Transcriptomic analysis of brine shrimp Artemia franciscana across a wide range of salinities

Brine shrimp Artemia franciscana , a commercially important species, can thrive in a wide range of salinities and is commonly found in hypersaline lakes and solar salterns. Transcriptome analysis can enhance the understanding of the adaptative mechanisms of brine shrimp in aquaculture. RNA sequencing (RNAseq) data was generated from A. franciscana adults that were salt-adapted for 2 – 4 weeks at five salinities: 35, 50, 100, 150, and 230 psu. Long-read isoform sequencing (IsoSeq) data was used to construct a high-quality transcriptome assembly. Also, the gene expression patterns in A. franciscana adults were examined. Notably, the transcriptional response of A. franciscana 's acclimation to intermediate salinities (50 – 150 psu) displayed frequently and differentially U-shaped or inverted U-shaped expression patterns. In addition, the types of genes showing two nonmonotonic expression patterns were distinct from each other. The coordinated shifts in gene expression suggest different homeostatic strategies of A. franciscana at specific salinities; such strategies may enhance population fitness at extreme salinities. Our study should promote a scientific concept for the gene expression patterns of A. franciscana along a broad salinity gradient, and a variety of salinity and prey should be monitored for testing the gene expression pattern of this important aquaculture species.


Introduction
Hypersaline environments have provided extraordinary habitats for all domains of life on Earth since the Precambrian period (Javor, 1989).Brine shrimp Artemia (Branchiopoda; Pancrustacea; Crustacea) is frequently found in natural hypersaline lakes, such as the Great Salt Lake in the United States, and artificial solar salterns (Triantaphyllidis et al., 1998;Gajardo and Beardmore, 2012).Artemia franciscana, first found at Redwood City, San Francisco Bay, has since been identified as an invasive species in numerous countries (Kellogg, 1906;Sheir et al., 2018).Artemia is a critical species as prey for larval fish in aquaculture and an experimental organism in ecotoxicology (Hwang et al., 2010;Lenormand et al., 2018).A. franciscana cysts are estimated to account for up to 90% of the global Artemia trade (Treece, 2000).
Artemia is bisexual or parthenogenetic filter feeders that consume phytoplanktons, such as Dunaliella spp.and prokaryotes, in high-salinity waters.Since Artemia species regularly ingest high-salinity water along with their preys, they must be strict osmoregulators (Copeland, 1967;Holliday et al., 1990;Bradley, 2009;Sellami et al., 2020).An increase in salinity has been reported to stimulate Na + /K + -ATPase activity in multiple organs, including metepipodites, maxillary glands, and gut, of adult Artemia to excrete the accumulated salts (Copeland, 1967;Holliday et al., 1990;Bradley, 2009;Sellami et al., 2020).Furthermore, Artemia hemolymph is maintained at approximately 300 mOsm/kg, equivalent to 9 psu and similar to the osmolality of the teleosts exposed to a range of external high salinities (Bradley, 2009;Croghan, 1958).
Artemia spp.grow best in a salinity range of 100-170 psu and can tolerate <5 psu or > 300 psu (Post and Youssef, 1977;Wear and Haslett, 1986;Browne and Wanigasekera, 2000;Sorgeloos et al., 2001).These findings suggest that some genes may be expressed in a quadratic profile.For example, some gene expression patterns may be a U-shaped curve when the genes are up-regulated at the two salinity extremes or an inverted U-shaped curve when they are down-regulated at the two salinity extremes.These patterns have been identified in euryhaline fishes at sublethal salinity or temperature thresholds (Jensen et al., 1998;Tine et al., 2012;Jeffries et al., 2018).
Crustaceans, including Artemia, can tolerate a wider range of salinities at the adult stage than other life stages (Péqueux, 1995).In contrast, the adaptive capacity of Artemia nauplii to high salinities decreases with the deterioration of the salt gland (Mahfouz et al., 2013).While transcriptome profiles of A. franciscana nauplii at 30 and 200 psu have been determined (De Vos et al., 2019, 2021), the differential transcriptome patterns in adult A. franciscana along a salinity gradient have not been investigated.In addition, little is known about the quadratic expression patterns in adult A. franciscana.Moreover, other adaptive osmoregulatory mechanisms of A. franciscana in response to the variations in salinity remain poorly understood.
In this study, a de novo transcriptome assembly of adult A. franciscana was generated using a hybrid assembly approach, with long-read isoform sequencing (IsoSeq) and high-throughput RNA sequencing (RNAseq), to analyze the differential gene expression patterns at a variety of salinities, i.e., 35, 50, 100, 150, and 230 psu.

Sample preparation and isoform and RNA sequencing
A. franciscana was isolated from 140 psu saline water collected from a solar saltern in Uiseong on the west coast of the Republic of Korea, in April 2018 (36 • 60 ′ 20.86 ′′ N, 126 • 29 ′ 71.16 ′′ E).A population was maintained in a 20 L aquarium containing the sediment from the saltern and 15 L of 150 psu artificial saline water made by diluting 300 psu Medium V (300 psu; 272 g of NaCl, 7.6 g of KCl, 17.8 g of MgCl 2 , 1.8 g of MgSO 4 ⋅7H 2 O, and 1.3 g of CaCl 2 L − 1 ) (Park, 2012) at 25 • C for 2 years.
The aquarium was illuminated using a 60 μE m − 2 s − 1 cool white fluorescent light at 14 h light /10 h dark cycles.Approximately 20 adults were then transferred to and maintained in artificial saline water at 35, 50, 100, 150, or 230 psu for 2-4 weeks; their active movement was checked.The dissolved oxygen (DO) concentration was monitored using the YSI ProSolo ODO/T meter (Yellow Springs, USA) while illumination and temperature conditions remained constant.Adults, five males and five females, were collected from each salinity condition, and their gut content was cleared for 3 days before the transcriptome analyses of isoforms to reduce the sequencing bias originated from preys.This procedure might partially affect the gene expression patterns.
Total RNA was extracted from A. franciscana using the RNeasy Plus Mini Kit, including an on-column DNase I treatment to eliminate residual genomic DNA (Qiagen, Germany) following the manufacturer's protocol.RNAseq libraries were constructed using an Illumina Truseq Standard mRNA Prep Kit (Illumina, USA), and the library qualities were checked with a 2100 Bioanalyzer (Agilent Technologies, USA).The RNAseq libraries were sequenced using 100 bp paired-end reagents with an Illumina Novaseq6000 (Supplementary Table S1).The IsoSeq library was constructed using a SMARTer PCR cDNA Synthesis Kit and DNA Template Prep Kit 1.0.IsoSeq data were generated (1.0 Gbp; 780,667 reads) using the PacBio Sequel platform (Pacific Biosciences, Menlo Park, CA, USA).The high-quality (HQ) isoform consensus (16.6 Mbp; 16,555 reads) of the IsoSeq data was constructed using SMRT Link 5.1.0with the Iso-Seq2 application platform.The raw reads of RNAseq and HQ isoform consensus data were uploaded to the NCBI sequence read archive (SRA) database with accession numbers (SRR12358297-SRR12358302; BioProject PRJNA649341).

De novo transcriptome assembly, protein predictions, and functional annotations
A hybrid de novo transcriptome assembly of A. franciscana was generated to construct HQ isoform sequences using a long-read HQ isoform consensus (16.6 Mbp; 16,555 reads; PacBio Sequel) and highthroughput RNAseq (34.6 Gbp; 100 × 100 bp Illumina Novaseq6000 reads; Supplementary Table S1).The error correction step of the IsoSeq consensus was performed to improve sequence accuracy using the highaccuracy Illumina sequencing reads aligned by Bowtie2, and variant calling in the alignment was processed using Samtools (Fig. 1; Li et al., 2009;Langmead and Salzberg, 2012).The corrected IsoSeq reads were assembled with the short-read Illumina sequencing data using Trinity assembler v2.8.4 with the default option of de novo transcriptome assembly with '-long_reads' (Grabherr et al., 2011;Haas et al., 2013).The assembled transcripts were clustered to reduce redundancy using the cdhit-EST program v4.6 (-c 0.95 -aS 0.95 -n 10; https://github.com/weizhongli/cdhit).The RNAseq raw reads were mapped into the assembled transcripts using the Salmon program with default options (Patro et al., 2017); the potential false-positive transcripts, including 0 tpm (transcripts per million) values, were excluded.Potential ribosomal RNA fragments in the assembled transcripts were excluded based on the results of a local BLASTn search (e-value cutoff = 1.e − 10) using Artemia salina ribosomal RNA sequences (X01723.1 and AF169697.1).

Analyses of conserved eukaryotic gene sets and phylogeny
The completeness analyses of conserved eukaryotic gene set completeness analyses were conducted using BUSCO (Benchmarking Universal Single-Copy Orthologs; v3.0.2;Simão et al., 2015;Waterhouse et al., 2018).Sixteen taxa from Pancrustacea, including A. franciscana, were used to reconstruct the molecular phylogenetic tree.The phylogenetic position of A. franciscana was inferred using 4937 concatenated protein sequences related to Pancrustacea (local BLASTp e-value cutoff = 1.e − 05).All the protein sequences were aligned using MAFFT v7.313 (default options: -auto; Yamada et al., 2016).Each alignment was trimmed when an aligned locus included more than 70% of gap sequences.The phylogenetic analysis was inferred using maximum likelihood (ML) analysis.The ML tree was estimated using IQ-tree v.1.6.12 (Nguyen et al., 2015) with the best-fit evolutionary model, and statistical support was estimated using bootstrapping with 1000 replicates.

Differentially expressed gene analysis of A. franciscana
The RNAseq raw reads were mapped using Salmon (default option; Patro et al., 2017) for the differentially expressed gene (DEG) analyses of A. franciscana at 35, 50, 100, 150, and 230 psu.Up-and down-regulated gene expression patterns were analyzed using the tpm value from the mapping results.The genes were sorted by salinity-dependent gene expression patterns, including those with the following criteria: maximum fold-change ≥2-fold between the highest and the lowest tpm value and the highest tpm value ≥5 to collect strict DEG candidates.The tpm values were normalized by the z-score [('Expression' -'Average expression of all conditions in each gene') / 'Standard deviation of all conditions in each gene'].Gene ontology (GO) terms were extracted from the functional annotations of the target genes described by the eggNOG-mapper results, and the enrichment of GO terms was performed using the topGO package in R (Fisher test, P-value <0.05).

De novo transcriptome assembly and phylogenomic analysis
A total of 18,301 protein-coding sequences from A. franciscana were obtained by filtering the assembled transcripts to remove the transcripts that were too short, redundant, or false-positive.Reference data were used for the homologous pancrustacean sequences (Fig. 1).Based on the BUSCO analysis using the 'eukaryota_odb10' database, the protein profiles of A. franciscana included 95.3% of the conserved eukaryotic gene sets, with a reliability quality comparable to those of other crustaceans at 85%-98% of the BUSCOs (Supplementary Table S3).Furthermore, our coverages of core eukaryotic genes and fragmented BUSCOs had higher accuracy than previous datasets of A. franciscana (Supplementary Table S3).The class Branchiopoda, including the sequences of our isolate, A. franciscana strain KOPRI (the database of Korean Polar Research Institute; https://antagen.kopri.re.kr/), and A. franciscana strain 'Specimen 9' (De Vos et al., 2021; the F2 progeny of female and male A. franciscana collected from the Great Salt Lake, Nambu et al., 2007), was the sister group to the class Hexapoda (ML: 100%), forming a monophyletic group (Fig. 2). A. franciscana in this study was closest to two other A. franciscana, forming a maximally supported clade (Fig. 2).In the 4937 homologous gene datasets, previous A. franciscana strains KOPRI (79.7%) and 'Specimen 9' (76.1%) has less fidelity in full-length protein sequences than our strain, suggesting that the completeness of coding-gene construction in A. franciscana was considerably improved in the present study.
The mitochondrial cytochrome oxidase subunit I gene (COI) sequence is a suitable biomarker for identifying and classifying Artemia species (Sheir et al., 2018;Eimanifar et al., 2015).The 1539 bp COI sequence retrieved from the transcript of the isolate was 99.94%-100% identical to several A. franciscana sequences (GenBank accession numbers NC001620, X69067, KX925412, and MG572074).Furthermore, the male A. franciscana species of this study displayed basal spinelike projections on the gonopod (data not shown) characteristic of A. franciscana strains (Mura and Brecciaroli, 2004), suggesting that this isolate from a Korean solar saltern was identical to the A. franciscana species reported in other countries.

Differentially expressed gene patterns along the salinity gradient
Due to the broad salinity tolerance of A. franciscana, the variations in gene expression patterns of the species' salt-adapted cells along a salinity gradient were analyzed (35-230 psu) based on the z-scores (Fig. 3 and Supplementary Table S4).The enriched GO terms of the salinitydependent genes (topGO package in R, Fisher test, P-value <0.05) revealed that A. franciscana significantly up-regulated genes related to Na + /K + -ATPases, mannose and carotenoid metabolism, and monocarboxylate transmembrane transporters with increasing salinities from 35 to 230 psu (Fig. 3 and Supplementary Table S5).In contrast, the genes associated with galactose and glucose metabolism, N-acetylgalactosamine-4-sulfatases, amino acid symporters, and amino acid transmembrane transporters were down-regulated (Fig. 3).
In addition to the genes involved in osmoregulation in A. franciscana (e.g., Na + /K + -ATPase), the genes related to vision and mitochondrial morphogenesis were also up-regulated with increasing salinities from 35 to 230 psu (Fig. 3 and Supplementary Table S5).Furthermore, the genes involved in oxidative stress regulation and glycoprotein catabolic processes were up-regulated at high salinities (Supplementary Table S5).Meanwhile, the genes related to diverse transporter activities were down-regulated in response to high salinity (Fig. 3).These data suggest that the transporter systems in adult A. franciscana may be repressed in high-salinity environments.

U-shaped or inverted U-shaped expression patterns
In the U-shaped curves, a total of 326 A. franciscana genes had their lowest level of expression at 50 psu, 222 genes at 100 psu, and 411 genes at 150 psu (Fig. 4 and Supplementary Table S6).On the other hand, in the inverted U-shaped expression patterns, 379 genes were expressed at the highest level at 50 psu, 289 genes at 100 psu, and 63 genes at 150 psu (Fig. 4 and Supplementary Table S7).
Based on the enriched GO terms, the U-shaped expression patterns observed in A. franciscana represented the cellular responses related to cell signaling, catabolism, morphogenesis, and development (Supplementary Table S8).Meanwhile, the inverted U-shaped expression patterns were related to diverse environmental response factors, such as salt aversion, sensory perception of salty taste, cellular response to light intensity, UV-B, ozone, response to oxidative stress, and starvation (Supplementary Table S8).The dissolved oxygen concentrations of saline waters containing approximately 20 adults were 6.7 ± 0.3 mg L − 1 at 35 psu, 6.2 ± 0.2 mg L − 1 at 50 psu, 4.7 ± 0.5 mg L − 1 at 100 psu, 3.1 ± 0.5 mg L − 1 at 150 psu, and 1.8 ± 0.3 mg L − 1 at 230 psu during the incubation period.
Certain Artemia genes are up-or down-regulated in response to high salinity (Mahfouz et al., 2013;De Vos et al., 2019;Jorgensen and Amat, 2008;Zhang et al., 2013;Maniatsi et al., 2015;Vikas et al., 2016;Zhao et al., 2016).The proteins encoded by the genes identified to be upregulated included the chloride channel protein, ecdysone receptor isoform A, fatty acid hydroxylase domain-containing protein 2,  pancreatic triacylglycerol lipase, sphingomyelin phosphodiesterase, cyclin-dependent kinase 14, tribbles homolog 2, and putative inorganic phosphate cotransporter: These genes displayed U-shaped patterns along the salinity gradient (Supplementary Table S9).On the other hand, the proteins encoded by the genes identified to be up-regulated included the FK506 binding protein, CCAA/enhancer-binding protein, epithelial discoidin domain-containing receptor, PAS domaincontaining serine/threonine-protein kinase, copper-zinc superoxide dismutase, and putative inorganic phosphate cotransporter: These genes displayed an inverted U-shaped expression pattern along the salinity gradient (Supplementary Table S9).Lastly, the gene encoding group 3 late embryogenesis abundant protein is down-regulated at high salinities while displaying a U-shaped pattern; thus, this gene constitutes the third group (Supplementary Table S9).

Discussion
This study presents the gene expression patterns of adult A. franciscana along a salinity gradient.U-shaped and inverted U-shaped expression patterns were identified in A. franciscana adults responding to salinity, similar to those found in euryhaline fishes (Jensen et al., 1998;Tine et al., 2012;Jeffries et al., 2018).These transcriptome results provide the first evidence that the cellular response of adult A. franciscana to acclimate to intermediate salinities at 50, 100, and 150 psu is dependent on the number and type of genes expressed.
Based on the enriched GO terms and KEGG metabolic pathways, the functional annotations of the U-shaped expression pattern genes were different from those of the inverted U-shaped expression pattern genes (Fig. 5 and Supplementary Table S10).This result indicates that the genes related to the U-shaped and inverted U-shaped patterns are differentially expressed, and their expression pattern at a specific salinity may be related to homeostasis in A. franciscana.The genes expressed in the U-shaped pattern of A. franciscana indicate the promotion of cellular signaling, catabolic processes, morphogenesis, and development, for survival at 35 and 230 psu.On the other hand, the inverted U-shaped gene expression may represent a reduction of the potential impacts of diverse environmental response factors on the survival of A. franciscana at the two salinity extremes.Presumably, the coordinated shifts in gene expression enhance population fitness in extreme salinity conditions (Gajardo and Beardmore, 2012;Jeffries et al., 2018;Hebert et al., 2002).
Halophilic and halotolerant organisms, well-adapted to high-salinity waters, maintain their internal osmotic equilibrium to thrive in harsh environments.The salt-in and salt-out processes are regarded as two distinct osmoregulation strategies (Oren, 2002;Foissner et al., 2014;Harding et al., 2016).Artemia spp.can express genes related to ion homeostasis; for example, it can up-regulate the genes encoding the chloride channel protein (CIC) and Na + /K + -ATPase genes.As a result, it can transport ions through the salt-out processes (Copeland, 1967;Holliday et al., 1990;Bradley, 2009;Thabet et al., 2017;Sellami et al., 2020).CIC, which plays an important role in anion regulation, including chloride ions in prokaryotes and eukaryotes, displays a U-shaped expression pattern with the lowest expression level at 50 psu (Supplementary Table S9).This finding is consistent with Zhang et al. (2013), who determined that CIC in A. sinica nauplii displayed a U-shaped expression curve with the lowest expression level at 50 psu among the exposure range of 28, 50, 100, 150, and 200 psu, thus positing that CIC gene might contribute to cell volume regulation.These data suggest that CIC is likely to be widely expressed in Artemia spp.and responsive to salt stress and changes in cell volume regardless of life stages and species.
On the other hand, in marine crustaceans, Na + /K + -ATPase are responsible for maintaining sodium and potassium ion homeostasis across cell membranes (McNamara and Torres, 1999;Thabet et al., 2017;Faleiros et al., 2018;Bozza et al., 2019).Several studies have used the silver staining method to demonstrate that the metepipodites, digestive gut, and maxillary glands in A. salina exhibit high Na + /K + -ATPase activity under highly saline conditions (Holliday et al., 1990;Conte, 2008;Sellami et al., 2020).These findings are supported by the transcriptomic analysis in this study, which confirms the up-regulation of Na + /K + -ATPase in A. franciscana from 35 to 230 psu, and A. franciscana is a strict hyporegulator (Holliday et al., 1990;Conte, 2008;Bradley, 2009;Sellami et al., 2020).
This study also finds the genes related to beta-mannosidase and other enzymes involved with mannose metabolism to be up-regulated with Fig. 5. Analysis of the KEGG (Kyoto Encyclopedia of Genes and Genomes) metabolic pathways in up-and down-regulated genes at increasing salinities of 35, 50, 100, 150, and 230 psu, showing genes with U-shaped and inverted Ushaped expression patterns along a salinity gradient.up-inc: up-regulated genes with increasing salinity; dn-inc: down-regulated genes with increasing salinity; U-50, U-100, and U-150: genes with U-shaped expression patterns having minimum at 50, 100, and 150 psu, respectively; iU-50, iU-100, and iU-150: genes with inverted U-shaped expression patterns having a maximum at 50, 100, and 150 psu, respectively.increasing salinity.Thus, the metabolism of mannose and its derivatives at high salinities may reflect the potential effects of high salinity.As glucose and galactose-related activities are suppressed at high salinities, mannose can be a primary sugar in the glycolytic pathway.Furthermore, at high salinities, A. franciscana has been found to down-regulate the gene encoding UDP-glucose:hexose-1-phosphate uridylyltransferase (synonym: galactose-l-phosphate uridylyltransferase), which converts galactose-1-phosphate into glucose-1-phosphate.This result further supports the hypothesis that mannose is the main energy source of A. franciscana in highly saline environments.
Life in highly saline waters also involves unavoidable exposure to low-oxygen tension (Javor, 1989;Sherwood et al., 1991;Thabet et al., 2017).Theoretically, in high-salinity waters of more than 220 psu, dissolved oxygen (DO) is more than 3.2 times lower than that in seawater (Sherwood et al., 1991), similar to the ratio of DO concentration of approximately 3.7 at 35 to 230 psu in this study.Therefore, the organisms living hypersaline environments inevitably experience high salinity and oxidative stress.Crustaceans can usually up-regulate the antioxidative genes, such as those encoding superoxide dismutase (SOD), catalase, glutathione peroxidase, and glutathione S-transferase to adapt to higher salinity conditions (Thabet et al., 2017).De Vos et al. ( 2019) also reported the up-regulation of SOD in A. franciscana nauplii when they were exposed to two different salinities at 30 and 200 psu.The up-regulation of the genes related to the cellular response to oxidative stress, based on the enriched GO terms (Supplementary Table S9), suggests that A. franciscana can simultaneously adapt to highsalinity and low-oxygen environments.However, SOD displayed an inverted U-shaped expression pattern (K04565; Supplementary Table S10).The different expression patterns are probably due to variations in life stages and salinity exposures.
While Artemia can maintain cellular ion homeostasis at increasing external salinity, it requires more energy for growth and reproduction (Dana and Lenz, 1986).In particular, vision, one of the most energetically demanding functions of marine invertebrates, is highly susceptible to reducing DO availability (McCormick et al., 2019).Many diverse metazoan species contain ninaB homologs encoding carotenoid isomerooxygenase (Fig. 6), which are associated with the synthesis of visual pigment and oxidative stress (Arnouk et al., 2011;Jahng, 2012;Poliakov et al., 2020).Moreover, these previous studies suggest the upregulation of ninaB homologs in A. franciscana under highly saline conditions.As expected, our study confirms the up-regulation of ninaB at increasing salinity, demonstrating the importance of the vision in Artemia at high salinities.
Within the genus Artemia, the gene expression patterns at increasing salinity observed in this study are dissimilar to the previously identified patterns from assembled transcriptome data on the up-regulation or down-regulation of A. franciscana genes at 30 and 200 psu (Post and Youssef, 1977).The patterns observed in this study also differ from the immunomodulatory responses of A. sinica to Micrococcus lysodeikticus infection at 28 psu (Zhang et al., 2018) and the different transcriptomic patterns of the male and female of A. franciscana species collected from the Cejar lagoon in Chile (Valenzuela-Miranda et al., 2014).The nonmonotonic transcriptome patterns identified in adult A. franciscana in this study can be mainly due to the exposure to five different salinities or the differences in species, life stages, and exposure times (Jorgensen and Amat, 2008;Mahfouz et al., 2013;Zhang et al., 2013;Maniatsi et al., 2015;Vikas et al., 2016;Zhao et al., 2016;De Vos et al., 2019).
Our study also suggests that A. franciscana can be a model organism for studying locally adapted populations (Gajardo et al., 2002;Campillo et al., 2011;Gajardo and Beardmore, 2012).The gene expression of A. franciscana can fluctuate depending on the specific salinity faced by a population.Therefore, it is also worthwhile to explore whether the Ushaped and inverted U-shaped expression patterns in adult A. franciscana will be conserved when the animals are exposed to diverse salinities and preys using real-time qPCR.The high-quality isoform sequencing data from this study on the transcriptome analysis of A. franciscana provide essential information for future studies.

Conclusions
A. franciscana adults exhibit a complicated, nonmonotonic pattern of salt-adaptation at five different salinities (35, 50, 100, 150, and 230 psu), according to transcriptomic analysis.The types of genes displaying U-shaped and inverted U-shaped expression patterns suggest that A. franciscana has substantially different adaptive homeostatic mechanisms to intermediate salinities to enhance its fitness when exposed to various salinities.

Fig. 2 .
Fig. 2. Phylogenetic tree of Pancrustacea estimated from the 4937-protein dataset.Bootstrap support values are indicated at the nodes.

Fig. 3 .
Fig. 3.The heatmap displaying the expression patterns of up-and down-regulated genes at increasing salinities of 35, 50, 100, 150, and 230 psu.The expression values were normalized as Z-scores, and the annotated contigs were assigned to the molecular functions of the enriched gene ontology terms (Fisher test, Pvalue <0.05).

Fig. 4 .
Fig. 4. Hypothetical U-shaped and inverted U-shaped responses in gene expression at increasing salinities of 35, 50, 100, 150, and 230 psu.Normalized values (z-scores) were used to identify the U-shaped or inverted U-shaped patterns.

Fig. 6 .
Fig. 6.Phylogenetic tree of carotenoid isomerooxygenases in metazoans.Homologous proteins of Artemia franciscana were retrieved from the BLASTp results (NR database, top 500 hits; e-value cutoff = 1.e− 05).The bootstrap support values are shown at the nodes.