Assessment of reproductive parameters
Based on our overall experience in hatchery practices, the cumulative percentage of surviving embryos stabilizes prior to hatching, by 12 days post fertilization (dpf). Therefore, embryo survival at this stage was utilized as the measure of egg quality in this study. The actual survival rates in the overall sample inventory ranged from 93% for good quality eggs to 25% for the poor quality eggs. Egg batches with an embryo survival rate of ≥72% were considered to be of good quality and those spawns with ≤71% embryo survival were considered to be of poor quality in all seasons. Egg batches from halibut females employed in this study showed high variation in fecundity, buoyancy, fertilization, and normal cell division. Some egg batches exhibited low embryo survival rates despite a high percentage of fertilization and embryo progression through early stages of cell division (see batches marked with asterisks in Table S1). Correlation assessments between other measured reproductive parameters (female fecundity, egg buoyancy, fertilization rate, normal cell division) and embryo survival rate at 12 dpf were made using Spearman’s correlation analysis. Results indicate no obvious link of fecundity and egg buoyancy but strong correlation of fertilization rate and normal cell division rate to the proportion of embryos surviving before hatching up to 12 dpf (p < 0.01, Fig. S1, Panel A). When the same parameters were subjected to Spearman’s correlation with egg quality ascertained in this way, the stated relationships were apparent and the range of observations for each parameter in the two egg quality groups were made readily evident (p < 0.01, Fig. S1, Panel B). The range of survival observations for good and poor quality eggs, and a summary of the referenced Spearman’s correlation coefficients and corresponding significance values are shown in Fig. S1, Panel C.
TMT labeling based LC-MS/MS
A total of 1619 out of 1886 identified proteins were considered to be valid if they were detected in at least four biological samples. A total of 115 valid proteins were found to be differentially abundant between good and poor quality eggs (Independent samples t-test, p < 0.05 followed by Benjamini Hochberg correction for multiple testing, p < 0.05). Detailed information on these proteins is given in Table S2. In this study, proteins with higher abundance in good quality eggs are indicated as down-regulated in poor quality eggs (N = 64), and those with higher abundance in poor quality eggs are indicated as up-regulated in poor quality eggs (N = 51). Fig. 1A shows a volcano plot of these proteins based on p values obtained from Student’s t-test, p < 0.05 followed by Benjamini Hochberg correction for multiple testing, p < 0.05. A heatmap representation of the clustering of differentially regulated proteins based on their abundance in good versus poor quality eggs is given in Fig. 1B.
Frequency distributions of differentially abundant proteins among thirteen arbitrarily chosen functional categories accounting for > 90% of the proteins are shown in Fig. 2. Proteins that were down-regulated in poor quality eggs (N = 64) (Fig. 2 Panel A) were mainly related to cell cycle, division, growth and fate (26%), protein folding (14%), energy metabolism (12%), translation (11%), protein transport (8%), and lipid metabolism (8%) with the remaining categorized proteins being related to protein degradation and synthesis inhibition (5%), transcription (5%), mitochondrial biogenesis (5%), metabolism of cofactors and vitamins (3%), and redox/detox activities (1%). Only 2 % of proteins that were down-regulated in poor quality eggs were placed in the category ‘others’. Proteins that were up-regulated in poor quality eggs (N = 51) (Fig. 2 Panel B) were mainly related to cell cycle, division, growth and fate (19%), protein degradation and synthesis inhibition (18%), mitochondrial biogenesis (17%), transcription (16%), energy metabolism (8%), and protein transport (8%) with the remaining categorized proteins being related to lipid metabolism (4%), protein folding (2%), translation (2%), redox/detox activities [2], and immune response related (2%). Two percent of proteins that were up-regulated in poor quality eggs were placed in the category ‘others’. The distribution of these differentially regulated proteins among functional categories significantly differed between egg quality groups (χ2 p < 0.05). Accordingly, good quality eggs seem to contain a significantly higher proportion of proteins related to protein folding (14%), while poor quality eggs contain significantly higher proportions of proteins related to transcription (16%), protein degradation and synthesis inhibition (18%), and mitochondrial biogenesis (17%) (Fig. 2).
Gene ontology (GO) enrichment analyses based on overrepresentation tests (p < 0.05), with the human database used as a reference, revealed significant Biological processes, Molecular functions and Cellular components with a close relation to findings of the frequency distribution analyses (Fig. 3). Biological processes that were overrepresented by proteins down-regulated in poor quality eggs were as follows; protein folding, small molecule catabolic process, ribonucleoprotein (RNP) complex biogenesis, RNP complex subunit organization, cofactor biosynthetic process, coenzyme metabolic process, organophosphate (OP) catabolic process, nuclear transport, nucleobase-containing small molecule biosynthetic process, and RNA catabolic process. Molecular functions overrepresented by proteins down-regulated in poor quality eggs were related to isomerase activity, oxidoreductase activity (acting on the aldehyde or oxo group of donors), oxidoreductase activity (acting on the CH-CH group of donors), oxidoreductase activity (acting on a sulfur group of donors), RNP complex binding, kinesin binding, translation factor activity (RNA binding), snRNA binding, mRNA binding, and Ran GTPase binding. Cellular components overrepresented by these proteins were RNP complex, Sm-like protein family complex, mitochondrion, cytoplasmic region, mitochondrial part, cytoplasmic RNP granule, P-body, mitochondrial matrix, neuron projection cytoplasm, and tertiary granule lumen. KEGG pathways that were significantly overrepresented by this same set of proteins were RNA degradation, metabolic pathways, fatty acid degradation, valine, leucine and isoleucine degradation, glycolysis/gluconeogenesis, necroptosis, propanoate metabolism, glycine, serine and threonine metabolism, tryptophane metabolism, and ferroptosis (Fig. 3A, B, C and D Left sides).
In contrast, Biological processes that were overrepresented by proteins up-regulated in poor quality eggs were as follows; generation of precursor metabolites and energy, mitochondrial transport, mitochondrial respiratory chain complex assembly, tricarboxylic acid metabolic process, nucleoside phosphates metabolic/biosynthetic process, ribonucleotide metabolic process, RNA splicing, and mRNA processing. Molecular functions overrepresented by proteins up-regulated in poor quality eggs were related to cofactor binding, metal cluster binding, pattern binding, modification-dependent protein binding, kinesin binding, TBP-class protein binding, electron transfer activity, pre-mRNA binding, single-stranded RNA binding, and mRNA binding. Cellular components overrepresented by these proteins were respiratory chain, cytochrome complex, Sm-like protein family complex, oxidoreductase complex, mitochondrial protein complex, spliceosomal complex, polysome, mitochondrial membrane part, mitochondrial inner membrane and ficolin-1-rich granule. KEGG pathways that were significantly overrepresented by the same set of proteins were Alzheimer’s disease, oxidative phosphorylation, thermogenesis, Parkinson’s disease, cardiac muscle contraction, citrate cycle (TCA cycle), Huntington disease, metabolic pathways, spliceosome, and non-alcoholic fatty liver disease (NAFLD) (Fig. 3A, B, C and D Right sides). Taken together, the congruent results of the GO enrichment analyses for Biological processes, Molecular functions, Cellular components and KEGG pathways clearly indicate for a struggle of the BQ eggs in RNA processing along with mitochondria generation and functioning.
When the 115 differentially regulated proteins with significant differences in abundance between good and poor quality eggs were submitted separately (down-regulated in BQ, N = 64, up-regulated in BQ, N = 51) to a functional protein association network analysis using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) and the human protein database, they resolved into a network with significantly and substantially greater numbers of known and predicted interactions between proteins than would be expected of the same size lists of proteins randomly chosen from the human database (Fig. 4). The subnetwork formed by proteins down-regulated in poor quality eggs is made up of three interrelated clusters mainly related to cytoskeletal regulation and energy and protein homeostasis (Fig. 4 Left side). The cluster to the far left includes proteins involved in cytoskeletal organization such as Adenylyl cyclase-associated protein 1 (CAP1), Actin beta (ACTB), Tubulin alpha 4a (TUBA4A), Kinesin family member 1B (KIF1B), Voltage dependent anion channel 1 (VDAC1), Deoxyuridine 5′-triphosphate nucleotidohydrolase, mitochondrial (DUT), and Adenosylhomocysteinase like 1 (AHCYL1), and in energy production and homeostasis such as Creatine kinase (M-type) (CKM), Phosphoglycerate mutase 1 (PGAM1), and Enolase (ENO1). Other proteins forming this cluster are the Complement component 1 Q subcomponent-binding protein, mitochondrial (C1QBP) and Prohibitin (PHB), which are related to mitochondrial structure, and Superoxide dismutase 1 (SOD1), which is related to redox/detox activities.
The central cluster in this subnetwork includes proteins related to mRNA biogenesis and transcription (LSM6 homolog, U6 small nuclear RNA and mRNA degradation associated (LSM6), ATP-dependent RNA helicase DDX6 (DDX6), Small nuclear ribonucleoprotein U1 subunit 70 (SNRNP70), and Mago homolog, exon junction complex subunit (MAGOH)), protein translation (Gem-associated protein 5 (GEMIN5), Eukaryotic translation initiation factor 3 subunit L (EIF3L), Ribosomal protein L17 (RPL17), G elongation factor mitochondrial 1 (GFM1), and Translation initiation factor IF-3, mitochondrial (MTIF3)), protein folding (Dolichyl-diphosphooligosaccharide-protein glycosyltransferase subunit 2 (RPN2), Methyltransferase like 7A (METTL7A), Peptidylprolyl isomerase like 4 (PPIL4), DnaJ heat shock protein family (Hsp40) member A4 (DNAJA4), DnaJ heat shock protein family (Hsp40) member C19 (DNAJC19), Peptidylprolyl isomerase D (PPID), Protein disulfide isomerase family A member 4 (PDIA4), Quiescin sulfhydryl oxidase 1 (QSOX1)), protein transport (Importin (IPO9)) and Huntingtin-interacting protein 1-related protein (HIP1R)). Three other proteins included in this cluster are CCR4-NOT transcription complex subunit 1 (CNOT1), a transcription suppressor associated with DNA damage, Lipopolysaccharide-induced tumor necrosis factor-alpha factor homolog (LITAF), which targets proteins for lysosomal degradation, and Ferritin heavy chain 1 (FTH1), which is related to cellular iron homeostasis. The cluster in the upper right of this subnetwork includes proteins whose major functions are mostly related to fatty acid degradation (Propionyl-CoA carboxylase beta chain, mitochondrial (PCCB), Alpha-aminoadipic semialdehyde dehydrogenase (ALDH7A1), Glutaryl-CoA dehydrogenase, and mitochondrial (GCDH)), and amino acid catabolism in mitochondria (2-oxoisovalerate dehydrogenase subunit beta, mitochondrial (BCKDHB)), and a redox factor employed for respiration in the electron transport chain (Fumarylacetoacetate hydrolase domain containing 2A (FAHD2A)).
Proteins that were found to be up-regulated in poor quality eggs formed a subnetwork made up of two major clusters (Fig. 4 Right side). The cluster on the top left includes proteins mainly involved in mitochondrial structure (Ubiquinol-cytochrome c reductase binding protein (UQCRB), Cytochrome b-c1 complex subunit Rieske, mitochondrial (UQCRFS1), Cytochrome c1 (CYC1), and ATP synthase F1 subunit beta (ATPF5B)), complex assembly factors (Complex I assembly factor ACAD9, mitochondrial (ACAD9), NADH:ubiquinone oxidoreductase complex assembly factor 4 (NDUFAF4)), and proteins related to mitochondrial energy generation (Fumarate hydratase, mitochondrial (FH), ATP citrate lyase (ACLY), Fructose-bisphosphatase 1 (FBP1), and Glycogen debranching enzyme (AGL)). This cluster includes two other proteins, Glycine-tRNA ligase (GARS), which is related to protein translation, and Hsp70-binding protein 1 (HSPBP1), which is related to protein degradation and synthesis inhibition. The second cluster shown in the bottom right of this subnetwork includes proteins mainly related to mRNA biogenesis and transcription (Cleavage stimulation factor subunit 2 (CSTF2), Splicing factor 3b subunit 2 (SF3B2), RNA-binding protein FUS (FUS), Pre-mRNA processing factor 8 (PRPF8), Heterogeneous nuclear ribonucleoprotein A1 (HNRNPA1) and TAR DNA binding protein (TARDBP)). Some other proteins within this cluster are the Mitochondrial import receptor subunit TOM34 (TOMM34) and Kinectin 1 (KTN1), which are involved in mitochondrial biogenesis, Karyopherin subunit alpha 2 (KPNA2), a nuclear protein importer, the GCN1 activator of EIF2AK4 (GCN1) related to protein degradation and synthesis inhibition, and the SET nuclear proto-oncogene (SET) involved in DNA replication and chromatin binding.
Enrichment results for the revealed networks are given in Table S3. Aside from being in complete accordance with findings of the GO enrichment analyses for Biological processes, Molecular functions and Cellular components, these results include some interesting KEGG and Reactome pathway enrichment signatures. On the one hand, proteins down-regulated in poor quality eggs were enriched in metabolic pathways, RNA degradation, valine, leucine, and isoleucine degradation, fatty acid degradation, necroptosis and glycolysis/gluconeogenesis KEGG pathways, and also the metabolism Reactome pathway (PPI network enrichment value p = 1.70 × 10− 9). On the other hand, proteins up-regulated in poor quality eggs were enriched in Alzheimer’s disease, Parkinson’s disease, thermogenesis, metabolic pathways, oxidative phosphorylation, cardiac muscle contraction, Huntington’s disease, citrate cycle (TCA cycle), spliceosome and non-alcoholic fatty liver disease (NAFLD) KEGG pathways, and also the citric acid (TCA) cycle and respiratory electron transport, ATP synthesis by chemiosmotic coupling, and heat production by uncoupling proteins, respiratory electron transport, processing of capped intron containing pre-mRNA, mRNA splicing - major pathway, metabolism, ISG 15 antiviral mechanism, and metabolism of RNA Reactome pathways (PPI network enrichment value p = 0.000574).
Taking into account the overall results obtained by TMT labeling-based LC-MS/MS, a total of 21 proteins that significantly differed in abundance between good and poor quality eggs were chosen as candidate markers of egg quality in this study. Thirteen proteins down-regulated in poor quality eggs were chosen to represent the majority of functional categories, with a special emphasis on proteins related to mitochondrial biogenesis and energy metabolism related proteins. The remaining proteins up-regulated in poor quality eggs were chosen to mainly represent the functional categories of mitochondrial biogenesis and energy metabolism. Fold differences in abundance of candidate proteins between good and poor quality eggs varied between 1.07 and 1.85 for proteins down-regulated in poor quality eggs and between 1.07 and 4.67 for proteins up-regulated in poor quality eggs. Comparisons between good and poor quality halibut eggs in the abundance of these proteins are given in Fig. S2.
qPCR
Gene expression levels for the 20 candidate markers with a significant difference in protein abundance between good and poor quality eggs, in addition to mt-atp6, a mitochondrial gene utilized as marker for high quality eggs in previous studies [13], are given in Fig. S3. Eight out of the 21 genes (cyc1, fh, uqcrb, gcn1, ghitm, uqcrfs1, fbp1a, and atp5f1a) exhibited increases in expression in poor quality eggs coinciding with increased abundance of the product protein and with significant differences between good and poor quality eggs (Independent samples t-test, p < 0.05 followed by Benjamini Hochberg correction for multiple testing, p < 0.05). Four genes (mt-nd5, mt-atp6, acly1, and dhrs9) appeared to show increased expression in poor quality eggs, the same tendency shown by product protein abundance, but these differences were not statistically significant. Finally, nine genes (gcdh, ppid, gatd3a, gfm1, cap1, phb, sod1, mecr, and vdac) appeared to exhibit changes in expression converse to the trend of product protein abundance, and these differences were also not statistically significant.
PRM based LC-MS/MS
Differential abundance of 8 of the 21 candidate marker proteins (MT-ND5, DHRS9, GATD3A, CAP1, GCN1, FBP1, UQCRFS1, GHITM) between good and poor quality eggs was confirmed by PRM-based LC-MS/MS (Fig. S4). The number of proteins targeted by this method was limited by the availability of peptides that were suitable for use as references for this study (see Material and Methods section for details). Results revealed all candidate marker proteins, except GHITM, to exhibit the same tendency of regulation relative to egg quality as was detected by TMT-labeling-based LC-MS/MS. However, only five candidate proteins (MT-ND5, DHRS9, GATD3A, FBP1, UQCRFS1) were found to significantly differ in abundance between good and poor quality eggs. Results were consistently stable for all representative heavy peptides, which varied from 1 to 3 in number of cases per candidate protein. Respectively, FBP1 and UQCRFS1 were up-regulated while MT-ND5, DHRS9 and GATD3A were down-regulated in poor quality eggs (Fig. S4). Comparisons of protein abundance quantified via TMT- versus PRM-based LC-MS/MS, and of gene expression quantified by qPCR, for the eight candidate marker proteins, are given in Fig. 5.
Transmission electron microscopy and mtDNA levels
Transmission electron microscopy (TEM) was conducted to detect possible differences in mitochondrial morphology or numbers between good and poor quality eggs. Number of vesicles with double membranes, as seen in intact mitochondria, and the number of intact mitochondria (those with ≥5 cristae) were significantly higher in poor quality eggs (p = 0.000724 and p = 0.010729, respectively) (Fig. S5). Poor quality eggs contained a ~ 1.3 x higher number of vesicles and a ~ 1.2 x higher number of intact mitochondria relative to good quality eggs. Poor quality eggs additionally exhibited a significantly higher (~ 1.3 x) number of cristae per mitochondria on average in comparison to good quality eggs (p = 9.21E-15). In contrast, good quality eggs contained larger, well-formed mitochondria with significantly higher mitochondrial area (μm2) and mitochondria circularity (p = 1.15E-08 and p = 0.016094, respectively) (Fig. S5). There was no significant difference between good and poor quality eggs in total mitochondrial area per unit of cytoplasmic area (p = 0.408). A high variation among females of the same quality group and within eggs from the same batch was observed. Some eggs from good quality batches contained irregularly-shaped, empty vesicles (possibly former mitochondria) while some others from poor quality batches exhibited well-formed mitochondria with well-defined cristae. Moreover, evidence of possible mitochondrial fusion was observed in both good and poor quality eggs (Fig. 6, Fig. S6).
The higher incidence of small and poorly formed mitochondria containing higher numbers of cristae in poor quality halibut eggs led us to quantify and compare genomic mitochondrial DNA levels (mt-nd5 and mt-atp6) in good versus poor quality eggs. Results did not reveal any statistically significant differences in mtDNA levels at 1 hpf or 24 hpf stages (p > 0.05) (Fig. 7).