Integrated Application of Transcriptomics and Metabolomics Provides Insights Into Unsynchronized Growth in Pearl Oyster Pinctada Fucata Martensii

Science of the Total Environment 666 (2019) 46–56

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Integrated application of transcriptomics and metabolomics provides insights into unsynchronized growth in pearl oyster Pinctada fucata martensii

Ruijuan Hao a, Xiaodong Du a,b, Chuangye Yang a,YuewenDenga,b,⁎, Zhe Zheng a,⁎, Qingheng Wang a,b a Fisheries College, Guangdong Ocean University, Zhanjiang 524088, China b Pearl Breeding and Processing Engineering Technology Research Centre of Guangdong Province, Zhanjiang 524088, China

HIGHLIGHTS GRAPHICAL ABSTRACT

• Metabolomics and transcriptomic were applied to analyze the bivalve unsynchronized growth. • Fast-growing group hold high biomineralization activity. • Slow-growing group consumed more energy in response to environment stress. • Fast-growing group exhibited high digestion and anabolic ability. • Fast-growing group possessed high osmotic regulation ability.

article info abstract

Article history: Similar to other marine bivalves, Pinctada fucata martensii presents unsynchronized growth, which is one of the Received 5 October 2018 problems farmers currently face. However, the underlying mechanisms have not been studied. In the present Received in revised form 13 February 2019 study, pearl oyster P. f. martensii from cultured stocks were selected to produce a progeny stock. At 180 days, Accepted 14 February 2019 the stock was sorted by size, and fast- and slow-growing individuals were separately sampled. Then, Available online 15 February 2019 metabolomic and transcriptomic approaches were applied to assess the metabolic and transcript changes be- Editor: Julian Blasco tween the fast- and slow-growing P. f. martensii groups and understand the mechanism underlying their unsynchronized growth. In the metabolomics assay, 30 metabolites were considered signiﬁcantly different metabolites Keywords: (SDMs) between the fast- and slow-growing groups and pathway analysis indicated that these SDMs were in- Unsynchronized growth volved in 20 pathways, including glutathione metabolism; sulfur metabolism; valine, leucine, and isoleucine bio- Metabolomics synthesis; and tryptophan metabolism. The transcriptome analysis of different growth groups showed 168 Transcriptomic differentially expressed genes (DEGs) and pathway enrichment analysis indicated that DEGs were involved in ex- Pinctada fucata martensii tracellular matrix-receptor interaction, pentose phosphate pathway, aromatic compound degradation. Inte- grated transcriptome and metabolome analyses showed that fast-growing individuals exhibited higher biomineralization activity than the slow-growing group, which consumed more energy than the fast-growing group in response to environmental stress. Fast-growing group also exhibited higher digestion, anabolic ability, and osmotic regulation ability than the slow-growing group. This study is the ﬁrst work involving the integrated metabolomic and transcriptomic analyses to identify the key pathways to understand the molecular and metabolic mechanisms underlying unsynchronized bivalve growth. © 2019 Elsevier B.V. All rights reserved.

⁎ Corresponding authors at: Guangdong Ocean University, China. E-mail addresses: [email protected] (Y. Deng), [email protected] (Z. Zheng).

1. Introduction worldwide. However, the growth of this species is also unsynchronized. It is reasonable to hypothesize that the unsynchronized growth of P. f. Growth performance is a major problem for aquaculture to increase martensii might result from the endogenic metabolic level differences. the production of shellfish farms. Researchers explored the factors Therefore, we utilized gas chromatography time-of-flight mass spec- influencing the growth performance of bivalves, such as genotype trometry (GC-TOF/MS)-based metabolomics approach, together with (Wada and Komaru, 1994; J.M. Yang et al., 2018), stocking density transcript analysis, to investigate the difference in metabolite profiles (Taylor et al., 1997), mesh covers (Taylor et al., 1998), salinity (Taylor between the fast- and slow-growing P. f. martensii. The connection net- et al., 2004), temperature (Wang et al., 2012), and food (Mills, 2000). works were mapped based on correlations between metabolites and However, for many marine invertebrates, bivalve growth is unsynchro- regulatory genes associated with growth performance difference. The nized and subject to considerable variability. Namely, differences in findings will provide new insights into the study of the molecular mech- growth rate are exhibited by different individuals under similar envi- anisms associated with growth traits and highlight the significance of an ronmental conditions (Brown, 1988; Tamayo et al., 2011; Hao et al., integrated approach for this research. 2018; Venter et al., 2018; Wang et al., 2019), which are one of the problems farmers currently face. These differences have often been attrib- 2. Materials and methods uted to unexplained variance and given comparatively less research attention (Tamayo et al., 2011). However, a comprehensive overview 2.1. Experimental animals between the genes and the metabolites controlling pearl oyster growth performance is unclear. And this require an in-depth study to shorten The fifth-generation selected line of P. f. martensii for rapid growth the growth duration and eliminate the variability which would indeed was used in the experiment. The selected line was designated as reduce production costs and increase turnover in the long run (Ten “Haixuan NO 1”, and the development of the line was described in detail Doeschate and Coyne, 2008). by Du et al. (2015). A total of 48 mature animals (female: male = 28:20) Rapidadvancesintechnologyhavemadenext-generationsequenc- were used to produce the progeny stock. In October 2017, breeders ing suitable to conduct large-scale studies to identify candidate genes were mass spawned in the hatchery. The procedures for larval, juvenile, related to different phenotypes (Rondon et al., 2016; Wang et al., and adult rearing were detailed by Deng et al. (2009).Briefly, fertilized 2018a). However, phenotype is not only regulated by genes but also eggs were incubated in 200 L polyethylene tanks until the D stage, regulated at multiple other levels, including at the metabolite level. which occurred 24 h after fertilization. Then, larvae were transferred Transcriptome sequencing can provide important information on the to 1000 L polyethylene tanks. The density was maintained at 1 individ- gene expression level, which facilitates functional genomic studies, in- ual/mL. The water temperature was set at 28 ± 1 °C, and salinity was set cluding global gene expression, novel gene discovery, and full-length at 30.0 ± 0.5‰. Daily feeding consisted of Isochrysis galbana from day 2 gene assembly (Vera et al., 2008; Salem et al., 2010; Huang et al., to day 7 and a mixture of I. galbana and I. zhanjiangensis from day 7 to 2011; Qiao et al., 2018; Zhao et al., 2012). However, transcriptome se- day 60. At day 25, plastic plates were provided as the substrate for meta- quencing fails to study the real metabolite levels in organisms, thereby morphosis. At day 60, large individuals that are 3–5 mm in size were re- causing difficulty in confirming the critical pathways responsible for moved from the plates, transferred into net cages, and suspended in the regulating specific traits. Meanwhile, metabolomics is an analytical ap- sea of Liushagang, Zhanjiang. The shells were cleaned and placed in new proach used to study metabolites and understand the physiological nets at appropriate intervals. and biochemical statuses of a biosystem relative to phenotype (Cappello et al., 2017a; Hao et al., 2018; C.Y. Yang et al., 2018; Yang et al., 2019; Hayden et al., 2019; Jiang et al., 2019). For example, Hao 2.2. Growth measurement and sampling et al. (2018) reported that pearl oyster Pinctada maxima changes its metabolic status with different growth performances and showed that At 180 days, the stock was sorted by size, and fast- and slow- glycine, serine, and threonine metabolism and glutathione metabolism growing individuals were separately sampled. The shell length, shell made their contribution for the growth of oysters. This approach has width, and shell height of each sample were measured using a digital been used to study the global metabolites in cells, tissues, and biofluids caliper (0.02 mm accuracy). The shell weight and total weight of each of living systems and understand the physiological and biochemical sta- sample were measured using an electronic balance (0.01 g accuracy). tuses of biosystems with further biological principle interpretation (Cappello et al., 2017a; Hao et al., 2018; C.Y. Yang et al., 2018). There- 2.3. Metabolite extraction, detection, and analysis fore, using metabolomics together with transcriptomics to study the variation in growth is an effective method to provide insight into the Adductor muscles from each sample were dissected from fast- and unsynchronized growth of pearl oyster. The integration of these large- slow-growing individuals and stored in liquid N. A total of 10 adductor scale datasets, including transcriptome and metabolome, has been ap- muscles were collected from each group. Approximately 50 ± 1 mg plied successfully in other aquatic animals (Gracey and Connor, 2016; sample was placed into the 2 mL microcentrifuge tubes and extracted

Xu et al., 2016; Li et al., 2017). Gracey and Connor (2016) utilized tran- with 480 μL of extraction liquid (VMethanol:VChloroform = 3:1). Subse- scriptome and metabolome to characterize spontaneous metabolic cy- quently, 10 μL of L-2-chlorophenylalanine (1 mg/mL stock in dH2O) cles in Mytilus californianus under subtidal conditions. Xu et al. (2016) was added as an internal standard and vortex mixed for 30 s. The solu- provided insights into population-asynchronous ovary development in tion was homogenized in a ball mill for 4 min at 45 Hz and then ultra- Coilia nasus by using the same integrated analysis of transcriptomics sound treated for 5 min (incubated in ice water). The solution was and metabolomics. Li et al. (2017) generated the regulatory networks centrifuged for 15 min at 12000 rpm and 4 °C. Approximately 350 μL associated with the differences in glycogen content by using integrated of the supernatant was transferred into 1.5 mL microcentrifuge tubes metabolome and transcriptome analyses in pacific oysters with differ- (every two tubes constitute 1 mixed sample). The sample was dried ent intrinsic glycogen levels. completely in a vacuum concentrator without heating. Subsequently, Pearl oyster Pinctada fucata martensii is naturally distributed in the 80 μL of methoxy amination hydrochloride (20 mg/mL in pyridine) equatorial zone between Tropic of Cancer and the Tropic of Capricorn was incubated for 30 min at 80 °C. Approximately 100 μLofthebis- of the Indo-Pacific and Western Atlantic regions, such as South China, (trimethylsilyl)-trifluoroacetamide regent (1% trimethylchlorosilane, Japan, Australia, and Southeast Asia (Southgate and Lucas, 2008). P. f. v/v) was added to the sample aliquots and incubated for 1.5 h at 70 martensii is cultured for round pearl production. In China, the pearls °C. All samples were analyzed by a gas chromatograph system coupled produced by this species are called South China sea pearls and popular with a Pegasus HT time-of-flight mass spectrometer (GC-TOF-MS). 48 R. Hao et al. / Science of the Total Environment 666 (2019) 46–56

Fig. 1. Growth trait comparison of Pinctada fucata martensii between fast- and slow-growing groups. “**” indicated the signiﬁcant difference among the growth traits in the two groups (P b 0.01). t-test was used to analyze the difference in the growth traits between fast- and slow-growing groups.

GC-TOF-MS analysis was performed using an Agilent 7890 gas chro- data distribution. To obtain a high level of group separation and im- matograph system and the system utilized a DB-5MS capillary column proved understanding of the variables responsible for classification, coated with 5% diphenyl crosslinked with 95% dimethylpolysiloxane we conducted supervised OPLS-DA. Sevenfold cross validation was (30 m × 250 μm inner diameter, 0.25 μm film thickness; J&W Scientific, used to estimate the model robustness and predictive ability and vali- Folsom, CA, USA). A 1 μL aliquot of the analyte was injected in splitless date the model further. The OPLS-DA model was used with the first mode. He was used as the carrier gas. The front inlet purge flow was principal component of variable importance in projection (VIP) values 3mLmin−1,andthegasflow rate through the column was (VIP N 1) combined with Student's t-test (t-test) (P b 0.05) to determine 1mLmin−1. The initial temperature was maintained at 50 °C for the SDMs among the pairwise comparison groups. Kyoto Encyclopedia 1 min, increased to 310 °C at a rate of 10 °C min−1, and then maintained of Genes and Genomes (KEGG, http://www.genome.jp/kegg/) was uti- for 8 min at 310 °C. The injection, transfer line, and ion source temper- lized to search for the metabolite pathways. MetaboAnalyst, which is a atures were 280 °C, 280 °C, and 250 °C, respectively. The energy was free and web-based tool that uses high-quality KEGG metabolic path- −70 eV in the electron impact mode. The mass spectrometry data way as the backend knowledge base, was used for the pathway analysis were acquired in full-scan mode with the m/z range of 50–500 at a (http://www.metaboanalyst.ca). rate of 20 spectra per s after a solvent delay of 6.17 min. Chroma TOF 4.3× software of LECO Corporation and LECO-Fiehn 2.5. RNA sequencing data analysis Rtx5 database were used for raw peak extraction, data baseline filtra- tion, baseline calibration, peak alignment, deconvolution analysis, The adductor muscle of ten pearl oysters from each group (every peak identification, and integration of the peak area (Kind et al., two individuals for one sample) was used to extract RNA for the RNA se- 2009). The retention time index (RI) method was used for peak identi- quencing of fast- and slow-growing groups. mRNA from the pearl oyster fication, and the RI tolerance was 5000. The similarity value (SV) for tissues (i.e., fast- and slow-growing oysters) was enriched by Oligo(dT) evaluating the accuracy of the discriminating compound was obtained beads. Then, the enriched mRNA was fragmented into short fragments from the LECO/Fiehn Metabolomics Library to identify the compounds. by using a fragmentation buffer and reverse transcribed into cDNA An SV N700 indicated that metabolite identification was reliable, with random primers. Second-strand cDNA was synthesized by DNA whereas an SV b200 meant that the identified metabolite as “unreli- polymerase I, RNase H, dNTP, and buffer. Then, cDNA fragments were able.” A compound with a SV between 200 and 700 was considered a purified with QiaQuick PCR extraction kit, end repaired, added with putative annotation. poly(A), and ligated to Illumina sequencing adapters. The ligation prod- ucts were size-selected by agarose gel electrophoresis, amplified by 2.4. Identification of significantly different metabolites (SDMs) PCR, and sequenced using Illumina HiSeq™ 2500. To obtain high- quality clean reads, we filtered the reads by removing those containing The resulting 3D data involving the peak number, sample name, and adapters, those with N10% of unknown nucleotides (N), and those with normalized peak area were inputted to the SIMCA 14.1 software pack- N50% of low-quality (Q-value≤20) bases. Short reads alignment tool age (V14.1, MKS Data Analytics Solutions, Umea, Sweden) for principal Bowtie2 (V 2.2.8, https://www.nature.com/articles/nmeth.1923) component analysis (PCA) and orthogonal projections for latent (Langmead and Salzberg, 2012) was used to map reads to the rRNA da- structures-discriminant analysis (OPLS-DA). PCA showed the original tabase and remove rRNA-mapped reads. Then, the rRNA-removed reads

Fig. 2. Derived PCA score plot, derived OPLS-DA score plots, and corresponding validation plots of OPLS-DA from the GC-TOF-MS metabolite profiles of P. f. martensii. FG represents the fast- growing group, and SG represents the slow-growing group. R. Hao et al. / Science of the Total Environment 666 (2019) 46–56 49 50 R. Hao et al. / Science of the Total Environment 666 (2019) 46–56 of each sample were mapped to the reference genome (Du et al., 2017) 3. Results by TopHat2 (V2.1.1) (Kim et al., 2013). The alignment parameters were as follows: maximum read mismatch as 2, the distance between mate- 3.1. Growth performance of P. f. martensii pair reads as 50 bp and the error of distance between mate-pair reads as ±80 bp. The mean shell length, shell width, shell height, total weight, and shell weight of the fast-growing group were significantly higher than those of the slow-growing group at the end of the experiment (P b 2.6. Transcript reconstruction and abundance quantification 0.01; Fig. 1 and Table S1).

The reconstruction of transcripts was carried out with software 3.2. Metabolome analysis of fast- and slow-growing P. f. martensii groups Cufflinks (V2.2.1) (Trapnell et al., 2012). The program reference annotation-based transcripts were preferred (Du et al., 2017; http:// A total of 706 valid peaks were extracted via GC-TOF/MS. Gross gigadb.org/dataset/100240). Cufflinks constructed faux reads according changes in metabolic physiology were easily detected using the unsu- to the reference to compensate for the influence of low coverage pervised PCA of the entire set of measured analytes (Fig. 2A). The R2X sequencing. During the last step of assembly, all reassembled fragments value of the PCA model accounting for the variance was 0.562, which were aligned with reference genes, and then similar fragments were showed considerable separation between fast- and slow-growing removed. Then, we used Cuffmerge (V1.0.0) to merge transcripts from groups. To maximize the discrimination between the two groups, we different replicates of a group into a comprehensive set of transcripts used OPLS-DA to elucidate the different metabolic patterns (Fig. 2B). and the transcripts from multiple groups into a final comprehensive The parameters considered for classification from the software were set of transcripts for further downstream differential expression R2X = 0.403, R2Y = 0.987, and Q2 = 0.696, all of which were stable analysis. Gene abundances were quantified by the software RSEM and effective for fitness and prediction. The R2 and Q2 intercept values 1.2.19 (Li and Dewey, 2011) and the gene expression level was normal- determined after 200 permutations were 0.99 and −0.26, respectively ized by using FPKM (Fragments Per Kilobase of transcript per Million (Fig. 2C). The low Q2 intercept values indicate that the robustness of mapped reads) method. The FPKM method is able to eliminate the the models presented low overfitting and reliability risks. All samples influence of different gene lengths and sequencing data amount on in the score plots were within the 95% Hotelling's T-squared ellipse, the calculation of gene expression. Therefore, the calculated gene and clear separation and discrimination were found between the expression can be used to compare the difference in the gene expression pairwise groups. The OPLS-DA model can identify differences between among samples. the pairwise groups and be used in subsequent analyses. A total of 264 metabolites were quantified, and 85 metabolites were identified by mass spectrum matching, with a spectral similarity value 2.7. Transcriptome data analysis of N700 (Table S2). FC values were utilized to determine the specific variable quantities between fast- and slow-growing groups. Metabolite dis- Principal component analysis (PCA) was performed with R pack- tribution can be visually divided into up- and downregulated age gmodels (http://www.r-project.org/) for sample relationship metabolites. On the basis of the OPLS-DA results, among 264 metabo- analysis. To identify differentially expressed genes across groups, lites, we determined 30 SDMs (VIP N 1andP b 0.05) between fast- the edgeR package (V3.12.1, http://www.rproject.org/)wasused. and slow-growing groups (Fig. 3A and Table S3). Among these 30 We identified genes with FDR b 0.05 and |log2FC| N 1inacompari- SDMs, 4 yielded higher concentrations in fast-growing group than in son as significant DEGs. Then, DEGs were then subjected to enrich- the slow-growing group (Fig. 3A and Table S3). By contrast, 26 metabo- ment analysis of KEGG pathways. KEGG is the major public lites in the fast-growing group were significantly downregulated com- pathway-related database (Kanehisa et al., 2008). Pathway enrich- pared with those in the slow-growing group. ment analysis (http://www.kegg.jp/kegg/kegg1.html)identified To explore the potential metabolic pathways affected by different significantly enriched metabolic pathways in DEGs compared with growth performances, we introduced the SDMs into MetaboAnalyst the whole genome background, and the calculated P-value was de- 4.0. A total of 20 pathways were found when the SDMs between fast- termined through FDR correction, taking FDR ≤ 0.05 as a threshold. and slow-growing groups were introduced into KEGG (Table S4). On Pathways meeting this condition were defined as significantly the basis of both -ln(P-value) and pathway impact scores, we identified enriched pathways in DEGs. Raw transcriptome sequences data that the most relevant metabolic pathways were “glutathione metabo- have been deposited at NCBI Sequence Read Archive (SRA) under lism”, “sulfur metabolism”, “valine, leucine and isoleucine biosynthesis”, accession SRP174853. “glycine, serine and threonine metabolism”,and“tryptophan metabolism” (Fig. 3B).

2.8. Integrative analysis of metabolome and transcriptome 3.3. Transcriptome analysis of the fast- and slow-growing P. f. martensii groups DEGs (FDR b 0.05 and |log2FC| N 1) and SDMs (VIP N 1andP b 0.05) were used for the integrative analysis of fast- and slow-growing cDNAs prepared from the pearl oysters representing the two groups groups. Spearman method was used to analyze the correlation coef- were sequenced individually using an Illumina HiseqTM platform. The ﬁcients for the metabolome and transcriptome data integration. averages of 53.9 and 55.9 million high-quality 150 bp paired-end Heatplot was used to show the connection between genes and reads were mapped to the P. f. martensii genome from the fast- and metabolites. slow-growing oysters, respectively (Table S5). Quality analysis of fast- and slow-growing group transcriptome showed that fast- and slow- growing oysters had an average of 43.78% and 43.46% GC (%) content 2.9. Statistical analysis of clean reads, Q20 of 98.69% and 98.68% and Q30 of 96.02% and 96.03%, respectively. Genes were mapped to the P. f. martensii genome, The growth traits of the fast- and slow-growing groups were and the average map rate were 67.84% for fast-growing group and compared by t-test. The signiﬁcance level in the analyses was con- 68.43% for slow-growing group, respectively (Table S5). The data sidered at P b 0.05. The analysis was carried out with SPSS 19.0 above indicated good sequencing quality, suggesting that the subse- software. quent transcriptome analysis results were reliable. The transcriptomes .Hoe l cec fteTtlEvrnet66(09 46 (2019) 666 Environment Total the of Science / al. et Hao R. – 56

Fig. 3. Hierarchical clustering analysis for SDMs and the metabolomic view map of the significant metabolic pathways of the fast- and slow-growing P. f. martensii groups. (A) The relative metabolite level is depicted according to color scale. Red indicates upregulation, and blue indicates downregulation. FG represents fast-growing group, and SG represents slow-growing group. (B) Significantly changed pathways on the basis of the enrichment and topology analyses. The x axis represents pathway enrichment, and the y axis represents the pathway impact. Large sizes and dark colors represent the major pathway enrichment and high pathway impact values, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) 51 52 R. Hao et al. / Science of the Total Environment 666 (2019) 46–56 of the two pearl oyster groups were compared using PCA, which 4. Discussion allowed an unbiased analysis in a format, in which the groups can be visually and quantitatively compared. Fig. 4A shows that the fast- and Growth performance vary among individuals in natural bivalve pop- slow-growing groups can be well separated. ulations even when exposed to similar environmental conditions A total of 168 DEGs (differentially expressed genes) were identi- (Tamayo et al., 2014; Tamayo et al., 2016). To explore the molecular fied between the two groups (FDR b 0.05 and |log2FC| N 1). The ex- and metabolic mechanisms underlying the interindividual growth vari- pression levels of the genes mentioned are shown in Table S6. ability further, we selected the pearl oyster P. f. martensii from the cul- Compared with the slow-growing group, the fast-growing group tured stocks to produce the progeny stock, which was cultured in the showed 107 and 61 up- and downregulated genes, respectively same environment. After 180 days, the stock was sorted by size, and (Fig. 4B). the fast- and slow-growing individuals were separately sampled. In Pathway enrichment analysis identified significantly enriched order to analyze the pattern of unsynchronized growth for fast- and metabolic pathways in DEGs and showed significantly altered path- slow-growing group, adductor muscle was sampled from each group ways in the transcriptome. We mapped all DEGs to KEGG pathways, for the transcriptome and metabolome detection. Adductor muscles of and 27 significant specific pathways were represented (Table S7 shellfish are the main muscular system in bivalve molluscs and con- and Fig. 4C). Among these 27 pathways, “extracellular matrix nected to their shells and thus control “turn on” and “turn off” actions (ECM)-receptor interaction”, “pentose phosphate pathway” and (Ruegg, 1971; Zheng et al., 2016). Plenty reports showed that many “degradation of aromatic compounds” were the three most repre- genes which expressed highly in the adductor muscle were associated sented pathways. ECM-receptor interaction pathway is an impor- with the growth of oysters (Wang et al., 2018b; Guo et al., 2012)andre- tant pathway in signal transfer, and ECM receptors (collagen) searchers focused on the growth of shellfish also utilized the adductor (XLOC_041210, Pma_10019836) widely exist in the mollusk shell muscle as the detected tissue for the omics analysis (Li and He, 2017; matrix proteins, which showed high expression level in the fast- J.M. Yang et al., 2018). The present paper is aimed to understand the growing group (Table S6). mechanism underlying their unsynchronized growth, so the adductor muscle was selected. On the other hand, haemocytes for example, which play a key role in the innate immune system (e.g., phagocytosis, 3.4. Integrated analysis of metabolome and transcriptome in terms of pearl encapsulation and nacrezation of foreign particles), are commonly used oyster growth difference to investigate immune responses of bivalves to pathogens (Pruzzo et al., 2005; Allam et al., 2006). Bivalve gills are also often analyzed in immuno- Correlation analysis utilized Spearman calculation to show the cor- logical studies, since they accumulate marine pathogens and are targets of relation of the transcriptome and metabolomic data. An overview of infection (Liu et al., 2014; Lu et al., 2016). The whole organism including the DEGs and metabolites found between fast- and slow-growing oys- gonad, whose fat content varies with the development of the gonads. ters is shown in Fig. 5 and showed strong correlations between each And visceral mass and gonad are closely connected and hard to be sepa- transcript and metabolites, which further highlighted the association rated. In order to rule out the interference of the gonads, the whole organ- of each transcript with specific metabolites. ism was also not selected. Therefore, we chose adductor muscle to discuss

Fig. 4. Transcriptome analysis of the fast- and slow-growing pearl oyster groups. (A) PCA score plot derived from the P. f. martensii transcriptome. FG represents fast-growing group, and SG represents slow-growing group. (B) Differential gene expression analyses between fast- and slow-growing groups. Red indicates the upregulated gene expression level of the fast-growing group, green indicates the upregulated gene expression level of the slow-growing group, and black indicated insigniﬁcant differential expression. (C) Pathway enrichment of the differentially expressed genes was analyzed by the pathway enrichment statistical scatter plot (top 20). (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.) R. Hao et al. / Science of the Total Environment 666 (2019) 46–56 53

growth of pearl oyster. And an integrative analysis of the transcriptome and metabolome presented a comprehensive overview between the genes and the metabolites controlling oyster growth performance, as shown in Fig. 6.

4.1. Fast-growing oysters possessed high biomineralization activity

On the basis of the pathway enrichment analysis of DEGs from different pearl oysters, we found that the ECM-receptor interaction pathway was significantly upregulated in the fast-growing group and then upregulated the whole pathway. Integrins are heterodimeric cell adhesion molecules that link the ECM to the cytoskeleton and play important roles in controlling various steps in the signaling pathways that regulate processes such as proliferation, differentiation, apoptosis, and cell migration and implantation, in various species (Bosman and Stamenkovic, 2003; Reddy and Mangale, 2003). Specific interactions between cells and ECM are mediated by integrins. In the present study, integrin-related genes (i.e., XLOC_036496, Pma_10015864, Pma_10027571) were upregulated in the fast-growing group, which indicated an active interaction between the cell and ECM. Cell regulation plays a key role in the mineral formation (Reddy and Mangale, 2003). The WGCNA of the 234 nacre matrix protein genes from P. f. martensii revealed that ECM related proteins exist as hubs of the network, and these coexpressed genes are significantly enriched in the ECM- receptor interaction involved in shell formation (Du et al., 2017). ECM consists of a complex mixture of structural and functional mac- romolecules, including glycosaminoglycans and fibrous proteins (e.g., collagen, elastin, fibronectin, and lammin) (Timpl, 1996; Flier and Sonnenberg, 2001; Mariman and Wang, 2010). Many studies have reported that ECM proteins are associated with biomineralization in many species (Krammer et al., 1999; Little et al., 2008; Bradshaw and Smith, 2014; Huang, 2016). Bone, teeth, and cartilage-containing mineral structures incorporated into the ECM, where a network of protein fibers holding cells together serves as a template for biomineralization (Greene, 2006). Huang (2016) compared the shell matrix proteins of P. f. martensii, Lottia gigantea,andCrassostrea gigas and showed that the multicopy genes of the consensus gene family are mainly ECMs, including laminin, collagen, and cytoskeletal protein myosin. Liuetal. (2015) indicated that the ECM-related proteins of the shell matrix play important roles in shell repair and cell regulation. Therefore, different ECM-related proteins (e.g., Pma_10010976, XLOC_020586, XLOC_041210, and Pma_10019836) and integrins were upregulated in the fast-growing group, which indicated that the fast-growing pearl oyster group possessed high biomineralization activity.

4.2. Slow-growing group consumed high energy in response to environmental stress

Intertidal marine invertebrates, such as pearl oysters, are exposed to a range of environmental contaminants as stressors, including both natural (like marine toxins, water temperature, salinity, tidal/wave action, food availability, and predation) and anthropogenic (for example emerging contaminants), which can both affect their physiology (Kuchel et al., 2012; Cappello et al., 2015; Cappello et al., 2017b). Nonenzymatic small organic molecules, such as glutathione, play important roles in maintaining the cellular redox status (Wen et al., 2018). Pentose phosphate pathway is an alternative route of glucose

Fig. 5. Heatplot of the correlations between P. f. martensii metabolome and transcriptome. The heatplot with genes in columns and metabolites in rows to show the connection between genes and metabolites. The red and blue showed the positively and negative correlation among transcriptomics and metabolomics, respectively. “*” indicates signiﬁcant correlative between the gene and metabolites (P b 0.05). (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.) 54 R. Hao et al. / Science of the Total Environment 666 (2019) 46–56

Fig. 6. Pathway overview of the unsynchronized growth enriched by metabolome and transcriptome in pearl oysters. The differentially expressed gene abbreviations identified by the transcriptome analysis are shown in either red (highly expressed in the fast-growing group) or blue boxes (highly expressed in the slow-growing group). Differentially abundant metabolites identified by metabolome analysis are shown either in red (highly expressed in the fast-growing group) or blue letters (highly expressed in the slow-growing group). Abbreviations: COL, collagen; FN, fibronectin; RELN, reelin; SDC, syndecan; GST, glutathione S-transferase; gnl, gluconolactonase; FBP, fructose-1,6-bisphosphatase. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) catabolism that functions in NADPH formation for biosynthetic that protein catabolism was low in the fast-growing group or that reactions (Moreira et al., 2016). The activity of pentose phosphate they have enhanced protein anabolism, which resulted in the depletion pathway increases with environment stress, such as decreasing of free valine pools, thereby causing low levels of available valine as temperatures (Johnston and Dunn, 1987). In the present study, pentose seen in this study. Previous studies also revealed that the changes in va- phosphate pathway was activated in the slow-growing group along line are consistent with the growth performance of Pinctada maxima with high expression levels of fructose-1,6-bisphosphatase (Hao et al., 2018). Hence, fast-growing group exhibits high digestion (Pma_10006936) and gluconolactonase (Pma_10015184 and and anabolic abilities. Pma_10015186), thereby resulting in large amounts of NADPH, which can promote glutathione regeneration (Fan et al., 2014). This result 4.4. Fast-growing group possessed high osmotic regulation ability agreed with the increase in the glutathione levels in the slow-growing group on the basis of metabolome analysis. Moreover, glutathione S- Free amino acids and their catabolites are used in marine mollusks, transferase (Pma_10015580) is a multifunctional dimeric protein, as well as in other marine invertebrates, as the major osmolytes to bal- which catalyzes the conjugation of the reduced form of glutathione to ance their intracellular osmolarity with the environment (Yancey et al., xenobiotic substrates that are involved in the cellular detoxification of 1982). Glycine is a common osmolyte in marine bivalves (Kube et al., the reactive electrophilic compounds and protect tissues against oxida- 2006). Recent metabolomic studies revealed that glycine content is re- tive damage (Chen et al., 2011). The results of the present study showed lated to the health states of oysters and clams, with reductions in this that the enzyme glutathione S-transferase was highly expressed in the amino acid occurring after infection with pathogens (Liu et al., 2013), slow-growing group. Therefore, as soon as the metabolic energy is con- after exposure to heavy metals (Ji et al., 2013), and under hypo- sumed on maintenance or recovering from stress, minimal energy is osmotic conditions (Meng et al., 2013). Young et al. (2015) and Hao available for the growth of the slow-growing group. The reason for et al. (2018) demonstrated that glycine content is substantially reduced this phenomenon maybe that the slow-growing group individuals be- in poor-quality cohorts, which is similar to the result of the present haved more sensitive to the environmental stressors. study. Thus, glycine contents may influence glutathione metabolism. This result indicated that the fast-growing group possessed high os- 4.3. Fast-growing group exhibited high digestion and anabolic ability motic regulation ability in response to salinity changes in the environment. Myristic acid is a C:14 fatty acid found in high concentrations (13%– 15% of lipids) in the microalgae diet fed to bivalves (Ragg et al., 2010). 5. Conclusions The mean relative abundances of the free myristic acid in the slow- growing group were lower than in the fast-growing group. This result GC–MS-based metabolomics and transcriptomic approaches were may reflect that fast-growing group exhibits high lipid digestive ability. applied to assess metabolite and transcript changes between fast- and Similar phenomenon was also observed in Perna canaliculus (Young slow-growing groups to provide insight into the mechanism underlying et al., 2015). Meanwhile, valine has been identified as an essential the unsynchronized growth of pearl oyster P. f. martensii. Pearl oyster P. amino acid, and is an α-amino acid that is used in the biosynthesis of f. martensii changed its metabolic and transcriptomic statuses with dif- proteins. Valine levels are increased during conditions of increased pro- ferent growth performances. The integrated transcriptome and metab- tein catabolism when mussel is subjected to environmental stress olome analyses showed that fast-growing individuals possessed high (Cappello et al., 2017a). Thus, low valine levels can merely indicate biomineralization activity, and slow-growing group consumed high R. Hao et al. / Science of the Total Environment 666 (2019) 46–56 55 energy in response to environmental stress. Fast-growing group exhib- Greene, M.E., 2006. Biomineralization observed through shear force. Mater. Today 9 (12), 18. ited high digestion, anabolic ability, and osmotic regulation ability. Guo, H., Bao, Z., Li, J., Lian, S., Wang, S., He, Y., Fu, X., Zhang, L., Hu, X., 2012. Molecular char- Therefore, P. f. martensii exhibits unsynchronized growth. This study is acterization of TGF-β type I receptor gene (Tgfbr1) in Chlamys farreri,andtheassoci- the first work on the integrated metabolomic and transcriptomic analy- ation of allelic variants with growth traits. PLoS One 7 (11), e51005. Hao, R.J., Wang, Z.M., Yang, C.Y., Deng, Y.W., Zheng, Z., Wang, Q.H., Du, X.D., 2018. ses to identify the key pathways and provide insight into the molecular Metabolomic responses of juvenile pearl oyster Pinctada maxima,todifferentgrowth and metabolic mechanisms underlying the unsynchronized bivalve performances. Aquaculture 491, 258–265. growth. Hayden, H.L., Rochfort, S.J., Ezernieks, V., Savin, K.W., Mele, P.M., 2019. Metabolomics ap- Supplementary data to this article can be found online at https://doi. proaches for the discrimination of disease suppressive soils for Rhizoctonia solani AG8 in cereal crops using 1H NMR and LC-MS. Sci. Total Environ. 651, 1627–1638. org/10.1016/j.scitotenv.2019.02.221. Huang, R.L., 2016. The Shell Proteomics and Stress Network of Biomineralization in Pearl Oyster. Doctoral dissertation. University of Chinese Academy of Sciences (Institute of Acknowledgement Oceanology). Huang, Y.H., Huang, X.H., Yan, Y., Cai, J., Ouyang, Z.L., Cui, H.C., Wang, P., Qin, Q.W., 2011. Transcriptome analysis of orange-spotted grouper (Epinephelus coioides) spleen in re- This work was supported by Science and Technology Program of sponse to Singapore grouper iridovirus. BMC Genomics 12 (1), 556. Guangdong Province (Grant No. 2017A030307024 and Ji, C.C., Wu, H.F., Liu, X.L., Zhao, J.M., Yu, J.B., Yin, X.L., 2013. The influence of salinity on tox- icological effects of arsenic in digestive gland of clam Ruditapes philippinarum using 2017A030303076); the Graduate Education Innovation Program of metabolomics. Chin. J. Oceanol. Limnol. 31 (2), 345–352. Guangdong Ocean University (Grant No. 201828); Department of Edu- Jiang, W.W., Tian, X.L., Fang, Z.H., Li, L., Dong, S.L., Li, H.D., Zhao, D., 2019. Metabolic re- cation of Guangdong Province (Grant No. 2017KCXTD016); China Agri- sponses in the gills of tongue sole (Cynoglossus semilaevis) exposed to salinity stress using nmr-based metabolomics. Sci. Total Environ. 653, 465–474. culture Research System (Grant No. CARS-049). Our data analysis was Johnston, I.A., Dunn, J., 1987. Temperature acclimation and metabolism in ectotherms assisted by Biotree Biotech Co., Ltd. (Shanghai, China). with particular reference to teleost fish. Symp. Soc. Exp. Biol. 41, 67–93. Kanehisa, M., Araki, M., Goto, S., Hattori, M., Hirakawa, M., Itoh, M., Katayama, T., Kawashima, S., Okuda, S., Tolimatsu, T., Yamanishi, Y., 2008. KEGG for linking ge- Authors contributions nomes to life and the environment. Nucleic Acids Res. 36, 480–484. Kim, D., Pertea, G., Trapnell, C., Pimentel, H., Kelley, R., Salzberg, S.L., 2013. Tophat2: accu- Yuewen Deng and Zhe Zheng conceived and supervised the ex- rate alignment of transcriptomes in the presence of insertions, deletions and gene fu- periments. Ruijuan Hao performed the experiments, contributed to sions. Genome Biol. 14 (4), R36. Kind, T., Wohlgemuth, G., Lee, D.Y., Lu, Y., Palazoglu, M., Shahbaz, S., Fiehn, O., 2009. data analysis, and wrote the paper. Chuangye Yang, Xiaodong Du, FiehnLib: mass spectral and retention index libraries for metabolomics based on Yuewen Deng, Zhe Zheng and Qingheng Wang gave valuable advice quadrupole and time-of-flight gas chromatography/mass spectrometry. Anal. Chem. – for the modifications of the paper. All authors read and approved 81 (24), 10038 10048. Krammer, A., Lu, H., Isralewitz, B., Schulten, K., Vogel, V., 1999. Forced unfolding of the fi- the manuscript. bronectin type III module reveals a tensile molecular recognition switch. Proc. Natl. Acad. Sci. U. S. A. 96 (4), 1351–1356. References Kube, S., Gerber, A., Jansen, J.M., Schiedek, D., 2006. Patterns of organic osmolytes in two marine bivalves, Macoma balthica,andMytilus spp. along their european distribution. Mar. Biol. 149 (6), 1387–1396. Allam, B., Paillard, C., Auffret, M., Ford, S.E., 2006. Effects of the pathogenic vibrio tapetis Kuchel, R.P., Nair, S., Raftos, D.A., 2012. Changes in the transcriptional expression of oxida- on defence factors of susceptible and non-susceptible bivalve species: II. Cellular tive stress response genes in akoya pearl oysters (Pinctada fucata) exposed to air and and biochemical changes following in vivo challenge. Fish Shellfish Immunol. 20 mechanical agitation. Aquaculture 362-363 (5), 33–38. (3), 384–397. Langmead, B., Salzberg, S.L., 2012. Fast gapped-read alignment with bowtie 2. Nat. Bosman, F.T., Stamenkovic, I., 2003. Functional structure and composition of the extracel- Methods 9 (4), 357–359. lular matrix. J. Pathol. 200 (4), 423–428. Li, B., Dewey, C.N., 2011. RSEM: accurate transcript quantification from RNA-Seq data Bradshaw, M.J., Smith, M.L., 2014. Multiscale relationships between fibronectin structure with or without a reference genome. BMC Bioinf. 12 (1), 323. and functional properties. Acta Biomater. 10 (4), 1524–1531. Li, Y.G., He, M.X., 2017. Correlation between genome methylation level and growth trait of Brown, J.R., 1988. Multivariate analyses of the role of environmental factors in seasonal pearl oyster, Pinctada fucata. Pak. J. Zool. 49 (3), 833–839. and site-related growth variation in the pacific oyster Crassostrea gigas. Mar. Ecol. Li, B., Song, K., Meng, J., Li, L., Zhang, G.F., 2017. Integrated application of transcriptomics Prog. Ser. 45 (3), 225–236. and metabolomics provides insights into glycogen content regulation in the pacific Cappello, T., Maisano, M., Giannetto, A., Parrino, V., Mauceri, A., Fasulo, S., 2015. oyster Crassostrea gigas. BMC Genomics 18 (1), 713. Neurotoxicological effects on marine mussel Mytilus galloprovincialis caged at petro- Little, W.C., Smith, M.L., Ebneter, U., Vogel, V., 2008. Assay to mechanically tune and opti- chemical contaminated areas (eastern Sicily, Italy): 1H NMR and immunohistochem- cally probe fibrillar fibronectin conformations from fully relaxed to breakage. Matrix ical assays. Comp. Biochem. Physiol. C 169, 7–15. Biol. 27 (5), 451–461. 1 Cappello, T., Maisano, M., Mauceri, A., Fasulo, S., 2017a. H NMR-based metabolomics Liu, X.L., Ji, C.L., Zhao, J.M., Wu, H.F., 2013. Differential metabolic responses of clam investigation on the effects of petrochemical contamination in posterior adduc- Ruditapes philippinarum,toVibrio anguillarum,andVibrio splendidus,challenges. tor muscles of caged mussel Mytilus galloprovincialis. Ecotoxicol. Environ. Saf. Fish Shellfish Immunol. 35 (6), 2001–2007. 142, 417–422. Liu, X.L., Ji, C.L., Zhao, J.M., Wang, Q., Li, F., Wu, H.F., 2014. Metabolic profiling of the tissue- Cappello, T., Fernandes, D., Maisano, M., Casano, A., Bonastre, M., Bebianno, M.J., Mauceri, specific responses in mussel Mytilus galloprovincialis towards Vibrio harveyi chal- A., Fasulo, S., Porte, C., 2017b. Sex steroids and metabolic responses in mussels, lenge. Fish Shellfish Immunol. 39 (2), 371–373. Mytilus galloprovincialis, exposed to drospirenone. Ecotoxicol. Environ. Saf. 143, Liu, C., Li, S.G., Kong, J.J., Liu, Y.J., Wang, T.P., Xie, L.P., Zhang, R.Q., 2015. In-depth proteomic 166–172. analysis of shell matrix proteins of Pinctada fucata.Sci.Rep.5,17269. Chen, J.H., Xiao, S., Deng, Y.W., Du, X.D., Yu, Z.N., 2011. Cloning of a novel glutathione s- Lu, J., Shi, Y.Y., Wang, S.H., Chen, H., Cai, S.H., Feng, J.H., 2016. NMR-based metabolomic transferase 3 (GST3) gene and expression analysis in pearl oyster, Pinctada martensii. analysis of Haliotis diversicolor exposed to thermal and hypoxic stresses. Sci. Total En- Fish Shellfish Immunol. 31 (6), 823–830. viron. 545-546, 280–288. Deng, Y.W., Du, X.D., Wang, Q.H., 2009. Selection for fast growth in the Chinese pearl oys- Mariman, E.C.M., Wang, P., 2010. Adipocyte extracellular matrix composition, dynamics ter, Pinctada martensii:responseofthefirst generation line. J. World Aquacult. Soc. 40 and role in obesity. Cell. Mol. Life Sci. 67 (8), 1277–1292. (6), 843–847. Meng, J., Zhu, Q.H., Zhang, L.L., Li, C.Y., Li, L., She, Z.C., Huang, B.Y., Zhang, G.F., 2013. Ge- Du, X.D., Deng, Y.W., Wang, Q.H., Xie, S.H., Liu, D., 2015. Haixuan NO 1 stock of pearl oyster nome and transcriptome analyses provide insight into the euryhaline adaptation Pinctada martensii. China Fish. 10, 53–55. mechanism of Crassostrea gigas. PLoS One 8 (3), e58563. Du,X.D.,Fan,G.Y.,Jiao,Y.,Zhang,H.,Guo,X.M.,Huang,R.L.,Zheng,Z.,Bian,C.,Deng, Mills, D., 2000. Combined effects of temperature and algal concentration on survival, Y.W.,Wang,Q.H.,Wang,Z.D.,Liang,X.M.,Liang,H.Y.,Shi,C.C.,Zhao,X.X.,Sun, growth and feeding physiology of Pinctada maxima (Jameson) spat. J. Shellfish Res. F.M.,Hao,R.J.,Bai,J.,Liu,J.L.,Chen,W.B.,Liang,J.L.,Liu,W.Q.,Xu,Z.,Shi,Q.,Xu, 19 (1), 159–166. X.,Zhang,G.F.,Liu,X.,2017.The pearl oyster Pinctada fucata martensii genome Moreira, A., Figueira, E., Soares, A.M., Freitas, R., 2016. The effects of arsenic and seawater and multi-omic analyses provide insights into biomineralization. Gigascience 6 acidification on antioxidant and biomineralization responses in two closely related (8), 1–12. crassostrea species. Sci. Total Environ. s545–546, 569–581. Fan, J., Ye, J., Kamphorst, J.J., Shlomi, T., Thompson, C.B., Rabinowitz, J.D., 2014. Quantita- Pruzzo, C., Gallo, G., Canesi, L., 2005. Persistence of vibrios in marine bivalves: the role of tive flux analysis reveals folate-dependent NADPH production. Nature 510 (7504), interactions with haemolymph components. Environ. Microbiol. 7 (6), 761–772. 298–302. Qiao, F., Lei, K., Li, Z., Wei, Z., Liu, Q., Yang, L., He, J., An, L., Qi, H., Song, C., 2018. Flier, A.V.D., Sonnenberg, A., 2001. Function and interactions of integrins. Cell Tissue Res. Transcriptomic responses of the freshwater snail (Parafossarulus striatulus) following 305 (3), 285–298. dietary exposure to cyanobacteria. Sci. Total Environ. 624, 153–161. Gracey, A.Y., Connor, K., 2016. Transcriptional and metabolomic characterization of spon- Ragg, N.L.C., King, N., Watts, E., Morrish, J., 2010. Optimising the delivery of the key dietary taneous metabolic cycles in Mytilus californianus, under subtidal conditions. Mar. Ge- diatom Chaetoceros calcitrans, to intensively cultured greenshell™ mussel larvae, nomics 30, 35–41. Perna canaliculus.Aquaculture306(1),270–280. 56 R. Hao et al. / Science of the Total Environment 666 (2019) 46–56

Reddy, K.V.R., Mangale, S.S., 2003. Integrin receptors: the dynamic modulators of endo- Wang, H., Zhu, X.W., Wang, Y.N., Luo, M.M., Liu, Z.G., 2012. Determination of optimum metrial function. Tissue Cell 35 (4), 260–273. temperature and salinity for fertilization and hatching in the Chinese pearl oyster Rondon, R., Akcha, F., Alonso, P., Menard, D., Rouxel, J., Montagnani, C., Mitta, G., Cosseau, Pinctada martensii, (Dunker). Aquaculture 358-359 (2), 292–297. C., Grunau, C., 2016. Transcriptional changes in Crassostrea gigas,oysterspatfollow- Wang, Q.H., Liu, Y., Zheng, Z., Deng, Y.W., Jiao, Y., Du, X.D., 2018a. Adaptive response of ing a parental exposure to the herbicide diuron. Aquat. Toxicol. 175, 47–55. pearl oyster Pinctada fucata martensii to low water temperature stress. Fish Shellfish Ruegg, J.C., 1971. Smooth muscle tone. Physiol. Rev. 51, 201–248. Immunol. 78, 310–315. Salem, M., Rexroad, C.E., Wang, J., Thorgaard, G.H., Yao, J., 2010. Characterization of the Wang, Q.H., Hao, R.J., Zhao, X.X., Huang, R.L., Zheng, Z., Deng, Y.W., Chen, W.Y., Du, X.D., rainbow trout transcriptome using sanger and 454-pyrosequencing approaches. 2018b. Identification of EGFR in pearl oyster (Pinctada fucata martensii) and correla- BMC Genomics 11 (1), 564. tion analysis of its expression and growth traits. Biosci. Biotechnol. Biochem. 82 (7), Southgate, P.C., Lucas, J.S., 2008. The Pearl Oyster. 1073–1080. Tamayo, D., Ibarrola, I., Urrutia, M.B., Navarro, E., 2011. The physiological basis for inter- Wang, Z.M., Liang, F.L., Huang, R.L., Deng, Y.W., Li, J.H., 2019. Identification of the differen- individual growth variability in the spat of clams (Ruditapes philippinarum). Aquacul- tially expressed genes of Pinctada maxima individuals with different sizes through ture 321 (1), 113–120. transcriptome analysis. Reg. Stud. Mar. Sci. 26, e100512. Tamayo, D., Ibarrola, I., Urrutxurtu, I., Navarro, E., 2014. Physiological basis of extreme Wen, B., Jin, S.R., Chen, Z.Z., Gao, J.Z., 2018. Physiological responses to cold stress in the growth rate differences in the spat of oyster (Crassostrea gigas). Mar. Biol. 161 (7), gills of discus fish (Symphysodon aequifasciatus) revealed by conventional biochemi- 1627–1637. cal assays and GC-TOF-MS metabolomics. Sci. Total Environ. 640, 1372–1381. Tamayo, D., Azpeitia, K., Markaide, P., Navarro, E., Ibarrola, I., 2016. Food regime modu- Xu, G.C., Du, F.K., Yan, L., Nie, Z.J., Xu, P., 2016. Integrated application of transcriptomics lates physiological processes underlying size differentiation in juvenile intertidal and metabolomics yields insights into population-asynchronous ovary development mussels Mytilus galloprovincialis. Mar. Biol. 163, 1–13. in Coilia nasus.Sci.Rep.6,31835. Taylor, J.J., Rose, R.A., Southgate, P.C., Taylor, C.E., 1997. Effects of stocking density on Yancey, P.H., Clark, M.E., Hand, S.C., Bowlus, R.D., Somero, G.N., 1982. Living with water growth and survival of early juvenile silver-lip pearl oysters, Pinctada maxima, (Jame- stress: evolution of osmolyte systems. Science 217 (4566), 1214–1222. son), held in suspended nursery culture. Aquaculture 153 (1–2), 41–49. Yang, J.M., Luo, S.J., Li, J.H., Zheng, Z., Du, X.D., Deng, Y.W., 2018. Transcriptome analysis of Taylor, J.J., Southgate, P.C., Rose, R.A., 1998. Effects of mesh covers on the growth and sur- growth heterosis in pearl oyster Pinctada fucata martensii. FEBS Open Bio. 8 (11), vival of silver-lip pearl oyster (Pinctada maxima, Jameson) spat. Aquaculture 162 1794–1803. (162), 241–246. Yang, C.Y., Hao, R.J., Du, X.D., Wang, Q.H., Deng, Y.W., Sun, R.J., Zheng, Z., 2018. GC–TOF/ Taylor, J.J., Southgate, P.C., Rose, R.A., 2004. Effects of salinity on growth and survival of MS-based metabolomics studies on the effect of protein sources in formulated diet silver-lip pearl oyster, Pinctada maxima, Spat. J. Shellfish Res. 23, 375–377. for pearl oyster Pinctada fucata martensii. Aquaculture 486, 139–147. Ten Doeschate, K.I., Coyne, V.E., 2008. Improved growth rate in farmed Haliotis midae Yang, C.Y., Hao, R.J., Du, X.D., Wang, Q.H., Deng, Y.W., Sun, R.J., 2019. Response to different through probiotic treatment. Aquaculture 284 (1–4), 174–179. dietary carbohydrate and protein levels of pearl oysters (Pinctada fucata martensii)as Timpl, R., 1996. Macromolecular organization of basement membranes. Curr. Opin. Cell revealed by GC-TOF/MS-based metabolomics. Sci. Total Environ. 650, 2614–2623. Biol. 8 (5), 618–624. Young, T., Alfaro, A.C., Villas-Boas, S., 2015. Identification of candidate biomarkers for Trapnell, C., Roberts, A., Goff, L., Pertea, G., Kim, D., Kelley, D.R., Pimentel, H., Salzberg, S.L., quality assessment of hatchery-reared mussel larvae via GC/MS-based metabolomics. Rinn, J.L., Pachter, L., 2012. Differential gene and transcript expression analysis of N. Z. J. Mar. Freshw. Res. 49 (1), 87–95. RNA-seq experiments with TopHat and cufflinks. Nat. Protoc. 7 (3), 562–578. Zhao, X., Wang, Q., Jiao, Y., Huang, R., Deng, Y., Wang, H., Du, X., 2012. Identification of Venter, L., Loots, D.T., Vosloo, A., van Rensburg, P.J., Lindeque, J.Z., 2018. Abalone growth genes potentially related to biomineralization and immunity by transcriptome anal- and associated aspects: now from a metabolic perspective. Rev. Aquac. 10 (2), ysis of pearl sac in pearl oyster Pinctada martensii.Mar.Biotechnol.14,730–739. 451–473. Zheng, Z., Huang, R.L., Tian, R.R., Jiao, Y., Du, X.D., 2016. Pm-miR-133 hosting in one poten- Vera, J.C., Wheat, C.W., Fescemyer, H.W., Frilander, M.J., Crawford, D.L., Hanski, I., Marden, tial lncRNA regulates RhoA expression in pearl oyster Pinctada martensii. Gene 591 J.H., 2008. Rapid transcriptome characterization for a nonmodel organism using 454 (2), 484–489. pyrosequencing. Mol. Ecol. 17 (7), 1636–1647. Wada, K.T., Komaru, A., 1994. Effect of selection for shell coloration on growth rate and mortality in the Japanese pearl oyster, Pinctada fucata martensii. Aquaculture 125 (1–2), 59–65.