spines of E. keyserlingi, which are well known 19 , are smooth and have much smaller spinules than those of E. guadalupensis 19 . They lack the clavate spine morphotype of E. guadalupensis and are much shorter. E guadalupensis also differs significantly from E. connorsi 24

Results
Phylogenetic analyses support the hypothesis that this taxon is a member of the cidaroidea (Fig. 2 Fig. S3) and Eotiaris guadalupensis is sister group to E. keyserlingi with a bootstrapped confidence interval of 83%. Because Eotiaris guadalupensis had apophyses and two columns of interambulacral plates, and plots as a cidaroid in the phylogenetic analyses ( Supplementary  Fig. S2, S3), then the strata from which it is known must be younger than the divergence time of euechinoids and cidaroids. Furthermore this provides a new basis upon which to obtain the hard minimum divergence date and thus is used to date the gene regulatory changes associated with this divergence. Following the best practices approach of Parham et al. 39 a hard minimum divergence time was established for the divergence of the euechinoids and cidaroids. The oldest known occurrence of Eotiaris guadalupensis is the Road Canyon Formation of the Glass Mountains of west Texas. Based upon the presence of the transitional form between the conodonts Jinogondolela idahoensis and J. nankingensis and the presence of J. nankingensis, the Road Canyon Formation was determined to be Kungurian to Roadian in age 34,36,40 . Because the exact stratigraphic horizon within the Road Canyon Formation from which the specimen of E. guadalupensis was collected is unknown, the top of the Roadian stage was chosen as the hard minimum for the divergence of the cidaroids and euechinoids, following the conservative practices for establishing hard minima set forth by Parham et al. 39 . The top of the Roadian stage is set at 268.8 Ma based upon a smoothed cubic spline interpolation fit to the existing radiometric age dates for the Carboniferous and Permian, 41 thus making the hard minimum divergence time for the euechinoids and cidaroids 268.8 Ma (Fig. 2). The discovery of this new taxon extends the minimum divergence time of the euechinoids and the cidaroids ten million years older than previously demonstrated 19,42 , shifting the minimum divergence time between these two taxonomic groups from Wuchiapingian (Lopingian) to Roadian (Guadalupian) (Fig. 2) and establishing that gene regulatory changes associated with this divergence must have also occurred by the Roadian.

Discussion
The euechinoidea and cidaroidea are differentiated, in part, because of the structure of their Aristotle's lanterns and perignathic girdles. The Aristotle's Lantern operates as the "jaws" of the echinoid, and contains numerous calcareous elements including the teeth. The perignathic girdle comprises skeletal protrusions on the interior of the test that the retractor and protractor muscles, which move the lantern in and out of the test, attach to. Based upon the lantern and perignathic girdle structure of Eotiaris keyserlingi, Smith & Hollingworth 19 determined that the euechinoids and cidaroids must have diverged prior to the Wuchiapingian stage (259.8 Ma). The perignathic girdle structures in the euechinoids and cidaroids are developmentally different, with the euechinoid auricles forming as protrusions from ambulacral plates and cidaroid apophyses developing from interambulacral plates [43][44][45] . Although euechinoids and cidaroids have differing perignathic girdle structures, neither structure is basal with respect to the other. This is known to be the case, because archaeocidarids, from which both the cidaroids and euechinoids likely evolved 3,19 , possessed the basal character state of having no perignathic girdle. Eotiaris guadalupensis also has two columns of interambulacral plates, and, through phylogenetic inference likely had two columns of ambulacral plates, as this character had been fixed in Archaeocidaris and its predecessors for approximately 90 Myr, since the Devonian 46 . These characters are synapomorphies of the crown group echinoids. As demonstrated in Fig. 2. and Supplementary Figures S2 and S3, the presence of apophyses, paired with two columns of interambulacral plates, indicates that Eotiaris guadalupensis is definitively a cidaroid, and thus the cidaroid lineage and euechinoid lineage must have already diverged prior to the appearance of this taxon in the rock record. The presence of this taxon in Guadalupian rocks not only reinforces that the cidaroid-euechinoid divergence happened prior to the Permo-Triassic mass extinction 19 , but indicates that it had occurred by the Roadian (268.8 Ma; Fig. 2) at least 10 Myr earlier than previous estimates. Furthermore, the potential exists for new discoveries to show that it may be even earlier, especially given that Eotiaris guadalupensis does not plot as the most basal cidaroid in the phylogenetic analyses ( Supplementary Fig. S2). In addition, this indicates that crown-group echinoids may have been established by the Guadalupian and were certainly biogeographically widespread by the Lopingian 24 . The appearance of Eotiaris guadalupensis in the Roadian also extends the inferred range of euechinoids prior to the Permian-Triassic boundary. The oldest definitive euechinoids, Hemipedina hudsoni and Diademopsis heberti are not known until the Norian (Late Triassic) 47-49 thus making the implied fossil gap a minimum of 40 Myr. This new species also likely has profound impacts on the molecular clock divergence dating for all echinoid clades. As the divergence of the cidaroids and euechinoids is the root divergence node used for all divergence-dating analyses of echinoids 42,50 , this new taxon has pushed back the basal node for divergence analyses 10 Myr. Future work will attempt to incorporate this new basal divergence node into molecular clock analyses.
Underlying this phylogenetic divergence must have been large-scale reorganization of the developmental GRNs of cidaroids and euechinoids, with profound impacts on the differential development of these clades. With regard to post-larval development, E. guadalupensis and other basal stem-group cidaroids are morphologically very similar to even the most derived members of the crown group cidaroidea, due to the conserved nature of the cidaroid body plan. Developmentally, this poses an interesting comparison with the euechinoidea, which have a much higher degree of post-larval morphological disparity relative to the cidaroids 1,2 . New evidence has also shed light on the gene regulatory development of juvenile skeletal structures, particularly with regard to the development of apophyses and auricles. Both apophyses and auricles develop through the expression of specific genes known to be required for skeletogenic specification in embryonic and post-embryonic development: sm37, alx1, and vegfR 45 . In particular, sm37 is a well-understood biomineralization gene 51,52 the expression of which is regulated by the upstream transcription factor alx1 6,53 . The differential spatial deployment of these genes during skeletogenesis is controlled by vegfR in the embryo 54 , and as such, this gene may be responsible for the differential spatial expression of alx1 and sm37 during the formation of apophyses and auricles 45 . Because of the presence of Eotiaris guadalupensis, which has definite apophyses, in the Roadian, the fixation of the differential deployment of these biomineralization genes must have at least begun by 268.8 Ma.
Additionally, there are a number of larval and embryonic developmental differences between modern cidaroids and euechinoids that must have arisen with the divergence of these two clades in the Permian. Euechinoid embryos possess four micromeres, and their larval skeleton arises from primary mesenchymal cells, which ingress at the vegetal pole of the embryo 10 . Cidaroids, however, have a variable number of micromeres 8,9,55 and lack primary mesenchymal cells, instead deriving their larval skeleton from skeletogenic cells emerging along with other mesodermal cells from the tip of the archenteron 8,9,56 . In euechinoids, the specification of skeletogenic mesenchyme is regulated by the double-negative gate, whereby in the micromere lineage, pmar1 represses hesC, which then allows for the expression of downstream genes responsible for micromere specification such as alx1, ets1, and tbr 6,57 . The double negative gate appears to be responsible for skeletogenic micromere specification across numerous phylogenetically diverged euechinoid lineages, including the stomopneustoids, spatangoids, clypeasteroids and camaradonts 58 such that it is very likely present throughout all indirect developing euechinoids. Contrary to euechinoids, it has been demonstrated that cidaroids lack the hesC mediated double negative gate 7 and that tbr plays no role in skeletogenesis 7 . Many of the genes encoding transcription factors and biomineralization genes responsible for micromere specification and embryonic skeletogenesis in euechinoids are also involved in juvenile euechinoid skeletogenesis and were likely co-opted by the skeletogenic micromere lineage 59 . As the euechinoids alone possess a larval skeleton that is derived from primary mesenchymal cells, it is likely that this co-option of juvenile skeletogenic genes occurred with the divergence of cidaroids and euechinoids. It is unknown as to whether the euechinoid or cidaroid suites are ancestral, however, this new fossil evidence indicates that the acquisition of one of these two differential character suites must Scientific RepoRts | 5:15541 | DOi: 10.1038/srep15541 have occurred since the divergence of the euechinoids and cidaroids in the Roadian (268.8 Ma) and is potentially very ancient.

Conclusions
Eotiaris guadalupensis, the geologically oldest cidaroid, is the oldest known probable crown-group echinoid in the fossil record. This taxon pushes back the divergence of the crown-group echinoids, the cidaroids and the euechinoids, to at least 268.8 Ma in the Roadian stage of the Permian. It furthermore extends the inferred range of early euechinoids and establishes a new hard minimum divergence for the basal node of all divergence dating studies regarding the echinoidea. In light of recent discoveries of differential cidaroid and euechinoid embryonic and juvenile development, this taxon also provides strong evidence for fixation of disparate gene expression systems by the Roadian. Eotiaris guadalupensis provides direct evidence for the differential spatial expression of specific genes in euechinoid and cidaroid post-metamorphosis skeletogenesis and indicates that this differential spatial expression must have been established by at least 268.8 million years ago.

Methods
Specimens of Eotiaris guadalupensis were analysed using dissecting microscopes and ESEM microscopy was used to determine mineralogy of specimens. Measurements were taken with calipers. Phylogenetic analyses were undertaken to rigorously demonstrate the phylogenetic relationships of this species with respect to other Permian and Triassic echinoids. Permian and Triassic euechinoids (three species; all from the family Pedinidae) and cidaroids (three species; two from the family Miocidaridae and one from the Triadotiaridae) were included in the analysis, in addition to E. guadalupensis. The outgroup of the analysis was Archaeocidaris whatleyensis, a well-known, stem-group echinoid, which has been used as outgroup to all crown group echinoids in previous analyses 1,2,49 . The characters used in the phylogenetic analysis in Supplementary Figures S2 and S3 consisted of 24 characters, 20 were binary and 4 were multistate. Characters and character states are in supplementary information. All characters were unordered and unweighted in original analyses and character matrix is listed in Supplementary Table S1. Corresponding Nexus file is in supplementary information. Initial phylogenetic analysis was run in PAUP version 4 60 and consisted of an exhaustive search of all possible trees. This analysis resulted in 2 most parsimonious trees with length 31 consistency index (CI) .806 and retention index (RI) .750. Characters were then reweighted by their maximum retention indices and analyses were rerun. This resulted in one most parsimonious tree, equal to one of the two resultant trees from the unweighted search and with length 22.5, CI .911 and RI .875 ( Supplementary Fig. S2). In order to estimate branch support we ran a heuristic search with 1000 RASs and TBR with 1000 bootstrap replicates on the reweighted character matrix. Bootstrapped confidence intervals are shown with appropriate branches in Supplementary Figure S3.