Rapid expansion of pigmentation genes in penaeid shrimp with absolute preservation of function

The carotenoprotein complex called crustacyanin (CRCN) is an evolutionary novelty among the animals in class Crustacea as the mechanism for producing shell colour and patterns (Wade et al., 2009). The carotenoid astaxanthin forms the central chromophore of crustacyanin (Britton and Goodwin, 1982), where it is non-covalently bound but twisted by the protein interaction to produce the diverse array of crustacean colours (Chayen et al., 2003; Helliwell, 2010). Five distinct carotenoid-binding proteins (CBPs) were originally identified from the European lobster (Homarus gammarus) and these proteins were grouped into the broad classes CRCN-A and CRCN-C based upon amino acid composition, electrophoretic mobility and peptide mapping (Cheesman et al., 1966; Quarmby et al., 1977). The protein forms a dimer composed of two 20 kDa CRCN subunits and two astaxanthin (Axn) molecules, known as β-crustacyanin, which in turn form a large multimeric complex, called α-crustacyanin, composed of eight dimeric β-crustacyanin subunits (Chayen et al., 2003; Cheesman et al., 1966; Habash et al., 2004; Helliwell, 2010; Zagalsky and Cheesman, 1963).

Within the shell and hypodermal tissue of crustaceans, two CRCN proteins and two Axn molecules dimerise to form β-crustacyanin, and these dimers form the building blocks of the much larger multimeric α-crustacyanin (Chayen et al., 2003; Zagalsky, 2003). The 3D crystal structures of both α-crustacyanin and β-crustacyanin have been resolved on several independent occasions, as summarised in Table 1. Current research suggests that the colour shift from red to blue when astaxanthin is bound to CRCN is induced by acid–base change in carotenoid conformation, associated with the relocation of a proton on the end ring of Axn (Begum et al., 2015). This theory is distinct from the physical twisting of Axn or electron coupling within the protein-bound form investigated in other studies (Christensson et al., 2013; Strambi and Durbeej, 2009).

The full protein sequence was identified by protein purification and direct sequencing methods for the predominant CRCN A and C subunits (Keen et al., 1991a,b). It is thought that post-translational modifications, for example amidation or glycosylation, are responsible for the differences in electrophoretic mobility of subunits (Habash et al., 2004). In the American lobster (Homarus americanus) only two protein isoforms H1 and H2 have been identified (Zagalsky and Tidmarsh, 1985). However, there are only two genes reported to encode the different CRCN protein subunits (Chayen et al., 2003; Wade et al., 2009). A single CRCN-A and CRCN-C gene has been identified from a number of different crustaceans, including spiny lobsters (Wade et al., 2009), clawed lobsters (Ferrari et al., 2012; Wade et al., 2009), hermit crabs (Wade et al., 2009) and penaeid shrimp (Ertl et al., 2013; Wade et al., 2009). Only one gene (either A or C) has been identified in some species (Wade et al., 2009; Yang et al., 2011), although the inability to identify genes does not definitively indicate the absence of that gene in that species. Meanwhile, a duplication of the CRCN-A gene was identified from the spiny lobster Panulirus cygnus, with no corresponding CRCN-C gene identified (Wade et al., 2009). As such, the number of genes encoding functional CRCN proteins in crustaceans remains unclear.

In the present study, 35 new genes that encode CRCN proteins were identified across 12 species of penaeid shrimp, using a combination of degenerate PCR and bioinformatics data mining. These sequences were used to establish potential lineage-specific gene duplications, and predict three-dimensional (3D) structures of the encoded proteins using known crystal structures. The giant tiger shrimp (Penaeus monodon Fabricius 1798) was used as a model species to understand the functional regulation of three of the putative CRCN isoforms (PmonCRCN-A, PmonCRCN-C1 and PmonCRCN-C2) using reverse transcription PCR (RT-PCR) in pigmented hypodermal tissue and across the moult cycle. The systemic injection of double-stranded RNA (dsRNA) against each specific PmonCRCN isoform was used to trigger the RNA interference (RNAi) pathway, and the isoform-specific pigment phenotype was tracked in live animals using colour quantification from digital images. The functional integration of each of the PmonCRCN isoforms within CRCN complexes was confirmed using microscopy of pigmentory chromatophores.

Animal samples were opportunistically collected as part of CSIRO trawl surveys in the Gulf of Carpentaria in 2013 and 2014 (Kenyon et al., 2015). Animals were euthanized by immersion in an ice–seawater slurry for several minutes. Species were identified by experienced researchers equipped with species-specific reference material (Grey et al., 1983). Once collected and identified, small pieces of hypodermal tissue were removed and stored in RNAlater^TM (Invitrogen, Carlsbad, CA, USA). Samples were preserved and extracted from three to four individuals, with the exception of the red spot king prawn (Penaeus longistylus Kubo 1943), where only one animal was captured.

Total RNA was extracted using Trizol reagent (Invitrogen) according to the manufacturer’s instructions, and precipitated by adding 0.5 volumes of isopropyl alcohol and 0.5 volumes of RNA precipitation solution for purity improvement (Green et al., 2012). Total RNA was DNase digested with the Turbo DNA-free kit (Applied Biosystems, Foster City, CA, USA) and quality and quantity were assessed by gel electrophoresis and on a NanoDrop spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). All RNA samples were diluted to 200 ng µl⁻¹ using an epMotion 5070 (Eppendorf, Hamburg, Germany). Reverse transcription was performed on 1 µg total RNA using Superscript III (Invitrogen) with 25 µmol l⁻¹ oligo(dT)₂₀ and 25 µmol l⁻¹ random hexamers.

Partial fragments of CRCN gene sequences were isolated using several penaeid degenerate CRCN-A or CRCN-C primer pairs (Table 2) designed on a region of sequence conserved across the Pacific white shrimp [Litopenaeus vannamei (Boone 1931): LvanCRCN-A, CV468194; LvanCRCN-C, CV468290] and the banana prawn [Fenneropenaeus merguiensis (de Man 1888): FmerCRCN-A, HM370278; FmerCRCN-C, HM370279]. Hypodermal cDNA from each species was prepared as described below, and a PCR amplification product of approximately 560 and 460 bp was obtained for CRCN-A and CRCN-C, respectively (data not shown). PCR fragments were cloned into a pGEM-T Easy (Promega, Madison, WI, USA) and sequenced in both directions using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems). Several other CRCN gene sequences were reported in the NCBI database, from two previous studies in shrimp unrelated to pigmentation (Leu et al., 2007; Robinson et al., 2014). Sequence editing was performed in CLC Main Workbench (CLC Bio, Aarhus, Denmark). New sequences identified in this study have been submitted to GenBank using accession numbers KP790005–KP790007 and MF627611–MF627642, as shown in Table 3.

Protein and nucleotide alignments were performed using CLC Main Workbench (CLC Bio). Nucleotide and protein sequence comparisons across all species were performed on the minimum shared sequence across both A and C genes (Fig. S1). Only short transcripts could be amplified for several CRCN-C isoforms; therefore, the Melicertus latisulcatus MlatCRCN-C1 and MlatCRCN-C2, the Penaeus esculentus PescCRCN-C1 and PescCRCN-C2, and the PmonCRCN-C4 sequences were compared only across the shorter sequence identified. Sequences were trimmed to equal length for phylogenetic tree construction, and separate alignments were performed for the CRCN-A (Fig. S2) and CRCN-C (Fig. S3) protein sequences. Maximum likelihood trees from protein alignments were created using the neighbor joining model and WAG protein substitution models over 100 bootstrap replicates, and the H. americanus CRCN-A (DV774018) or CRCN-C (CN951851) protein translation as outgroups. The tertiary structures of the identified putative carotenoid-binding proteins were generated using the Modeller software from HHPRED alignments on HHpred servers (Eswar et al., 2007; Soding et al., 2005). Verify3D was used to check the quality of the models generated (Eisenberg et al., 1997), where compatibility of a 3D model was analyzed against the corresponding 1D amino acid sequence. The 3D structures were visualized and superimposed using the Matchmaker application of the UCSF Chimera software (Pettersen et al., 2004).

Three CRCN isoforms identified by degenerate PCR from P. monodon (PmonCRCN-A, PmonCRCN-C1 and PmonCRCN-C2) were targeted for quantitative RT-PCR analysis. Expression of each CRCN gene across the moult cycle was assessed in hypodermal tissue only from three P. monodon at each of the moult stages A, B, C, D₀, D₁, D₂ and D_3/4 [determined according to setal staging and epidermal withdrawal in uropods (Promwikorn et al., 2004)], as used previously (Wade et al., 2012). Expression of PmonCRCN genes was analyzed by quantitative RT-PCR using specific Taqman^® (Applied Biosystems) primers and probes as shown in Table 2 using primers located outside the region used to create dsRNA constructs. The relative expression of each gene was normalized across samples using PmonEF1a, then to the average of a pool of all samples (tissue expression and dsRNA) or to the average expression at moult stage A (moult stage samples) and log₂ transformed.

Partial sequences specific to the PmonCRCN-A, PmonCRCN-C1 or PmonCRCN-C2 genes were amplified using primers shown in Table 2, cloned into pGEM-T easy (Promega) then sub-cloned into the dual T7 dsRNA expression vector L4440 (Addgene, Cambridge, MA, USA) according to the manufacturer’s instructions. To synthesize dsRNA, vector primers were designed to amplify the target genes contained within each plasmid, then the specific PCR products were converted to dsRNA using the MEGAscript T7 kit (Applied Biosystems) according to the manufacturer’s instructions. Integrity and size of newly synthesized dsRNAs were quantified by 1.5% agarose/EtBr gel electrophoresis (data not shown). Luciferase (Luc) dsRNA was used as an exogenous downregulation control, as used previously (De Santis et al., 2011; Sellars et al., 2011).

Juvenile shrimp, P. monodon, were obtained from commercial farms and maintained at CSIRO Agriculture and Food laboratories at Bribie Island Research Centre. For all trials, filtered seawater was heated then pumped through the tanks at 0.8 l min⁻¹, maintaining water temperatures at 28°C and salinity at 35 g l⁻¹. Ten animals were held in each of six red polyethylene tanks that held 80 litres seawater in each. Red tanks were used as they had previously been observed to produce an intermediate coloured shrimp. The experiment was conducted indoors under low artificial light conditions and a 12 h:12 h light:dark photoperiod. Initial mass of the animals across all treatments was 6.13±0.63 g and shrimp in each tank were fed once per day on a commercial diet.

Shrimp (15 per treatment) were tail-muscle injected with the four experimental treatments: Luc-dsRNA (5 µg), PmonCRCN-A-dsRNA (5 µg), PmonCRCN-C1-dsRNA (5 µg) and PmonCRCN-C2-dsRNA (5 µg) in a total of 50 µl of shrimp saline solution (10 mmol l⁻¹ HEPES, 450 mmol l⁻¹ NaCl, 10 mmol l⁻¹ KCl, 10 mmol l⁻¹ EDTA pH 7.2–7.5). Shrimp colour change over time was quantified on days 0, 2, 4 and 7 using digital images as outlined previously (Wade et al., 2015, 2014). Two shrimp were randomly sampled from each treatment on day 4 and dissected for RNA extraction and quantitative RT-PCR analysis as outlined above. The remaining animals were photographed for colour quantification on day 7 and retained frozen. For microscopy analysis, the shell was removed from abdominal segments and chromatophores were photographed immediately using a Leica M165C stereomicroscope fitted with a Canon EOS 5D digital camera.

Where comparison between individual measurements was required, statistical significance was assessed by single-factor ANOVA, followed by Fisher's test allowing 5% error. Statistical significance of the type of dsRNA injection, time after injection or an interaction of these two variables was assessed by two-way ANOVA, followed by Fisher's test allowing 5% error. All statistical analyses were performed using StatPlus:mac 2009 (AnalystSoft Inc., Walnut, CA, USA).

A 518 bp fragment was isolated by degenerate PCR from P. monodon and sequenced. Seven published sequences (GO068457, GO069205, GO071154, GO077931, GO080976, JR225989 and FJ498898) aligned with CRCN-A1 identified in this study, and were combined with the identified sequence to produce a 607 bp fragment, which included a 555 bp region that encoded a 185 amino acid (aa) PmCRCN-A1 protein. Evidence of another CRCN-A isoform was also identified in silico, with a group of five sequences (GO077347, GO078698, GO079441, GO080966 and JR220102) that aligned to produce a 622 bp CRCN-A2 sequence, which included a 558 bp region that encoded a 186 aa protein. The CRCN A1 and A2 proteins shared 87% identity, including a small 2 aa deletion after residue D71. A further four sequences (DT624270, GO076474, JR226983 and JR215171) were identified that were strongly homologous with CRCN-A. Together they aligned to form a 622 bp fragment that encoded a 189 aa protein that was highly similar to PmonCRCN-A2 except it did not contain the deletion noted above. However, this putative PmonCRCN-A3 isoform contained a large amount of sequence variation and resulted in five unknown amino acids within the coding sequence, which made alignment with each other and other isoforms difficult. Despite the presence in silico, no evidence of the expression of the potential PmonCRCN-A2 or PmonCRCN-A3 isoforms was detected by degenerate PCR of hypodermal tissue in this study.

Two PmonCRCN-C sequences were identified by degenerate PCR, and combined with other publicly available sequences to resolve four putative CRCN-C isoforms, named PmonCRCN-C1, PmonCRCN-C2, PmonCRCN-C3 and PmonCRCN-C4. Four published sequences (JR203011, JR227352, GO070224 and FJ498904) matched with PmonCRCN-C1, while one (GO080326) matched with PmonCRCN-C2. Five EST sequences (GO070135, GO075578, GO075579, GO076593 and GO077222) formed a distinct potential PmonCRCN-C3 isoform, and two remaining sequences (JR203011 and FJ498904) formed a potential PmonCRCN-C4 isoform (Table 3). For the regions of overlap, PmonCRCN-C1 shared 75%, 74% and 78% protein identity with the PmonCRCN-C2, PmonCRCN-C3 and PmonCRCN-C4 isoforms, respectively (Fig. S1). Meanwhile, PmonCRCN-C2 shared 90% protein identity with the PmonCRCN-C3 isoform and 72% identity with the PmonCRCN-C4 isoform. The PmonCRCN-C3 isoform was 68% similar to the PmonCRCN-C4 isoform at the protein level. Similar to the PmonCRCN-A2 isoform, the presence in silico of the PmonCRCN-C3 or PmonCRCN-C4 isoforms could not be confirmed by RT-PCR.

Degenerate PCR products were amplified from hypodermal tissue from 12 species of penaeid shrimp. These products were sequenced, and along with in silico analysis of transcriptomic data produced a total of 35 unique CRCN sequences (Table 3). A CRCN-A and CRCN-C subunit was identified in eight of those species, whereas only CRCN-C was identified in Metapenaeus bennettae Racek and Dall 1965, Parapenaeopsis sculptilis (Heller 1862) and Trachypenaeus anchoralis (Bate 1888), and only CRCN-A was identified in Metapenaeus endeavouri (Schmidt 1926). However, the absence of detection of either isoform does not indicate that the gene does not exist in that species, just that it was not found in this study.

In 10 of the 12 species analyzed, more than one copy of the CRCN-A or CRCN-C gene was identified by degenerate PCR. This included up to a total of six isoforms in Metapenaeus ensis (de Haan 1850) and P. monodon, and four isoforms in three other species. Only three isoforms were identified by degenerate PCR in P. monodon, while others were identified by resolution of high-throughput sequence data as outlined above. Regardless of species, the CRCN-A and CRCN-C genes shared approximately 50% nucleotide similarity and 30% amino acid similarity, in line with levels observed in clawed lobsters (Fig. S1). Sequences duplicated within the CRCN-A or CRCN-C isoforms displayed between 72.1 and 99.8% nucleotide and 68.2 and 99.3% protein similarity with other sequences from the same species. In several examples, sequences were more similar between species than within species. Indeed, no nucleotide or amino acid variance was detected between the sequences determined for PlonCRCN-C1 and MensCRCN-C1.

The aligned protein sequences were trimmed and used to create two independent maximum likelihood trees for the CRCN-A and CRCN-C genes using the corresponding clawed lobster H. americanus (Hamer) sequence as a root node (Fig. 1). There was clear separation between intraspecific isoforms, including those with high sequence homology such as MensCRCN-A2 and A3, PscuCRCN-C1 and C2, and PmonCRCN-C2 and C3. There was evidence of both lineage-specific and cross-lineage duplication of both the CRCN-A and CRCN-C genes. There did not appear to be a clear duplication event that reflected a recent common ancestor, and the present nomenclature did not attempt to reflect that evolution. One prevailing observation was the broad separation of the CRCN-C genes into two clades (Fig. 1B). This may suggest that the majority of penaeid shrimp would be expected to have one copy of the CRCN-A gene and at least two copies of the CRCN-C gene, with clear evidence that more are present in some species.

Protein sequences from the identified CRCN-A and CRCN-C isoforms were aligned with their respective sequences using the clawed lobster HamerCRCN-A and HamerCRCN-C genes as a reference (Figs S2 and S3, respectively). The translated CRCN sequences are known to contain a short signal peptide and these were removed from the alignments. Protein sequence similarity with other penaeid shrimp was typically high, between 80 and 90% for CRCN-A and 70 and 75% for CRCN-C. This included absolute conservation of critical cysteine (HamerCRCN-A: C12-C120, C47-C171; HamerCRCN-C: C12-C121, C51-C173) and tryptophan (HamerCRCN-A: Y52) and histidine (HamerCRCN-C: H92) residues required for protein folding and interaction with the carotenoid chromophore. Broad regions of absolute sequence conservation were identified within penaeid shrimp, such as FAAPYEVIETDYDSYSCVYSC for CRCN-A and APYVILDTDYENFSCIYSC for CRCN-C.

Given that the 3D structure has been resolved for clawed lobsters (PDB ID: 1i4u; Gordon et al., 2001), the 3D structure of the newly resolved sequences was predicted using Modeller. Structures for each gene were overlayed on each other using Chimera software and a different colour for each gene. Despite significant sequence changes, structural predictions revealed a complete preservation of tertiary structure for both the CRCN-A (Fig. 2A) and CRCN-C proteins (Fig. 2B). Any regions that contained variability or small deletions had minimal impact on tertiary structure, apart from slight shifts in loop structures on the exterior of the structure. The hydrophobic pocket that forms to enclose the carotenoid was completely preserved, as was the number and location of internally facing hydrophobic residues (Fig. 2C).

Three CRCN isoforms detected by degenerate PCR from P. monodon (A, C1 and C2) were used to determine any potential isoform-specific expression patterns, and specific quantitative PCR primers were designed for each isoform. Across the moult cycle, the expression of the PmonCRCN-A and PmonCRCN-C2 was stable (Fig. 3A), and was not significantly different. However, the expression of the PmonCRCN-C1 isoform was significantly less expressed during the D₀ and D₁ intermoult period (Fig. 3A). Relative gene expression within the same tissue showed that PmonCRCN-A was approximately twofold more highly expressed than PmonCRCN-C2 in shrimp hypodermal tissue, with twofold less expression of PmonCRCN-C1 relative to PmonCRCN-C2 (Fig. 3B).

The RGB colour space is an additive colour space based on varying levels of the pigments red (R), green (G) and blue (B). The RGB values can be combined to produce a single colour that is representative of the average colour of a particular treatment that includes all the biological replicates (each box shown in Fig. 4). The average RGB values quantified for each group of replicates demonstrated that injection of dsRNA specific for each of the PmonCRCN isoforms produced a strong phenotypic colour change (Fig. 4A). Shrimp colour visibly changed, both relative to the type of dsRNA injected and over time. Analysis of the RGB values by two-way ANOVA showed that the type of dsRNA injected produced a significant change in R (F=42.6, P<0.001), G (F=8.1, P<0.001) and B values (F=9.1, P<0.001; Table S1). Time after injection produced a significant change in R (F=29.7, P<0.001), G (F=8.1, P<0.001) and B values (F=9.1, P<0.001), while a significant interaction effect was found between R (F=8.6, P<0.001) and G values (F=3.0, P=0.01). Injection with PmonCRCN-A dsRNA caused a significant increase in R and B values compared with values prior to injection (Table S1). Injection with PmonCRCN-C1 dsRNA only caused a significant increase in R values, while injection with PmonCRCN-C2 dsRNA caused a significant increase in R and G values, relative to values prior to injection (Table S1). Injection of animals with Luc dsRNA had no significant effect on RGB values (Table S1).

The specific downregulation of each isoform by their respective dsRNA target was confirmed by quantitative RT-PCR (Fig. 4B). Injection of dsRNA against each gene caused a significant three- to fourfold reduction in expression of the target gene relative to the Luc dsRNA-injected control or the other PmonCRCN genes (Fig. 4B). Injection of PmonCRCN-C2 dsRNA also caused a significant reduction in expression of the PmonCRCN-C1 gene, although expression of the PmonCRCN-C2 gene was further reduced. As a result of dsRNA injection, hypodermal chromatophores reflected the hypodermal pigment changes (Fig. 5). Blue pigments were depleted in animals injected with PmonCRCN-A (Fig. 5D–F), PmonCRCN-C1 (Fig. 5G–I) or PmonCRCN-C2 dsRNA (Fig. 5J–L) compared with Luc dsRNA-injected controls (Fig. 5A–C). Injection of PmonCRCN-C1 dsRNA removed blue pigment from some chromatophores (Fig. 5G–I, white arrowheads), but most were unaffected. Injection of PmonCRCN-C2 dsRNA removed blue pigment from most of the chromatophores, with only a small number maintaining their distinct blue pigment (Fig. 5J–L, white arrowheads). A similar effect on pigmentation was recorded in the freshwater shrimp Macrobrachium rosenbergii, where injection of animals with the MrosCRCN-C isoform completely removed blue pigmentation from hypodermal chromatophores (Yang et al., 2011), but this pigmentation effect was not analyzed in response to CRCN-A dsRNA injection, or using specific CRCN isoforms.

This study demonstrated that the CRCN genes that control colour production have undergone significant expansion and diversification in penaeid shrimp. However, despite this duplication and sequence divergence, their structure and functional role in shrimp colour production has been strictly conserved. Using P. monodon as an example, we show that multiple CRCN isoforms play a role in forming colours in the pigmented hypodermis of shrimp, that there is differential expression of these isoforms, and that subtle differential phenotypes are produced by the specific downregulation of each isoform. These duplications are unlikely to be maintained in a population unless they differ in some aspect of their efficacy, mutation rate or function (Hughes, 1994; Nowak et al., 1997). In the case of CRCN, the sequence of both CRCN-A and CRCN-C genes have been strictly conserved, suggesting that both subunits are required to maintain overall function. Multiple copies of the CRCN genes within species, particularly evident for CRCN-C, may allow increased expression of CRCN and potentially be advantageous for colour production. The ability of crustaceans to produce cryptic colours is critical for adaptive camouflage (Wade et al., 2009), although may not be as critical for penaeid shrimp that are often buried in muddy or sandy substrates.

To date, only one CRCN-A and one CRCN-C gene was thought to encode the various CRCN protein isoforms that have been identified in a range of crustaceans (Chayen et al., 2003; Wade et al., 2009). In a comprehensive transcriptomic assessment of pigmentation genes in the banana shrimp (F. merguiensis), only single copies of the FmerCRCN-A and FmerCRCN-C genes were identified (Ertl et al., 2013). The present study has identified several duplications of the FmerCRCN-C gene that were expressed within hypodermal tissue. Under the current de novo assemblies of transcriptome data, this type of isoform-specific information is compressed and lost, to the detriment of our understanding of gene function. This study also putatively identified two further PmonCRCN-A (A2 and A3) and two PmonCRCN-C (C3 and C4) isoforms from in silico data. The presence of these other isoforms and any potential function in shrimp pigmentation awaits confirmation. The sequence difference between the MensCRCN-A2 and MensCRCN-A3 isoforms was eight nucleotides (99.8%) and four amino acids (99.3%). Such small sequence variation between isoforms or in silico assemblies may not have arisen from gene duplications, but may represent sequencing errors, normal polymorphic variation or the development of specific-sequence variations within individuals from disparate populations. A complete species genome may resolve some of these questions about isoforms, something that may be possible in the near future for some species. In any case, there is ample evidence from the number of sequences identified, and from the functional downregulation of the different P. monodon isoforms, to demonstrate that significant lineage-specific duplications have occurred.

The present study suggests that, at least in P. monodon, the β-crustacyanin is likely a heterodimer requiring the presence of both CRCN-A and CRCN-C isoforms. The downregulation of PmCRNC-C1 expression removed the blue pigment from a small number of hypodermal chromatophores (Fig. 5G–I, arrowheads). Meanwhile, the blue pigment was retained within a complementary number of chromatophores after downregulation of the PmCRCN-C2 isoform (Fig. 5J–L, arrowheads). This result was consistent with the lower level of endogenous expression of the PmonCRCN-C1 isoform, and the more subtle colour phenotype of the animals injected with PmonCRCN-C1 dsRNA. This result demonstrates that two different blue dimeric β-crustacyanin subunits are forming between PmonCRCN-A/C1 and PmonCRCN-A/C2, but essentially these CRCN-C subunits are performing the same function. This concept is supported by a past study with recombinant CRCN protein where both subunits combined to produce the strongest blue colour when reconstituted with astaxanthin (Ferrari et al., 2012), although β-crustacyanin homodimers of either CRCN-A or CRCN-C were also possible in vitro.

A small amount of blue colour also remained in the PmonCRCN-A dsRNA-injected animals, potentially because of incomplete downregulation of this gene in this tissue, as the expression of this gene was two- to fourfold higher than the other CRCN genes. Alternatively, one of the other in silico PmonCRCN-A isoforms may be present within these tissues that was not identified by degenerate PCR. The inability to detect other isoforms by PCR in this study does not indicate that they do not exist, simply that their expression is extremely low or that they have different spatial and/or temporal expression patterns. Irrespective of in silico analysis, this study demonstrates that there has been significant functional duplication of CRCN in penaeid shrimp. This study expands our understanding of the origin and evolution of pigmentation in crustaceans, and the mechanisms by which gene function is maintained, diversified or subfunctionalized.

The authors thank Richard Thaggard, Kiname Salee, Nick Polymeris and Megan Dearnley for trial maintenance, injections and sampling. The authors also thank Carmel McDougall and Melony Sellars for considered review of this manuscript.