Proteomics Analysis of Early Developmental Stages of Zebrafish Embryos

Zebrafish is a well-recognized organism for investigating vertebrate development and human diseases. However, the data on zebrafish proteome are scarce, particularly during embryogenesis. This is mostly due to the overwhelming abundance of egg yolk proteins, which tend to mask the detectable presence of less abundant proteins. We developed an efficient procedure to reduce the amount of yolk in zebrafish early embryos to improve the Liquid chromatography–tandem mass spectrometry (LC–MS)-based shotgun proteomics analysis. We demonstrated that the deyolking procedure resulted in a greater number of proteins being identified. This protocol resulted in approximately 2-fold increase in the number of proteins identified in deyolked samples at cleavage stages, and the number of identified proteins increased greatly by 3–4 times compared to non-deyolked samples in both oblong and bud stages. Gene Ontology and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis revealed a high number of functional proteins differentially accumulated in the deyolked versus non-deyolked samples. The most prominent enrichments after the deyolking procedure included processes, functions, and components related to cellular organization, cell cycle, control of replication and translation, and mitochondrial functions. This deyolking procedure improves both qualitative and quantitative proteome analyses and provides an innovative tool in molecular embryogenesis of polylecithal animals, such as fish, amphibians, reptiles, or birds.

Proteome in zebrafish is usually investigated in adult organs or tissues [12][13][14][15]. The overwhelming occurrence of vitellogenin yolk proteins is a limiting factor in a polylecithal embryo, such as in zebrafish, as it hinders global identification of less abundant proteins using mass spectrometry-based techniques [4,16]. Proteolytic peptides of yolk proteins can potentially subdue the ionization of the less abundant proteolytic peptides of non-yolk proteins [17,18]. Consequently, abundant yolk proteins can potentially interfere with the identification of cellular proteins, although the degree of such interference is unknown. To reduce the abundance of yolk proteins, deyolking protocols are employed; they have been used in a number of studies on zebrafish embryos and larvae from 3.3 h post-fertilization (hpf) to  [16] and the current deyolking protocols. The detailed information is given in the text.
We compared our protocol to the protocols by Link et al. [16], which were based on 1-step deyolking with or without subsequent washing steps. For a fair comparison, we compared the reference 1-step deyolking procedure without washing [16] to our 1-step deyolking procedure without washing, and the reference 1-step deyolking plus double wash procedure [16] to our 3-step deyolking plus single wash procedure.  [16] and the current deyolking protocols. The detailed information is given in the text.
We compared our protocol to the protocols by Link et al. [16], which were based on 1-step deyolking with or without subsequent washing steps. For a fair comparison, we compared the reference 1-step deyolking procedure without washing [16] to our 1-step deyolking procedure without washing, and the reference 1-step deyolking plus double wash procedure [16] to our 3-step deyolking plus single wash procedure.
Both methods resulted in a reduction of yolk proteins, and the washing steps further depleted the protein content. Nevertheless, our new protocol yielded a larger number of unique proteins from a smaller number of embryos. We obtained approximately 1.7-fold increase in protein concentration per embryo sample when applying the 1-step deyolking process. When using our 3-step deyolking + single wash protocol, the protein yields per embryo sample were 3.1-and 2.5-fold higher (1-cell and high stage embryos, respectively) than those obtained with the reference protocols [16] with 1-step deyolking + double wash (Figure 2A). The effective number of 1-cell stage embryos needed to collect a workable amount of protein (30 µg) was approximately 2 or 3 times lower when using our 1-step deyolking or 3-step deyolking + single wash protocols, respectively; for the high stage embryos, this number of embryos was approximately 2 times lower than that of the respective reference protocols ( Figure 2B). Also, compared to our new protocol, less amount of proteins was harvested with the reference protocol ( Figure 2B). Both methods resulted in a reduction of yolk proteins, and the washing steps further depleted the protein content. Nevertheless, our new protocol yielded a larger number of unique proteins from a smaller number of embryos. We obtained approximately 1.7-fold increase in protein concentration per embryo sample when applying the 1-step deyolking process. When using our 3-step deyolking + single wash protocol, the protein yields per embryo sample were 3.1-and 2.5-fold higher (1-cell and high stage embryos, respectively) than those obtained with the reference protocols [16] with 1-step deyolking + double wash (Figure 2A). The effective number of 1-cell stage embryos needed to collect a workable amount of protein (30 µg) was approximately 2 or 3 times lower when using our 1-step deyolking or 3-step deyolking + single wash protocols, respectively; for the high stage embryos, this number of embryos was approximately 2 times lower than that of the respective reference protocols ( Figure 2B). Also, compared to our new protocol, less amount of proteins was harvested with the reference protocol ( Figure 2B).  [16]; Lane 3-1-step deyolking + double wash reference protocol [16]; Lane 4-1-step deyolking (new method); and Lane 5-3-step deyolking + single wash (new method). At the bottom line, number of embryos is given for each sample, from which the proteins were extracted.

Proteome in Deyolked Versus Non-Deyolked Samples
Generally, the amount of extracted total protein per embryo increased with the developmental advancement of the embryo, and the deyolking procedure greatly reduced the protein concentration. However, this reduction was decreasing from over 25-fold in cleavage stages (1-cell, 16-cell, and 32cell stages) to approximately 15-fold in the bud stage (Table 1). high stage (right panel) using the reference protocols versus new protocols. Lane 1-non-deyolked embryo (control); Lane 2-1-step deyolking reference protocol [16]; Lane 3-1-step deyolking + double wash reference protocol [16]; Lane 4-1-step deyolking (new method); and Lane 5-3-step deyolking + single wash (new method). At the bottom line, number of embryos is given for each sample, from which the proteins were extracted.

Proteome in Deyolked Versus Non-Deyolked Samples
Generally, the amount of extracted total protein per embryo increased with the developmental advancement of the embryo, and the deyolking procedure greatly reduced the protein concentration. However, this reduction was decreasing from over 25-fold in cleavage stages (1-cell, 16-cell, and 32-cell stages) to approximately 15-fold in the bud stage (Table 1). Analysis of the digested protein samples using the one-dimensional (1D) mass spectrometry (MS)/MS shotgun proteomics approach (1D shotgun) consistently demonstrated that the deyolking procedure resulted in a greater number of proteins being identified (Supplementary File 1). In the non-deyolked samples, the total numbers of proteins identified throughout the developmental stages were relatively consistent, ranging from 338 to 434 proteins identified in the 1-cell and bud stages, respectively. By comparison, the numbers of proteins identified in deyolked samples in all the developmental stages were considerably higher than in the non-deyolked counterparts. In the cleavage stages, these differences were approximately 2-fold, and increased to over 3-fold in the later developmental stages, ranging from 696 to 1687 proteins identified in the 1-cell and bud stages, respectively ( Figure 3A, Supplementary File 1). In contrast to the non-deyolked samples, where there was no apparent correlation between the developmental progression and the total number of proteins identified, deyolked samples resulted in a consistent number of proteins identified throughout cleavage stages (1-cell, 16-cell, and 32-cell), which considerably increased in the later developmental stages ( Figure 3A). Most of the proteins identified in the non-deyolked samples were also found in the deyolked counterparts ( Figure 3A, Supplementary File 1). The number of proteins unique to the non-deyolked samples (that is, not found in the deyolked counterparts) was relatively stable throughout the developmental stages. In contrast, most of the proteins identified in the deyolked samples were unique, meaning that they were not found in the non-deyolked counterparts, and the number of unique proteins apparently increased throughout the embryonic development from the cleavage stages to the bud stage ( Figure   When looking only to the proteins shared between the non-deyolked and deyolked samples, representation of vitellogenin in the deyolked samples was substantially reduced (36-58 times, depending on developmental stage; Supplementary File 2). At the same time, the representation of non-vitellogenin proteins in the deyolked samples was considerably elevated (2-6 times, depending on developmental stage; Supplementary File 3).
Among the 504 proteins unique to non-deyolked samples, most of them were specific to the cleavage stages, and 42 proteins were found in all the developmental stages. By comparison, out of 2129 proteins unique to the deyolked samples, 420 proteins were found in the cleavage stages only, and 266 proteins were found commonly in all the deyolked samples. In contrast to the non-deyolked sample counterparts, a substantial proportion of unique proteins was found in either or both oblong and bud stages. In total, 465 proteins were present in both non-deyolked and deyolked samples across all the developmental stages ( Figure 3B, Supplementary File 1). Among the 504 proteins unique to non-deyolked samples, most of them were specific to the cleavage stages, and 42 proteins were found in all the developmental stages. By comparison, out of 2129 proteins unique to the deyolked samples, 420 proteins were found in the cleavage stages only, and 266 proteins were found commonly in all the deyolked samples. In contrast to the non-deyolked sample counterparts, a substantial proportion of unique proteins was found in either or both oblong and bud stages. In total, 465 proteins were present in both non-deyolked and deyolked samples across all the developmental stages ( Figure 3B, Supplementary File 1).

Functional Annotations of the Proteome
In both non-deyolked and deyolked samples, the identified proteins were substantially involved in metabolic, ribosome, and biosynthesis of secondary metabolite and proteasome pathways, while enrichments specific to sampling protocol and/or developmental stage were found in certain pathways, such as in proteasome, RNA transport, or thermogenesis pathways (Table 2). Analysis of representation of the identified proteins annotated to functional Gene Ontology (GO) terms revealed multiple processes, functions, and components overrepresented and underrepresented in both non-deyolked and deyolked samples, with some of them specific to the developmental stage ( Figure 4, Supplementary File 4).
To distinguish the effect of the extraction protocol (non-deyolked versus deyolked samples) from the biological features (natural representation of proteins at given developmental stage), we used functional annotations of proteins represented in both non-deyolked and deyolked samples from all the developmental stages as a filtering criterion. In this way, shared GO terms were established by: The same proteins identified in samples from both extraction methods ("Shared" dataset); different proteins in both datasets ("unique ND" and "unique DY" datasets) enriching the same terms; or partially the same and partially different proteins ("Shared" and "unique ND", "Shared" and "unique DY", and all the three datasets). Whereas, unique GO terms were established exclusively by proteins from either "unique ND" or "unique DY" datasets. To distinguish the effect of the extraction protocol (non-deyolked versus deyolked samples) from the biological features (natural representation of proteins at given developmental stage), we used functional annotations of proteins represented in both non-deyolked and deyolked samples from all the developmental stages as a filtering criterion. In this way, shared GO terms were established by: The same proteins identified in samples from both extraction methods ("Shared" dataset); different proteins in both datasets ("unique ND" and "unique DY" datasets) enriching the same terms; or partially the same and partially different proteins ("Shared" and "unique ND", "Shared" and "unique DY", and all the three datasets). Whereas, unique GO terms were established exclusively by proteins from either "unique ND" or "unique DY" datasets.
Clearly, the deyolking procedure yielded a considerable number of unique GO terms, which were not annotated with the proteins identified in the non-deyolked samples (Supplementary File 5). The most prominent, developmentally relevant examples included enrichment in: Cellular component organization, RNA splicing, DNA replication, intracellular transport, cell cycle, translational initiation, and mitochondrial organization, transport, and gene expression (biological process); ATP binding, GTP binding, NADH dehydrogenase activity, ribonucleoprotein complex binding, translation initiation factor activity, ribosome binding, and ligase activity (molecular function); chromosome, endoplasmic reticulum, Golgi-associated vesicle, polysome, spliceosomal complex, cytochrome, and mitochondrial ribosome, matrix, and respiratory chains I and II (cellular component). Similarly, underrepresentation in unique GO terms was developmentally relevant, and it included: Cell-cell signaling, chemical synaptic transmission, intracellular signal transduction, and immune response (biological process); DNA-binding transcription factor activity, transcription regulator activity, channel activity, G protein-coupled receptor activity, and kinase activity (molecular function); cell surface, extracellular region, and plasma membrane-bounded cell projection (cellular component). In contrast to the abundance of unique GO terms annotated with the "unique DY" dataset, there were very few unique GO terms associated with "unique ND" dataset, Clearly, the deyolking procedure yielded a considerable number of unique GO terms, which were not annotated with the proteins identified in the non-deyolked samples (Supplementary File 5). The most prominent, developmentally relevant examples included enrichment in: Cellular component organization, RNA splicing, DNA replication, intracellular transport, cell cycle, translational initiation, and mitochondrial organization, transport, and gene expression (biological process); ATP binding, GTP binding, NADH dehydrogenase activity, ribonucleoprotein complex binding, translation initiation factor activity, ribosome binding, and ligase activity (molecular function); chromosome, endoplasmic reticulum, Golgi-associated vesicle, polysome, spliceosomal complex, cytochrome, and mitochondrial ribosome, matrix, and respiratory chains I and II (cellular component). Similarly, underrepresentation in unique GO terms was developmentally relevant, and it included: Cell-cell signaling, chemical synaptic transmission, intracellular signal transduction, and immune response (biological process); DNA-binding transcription factor activity, transcription regulator activity, channel activity, G protein-coupled receptor activity, and kinase activity (molecular function); cell surface, extracellular region, and plasma membrane-bounded cell projection (cellular component). In contrast to the abundance of unique GO terms annotated with the "unique DY" dataset, there were very few unique GO terms associated with "unique ND" dataset, with the most notably enriched terms in molecular function: Carbohydrate binding and endopeptidase regulator activity (Supplementary Files 4 and 5).
A certain number of proteins was unique for a given developmental stage (that is, identified only in a single developmental stage), in both non-deyolked and deyolked samples. Interestingly, significantly enriched GO terms for these proteins were different for non-deyolked and deyolked samples, in all five developmental stages investigated (Supplementary File 6).

Discussion
The improved deyolking procedure resulted in a considerably high quantity of the extracted total protein (Figure 2A) We identified 2575 proteins in total. In the study by Link et al. [16], 57 proteins were found, but six of them were not identified, and two proteins had a duplicated ID. We manually retrieved these 50 IDs, and found that 47 (94%) proteins were present in our dataset. Two of the three proteins not found in our dataset were actually Cyprinus carpio and Drosophila melanogaster proteins, but their possible homologues in zebrafish were missing in our dataset as well. We used 3 times less embryos in our procedure ( Figure 2B) than in the reference procedure [16]. Consequently, we were able to conduct the proteomics analysis of zygotic and cleavage stages of zebrafish for the first time.
Most of the proteins identified in the cleavage stages were unique to these stages of development (786 out of 1494; Figure 2B). This indicates a massive dynamics of zebrafish developmental proteome. It needs to be noted that the protein sequence database, which we did not use for annotating MS data, does not include the sequences of micro-peptides. Therefore, we cannot determine whether the method is suitable for harvesting very small proteins and micro-peptides.
KEGG analysis showed that ribosome, biosynthesis of secondary metabolites, carbon metabolism, and proteasome pathways were detectable in all the samples (Table 2). Also, a number of GO terms were detected in both deyolked and non-deyolked datasets (Supplementary File 5). Nevertheless, we observed a substantial increase in the number of identified unique proteins in the deyolked samples as compared to the non-deyolked counterparts. Consequently, they enriched a number of developmentally relevant GO terms, such as the cell cycle, mitochondrial organization, and functions or translation initiation, which were not enriched in the non-deyolked samples (Supplementary File 5). These functional terms are essential for the proper growth and development of the early stage of embryos [25][26][27][28] Knowledge of developmentally relevant proteome will aid understanding the regulation of early embryonic development. The underrepresented GO terms in the deyolked samples were mainly related to cellular signaling, transcription, G protein-coupled receptor activity, and cell surface (Supplementary File 5). These terms were not found underrepresented in the non-deyolked samples. In contrast to the significant GO terms found uniquely in the deyolked samples, there were very few unique GO terms associated with "unique ND" dataset (Supplementary File 5); this indicates that the presence of many embryonic proteins is masked due to the high abundance of yolk.
The functional annotation of cleavage stage proteome is concordant with the canonical knowledge of the catabolism, cell cycle, subcellular organization, and the transcriptional quiescence of pre-MZT embryos [24,29]. Moreover, our data suggest active translation-related processes in the very early embryos. Since zygotic transcripts are not produced yet [8], maternally-provided mRNAs [30] were used to produce the translational machinery and perform the translation. Quantitative proteome analysis throughout the development would be needed to determine the extent of this process, though.
Although the dechorionation/deyolking procedure generally resulted in a substantial increase in the number of identified proteins, it also resulted in a loss of certain proteins as compared to the non-deyolked counterparts ( Figure 3A), similarly to a study on 5 dpf zebrafish larvae [21]. Most of the previous proteomic studies did not address the problem of protein depletion due to the deyolking process, and they only used deyolked embryos for the analyses [16,21,22]. In the present study, approximately 30% of the proteins at cleavage stages and 12% at oblong and bud stages were not identified after the deyolking (Figure 3, Supplementary File 1). The GO analysis revealed that these lost proteins are involved in a number of biological processes (translation, protein folding, and mitochondrial organization), molecular functions (generation of precursor metabolites and energy), and cellular component (ribosome and cytosol; Supplementary File 4). Moreover, developmental stage-unique proteins enrich GO terms different for non-deyolked and deyolked samples, in all investigated developmental stages (Supplementary File 6). Altogether, our results suggest that deyolked and non-deyolked samples should be analyzed in parallel to extract a reliable information on the proteome in embryonic development.

Fish
The samples were collected at the zebrafish facility of the Nord University, Bodo, Norway. The experimental process and husbandry were performed in agreement with the Norwegian Regulation on Animal Experimentation (The Norwegian Animal Protection Act, No. 73 of 20 December 1974). This was certified by the National Animal Research Authority, Norway, General License for Fish Maintenance and Breeding no. 17.
The maintenance of zebrafish was done using an Aquatic Habitats recirculating system (Pentair, Apopka, FL, USA) and following established protocols [31]. The fish were fed newly hatched Artemia sp. nauplii (Pentair) and SDS zebrafish-specific diet (Special Diet Services, Essex, UK) according to the manufacturers' instruction. The zebrafish used in the experiment were from the AB line.

Sample Collection
Embryos originated from natural spawning and were collected at five developmental stages ( Figure 5). Embryo development was monitored and staged according to Kimmel et al. [32]. For each developmental stage, embryo batches were divided into two variants: Non-deyolked and deyolked. The non-deyolked (intact) embryos were promptly snap-frozen in liquid nitrogen and subsequently stored at −80 • C. The deyolked embryo variants went through the process of dechorionation (removal of chorion) and deyolking. Additionally, the 1-cell (0.5 hpf) and high-stage (3.3 hpf) embryos were collected to compare our deyolking protocol with that by [16]. mitochondrial organization), molecular functions (generation of precursor metabolites and energy), and cellular component (ribosome and cytosol; Supplementary File 4). Moreover, developmental stage-unique proteins enrich GO terms different for non-deyolked and deyolked samples, in all investigated developmental stages (Supplementary File 6). Altogether, our results suggest that deyolked and non-deyolked samples should be analyzed in parallel to extract a reliable information on the proteome in embryonic development.

Fish
The samples were collected at the zebrafish facility of the Nord University, Bodo, Norway. The experimental process and husbandry were performed in agreement with the Norwegian Regulation on Animal Experimentation (The Norwegian Animal Protection Act, No. 73 of 20 December 1974). This was certified by the National Animal Research Authority, Norway, General License for Fish Maintenance and Breeding no. 17.
The maintenance of zebrafish was done using an Aquatic Habitats recirculating system (Pentair, Apopka, FL, USA) and following established protocols [31]. The fish were fed newly hatched Artemia sp. nauplii (Pentair) and SDS zebrafish-specific diet (Special Diet Services, Essex, UK) according to the manufacturers' instruction. The zebrafish used in the experiment were from the AB line.

Sample Collection
Embryos originated from natural spawning and were collected at five developmental stages ( Figure 5). Embryo development was monitored and staged according to Kimmel et al. [32]. For each developmental stage, embryo batches were divided into two variants: Non-deyolked and deyolked. The non-deyolked (intact) embryos were promptly snap-frozen in liquid nitrogen and subsequently stored at -80 °C. The deyolked embryo variants went through the process of dechorionation (removal of chorion) and deyolking. Additionally, the 1-cell (0.5 hpf) and high-stage (3.3 hpf) embryos were collected to compare our deyolking protocol with that by [16].

Dechorionation and Deyolking
Embryos were placed in a Petri dish in phosphate-buffered saline (PBS) supplemented with 1.0 mg/mL Pronase (Sigma Aldrich, St. Louis, MO, USA) [31]. The enzymatic digestion of chorion was performed for 5 min at 37 °C with gentle shaking. Embryos were washed a minimum of 5 times with PBS or until all visible chorion fragments were removed.
The dechorionated embryos were processed using our modified protocol with 3-step deyolking and a single wash. The embryos were transferred to 1.5 mL Eppendorf tubes containing 1.0 mL of deyolking buffer (55 mM NaCl, 3.6 mM KCl, and 1.25 mM NaHCO3) and were mechanically disrupted by pipetting repeatedly through a 100 µL tip. The content was gently mixed by inverting

Dechorionation and Deyolking
Embryos were placed in a Petri dish in phosphate-buffered saline (PBS) supplemented with 1.0 mg/mL Pronase (Sigma Aldrich, St. Louis, MO, USA) [31]. The enzymatic digestion of chorion was performed for 5 min at 37 • C with gentle shaking. Embryos were washed a minimum of 5 times with PBS or until all visible chorion fragments were removed.
The dechorionated embryos were processed using our modified protocol with 3-step deyolking and a single wash. The embryos were transferred to 1.5 mL Eppendorf tubes containing 1.0 mL of deyolking buffer (55 mM NaCl, 3.6 mM KCl, and 1.25 mM NaHCO 3 ) and were mechanically disrupted by pipetting repeatedly through a 100 µL tip. The content was gently mixed by inverting the tube several times before centrifugation at 13,000 RPM for 1 min at 4 • C. The supernatant containing the yolk was discarded, and the pellet was re-suspended with the deyolking buffer, vortexed, and centrifuged as above. The procedure was repeated two times. After this, the pellet was re-suspended with 10 mM Tris-HCl (pH 7.4), vortexed, and centrifuged as above. The supernatant was discarded and the pellet (deyolked embryos) was snap-frozen in liquid nitrogen and stored at −80 • C. Additionally, for comparison of our protocol with that of [16], the dechorionated embryos at 1-cell and high stage were subjected to two types of deyolking protocols reported by Link et al. [16]: (1) 1-step deyolking, and (2) 1-step deyolking with two additional wash steps.

Protein Extraction
Both intact (non-deyolked) and deyolked embryo samples were lysed by adding 100 µL of sodium dodecyl sulphate (SDS) lysis buffer (1% SDS; Sigma-Aldrich, St. Louis, MO, USA), 0.5 M triethylammonium bicarbonate buffer pH 8.5 (TEAB; Sigma Aldrich), and 1 × Protease Inhibitor cocktail (Thermo Scientific, Rockford, IL, USA)). The tubes were vortexed and incubated at 90 • C for 30 min, then cooled on crushed ice for 5 min. The lysed material was centrifuged at 13,000 RPM for 20 min at 4 • C. The supernatant, containing the proteins, was collected and transferred to a new Eppendorf tube. The total protein concentration was quantified using a Qubit ® 3.0 Fluorometer (Invitrogen, Eugene, OR, USA) and the Qubit™ Protein Assay Kit (Invitrogen) according to the manufacturer's instructions. After the quantification, the samples were freeze-dried (VirTis BenchTop™ K, Warminster, USA) at −80 • C for 18 h before being shipped to the Department of Biological Sciences, National University of Singapore for proteomics analysis.

Polyacrylamide Gel Electrophoresis
One-dimensional gel electrophoresis was performed to check the efficiency of deyolking protocol, as well as to compare the efficiency of our protocol with the previous ones. Approximately equal concentrations of proteins from each sample were supplemented with 2× SDS loading dye. The samples were denatured by incubation at 95 • C for 10 min and then the proteins were separated by SDS gel electrophoresis (4%-20% Mini-PROTEAN ® TGX™ Precast Protein Gels, Bio-Rad, Hercules, California, USA) in SDS running buffer for 1 h. Afterwards, the gel was washed with deionized water for 10 min. The gel was stained with Coomassie Blue (Coomassie Brilliant Blue R-250, Bio-Rad) for 20 min, and de-stained with de-staining solution (40% methanol + 10% acetic acid) overnight at room temperature.

Tube-Gel Digestion and Sample Clean up
For each sample, 30 µg of proteins were used for downstream proteomics analyses. The samples were polymerized in a 10% polyacrylamide gel containing 4% SDS and subsequently fixed with a fixing reagent (50% methanol, 12% acetic acid) for 30 min at room temperature. The gel was cut into small pieces (1 mm 3 ) before being washed three times with 50 mM TEAB/50 % acetonitrile (v/v) and dehydrated with 100% acetonitrile. Next, samples were reduced using 5 mM Tris(2-carboxyethyl) phosphine (TCEP) at 57 • C for 60 min, followed by alkylation with 10 mM methyl methanethiosulfonate (MMTS) for 60 min at room temperature with occasional vortexing. The gel pieces were washed in 500 µL of 50 mM TEAB, dehydrated in 500 µL acetonitrile, and re-hydrated with 500 µL of 50 mM TEAB. The final dehydration step was performed using 100 µL acetonitrile. Trypsinization (1.5 µg trypsin) was performed at 37 • C for 16 h. The digested peptides were centrifuged at 6000× g for 10 min to collect the supernatant and stored at −20 • C (protocol modified from [17]. The samples were lyophilized and 30 µL of the dissolution buffer (0.5 M TEAB, pH 8.5) was added to each sample.

Protein Identification and Quantification
Peptide identification was carried out with the ProteinPilot 5.0 software Revision 4769 (AB SCIEX) using the Paragon database search algorithm (5.0.0.0.4767) and the integrated false discovery rate (FDR) analysis function. The data were searched against protein sequence databases downloaded from UniProt on May 2018 (total 119,356 entries). The MS/MS spectra obtained were searched using the following user-defined search parameters: Sample Type: Identification; Cysteine Alkylation: MMTS; Digestion: Trypsin; Instrument: TripleTOF5600; Special Factors: None; Species: None; ID Focus: Biological Modification; Database for 2018_May_uniprot-zebrafish.fasta; Search Effort: Thorough; FDR Analysis: Yes. The MS/MS spectra were searched against a decoy database to estimate FDR for peptide identification. The decoy database consisted of reversed protein sequences from the UniProt zebrafish database. FDR analysis was performed on the dataset and peptides identified with a confidence interval ≥95% were taken into account.

KEGG and Gene Ontology (GO) Functional Pathways Analysis
To analyse functional pathways associated with protein identified from deyolked and non-deyolked samples, KEGG analysis was performed. The FASTA files were submitted to online server "KAAS -KEGG Automatic Annotation Server" (https://www.genome.jp/kegg/kaas/) in order to get KEGG Orthology (KO) assignments [34]. To map KEGG pathways, the obtained KO numbers were submitted to KEGG mapper web server (http://www.genome.jp/kegg/tool/map_pathway2.html) [35].
GO annotation results and pathway of differentially expressed proteins in pairwise comparisons were obtained using Panther (Panther14.0, 2018_04) [36]. The web conversion tool (https://biodbnetabcc.ncifcrf.gov) was used to convert unmapped UniProt Accession IDs to ZFIN ID. The web tool Biomart was used to convert unmapped ZFIN IDs to Gene stable ID and to manually identify the unmapped IDs by gene names [37]. UniProt was used to identify protein IDs discontinued (deleted) in the 2018_11 release [38].

Conclusions
We established an effective deyolking procedure for the proteome analysis of the early stages of zebrafish embryos. Elimination of most of the yolk from early stages of embryos significantly enhanced the identification of cellular proteins with LC-MS-based shotgun proteomics analysis. The improved protocol is applicable to low-input material, enabling investigation of the earliest stages of development. Also, we demonstrated that the deyolking procedure results in the depletion of certain parts of the proteome that can be important in embryonic development. Thus, we suggest that both deyolked and non-deyolked samples should be processed in parallel to ensure a reliable coverage of the proteome during the embryogenesis. Our deyolking procedure will improve both qualitative and quantitative proteome analyses throughout embryonic development of polylecithal animals, such as fish, amphibians, reptiles, and birds.

Conflicts of Interest:
The authors declare no competing interests. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.