Transcriptomic analysis of the lesser spotted catshark (Scyliorhinus canicula) pancreas, liver and brain reveals molecular level conservation of vertebrate pancreas function

Background Understanding the evolution of the vertebrate pancreas is key to understanding its functions. The chondrichthyes (cartilaginous fish such as sharks and rays) have often been suggested to possess the most ancient example of a distinct pancreas with both hormonal (endocrine) and digestive (exocrine) roles. The lack of genetic, genomic and transcriptomic data for cartilaginous fish has hindered a more thorough understanding of the molecular-level functions of the chondrichthyan pancreas, particularly with respect to their “unusual” energy metabolism (where ketone bodies and amino acids are the main oxidative fuel source) and their paradoxical ability to both maintain stable blood glucose levels and tolerate extensive periods of hypoglycemia. In order to shed light on some of these processes, we carried out the first large-scale comparative transcriptomic survey of multiple cartilaginous fish tissues: the pancreas, brain and liver of the lesser spotted catshark, Scyliorhinus canicula. Results We generated a mutli-tissue assembly comprising 86,006 contigs, of which 44,794 were assigned to a particular tissue or combination of tissues based on mapping of sequencing reads. We have characterised transcripts encoding genes involved in insulin regulation, glucose sensing, transcriptional regulation, signaling and digestion, as well as many peptide hormone precursors and their receptors for the first time. Comparisons to mammalian pancreas transcriptomes reveals that mechanisms of glucose sensing and insulin regulation used to establish and maintain a stable internal environment are conserved across jawed vertebrates and likely pre-date the vertebrate radiation. Conservation of pancreatic hormones and genes encoding digestive proteins support the single, early evolution of a distinct pancreatic gland with endocrine and exocrine functions in jawed vertebrates. In addition, we demonstrate that chondrichthyes lack pancreatic polypeptide (PP) and that reports of PP in the literature are likely due cross-reaction with PYY and/or NPY in the pancreas. A three hormone islet organ is therefore the ancestral jawed vertebrate condition, later elaborated upon only in the tetrapod lineage. Conclusions The cartilaginous fish are a great untapped resource for the reconstruction of patterns and processes of vertebrate evolution and new approaches such as those described in this paper will greatly facilitate their incorporation into the rank of “model organism”. Electronic supplementary material The online version of this article (doi:10.1186/1471-2164-15-1074) contains supplementary material, which is available to authorized users.

a key role in β-cell differentiation [81,82] and has also been shown to suppress the expression of 219 glucagon to maintain β-cell identity, as well as being able to regulate glucose-sensitive insulin secretion 220 [83]. Our discovery of Pdx1 (and Pdx2), NeuroD1 and its partner E47, Isl1, Hnf1a and Nkx6.1 transcripts 221 expressed in the catshark pancreas suggests an ancient role for these genes in vertebrate pancreas function 222 and hints at early establishment of the insulin gene regulatory network. Additionally, the presence of 223 transcripts encoding NeuroD1, e47, Isl1 and Nkx6.1 in the catshark brain highlights shared ancestry of 224 these tissues in the vertebrate neuroendocrine system. We do not find any transcripts for MafA, which has 225 been shown to be a key regulator of glucose-sensitive insulin secretion in humans and rodents [58,84,85], 226 although other studies have also had difficulty identifying transcripts of this gene and other pancreas 227 transcription factors in non-PCR based experiments [86,87], possibly because of the low level of 228 expression of transcription factors in general [88]. 229

Transcription factors 230
In addition to the transcription factors involved in insulin regulation discussed above, KEGG orthology 231 (KO) analysis [89-91] has identified 13 transcription factors expressed in just the pancreas (including 232 Pdx1 and Pdx2, FoxA1 (Hnf3a) and Pancreas-specific transcription factor 1a (Ptf1a)), 14 expressed in 233 pancreas and liver, 33 in pancreas and brain and 51 in all three tissues (Tables 2 and 3). 234 In a survey of the expression of 790 human DNA-binding transcription factors, Kong et al. [88] identified 235 80 with expression restricted to the fetal pancreas, 32 restricted to the adult pancreas and 18 shared by 236 both. Of the 31 adult-specific genes, we find evidence that 6 are also expressed in the adult catshark 237 pancreas, although this number increases to 15 if members of the same gene family are considered (the 238 possibility of divergent resolution of gene duplicates following the whole genome duplications [92] in 239 early vertebrate ancestry must be considered). Since transcription factors are known to be expressed at low 240 levels in cells (less than 20 copies per human adult cell [88]) it is likely that our figure is an underestimate 241 and a more comprehensive survey of candidate transcription factor expression in this species is needed. 242

Signalling 243
Our KEGG orthology analysis identified 38 transcripts involved in signal transduction that are expressed 244 only in the catshark pancreas, 11 in both pancreas and liver, 104 in pancreas and brain and 187 in all three 245 tissues (Tables 2 and 3). Among these are representatives of the major vertebrate signalling pathways, 246 including ligands and receptors for Fgf, Wnt, Notch, Vegf, Tgfβ and Pdgf. Members of all of these 247 pathways have previously been identified in the human pancreas transcriptome [87]. 248

Homeobox gene diversity 249
Homeobox genes are a group of transcription factors that encode a 60 amino acid DNA-binding 250 homeodomain and that are involved in a wide variety of gene regulatory events in embryonic and adult 251 tissues. A number of homeobox genes are known to be expressed during endodermal regionalisation and 252 pancreas development, including Islet 1 and 2 (Isl1, Isl2), Pancreatic and duodenal homeobox 1 (Pdx1), 253 (Hlx, Hhex), three in pancreas and brain (Arx, Zfhx3, Zfhx4) and one in all three tissues (Cut-like 2). These 266 include genes known to be restricted to, or highly expressed in, β-cells (Pdx1), α-cells (Arx) and acinar 267 cell types (Cut-like 2) [86]. 268

269
Digestion 270 In addition to its endocrine roles, the pancreas is also an important exocrine organ, fulfilling key functions 271 in the digestion of proteins, lipids and carbohydrates. In the carnivorous elasmobranchs protein and lipids 272 are the main energy sources [99] and it has been shown that ketone bodies and amino acids are the main 273 oxidative fuel source for muscles and several other tissues, in preference to fatty acids [24,28,99]. 274 Carbohydrates are thought to be utilised as oxidative fuels in elasmobranch heart muscle, as well as brain, 275 red muscle and rectal gland [28, 100, 101]. It is therefore perhaps reasonable to assume that proteases and 276 lipases are the most significant digestive enzymes produced by the elasmobranch pancreas and indeed this 277 appears to be the case. Some form of chymotrypsinogen and trypsinogen have long been known to be 278 produced by the elasmobranch pancreas, as has carboxypeptidase B, although these enzymes have not It has recently been suggested [31] that a high frequency of dinucleotide simple sequence repeats (SSRs, 298 microsatellites) is a general feature of shark genomes. We find 6,843 transcripts containing one or more 299 di-, tri-or tetranucleotide microsatellites of five perfect repeats or more in our catshark data, with 482 of 300 these only in pancreas, 3,083 only in brain and 473 only in liver ( Our analysis of the catshark pancreas transcriptome reveals the presence of genes known to be involved in 306 glucose sensing and regulation of the insulin gene in other vertebrates and illustrates that functional 307 conservation of these aspects of the vertebrate pancreas is reflected at the molecular-level. We therefore 308 propose that these molecular-level mechanisms are a common feature of jawed vertebrates and that this 309 lends support to the theory that the evolution of blood-glucose sensing and regulatory mechanisms may 310 13 have facilitated the evolution of the complex glucose-dependent brain of vertebrates [7-9]. We further 311 suggest that the early evolution and fixation of these mechanisms has imposed evolutionary constraints on 312 glucose sensing and insulin regulation in vertebrates, including in cartilaginous fish, even in the face of 313 their ability to tolerate extended periods of hypoglycaemia and likely relaxed requirements for these 314

processes. 315
We find that the catshark pancreas produces at least eight peptide hormones (insulin; glucagon; 316 somatostatin; peptide YY; gastrin-releasing peptide, neuromedin U, encephalin and vasoactive intestinal 317 polypeptide, Table 5) and expresses a wide variety of genes involved in digestion, especially the digestion 318 of proteins and lipids. The catshark pancreas therefore clearly has the features of a distinct pancreatic 319 gland with both endocrine and exocrine functions and as such will be of great use in reconstructing the 320 characteristics of the earliest vertebrate pancreas. The similarity in gene expression between the catshark 321 and other vertebrates with respect to hormones, digestive enzymes, transcription factors and signaling 322 pathways again provides support to the theory that there was a single, early origin of the pancreas at the 323 base of the jawed vertebrate radiation. The overlap in peptides produced by the catshark pancreas and 324 brain (Table 5)  to study these processes. The catshark PYY+ cells will therefore provide important insights into the 335 evolution of the vertebrate pancreas, and especially progenitors of α, β, δ and γ-cells. 336 Our experiments make clear that much of the previous work on the presence or absence of peptides in 337 basal vertebrate lineages may be suspect, with many false-positive signals resulting from cross-reacting 338 antisera. Previous schemes of pancreas evolution based on these and similar data, which posited the 339 restriction of various hormones to the alimentary canal (similar to the situation in protochordates such as 340 amphioxus), the accumulation of these into a two-or three peptide islet organ in jawless fish and finally 341 the "classic four-hormone islet tissue" of cartilaginous fish and other vertebrates [2] are therefore 342 incorrect. In fact, it appears that the three hormone (glucagon, insulin, somatostatin) islet organ was 343 established early in vertebrate evolution and remains today in the adult (but not larval) lamprey, 344 cartilaginous fish and actinopterygian (ray-finned) fish, and that it is only in the sarcopterygian (lobe-345 finned fish) lineage that a four hormone (the above, plus PP) pancreas was formed. 346 Our analysis of homeobox gene expression reveals a surprising level of variation between the genes 347 known to be expressed in the catshark pancreas, human islets [87] and rat [53] and hamster [95] cell lines. 348 It therefore seems likely that this particular class of transcription factors is extremely variable with respect 349 to their spatial or temporal expression pattern in the vertebrate pancreas (or more likely both) and this is 350 perhaps not too surprising given the variety of roles carried out by the pancreas in response to feeding, 351 digestion and the regulation of blood glucose. As expected we have identified transcripts of both Pdx1 and 352 Pdx2 in the catshark pancreas, although we do not find any evidence for the presence of additional 353 duplicates of other genes encoding proteins known to interact with PDX1 in other species. It therefore 354 seems unlikely that the maintenance of paralogous Pdx2 genes in some vertebrate lineages reflects a wider independent loss in others (ray-finned fish and tetrapods) remain unknown. 361 With the availability of whole genome sequence information for a greater number of taxa and improved 362 coverage of vertebrate pancreas transcriptomes a larger amount of data than ever before is now becoming 363 available. These data, together with an appreciation that early vertebrate evolution was characterised by 364 extensive genetic, developmental and morphological innovation facilitated by multiple whole genome 365 duplications [120, 121] will better enable us to reconstruct pancreas evolution. As an example, we propose 366 that the creation of the paralogous NPY and PYY during these duplications at room temperature and the next day the slides were rinsed 3 x 5min each in PBS and specific cross 417 absorbed donkey anti-mouse, rabbit, or guinea pig secondary antisera (Jackson Immunoresearch) were 418 added for 30mins. The slides were rinsed in PBS and mounted. Details of antisera are given in Table 1 in 419 Additional file 10. All pictures were taken on a Zeiss Meta510 confocal microscope. 420 421

Antibody Absorption 422
In order to test their specificity against the Pancreatic polypeptide family, the antisera were incubated 423 overnight at 4°C with 10µg of either pancreatic polypeptide (Sigma), Neuropeptide Y (Bachem) or 424 peptide YY (in-house synthesis) or no peptide. The next day the antisera were added to the slides and the 425 staining was performed as above. The staining intensity was compared to the no peptide control and given 426 a rating of 1-3 (+, ++, +++). The results are shown in Table 2 in Additional file 10. 427 428

Competing interests 429
The authors declare no competing interests.   Process' terms for catshark pancreas, brain and liver tissue-specific transcripts.     Table 5. Peptide diversity of the catshark pancreas, brain and liver. Our comprehensive transcriptomic 890 survey of the lesser spotted catshark pancreas highlights the disparity in the estimation of peptide diversity 891 in early vertebrates as previously suggested by immunohistochemical (IHC) studies and highlights the 892 similarity of pancreas and brain peptide complements.