3D models of the morphology of the digestive system during development were reconstructed from a series of histological sections. Location and size of the GI-tract and its associated organs, such as liver, endocrine and exocrine pancreas, and gallbladder, were observed from stage 3 (prior to first feeding) until the post-metamorphic stage 10 (Figure 1).

The GI-tract includes a narrow foregut (esophagus and presumptive stomach/stomach), midgut, and a short hindgut (rectum) (Figure 2). The anterior region of the midgut, just after the pyloric sphincter (PS), was larger in diameter, i.e. more voluminous, compared to the rest of the midgut. This feature was maintained during GI-tract ontogeny (Figures 1 and 2). Both PS (which separates the presumptive stomach from the anterior midgut) and ileorectal sphincter (which separates midgut and hindgut) were identified from stage 3 onwards (Figures 1 and 2). Pyloric caeca became evident as projections from the most anterior part of the midgut at stage 6 (Figures 1 and 2). The stomach was well-differentiated at stage 10 and the gastric glands were visible on histological sections (Additional file 1). The luminal volume of the GI-tract increased during development, particularly in the two last stages analysed (stages 9A and 10) (Figure 3, Table 1 and Additional file 2). The stomach volume from 9A to 10 increased from 415 to 4933 nl, respectively and corresponded to an 11 fold increment (Table 1).

The values were calculated from the 3D models using Imaris MeasurementsPro.

^aGI-tract tissue volume = GI-tract outer layer - GI-tract inner layer.

The liver was positioned under the foregut and anterior to the ascending loop of the midgut (Figure 1) and its volume steadily increased during development (Figure 3 and Table 1). The exocrine pancreas was observed between the presumptive stomach and the anterior part of the midgut at stage 3 and it surrounded this midgut area throughout ontogeny (Figure 1). In the endocrine pancreas, a clearly distinguishable islet of Langerhans was observed close to the gallbladder at stage 3 (Figure 1). In contrast to the other digestive organs, the increment in the normalized volume of endocrine and exocrine pancreas was low and negative, respectively, between stages 9A and 10 (Figure 3 and Table 1). The yolk-sac, positioned under the GI-tract at stage 3, decreased in size after the initiation of exogenous feeding and a small vestige remained besides the liver at stage 4 (6 days post first feeding, dpff). The gallbladder was observed on the right-hand side between the exocrine pancreas and the liver, and maintained this position in all the developmental stages analysed (Figure 1). The pancreatic duct and the bile duct opened next to each other into the lumen at the median plane of the anterior midgut, just after the PS (data not shown).

The complete coding sequence (CDS) of Atlantic halibut pepsinogen A2 was 1128 bp and was submitted to GenBank under accession no. KF184647 (Additional file 3: C). The amino acid (AA) sequence of pepsinogen is relatively well-conserved among teleost fish and, as expected, more variable when compared to other vertebrate pepsinogens. For instance, halibut pepsinogen A2 shared respectively 88% and 64% AA sequence identity with winter flounder (Pseudopleuronectes americanus) pepsinogen A form IIb and IIa, but only 52% and 48% identity with homologues from Xenopus laevis and human, respectively (data not shown).

The cDNA fragments cloned for Atlantic halibut H⁺/K⁺-ATPase α subunit (911 bp) and Na⁺/K⁺-ATPase α subunit (714 bp) were deposited in GenBank with the accession numbers KF184648 and KF184650, respectively (Additional file 3: B, D). The CDS for H⁺/K⁺-ATPase β subunit of 874 bp was cloned and submitted to GenBank with the accession no. KF184649 (Additional file 3: A). Phylogenetic analysis of the α subunit of the gastric proton pump and Na⁺/K⁺-ATPase, and vertebrate homologues (Additional file 4) generated two major clades, one corresponding to H⁺/K⁺-ATPase and the other to Na⁺/K⁺-ATPase. Phylogenetic analysis of the β subunit (Additional file 5) generated a tree with two major clades that shared the same general topology as the phylotree for the α subunit with the H⁺/K⁺-ATPase and Na⁺/K⁺-ATPase clustered independently.

Atlantic halibut H⁺/K⁺-ATPase α subunit clustered most closely with teleost homologues, with which it shared 94% AA sequence identity, and increased to 98% identity with winter flounder and Atlantic cod (Gadus morhua). Lower AA sequence identity (72%) was found when Atlantic halibut H⁺/K⁺-ATPase α subunit was compared to Atlantic halibut Na⁺/K⁺-ATPase α subunit (70%) and to other vertebrate counterparts (72%). The Atlantic halibut Na⁺/K⁺-ATPase α subunit clustered with an Antarctic eelpout (Pachycara brachycephalum) homologue (98%) and shared approximately 88% AA identity with other teleost gene homologues. H⁺/K⁺-ATPase β subunit clustered as expected within the teleost clade (overall identity about 80%) and shared rather low identity with its human homologue (50%). Atlantic halibut H⁺/K⁺-ATPase β subunit did not share more than 39% AA sequence identity with the Atlantic halibut Na⁺/K⁺-ATPase β subunit.

The developmental expression profiles of pepsinogen A2, H⁺/K⁺-ATPase α and β subunits, Na⁺/K⁺-ATPase α subunit and ghrelin were analysed by qPCR in the GI-tract of individual Atlantic halibut larvae (Figure 4). The gene expression of both gastric proton pump subunits were significantly (p < 0.05; adjusted R²: 0.773) correlated (Figure 5) and had parallel expression patterns, with a sharp and significant (p < 0.05) increase at climax and in post-metamorphic stages (Figure 4). Pepsinogen A2 was significantly (p < 0.05) correlated with the expression profile of the gastric proton pump α (adjusted R²: 0.9738) and β (adjusted R²: 0.7963) subunits (Figure 5). A significant (p < 0.05) increase during stage 8 was observed for pepsinogen A2 and its expression peaked in the post-metamorphic stage. Ghrelin mRNA transcript abundance increased gradually and significantly (p<0.05) during the proclimax/climax of metamorphosis, and attained a maximum in the post-metamorphic stage (Figure 4). Moreover, ghrelin transcript abundance and proteolytic activity during GI-tract ontogeny were significantly correlated (p < 0.05; adjusted R²: 0.9342, 0.8852, 0.9252 for pepsinogen A2, gastric proton pump α and β subunits, respectively; see Figure 5). Expression of Na⁺/K⁺-ATPase α subunit mRNA was detected in all developmental stages, with significantly (p < 0.05) more transcripts at stage 5.

The pH assessment in the lumen of the stomach and midgut/hindgut during post-embryonic development was based on the colour observed after the injection of pH indicator solutions (Figure 6 and Table 2). The pH in the midgut/hindgut remained alkaline (above pH 8) in all the developmental stages analysed (stage 5 to 9B). The presumptive stomach also had an alkaline pH with values above 7.5 until stage 8. Gradual acidification was observed in the stages corresponding to the climax of metamorphosis. Transition from an alkaline to an acidic pH in the stomach lumen was evident at stage 9A, when the injected sol CPR remained purple but the sol mCP gave a yellow coloration (pH6.5 – 7.5). The lumen of the stomach was clearly in the acidic range (pH < 3.5) at stage 9B, as revealed by the yellow colour in the stomach following administration of both CPR and BPB solutions.

The presented pH values are based on visual observations of colour changes after the administration of three pH indicator solutions.

Spontaneous propagating contractions were observed in the GI-tract at prometamorphosis (stage 6; 25 dpff) and climax of metamorphosis (stage 9A/B; 49 dpff) (Figure 7). Due to considerable individual variation, number and frequency of contractions could not be grouped and are presented for each individual analysed (Table 3 and Additional file 6). Two types of contractions were observed in the midgut region 1 (mg1; after the PS, descending part of the loop) and 2 (mg2): phasic and propagating waves (Additional file 7). The propagating contractions observed in mg2 were retrograde waves that originated in area “A” and moved towards the mouth. However, in mg1 most of the propagating contractions originated just under the PS and were anterograde waves that moved in an anal direction. Motility activity in both midgut regions was detected at stage 6 with a frequency ranging from 0.31 to 3.77 min^-1, depending on the individual and type of contraction. At stage 9, relatively few spontaneous contractions of short duration were observed in the midgut. During the climax of metamorphosis, contractions in the stomach were registered in all individuals, in contrast to stage 6 when motility in the presumptive stomach was only observed in one larva. The rectal contraction (or defecation reflex) was a mixture of retro- and anterograde contractions and were observed in both stages 6 and 9 with similar frequencies in most of the individuals analysed.

Propagating and phasic contractions are stated for midgut 1 and midgut 2 regions. Frequency is the number of contractions registered (n) per min.

*Not possible to quantify phasic and propagating wave contractions. The affected GI-tract segments were constantly (tonic) contracted during the whole observation period (see Additional file 7).

The material for the present study came from different batches of commercially produced Atlantic halibut larvae. Larvae used for 3D modeling were the same as previously described by Kamisaka et al. 62 except for the last developmental stage (stage 10), where complementary material was sampled at Nordic Halibut (Askøy, Norway). For all other analysis, larvae were sampled at Sterling White Halibut AS (Marine harvest, Rørvik, Norway) during March 2012. Larvae were reared according to standard industrial protocols, with light/dark cycles of 18:6 hours and water temperature 11°C. Feeding with Artemia enriched with commercial products took place twice a day (10:00 and 22:00) following standard rearing procedures 63 .

Classification of developmental stage was based on mytome height (MH) and standard length (SL), according to a modified version of Sæle et al. 64 . The following stages were used in the functional studies: 5 - premetamorphic; 6 and 7 - prometamorphic; 8 - proclimax metamorphosis; 9A and 9B - climax metamorphosis; and 10 - post-metamorphosis. For the morphological studies (3D models) two extra stages were included, stage 3 and 4, based on morphological classification criteria of Pittman et al. 25 . Larvae intended for gene expression analysis were sampled 2 h after feeding (12:00) and euthanized with a lethal dose of MS222 (Tricaine methanesulfonate, Sigma-Aldrich, St. Louis, USA). Photos of each larva were taken in order to categorise them into different developmental stages. The GI-tract from each larva was dissected, rapidly transferred to RNAlater (Life Technologies, Carlsbad, USA) and stored at − 80°C. Atlantic halibut larvae used for in vivo studies (pH and motility analysis) were staged based on the photographs of living individuals.

To clone and study the expression profiles of pepsinogen, H⁺/K⁺-ATPase α and β subunit, Na⁺/K⁺-ATPase subunit α and ghrelin, Atlantic halibut juveniles (147.7 ± 15.1 g wet weight; 23.4 ± 1.1 cm total length; n = 6) were sampled at the Institute of Marine Research, Austevoll, Norway. The fish were euthanized with a lethal dose of MS222. The GI-tract was dissected into stomach, pyloric caeca, midgut and hindgut and stored in RNAlater at − 80°C until further analysis.

The experimental procedures and sampling protocols in the study were approved by an ethical committee (No. 2679; IMR Austevoll, Norway). All procedures were performed by scientists licensed by the Norwegian Animal Research Authority (NARA) to work on animals and under due consideration of the NARA guidelines.

For reconstruction of the digestive organs, six high quality preserved larvae were used for each stage studied (stages 3, 4, 5, 6, 9A and 10) and then the most representative larvae from each stage was used to construct the 3D model. Detailed material information about the approach taken is given in 62 . In summary, sampled larvae were fixed in Bouin’s solution overnight, stored in 70% EtOH at 4°C, dehydrated and embedded in paraffin. Serial sections were cut at 5 μm thickness and counterstained with hematoxylin. For the oldest stage, halibut larvae were fixed in 4% paraformaldehyde, dehydrated through an ethanol series and embedded in Technovit 7100 (Heraeus Kulzer GmbH, Hanau, Germany). Semi-thin (2 μm) serial sections were stained with Toluidin blue.

Photographs were taken every fifth section (10 μm between used sections) using a Nikon Digital Sight DS-U1 camera mounted on a Zeiss Axioscope 2 Plus microscope. The 3D reconstruction of the digestive system was performed as described by Kamisaka and Rønnestad 43 . In brief, manually defined contour lines of the digestive organs were made based on aligned images of serial sections, and contour surfaces were calculated using the software Imaris 6.2.0. (Bitplane AG Zurich, Switzerland). After generating a surface object, the same software (Imaris MeasurementPro) automatically calculated a range of statistical parameters including surface area and volume of the different organs. The volume increase of the digestive organs between stages was calculated and normalized to the overall mean of volume increase for each tissue (see Additional file 2).

Total RNA was isolated from the GI-tract of juvenile Atlantic halibut using TRI reagent (Sigma-Aldrich, St. Louis, USA) according to the manufacturer’s instructions. Samples were treated with TURBO DNA-free (LifeTechnologies, Austin, USA) to eliminate genomic DNA contamination. Quality of DNase treated total RNA was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies). cDNA was synthesized from 2.0 μg of DNase treated total RNA using oligo (dT) primer from SuperScript III First-Strand Synthesis system for RT-PCR kit (Invitrogen, Carlsbad, USA).

Transcript fragments of pepsinogen A2, ghrelin [GenBank: EF493849], gastric proton pump subunits and Na⁺/K⁺-ATPase subunit α were amplified using gene specific primers as listed in Table 4 designed with Primer Premier 5 software (Premier Biosoft Int., Palo Alto, USA). For pepsinogen A2 and H⁺/K⁺-ATPase β subunit, a PCR homology-cloning approach was used with primers designed in putative conserved N and C terminus regions of the winter flounder [GenBank: AF156788] and stickleback (Gasterosteus aculeatus, [Ensembl: ENSGACT00000020259]) homologue genes, respectively. The H⁺/K⁺-ATPase α subunit was cloned taking a comparative homology approach using the winter flounder homologue gene [GenBank: AF156789.1]. The Na⁺/K⁺-ATPase α subunit was cloned based on two ESTs from Atlantic halibut [GenBank: EB031798 and EB031117]. Amplifications were performed in a thermocycler Gene Amp PCR system 2700 (Applied Biosystems) using GoTaq DNA polymerase (Promega, Madison, USA) according to the manufacturer’s instructions and using the following conditions: 95°C for 2 min; 30 cycles of 95°C for 30 s, 58°C for 30 s, 72°C for 30 s; and a final step at 72°C for 5 min. Amplified PCR products were resolved on a 1% agarose gel and purified using E.Z.N.A. Gel Extraction Kit (Omega bio-tek, Norcross, USA). Purified fragments were cloned into the pGem-T easy vector system I (Promega, Madison, USA) and sequenced at the University of Bergen Sequencing Facility (Bergen, Norway). Sequence identity was confirmed by BLASTx (http://blast.ncbi.nlm.nih.gov/Blast.cgi) analysis against the GenBank database.

Multiple sequence alignments of H⁺/K⁺-ATPase subunit α and β and Na⁺/K⁺-ATPase subunit α protein sequence were performed with ClustalX (Gonnet 250 series matrix, Gap opening penalty 10, Gap extension 0.2) 65 . Alignments were displayed in GeneDoc (http://www.nrbsc.org/gfx/genedoc/) and percentage of sequence identity and similarity calculated. Phylogenetic analyses were performed using the Maximum Likelihood method 66 with 1000 bootstrap replicates 67 , using MEGA 5.2 software 68 .

Total RNA was isolated from the GI-tract of the larvae at each developmental stage and cDNA synthesized as described above. For expression pattern analysis, specific primers were designed for the target genes (Table 4) and the target amplified using a Bio-Rad CFX96™ Real-Time System. The gene eEF1AI (Elongation factor 1 alpha, [GenBank: EU561357]) was used as the internal reference gene 69 . Relative gene quantification was performed using the mean normalized expression (MNE) method of the Q-Gene application 70 71 . Assay efficiency was determined using a 10-fold cDNA pool dilution curve ranging from 200 to 0.02 ng. Reactions for each sample were performed in duplicated using the following PCR conditions: 95°C for 3 min; 45 cycles of 95°C for 30 sec, 58°C for 30 sec and 72°C for 30 sec. Melting curve analysis over a range of 45-95°C (increment of 0.5°C for 4 sec) allowed the detection of nonspecific products and/or primer dimers.

The mRNA expression levels are presented as the mean ± SEM (n = 6). Data was log-transformed to achieve normal distribution. Statistical significance of relative gene expression between groups was analysed by one-way ANOVA when the data-set had a normal distribution. One-way ANOVA followed by a Student-Newman-Keuls (SNK) multiple range test was applied when data failed the normality test. SigmaStat v.3.1 (Systat software, Inc., USA) was used for the statistical analysis.

Correlation analysis were performed between: A) H⁺/K⁺-ATPase α subunit versus β subunit; B) pepsinogen A2 versus gastric H⁺/K⁺-ATPase α and β subunits; C) ghrelin versus pepsinogen A2 and gastric H⁺/K⁺-ATPase α and β subunits. Assuming that the relationship is linear, a linear model (lm) 72 73 was applied to the mean of the log-transformed MNE of the transcripts through development (stage 5 to 10). Plot graphs were constructed based on the linear model results. The correlation analysis were conducted in R 74 .

The pH in the lumen of the GI-tract was determined with an in vivo method where pH indicator solutions (from alkaline to acidic ranges) were administered by tube feeding 53 . The in vivo set-up comprised a stereo dissecting microscope with a Leica DFC295 camera and a micromanipulator. A nanoliter injector (World Precision Instruments) with a plastic capillary tube (O.D. 0.19 mm, Sigma-Aldrich, St. Louis, USA) was fastened to the micromanipulator. The larvae were anaesthetized (MS-222; ranging from 3 to 100 μg/ml final concentration) and placed on a microscope slide in a droplet of clean seawater. The correct position of the larvae for injection was assured by the water surface tension. The capillary tube was gently passed through the mouth and esophagus into the presumptive stomach/stomach area and one of three pH indicator solutions was injected into a total of 3 larvae at each developmental stage (Figure 6). The first solution (sol mCP) consisted of 0.1% m-Cresol purple (Sigma-Aldrich, St. Louis, USA) in sea water (pH range 7.5–9.5). The second solution (sol CPR) contained 0.1% of Chlorophenol Red (Sigma-Aldrich, St. Louis, USA) in sea water (pH range 6–9.5) and the third (sol BPB) was 1% Bromophenol Blue (Sigma-Aldrich, St. Louis, USA) in sea water (pH range 3.0-4.6). The colour of the intestinal fluid was then compared to a set of solutions mCP, CPR and BPB standards prepared in pH buffers from pH 2.0 to pH 9.0 in steps of 0.5. The standards were inside sealed glass capillaries and immersed in sea water with the same light and temperature conditions as the larvae under the dissecting microscope. All experimental fish were euthanized with an overdose of MS222 after treatment. The work with Atlantic halibut larvae was conducted in a cold room to ensure a constant temperature.

Analysis of the GI-tract motility pattern in halibut larvae was based on in vivo video recordings, using a Leica DFC295 camera connected to a stereo dissecting microscope. Video sequences were recorded from four larvae with a full or partially full GI-tract at two different developmental stages: stage 6 - prometamorphic and stage 9 - climax metamorphosis. The animals were anaesthetized and maintained immersed in seawater on a microscopic slide and captured on video for 30 min. From the video recordings, still images were captured for analysis and identification of the different motility patterns. The GI-tract movements were quantified using Etholog 2.2 software 75 . The number and frequency (number of contractions min^-1) of phasic and propagating wave contractions in two different regions of the midgut were determined. The frequency and number of contractions in the presumptive stomach/stomach area, hindgut and in the rectal area were also quantified.

All supporting data are included in the Additional files. In addition, nucleotide sequences have been deposited in GenBank under the accession numbers: KF184647 (pepsinogen A2); KF184648 (H⁺/K⁺-ATPase α subunit); KF184650 (Na⁺/K⁺-ATPase α subunit); KF184649 (H⁺/K⁺-ATPase β subunit) and the alignments for the phylogenetic tree construction are available in TreeBase: http://purl.org/phylo/treebase/phylows/study/TB2:S15435.