The Unique Features of Proteins Depicting the Chicken Amniotic Fluid

Abstract

In many amniotes, the amniotic fluid is depicted as a dynamic milieu that participates in the protection of the embryo (cushioning, hydration, and immunity). However, in birds, the protein profile of the amniotic fluid remains unexplored, even though its proteomic signature is predicted to differ compared with that of humans. In fact, unlike humans, chicken amniotic fluid does not collect excretory products and its protein composition strikingly changes at mid-development because of the massive inflow of egg white proteins, which are thereafter swallowed by the embryo to support its growth. Using GeLC-MS/MS and shotgun strategies, we identified 91 nonredundant proteins delineating the chicken amniotic fluid proteome at day 11 of development, before egg white transfer. These proteins were essentially associated with the metabolism of nutrients, immune response and developmental processes. Forty-eight proteins were common to both chicken and human amniotic fluids, including serum albumin, apolipoprotein A1 and alpha-fetoprotein. We further investigated the effective role of chicken amniotic fluid in innate defense and revealed that it exhibits significant antibacterial activity at day 11 of development. This antibacterial potential is drastically enhanced after egg white transfer, presumably due to lysozyme, avian beta-defensin 11, vitelline membrane outer layer protein 1, and beta-microseminoprotein-like as the most likely antibacterial candidates. Interestingly, several proteins recovered in the chicken amniotic fluid prior and after egg white transfer are uniquely found in birds (ovalbumin and related proteins X and Y, avian beta-defensin 11) or oviparous species (vitellogenins 1 and 2, riboflavin-binding protein). This study provides an integrative overview of the chicken amniotic fluid proteome and opens stimulating perspectives in deciphering the role of avian egg-specific proteins in embryonic development, including innate immunity. These proteins may constitute valuable biomarkers for poultry production to detect hazardous situations (stress, infection, etc.), that may negatively affect the development of the chicken embryo.

In oviparous species, embryonic development depends on the various components, nutrients and structures composing the eggshell, the egg yolk, the egg white and the vitelline membrane (1). It also relies on the proper development of the extra-embryonic structures, namely the yolk sac, the amniotic sac and the allantoic/chorioallantoic sac (1) (Fig. 1A,). These structures develop at the very early stages of development and originate from embryonic tissues but are not considered to be part of the embryonic body (2). They are discarded or resorbed at hatching. These living structures are partly preserved among amniote species, but exhibit evolutionary particularities depending on the embryonic development mode (3). The yolk sac, which appears in the first stages of development, degenerates rapidly in mammals (Fig. 1B,), whereas in some birds, it participates in digestive processes until the last stages of incubation prior to complete abdominal resorption at hatch. The yolk sac may have many other functions, which are temporally regulated during incubation: it resembles the liver in the synthesis of plasma proteins, the bone marrow in erythropoiesis, and the intestine, in digestion of nutrients and their transport to the embryo (4). Thus, the yolk sac plays different roles to support or replace the functions of several organs that have not yet reached their full functional capacity. The chorioallantoic sac is composed of the chorioallantoic membrane, which results from the fusion of the chorion and the allantois at day 5/6 of incubation (ED5/ED6), and it includes the allantoic fluid. It is a highly vascularized structure that performs many functions during chicken embryonic development: it collects nitrogenous and excretory products from the embryonic metabolism, it participates in respiratory exchange, in calcium transport from the eggshell toward the embryo, in ion and water reabsorption from the allantoic fluid and, thus, in acid-base homeostasis (2, 5). In humans, the allantoic sac forms only a part of the umbilical cord. Concerning the amniotic sac, it is described in all amniotes as a structure, with amniotic fluid (AF)¹, which protects the embryo against mechanical shocks, dehydration or adhesion to the other extra-embryonic membranes. It also serves as a source of nutrients (6). It provides a favorable environment for the development of the embryo: pH of about 7.1 to 7.3, stable temperature, and sensorial stimulation (taste, sense of smell and hearing) (7).

Fig. 1.

Schematic representation of the extraembryonic structures during the chicken (A,) and human embryonic development (B,); chicken embryo at mid-incubation (11 days) and human embryo at mid-gestation (21 weeks), respectively.

Human AF is a fluctuating milieu mainly composed of water (about 96.4%), minerals, trace elements, carbohydrates, hormones, glucose, lipids, urea, cells, free amino-acids, proteins and peptides (6, 8). Its biochemical composition changes with gestational age/developmental stage as a result of various physiological mechanisms including feto-maternal exchanges: AF swallowing and lung fluid production by the embryo, excretion of fetal urine and transfer of solutes and fluids across amniotic and uterine membranes (Fig. 1B,) and across skin especially before keratinization (6, 9). In humans, before skin keratinization (25 weeks of age), the composition of the AF is very similar to fetal blood plasma. As a dynamic milieu reflecting the physiological or pathological status of the embryo, the biochemical composition of the AF has been extensively characterized in humans since it contains molecular markers for the detection of embryonic abnormal development, inflammation, infection or pregnancy-related complications (10). Its role as a mechanical cushioning of the embryo, together with the fact that 25% of its total protein content is associated with the immune response or is related to defense (9), corroborate its primary role in the protection of the embryo. However, it seems that proteins of the human AF are not solely involved in the nutrition and protection of the embryo, but they display many other functions related to metabolism and development (11). In oviparous species, however, the functions of the AF are still poorly understood and there are several structural and biochemical particularities that suggest similar but also diverging functions. First, there are two major extraembryonic fluids in birds, the amniotic and allantoic fluids, which are physically separated, to ensure different functions. Like humans, the bird AF is contained in the amniotic sac and bathes the embryo, whereas the allantoic fluid is secreted in the chorioallantoic sac (Fig. 1A,), which is an intestinal intussusception of the embryo that receives disposable wastes directly from the embryonic kidneys. These anatomical specificities deeply influence the biochemical composition of the AF, which thus in birds does not collect fetal urine. The second main difference, as compared with mammals, is that after ED12, egg white proteins are massively transferred into the amniotic sac (12), where they are absorbed orally by the embryo as a source of amino acids to support its rapid growth (13–17) until hatching (ED21). Consequently, this process drastically impacts the protein concentration of the AF, which is barely measurable before the 11th day of incubation (ED11) (about 0.01 mg/ml) and reaches 200 mg/ml following egg white transfer (18). In fact, before ED11, the chicken AF is mainly composed of water and mineral elements, such as chloride, sodium, potassium, phosphorus, magnesium, calcium, iron, and sulfur, like human AF (19, 20). More recently, a study analyzing the major proteins of the chicken AF, revealed that it contains egg white proteins even before the massive egg white transfer at ED12 (18). Some of these proteins are egg yolk proteins while others may originate from the embryo (skin, feather) or its extra-embryonic membranes (amniotic, yolk, and chorioallantoic sacs). All these proteins have been associated with functions comparable to those described for human AF, including metabolism, immune system or tissue remodeling, but they also serve as a major source of amino acids and energy for the embryo, especially during the second half of incubation. The avian specificity of some egg white and yolk proteins (21) that are recovered in the AF, together with the physical separation of amniotic sac from the embryonic urinary system (chorioallantoic sac) and the transfer of egg white proteins at midincubation, suggest that this fluid may have very specific biological functions related to birds or even oviparity.

In such a stimulating context, we explored the avian AF proteins and specificities using the chicken as a model of birds. Similar to human AF, with albumin, immunoglobulins, transferrin and haptoglobin (9, 11), chicken AF contains a few major proteins (ovalbumin, ovotransferrin, etc.) that complicate proteome profiling and mask the presence of proteins of lower abundance. Therefore, we designed a bottom-up proteomic approach combining two complementary strategies as GeLC-MS/MS (protein samples fractionated by SDS-PAGE are analyzed by nanoLC-MS/MS after in-gel digestion) and shotgun (protein samples are directly analyzed by nanoLC-MS/MS after in-solution digestion) to give an exhaustive view of the AF proteome at ED11, before the substantial transfer of egg white (which completely changes the global protein profile). We performed a functional annotation of AF proteome using Gene Ontology approaches: 1) on the 10 most abundant proteins representing 66% to 81% of the total protein contents for GeLC-MS/MS and shotgun analyses, respectively, and 2) on the complete list of proteins resulting from both analyses. This approach was compulsory because many of the major proteins composing egg structures do not have assigned functions in databases yet and consequently, the functional annotation of the entire list would enrich known functions, whose biological significance may be overestimated. Considering the importance of human amniotic fluid in innate immunity and to better appreciate the contribution of chicken amniotic fluid to embryo defense against microorganisms, we used an in-gel antibacterial assay combined to mass spectrometry, to identify antibacterial proteins and peptides contained in AF or in enriched fractions of AF, before and after egg white transfer. As the protein composition of the chicken AF is temporally regulated during incubation, AF functions are thoroughly modified following egg white transfer. Finally, a comparison with the published proteomes of the human AF (11) and a phylogenetic study on all proteins identified in this study, allowed us to highlight chicken AF specificities, revealing a set of proteins, which are only present in birds and whose physiological roles remain fully opened.

EXPERIMENTAL PROCEDURES

A diagram describing the experimental design is presented in Fig. 2.

Fig. 2.

Diagram describing the experimental design. A,, preparation of biological samples, B,, proteomic analyses using Shotgun and GeLC-MS/MS analyses, C,, antibacterial potential of egg samples and identification of antibacterial candidates. AF, amniotic fluid; Cb, Coomassie Brilliant Blue staining; ED, Day of incubation; EW, egg white; HBP, heparin-binding proteins.

Samplings

Fluids were sampled as previously described by Da Silva et al., (18). Briefly, fertile eggs were incubated under standard conditions (45% relative humidity, 37.8 °C, automatic turning every hour), after a three-day storage at 16 °C and 85% relative humidity to ensure synchronization of developmental stages (UE1295, INRA, F-37380 Nouzilly, France). At ED11 and ED16, 40 eggs of comparable weight (62.4 ± 4.3 g) containing viable embryos (checked by candling) were tested with the Acoustic Egg Tester (KU Leuven, Belgium), and cracked eggs were discarded. By ED11, egg white was sampled with a syringe after drilling a hole in the eggshell. After removing the eggshell, the egg content was poured into a Petri dish and the AF was recovered with a syringe through the amniotic membrane. By ED16, the egg white was collected in the Petri dish with pliers due to its high viscosity. All samplings were performed under sterile conditions. Sex of individual embryos was determined using PCR (22).

Fluid Characterization

After samplings, AF were centrifuged at 3,000 g (10 min, 4 °C) to remove insoluble components. Volumes, pH (Microelectrode pH InLab 423, Fisher Scientific, Illkirch, France), osmolality (Fiske Mark 3 Osmometer, Advanced Instruments, Niederbronn-Les-Bains, France), and absorbance spectrum (Nanodrop, ND-100 Spectro, Wilmington, USA) were analyzed for each sample. The total protein concentration for each sample was assessed using BioRad DC Protein Assay Kit II (BioRad, Marnes-la-Coquette, France). All samples (25 μl) containing loading buffer (5X loading buffer: 0.25 m Tris-HCl, 0.05% bromphenol blue, 50% glycerol, 5% SDS, pH 6.8) were independently loaded on a 12.5% SDS-PAGE (1 mm) using a Mini-Protean II electrophoresis cell (BioRad), and further stained with Coomassie Brilliant Blue G250 or silver nitrate. This overall characterization (pH, osmolality, protein concentration, absorbance and electrophoretic patterns, and embryo's sex) helped us to select homogenous samples that were stored at − 20 °C for further analyses by mass spectrometry (supplemental Data S1).

In-Gel and In-Solution Tryptic Digestion

Twelve homogenous AF samples including six males and six females were pooled for protein identification. Proteins were either separated by a 4–20% SDS-PAGE followed by Coomassie Brilliant Blue G250 staining (GeLC-MS/MS analysis), or directly identified in solution (shotgun analysis) as described previously (18). The SDS-PAGE gel was cut in 20 sections (Fig. 3B,) and each slice was rinsed separately in water and then acetonitrile. Proteins were then reduced with dithiothreitol, alkylated with iodoacetamide, and incubated overnight at 37 °C in 25 mm NH₄HCO₃ with 12.5 ng/μl trypsin (Sequencing grade, Roche, Paris, France) as described by Shevchenko (23). Peptides were pooled and dried using a SPD1010 speedvac system. Peptide mixtures associated to each band and in-solution samples were analyzed using nanoLC-MS/MS.

Fig. 3.

SDS-PAGE analysis of the chicken amniotic fluid (AF) during incubation. A,, 12.5% SDS-PAGE analysis of the AF from days 8–11 (ED8–11) to days 12–16 (ED12–16) of the incubation (2 μg and 10 μg, respectively) followed by Coomassie Brilliant Blue staining. B,, 4–20% SDS-PAGE profile of AF at ED11 for GeLC-MS/MS analysis. Horizontal lanes and numbers indicate the position of gel slices (1–20) prepared for in-gel digestion by trypsin.

NanoLC-MS/MS

The resulting peptide mixtures were analyzed using a LTQ Orbitrap Velos Mass Spectrometer (Thermo Fisher Scientific, Germany) coupled to an Ultimate® 3000 RSLC chromatographer controlled by Chromeleon 6.8 software (Dionex, Amsterdam, The Netherlands). Five microliters of sample were desalted and preconcentrated on a trap column (Acclaim PepMap 100 C18, 100 μm inner diameter × 2 cm long, 3 μm particles, 100 Å pores) for 10 min at 5 μl/min with 4% solvent B (0.1% formic acid, 15.9% water, 84% acetonitrile) in solvent A (0.1% formic acid, 97.9% water, 2% acetonitrile). Separation was conducted using a nanocolumn (Acclaim PepMap C18, 75 μm inner diameter × 50 cm long, 3 μm particles, 100 Å pores) at 300 nL/min by applying a gradient of 4 to 55% of solvent B during 90 min for GeLC-MS/MS analyses and during 120 min for shotgun analyses.

Data were acquired using Xcalibur 2.1 software (Thermo Fisher Scientific). The instrument was operated in positive mode in data-dependent mode. Survey full scan MS spectra (from 400 to 1800 m,/z,) were acquired with a resolution set at 60,000. The 20 most intense ions with charge states ≥ 2 were sequentially isolated (isolation width: 2 m,/z,; 1 microscan) and fragmented using CID mode (energy of 35% and wideband-activation enabled). Dynamic exclusion was active during 30 s with a repeat count of one. Polydimethylcyclosiloxane (m,/z,, 445.1200025) ions were used for internal calibration.

The mass spectrometry proteomics data have been submitted to the ProteomeXchange Consortium and are available via, the PRIDE partner repository (24) with the data set identifiers PXD008046 and 10.6019/PXD008046.

Protein Identification and Data Validation

The reliability between replicates was investigated by comparing chromatograms and spectra with Xcalibur 2.1 software. MS/MS ion searches were performed using Mascot search engine v 2.3.2 (Matrix Science, London, UK) via, Proteome Discoverer 2.1 software (ThermoFisher Scientific) against National Center for Biotechnology Information (NCBI) database with Chordata taxonomy (782,473 entries, downloaded in January 2017). Fragments and parents' tolerances were set at 0.80 Da and 5 ppm, respectively. The search parameters included trypsin as a protease with two allowed missed cleavages and carbamidomethylcysteine, methionine oxidation and acetylation of N-term protein as variable modifications. Mascot results obtained from the target and decoy databases searches were subjected to Scaffold 4.8.2 software (Proteome Software, Portland, OR), and displayed as clusters. Peptide and protein identification and validation were performed using the Peptide and Protein Prophet algorithms, respectively (95.0% probability, at least two different unique peptides), from three technical replicates. The abundance of identified proteins was estimated by calculating the emPAI using Scaffold Q+ software (version 4.4, Proteome Software).

Keratins were not taken into consideration in the subsequent analysis as they might be either contaminant from human skin and/or chicken skin. The list of keratins identified in the analysis is however available (supplemental Data S2).

Functional Annotation Using Gene Ontology

Gene ontology terms annotations for biological processes and cellular components category provided by the GO consortium (http://www.geneontology.org/) were investigated using protein database with UniprotKB (http://www.uniprot.org), and genomic databases with Ensembl (http://www.ensembl.org) and NCBI (https://www.ncbi.nlm.nih.gov).

Comparison Between Human and Chicken AF Proteomes

IPI numbers and gene symbols corresponding to the proteins identified in the human AF were retrieved from the publication of Cho et al., (11), which integrated all previous studies on the human AF proteome. Earlier versions of the human protein sequences were recovered at the European Bioinformatics Institute website (ftp://ftp.ebi.ac.uk/pub/databases/IPI/) and updated according to the Ensembl database. Human and chicken gene symbols were compared and for associated genes, protein sequences were aligned using BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi) to identify similarity and identity percentages between both species. Human orthologues were systematically checked using Ensembl compara (http://www.ensembl.org) to confirm previous alignments.

Phylogenetic Analysis

Phylogenetic branches of emergence were defined in order to determine the moment of appearance of each gene in the tree of life. Eight possible branches for gene birth were considered: Opisthokontas, (animals and fungi ∼ 1215 million years ago (MYA)), Bilateria, (bilateral animals ∼ 937 MYA), Chordata, (chordates ∼ 722 MYA), Vertebrata, (vertebrates ∼ 535 MYA), Tetrapoda, (tetrapods ∼ 371 MYA), Amniota, (amniotes ∼ 296 MYA), Sauropsida, (reptiles and birds ∼ 276 MYA), and Mammalia, (mammals ∼ 220 MYA). For all identified proteins, the corresponding Ensembl protein ID was retrieved from the Ensembl database and the related phylogenetic trees were analyzed to highlight the specificity of proteins evolution (http://www.ensembl.org) (25). Eighty-nine trees were studied. To complement the phylogenetic trees, the conservation of synteny was systematically observed using Genomicus (http://www.genomicus.biologie.ens.fr) and Mapviewer (https://www.ncbi.nlm.nih.gov/mapview) (21, 26). The gene was defined as conserved if both surrounding/adjacent genes were the same.

Purification of Heparin Binding Proteins

Heparin-Sepharose chromatography was performed according to manufacturer's instructions. Briefly, 2 ml of beads (Heparin Sepharose 6 Flast Flow, GE Healthcare, Sweden) were loaded onto a polypropylene column (Qiagen, Courtaboeuf, France) and washed with 10 volumes of water and 10 volumes of washing buffer (50 mm Tris-HCl, 150 mm NaCl, pH 7.4). After loading the crude sample (pool of ten individuals, five males and five females), the beads were washed extensively with the washing buffer until the absorbance at 220 nm reached zero, as previously reported (27). Elution of bound proteins was achieved with 50 mm Tris-HCl, 2 m NaCl, pH 7.4, until the absorbance at 220 nm reached zero. Eluted fractions were desalted and concentrated by ultrafiltration (Ultracel-3K, Merck Millipore, Molsheim, France), and analyzed by 12.5% SDS-PAGE under nonreducing and nonboiling conditions to preserve protein integrity, followed by Coomassie Brilliant Blue G250 staining.

Antibacterial Assays

Antibacterial tests were conducted by direct detection of antibacterial activities after SDS-PAGE, a method adapted from Bhunia et al., (28). Pathogenic bacterial strains, Salmonella enterica, serovar Enteritidis ATCC 13076 (S.,E.) and Listeria monocytogenes, EGD strain (L.m,.) were provided by the International Centre for Microbial Resources (CIRM, https://www6.inra.fr/cirm_eng/Pathogenic-Bacteria) from the French National Institute for Agricultural Research (INRA, France). Precultures of S.,E. and L.m., were performed overnight in Trypticase Soy Broth (TSB, BD Biosciences, Le Pont de Claix, France) and in Brain Heart Infusion broth (BHI, BD Difco), respectively. This pre-culture was then used to inoculate a new culture broth (TSB or BHI) so that the midexponential phase was obtained after 3 or 4 h of incubation depending on strains, with shaking at 37 °C. Bacteria were centrifuged at 2000 × g, for 10 min at 4 °C, washed twice with cold 10 mm sodium phosphate buffer (pH 7.4), and resuspended in cold sodium phosphate buffer. Bacteria (7.5 × 10⁶ Colony Forming Unit) were introduced in 25 ml of autoclaved nutrient-poor agar (10 mm phosphate buffer containing 0.03% TSB medium, 1% low-endosmosis agarose (w/v) (Sigma-Aldrich, Saint-Quentin-Fallavier, France), and 0.02% Tween 20.

In parallel, maximum of 20 μg of protein pools (see above) were loaded on two identical gels and further separated by 15% SDS-PAGE under nonreducing and nonboiling conditions, while maintaining the electrophoresis system at 4 °C to avoid protein degradation. The first gel was stained with Coomassie Brilliant Blue G250 for further identification of antibacterial proteins by mass spectrometry. The second gel was dedicated to the antibacterial assay and was washed separately with 2.5% Triton X-100 and MilliQ water (4 × 15 min, 4 °C), to eliminate SDS and renature proteins. Thereafter, the washed gel was covered with the nutrient-poor agar containing bacteria and incubated for 3 h at 37 °C to allow protein diffusion in the agar. A second nutrient-rich agar (10 mm phosphate buffer containing 6% TSB medium, 1% low-endosmosis agarose (w/v)) was poured on the nutrient-poor agar to allow bacterial growth. After an overnight incubation at 37 °C, clear zones were defined as inhibition zones. These inhibition zones were superimposed on the Coomassie-stained gel to locate bands containing antibacterial proteins. These bands were cut from the Coomassie-stained gel and further processed as described above, for protein identification by mass spectrometry.

Beta-microseminoprotein-like (BMSP) and avian beta-defensin 11 (AvBD11), two recently characterized antibacterial proteins from egg white were used as positive external controls (2 μg of proteins/well, data not shown). They were obtained as previously described (27).

Experimental Design and Statistical Rationale

To address our scientific questions, we decided to use a conventional chicken laying strain (ISA Brown, Hendrix Genetics, St Brieuc, France) at 38 weeks of age, that were raised at the UE1295 Pôle d'Expe′rimentation Avicole de Tours (INRA, F-37380 Nouzilly, France). Eighty fertilized eggs were incubated under standard conditions after a three-day storage at 16 °C and 85% relative humidity to favor synchronization of developmental stages—knowing that conditions and duration of storage prior to incubation negatively impact embryo survival and development. The developmental stages were confirmed using the atlas of chicken developmental stages (29). Egg weight and eggshell quality were both checked and protein profiles of each individual sample were analyzed by SDS-PAGE (supplemental Data S1, sheets #1 and #2). In parallel, the sex of the embryo was determined for all corresponding samples as described under Experimental Procedures to generate a pool of combined male and females AF that could encompass most proteins composing AF, regardless of the sex. The combination of all these parameters allowed us to define 12 samples with an equilibrated sex ratio that were all homogenous in terms of stages of development, biochemical parameters (pH, osmolality, absorbance spectrum, etc.), and SDS-PAGE protein profiles. All experiments were conducted in compliance with the European legislation on the “Protection of Animals Used for Experimental and Other Scientific Purposes” (2010/63/UE) and under the supervision of an authorized scientist (S. Réhault-Godbert, Authorization no. 37–144). Two complementary bottom-up proteomic strategies (GeLC-MS/MS and shotgun analyses) were combined to give a representative overview of protein diversity in chicken AF samples. Protein sequences were retrieved from Scaffold and blasted to check their GenBank accession numbers (NCBI), and to find corresponding Ensembl and Uniprot IDs. Chicken proteins identified in other species were also searched by BLAST alignments. Functional annotation and search for antibacterial candidates were performed by combining Ensembl, Uniprot and NCBI data on chicken proteins and their orthologues. The identification of antibacterial candidates was completed by investigating protein and peptide antibacterial domains using specific sequence analyzing tools such as InterPro (https://www.ebi.ac.uk/interpro/). To produce enriched fractions, we defined optimal conditions (volume of samples, volume of heparin-Sepharose beads) depending on the initial concentration of proteins in AF and egg white samples from ED0, ED8, ED11, ED14 or ED16 stages. All samples (raw, flow-through and eluted fractions) were analyzed by SDS-PAGE. Protein profiles of egg white at ED11 and ED16 were compared with previously published data for experimental validation (27, 30). For antibacterial assays, at least three independent assays were performed on Listeria monocytogenes, and Salmonella enterica, Enteritidis strains, using two positive controls (one peptide and one protein) purified from chicken egg white. The phylogenetic analysis and corresponding figures were conducted combining data available in Ensembl databases and in literature.

RESULTS

Proteomic Profiling of AF

The global protein profile of AF assessed by SDS-PAGE is completely different when comparing the first half (ED8–11) to the second half of development (ED12–16), due to the massive transfer of egg white into the AF from ED12 onwards (Fig. 3A,). In this context, to better appreciate the intrinsic protein composition of chicken AF, a bottom-up proteomic study combining two different strategies was conducted on AF samples collected at ED11, the last day before egg white transfer. AF of 40 eggs corresponding to 17 females and 22 males (plus one embryo for which tissue sample was discarded) were collected as detailed under Experimental Procedures. Individual SDS-PAGE profiles (using both Coomassie Brilliant Blue and silver staining, supplemental Data S1, sheet #2), protein concentrations, pH, osmolality values, and spectra of absorbance were determined for each AF sample (supplemental Data S1, sheet #1). Because no significant differences could be detected between individuals regardless of the sex after integration of all parameters, we constituted a pool representing 12 homogenous samples with equilibrated sex ratio for further analyses by high-resolution mass spectrometry. We performed both in-gel and in-solution independent tryptic digestions followed by nanoLC-MS/MS analysis for protein identification and quantification. For in-gel digestion, a 4–20% SDS-PAGE gel was cut (Fig. 3B,) and each slice was further processed as mentioned under Experimental Procedures. We filtered data for keratins and other proteins that could reflect sample contamination by experimenters (supplemental Data S2). Ninety-one nonredundant proteins were identified, including 49 for shotgun analysis and 70 for GeLC-MS/MS analysis (Fig. 4). Twenty-eight proteins were common to both approaches (Fig. 4). The identified proteins sometimes refer to other bird species than Gallus gallus,, such as Coturnix japonica, (japanese quail) or fish Tachysurus fulvidraco, (yellow catfish) (supplemental Data S3), either because the Gallus gallus, counterpart is still missing in the genome annotation, or because these specific variants have not yet been identified in the chicken species. Raw files, proteins, peptides scores, information and quantitative values are compiled in supplemental Data S2 and S3.

Fig. 4.

Quantitative distribution of the 91 proteins identified in the chicken amniotic fluid at ED11, for shotgun and GeLC-MS/MS mass spectrometry analyses. Proteins are represented according to log10 of their Exponentially Modified Protein Abundance Index (emPAI) and log10 of the percentage of their protein sequence coverage. The most abundant proteins identified after compilation of data resulting from both approaches are illustrated in a dotted square. They were further analyzed for functional annotation as shown in Table I.

Fig. 4 illustrates the results using log10 of the Exponentially Modified Protein Abundance Index (emPAI), and log10 of the percentage of the protein sequence coverage (supplemental Data S3). According to this graph, a set of ten proteins are very abundant compared with the others. Regardless of the methods, apolipoprotein A1 (APOA1), ovotransferrin (TF) and alpha-fetoprotein (AFP) were three of the most abundant proteins found in AF at ED11 (Fig. 4). If we consider the ten proteins with the highest emPAI for both shotgun and GeLC-MS/MS approaches, seven were common to both approaches: APOA1, TF, AFP, TTR (transthyretin), OVAL (ovalbumin), LYZ (lysozyme) and ALB (serum albumin). APOC3 (apolipoprotein C-III), SPINK7 (ovomucoid) and MDK (midkine) were found in the top ten resulting from in-solution method whereas GC (vitamin-D binding protein), FSTL1 (follistatin-related protein A) and ACTB (actin, cytoplasmic 1) were identified in the top ten resulting from GeLC-MS/MS method (supplemental Data S3). The ten major proteins of AF after integration of quantitative values from both approaches are compiled in Table I. These proteins represent about 66% to 81% of the total protein content for GeLC-MS/MS and shotgun analyses, respectively. A primary analysis of the Gene Ontology terms of these major proteins revealed that six of them are associated with lipid, vitamin or hormone metabolisms (APOA1, AFP, ALB, GC, TTR, APOC3), while two additional proteins participate in embryo's defense against pathogens (TF and LYZ) (Table I). On the other hand, OVAL, which is specific to birds (31), is assumed to have a role in embryonic nutrition as a source of amino acids, while SPINK7 is a very potent protease inhibitor containing three inhibitory kazal-like domains. Despite the lack of information related to its physiological targets, SPINK7 presumably regulates endogenous proteolytic activities, preventing uncontrolled protein degradation and preserving overall protein integrity.

Table I. Top-ten abundant proteins found in the chicken amniotic fluid at ED11 (11th day of incubation) using shotgun and GeLC-MS/MS analysis. Quantitative values are expressed as the percentage of the Exponentially Modified Protein Abundance Index (emPAI). Words in italics refer to hypothetical/unknown functions. Proteins that are also listed in the top 15 high abundance proteins from the human amniotic fluid are indicated with an asterisk (9, 11). Proteins that were not detected in one or the other approach are indicated by a dash in the emPAI column.

Gene symbol	Gene Description	GeLC-MS/MS % emPAI	Shotgun % emPAI	Biological Process	Cellular component
APOA1*	Apolipoprotein A-I	27.8	39.1	Lipid metabolism	Secreted
OVAL	Ovalbumin	7.8	2.0	Nutrition,	Secreted
TF*	Ovotransferrin	7.8	17.6	Defense response; Iron metabolism	Secreted
AFP*	Alpha-fetoprotein	7.5	7.7	Blood pressure homeostasis; Development (folliculogenesis)	Secreted
LYZ	Lysozyme C	3.4	1.7	Defense response	Secreted
ALB*	Serum albumin	3.3	1.6	Blood pressure homeostasis; Fatty acid, DNA, ion metabolism	Secreted
GC*	Vitamin d-binding protein	3.2	0.4	Vitamin metabolism	Secreted
TTR	Transthyretin isoform 1	2.8	3.5	Vitamin and hormone metabolisms	Secreted
SPINK7	Ovomucoid isoform 1	–	2.4	Regulation of proteolytic processes,	Secreted
APOC3	Apolipoprotein C-III	–	7.1	Lipid metabolism	Extracellular region

Except APOC3, all these proteins were predicted to be secreted (Table I).

Gene Ontology Analysis of the Chicken AF Proteome

An analysis of the occurrence of the Gene Ontology terms related to the biological processes and molecular functions, revealed that the most representative function of AF proteins is associated with cellular adhesion and migration, followed by functions associated with metabolism and transport of lipids, vitamins and carbohydrates (Fig. 5). We defined six categories of general functions knowing that many proteins have multiple functions and appear in various groups (supplemental Data S3, sheet #2). Three processes seem to be emphasized when comparing the first approach on the ten major proteins (Table I) and the second performed on the whole list of proteins (Fig. 5): cellular migration and adhesion, metabolism (and transport) of lipids and vitamins, and immunity (defense). Twenty-one genes are classified in “immune response” (Fig. 5). This category includes effectors/antibacterial molecules such as TF and LYZ, but also immunoglobulins (CAO79236.1 and LOC107051274), mucins (MUC5ACL, MUC5B), protein TENP (BPIFB2), ovoinhibitor (SPINK5) or ovalbumin-related protein X (OVALX). It is noteworthy that ten proteins could not be assigned to any known function, including two of the major proteins identified in this work: OVAL and SPINK7 (Table I). Interestingly, among these 10 proteins, seven are potent protease inhibitors whose biological functions in the egg are still uncertain.

Fig. 5.

Functions of the proteins identified in the chicken amniotic fluid at the 11^th day of incubation (ED11). Using Gene Ontology terms, as described under “Experimental Procedures”, 80 genes could be assigned to five major functions whereas the biological function of 10 proteins are still not documented in this database (this latter group includes ovalbumin (OVAL) and ovomucoid (SPINK7), two proteins of high abundance (Table I, supplemental Data S3).

Comparison with the Human AF Proteome

The chicken AF proteome at ED11 was compared with the 936 proteins identified in the human AF proteome (11), to shed light on chicken AF specificities. Out of the 936 proteins identified by Cho et al., (11), after removal of redundancies and considering the updated human genome assembly (GRCh38.p10, GCA_000001405.25, Dec 2013), we ended with 842 proteins, plus 81 proteins retrieved from the other studies (11) (supplemental Data S3, sheet #4). This data integration led to 923 proteins that are identified so far in the human AF. When considering gene symbols, 48 proteins were found to be common to both chicken and human AF (Fig. 6). We confirmed their orthologous relationship using the Ensembl database. Although they do not share the same gene symbol, LOC107051274 and CAO79236.1 (immunoglobulins) can be likened to human immunoglobulins G, which are omnipresent in human AF (11).

Fig. 6.

Venn diagrams representing the specific and overlapping proteins identified in the human (during the whole gestation) and chicken (11^th day of incubation) amniotic fluid (AF). Human AF data were obtained from the work of Cho et al., (11), which combines results from nine publications (supplemental Data S3, sheet #3).

Assessment of the Antibacterial Potential of Chicken AF and Comparison to Egg White

To further assess the biological significance of AF as a source of antibacterial proteins, we adapted a method from Bhunia et al,. (28) that allows the direct detection of antibacterial activities after SDS-PAGE. Briefly, AF proteins and peptides were separated by SDS-PAGE under nonheating and nonreducing conditions. After removing SDS from the gel, a nutrient-poor agar containing bacteria was poured on the SDS-PAGE gel and left on the bench for 3 hours to allow for protein diffusion. Thereafter, a nutrient-rich agar was added to favor bacterial growth. Regions where bacteria could not grow, because of the presence of antibacterial proteins and peptides, were detected as translucent areas. Antibacterial activity was tested against Listeria monocytogenes, (L.m.,; Gram-positive) and Salmonella enterica, Enteritidis (S.,E.; Gram-negative), two bacteria against which egg proteins have already demonstrated their activity under similar experimental conditions (27). Because AF protein composition at ED11 is thoroughly modified from ED12 onwards, due to the presence of egg white proteins masking the intrinsic protein composition of AF, we explored the antibacterial potential of AF at both ED11 and ED16 (Fig. 7A, and 7B,). Assays were performed on both pools of total AF, but also on fractions enriched in antibacterial molecules obtained after heparin-affinity chromatography (HBP, heparin-binding proteins), as previously described (27). At ED11, results show only weak antibacterial activity against Gram-positive L.m., for crude AF, and for flow-through and eluted (HBP) fractions collected from heparin-chromatography (Fig. 7A,). Overall, the antibacterial activity against both L.m., and S.,E. seems to increase with the egg white transfer, with new inhibition zones appearing for both bacterial strains at ED16, particularly for the eluted fraction: one to five anti-L.m,. zones from ED11 to ED16, versus, only one anti-S.,E. zone at ED16 (Fig. 7A, and 7B,).

Fig. 7.

Antibacterial activity of the chicken amniotic fluid (AF) before and after the egg white transfer into the amniotic sac at ED11 (A,) and ED16 (B,). AF samples were analyzed by Coomassie Brilliant Blue staining (Cb) and their antibacterial activities were assessed against Listeria monocytogenes, (L.m.,) and Salmonella enterica, Enteritidis (S.,E.). Flow-through and eluted fractions were obtained after affinity chromatography on heparin-Sepharose as described under Experimental Procedures. Black arrows indicate inhibition zones corresponding to an impaired bacterial growth (due to bacteriostatic or bactericidal activities of AF proteins or peptides). Gray letters A to H correspond to bands that were cut from the Cb gel, prior analysis by mass spectrometry for protein identification (Table II).

The HBP fraction of egg white has been previously reported to contain antibacterial proteins and activities (27, 30). To explore the regulation of egg white antibacterial activities all along incubation, we performed antibacterial assays on egg white HBP, sampled and purified at ED0, 8 and 14 (Fig. 8). These stages of development were chosen knowing that egg white transfer into the AF is nearly complete at ED16 (almost no residual egg white at this stage). The antibacterial activity of the egg white HBP against both L.m., and S.,E. changes during embryonic development. Indeed, the strong activity detected at ED0 at 19 kDa (Fig. 8, ED0) against L.m.,, progressively disappears to the benefit of anti-L.m., activities in the higher molecular masses (>250 kDa) or lower molecular masses (about 14 kDa) (Fig. 8, ED8 and 14). In addition, three bands exhibiting anti-S.,E. activities at ED0 (19 kDa and 14 kDa), were barely detectable at later stages (ED8 and ED14, Fig. 8). Antibacterial activity of egg white HBP at ED14 is similar to the antibacterial activities of the HBP in the AF at ED16: one inhibition zone at the high mass range (>250 kDa), and three bands at the low mass range (<15 kDa) for L.m,., as opposed to one inhibition zone at 15 kDa for S.,E. An anti-L.m., zone between 25 and 37 kDa for the HBP at ED16 solely appears in AF, which might be associated with specific activation or expression of antibacterial proteins in this fluid at this stage.

Fig. 8.

Antibacterial activity of the heparin-bound proteins of egg white at ED0 (start of incubation), at ED8 (8^th day of incubation) and ED14 (14^th day of incubation). Proteins were analyzed by Coomassie Brilliant Blue staining (Cb) and their antibacterial activities were assessed against Listeria monocytogenes, (L.m.,) and Salmonella enterica, Enteritidis (S.,E.), as described under Experimental Procedures. Black arrows indicate inhibition zones corresponding to an impaired bacterial growth (due to bacteriostatic or bactericidal activities of egg white proteins or peptides).

Identification of Candidate Proteins Responsible for Antibacterial Activity in AF at ED11 and ED16, by Mass Spectrometry

In parallel to gels prepared for antibacterial assays, an SDS-PAGE performed under the same conditions was stained with Coomassie Brilliant Blue. The superimposition of both the stained SDS-PAGE gel and the gel exhibiting antibacterial activities allowed us to select SDS-PAGE bands (labeled A to H, Fig. 7A, and 7B,), which were susceptible to contain antibacterial proteins or peptides. These bands were analyzed by mass spectrometry at ED11 (band A, Fig. 7A,) and at ED16 (bands B to H, Fig. 7B,), which led to the identification of 29 proteins (Table II) (supplemental Data S4). Among the most likely candidates possessing antibacterial activity, we could only find LYZ for AF at ED11. For the sample at ED16, the most probable molecules are LYZ, but also AvBD11, vitelline membrane outer layer protein 1 (VMO1), avidin (AVD), beta-microseminoprotein-like (LOC101750704 G. gallus/,LOC104403769 N. notabilis,), OVALX, TF, BPIFB2, ovocleidin-17 (OC-17) and midkine (MDK) (27, 30, 32–34). The antibacterial activity of SPINK5 was previously demonstrated against Bacillus, but not against the two strains tested in the present study (35).

Table II. Potential antibacterial candidates identified in the heparin-binding fraction of the chicken amniotic fluid at ED11 and ED16. Proteins contained in the SDS-PAGE bands corresponding to bacterial inhibition zones (Listeria monocytogenes and Salmonella Enteritidis, Fig. 7) were analyzed using mass spectrometry. Bands where spectral counts were the highest are indicated in bold.

Identified Proteins (29)	Accession Number	Gene symbol	Molecular Mass (kDa)	Corresponding SDS-PAGE bands (Fig. 7)
Mutant cysteine-rich FGF receptor [G. gallus,]	AAB39211.1	GLG1	122	G
Avidin [G. gallus,]	CAC34569.1	AVD	17	B, C, D, E, F, G, H
OvoglobulinG2 type BB [G. gallus,]	BAM13273.1	BPIFB2	47	B, C, D, E, F
Ovomacroglobulin, ovostatin [G. gallus,]	CAA55385.1	OVST	164	C
Sodium/potassium-transporting ATPase subunit alpha-1, precursor [G. gallus,]	NP_990852.1	ATP1A1	112	A
Peptidyl-prolyl cis-trans isomerase B precursor [G. gallus,]	NP_990792.1	PPIB	19	E
Mucin-6, partial [G. gallus,]	XP_015142236.1	MUC6	98	B, C
Metalloproteinase inhibitor 2 precursor [G. gallus,]	NP_989629.1	TIMP2	22	E
Metalloproteinase inhibitor 3 precursor [G. gallus,]	NP_990818.1	TIMP3	20	E
Gallinacin-11 precursor [G. gallus,]	NP_001001779.1	AvBD11	12	C
Ovoinhibitor precursor [G. gallus,]	NP_001025783.2	SPINK5	52	B, C
Ovalbumin-related protein Y [G. gallus,]	NP_001026172.1	OVALY	44	B, C, D
Elongation factor 1-alpha 1 [G. gallus,]	NP_001308445.1	EEF1A1	50	G
Vitelline membrane outer layer protein 1 precursor [G. gallus,]	NP_001161233.1	VMO1	20	C, E, F, G
Astacin-like metalloendopeptidase precursor [G. gallus,]	NP_001292019.1	ASTL	46	D
Lysozyme C [G. gallus,]	P00698.1	LYZ	16	A, B, C, D, E, F, G, H
Ovalbumin [G. gallus,]	P01012.2	OVAL	43	A, B, C, D, E, F, G, H
Ovotransferrin [G. gallus,]	P02789.2	TF	78	B, C
Ovocleidin-17 [G. gallus,]	Q9PRS8.2	OC-17	15	G
Ovomucin [G. gallus,]	XP_003641415.1	MUC5B	234	B, C
Tumor necrosis factor receptor superfamily member 6B [G. gallus,]	XP_004947191.1	TNFRSF6B	38	D
Beta-actin [G. gallus,]	CAA25004.1	ACTB	46	A, G
Beta-microseminoprotein-like [N. notabilis],	XP_010010315.1,	LOC104403769,	15,	C, E, F, G,
Ovalbumin-related protein X [G. gallus,]	XP_015137660.1	OVALX	45	B, C, D, E, F, G
Clusterin [G. gallus,]	XP_015140573.1	CLU	62	C
Midkine [G. gallus,]	XP_015142525.1	MDK	16	G
Alpha-2-macroglobulin-like protein 1 [G. gallus,]	XP_015148230.1	A2ML1	160	B, C
ATP synthase subunit alpha, mitochondrial [G. gallus,]	NP_989617.1	ATP5A1Z	56	A
Deleted in malignant brain tumors 1 protein-like [G. gallus,]	XP_015156102.1	DMBT1L3	100	B, C

In italics,: LOC104103769 from N. Notabilis, has a G. Gallus, homologue (LOC101750704), which has been erroneously removed from Pubmed database following the release of Gallus_gallus-5.0 assembly in 2015. This chicken homolog has been unambiguously identified in egg white (25) and shares 88.7% sequence identity with LOC104403769 from N. notabilis,.

Phylogenetic Analysis

To further explore specificities of the chicken model, we analyzed the evolutionary history of the 91 proteins identified in chicken AF at ED11 (Fig. 9), combined to the 18 new genes identified from the antibacterial assays at ED16 (Table II). Among the 109 proteins analyzed, the corresponding genes of 100 of them were found in the Ensembl database, but the evolution of only 90 of them could be successfully investigated. The 19 genes without an Ensembl accession number may be in the nonsequenced or nonannotated region of the chicken genome or lack a related phylogenetic tree (AFP; APOA2; APOC3; ASTL; COL1A1; DCHS1; DMBT1L3; DSG4; LOC104403769/LOC101750704; LOC107051274; LOC107323914; MUC6; OC-17; OVSTL; PSAP; SLC3A2; SPINT4; immunoglobulin heavy chain variable region CAO79236.1 (no GeneID); transferrin ADK35120.1 (no GeneID)). Studying the moment of appearance in the tree of life of each gene coding for these proteins, revealed that most of them (88%) appeared before 371 MYA (tetrapods) with a peak between 937 MYA and 722 MYA (bilateria) for almost 50% of them (Fig. 9A,). Orthologues of most of these genes (80 genes) are present in primates and other mammals. Nine proteins lacking orthologues in mammals were identified. Four of these genes, AvDB11, OVAL, OVALY and OVALX are bird-specific and are also found in duck or turkey, but not in lizard or turtle (31, 36) (Fig. 9B,). These genes appeared 104 MYA after divergence between birds and reptiles. APOV1, present in bird and reptile genomes, emerged 276 MYA. AVD and RBP appeared in vertebrates and chordates, respectively (21, 37, 38). These genes are present in reptile and bird genomes but also in fishes or frogs, suggesting that they have likely been lost in the mammalian branch. Concerning RBP, the death of this gene in the mammalian genome could not be confirmed because we could not find any pseudogenes. Genes encoding VTG1 and VTG2, which appeared in bilateria, are found in all egg-laying species including platypus (VTG), which is the only oviparous mammal. Some authors have shown that these VTG1- and VTG2-encoding genes progressively lost their functions and became pseudogenes during mammalian evolution (38) concomitantly with the appearance of lactation and placentation. Several authors have described the process of pseudogenization of VTG genes during mammalian evolution (21, 38). OC-116 gene has been described as orthologous to the MEPE gene in other species, although OC-116 has likely acquired some tissue specificities (eggshell biomineralization) resulting from divergent and adaptive evolution (21). Regarding SPINK7 (ovomucoid) and SPINK5 (ovoinhibitor) genes (respectively ENSGALG00000003512 and ENSGALG00000031496), which appeared between 296 MYA and 276 MYA, the story is quite tricky as the data available in Ensembl were not sufficient to better define their evolutionary fate in various amniotes (Fig. 9). Both chicken proteins share 50.24% sequence identity. In Gallus gallus,, they are located right next to each other on chromosome 13 within a 33-kb region, close to other SPINK genes, which have all probably arisen by local duplication from a common ancestoral gene. All these proteins are characterized by one to 14 Kazal-like domains, which are known as evolutionary conserved domains, commonly occurring in tandem arrays, and usually associated with inhibitory activities against serine proteases. This SPINK gene family seems to have evolved very quickly, and unfortunately, synteny analysis of these genes does not help in strengthening the corresponding phylogenetic tree, because all these genes are located within the same locus. Nevertheless, both chicken genes seem to have specifically undergone duplication in Sauropsida, before the speciation of birds and reptiles. SPINK7 (ovomucoid) and SPINK5 (ovoinhibitor) genes encode proteins of 210 amino acids and 472 amino acids, respectively, the latter being characterized by a C-terminal extension of 252 amino acids containing four additional Kazal-like domains. A low percentage of sequence identity of chicken SPINK5 is also observed with mammalian SPINK5 proteins (less than 30%), associated with differences in the length of the protein sequence and the number of corresponding Kazal-like domains (seven for chicken SPINK5/ENSGALG00000031496 and 14 for human SPINK5). Ultimately, the phylogenetic tree strongly suggests that both genes have been lost in several mammals, in primates, hinting at pseudogenization in mammals, in parallel to speciation after duplication in birds.

Fig. 9.

Evolutionary scenario of the chicken amniotic fluid (AF) proteins. A,, Evolutionary frieze of 76 of the 91 proteins identified in the AF on the 11th day of incubation (ED11), and of 14 of the 29 potential antibacterial candidates identified in the AF at ED16 (in gray). The 10 genes underlined in yellow are specific to birds and/or reptiles. B,, Life tree of the bird-and/or reptile-specific 10 genes (underlined in yellow) that display bird and/or reptile specificities. The presence of a gene is depicted by a solid symbol, while the loss of a gene is depicted by a hollow symbol and marked with “Ψ.” The divergent time of lineages refer to Tian et al., (21) and phylogenetic analysis using Ensembl (http://www.ensembl.org/index.html). For SPINK5, MEPE and SPINK7 genes, which are marked with a question mark, we cannot confirm the distribution of these genes within amniotes, based on the data currently available in Ensembl databases and despite the synteny analysis of these genes. This issue is discussed in the Results section.

DISCUSSION

In amniotes, AF plays a crucial role in maintaining a stable and protective milieu. In humans, the exhaustive characterization of its biochemical components (including proteins) at various stages of gestation in normal and pathological situations constitutes an extremely valuable approach to help in the diagnosis of human fetal disorders and infections. In avian species, there is evidence that AF also participates in homeostasis around the embryo (18) and in its protection against physical constraints (19). Nevertheless, little is known about the physiological role of AF proteins in birds, and there is to date little proteomic data available on this fluid for avian species. Yet, we suspect that the development of the embryo in an egg, with no mother-connected tissues, requires some inherent specificities and structural particularities that may influence the AF protein profile and composition in birds.

In the present work, we identified 91 nonredundant proteins with high confidence, in the chicken AF at ED11, before egg white transfer, which is higher than the 47 proteins identified in a preliminary study (18). This number is however far less than the number of proteins identified in two surveys performed on human AF (219 (9) and 923 (11)). This difference can be partly explained by the various exchanges with maternal tissues that take place during gestation in humans, which result in some specific AF protein profiles, and the fact that in humans, metabolic wastes originating from the embryo are indeed recovered in the AF, which is not the case in the chicken model (6).

All combined results indicate that more than half of the proteins identified in the chicken AF at ED11 have orthologues that have been identified as components of the human AF, which highlights that the overall protein composition of the AF before egg white transfer exhibits high similarities with the human AF. This is strengthened by the observation that, five of the top-ten high abundant proteins identified in the chicken AF at ED11 (Table I - ALB, TF, AFP, GC, and APOA1) are also listed in the 15 high abundant proteins recovered in human AF (9, 11). It is notable that the TF identified in the chicken AF corresponds to ovotransferrin, which shares 51.76% and 51.73% sequence identity with human serotransferrin (P02787) and human lactotransferrin (P02788), respectively. This moderate sequence identity between ovotransferrin and its human homologs, together with some specific structural features, such as glycosylation and number of disulfide bridges (39), suggest similar functions with regard to their iron-binding capacity and storage, but also possibly diverging functions. As an example, both lactotransferrin and ovotransferrin, but not serotransferrin, are annotated as antibacterial proteins (40, 41). Interestingly, two of the top-ten abundant proteins have no assigned functions: OVAL and SPINK7. These two proteins are major proteins of egg white (42) and appear belatedly in the timeline of evolution, with OVAL and very likely SPINK7, being specific to birds (Fig. 9).

Because of these similitudes in protein composition, most functions attributed to the most abundant proteins composing the human AF resemble that of the chicken AF: (1) metabolism and transport of vitamins, lipids and hormones, (2) immune response, and (3) hemostasis and homeostasis. The analysis of the list of chicken AF proteins using automatic tools dedicated to Gene Ontology annotation, emphasized functions associated with many aspects of developmental biology such as morphogenesis and organogenesis, which include cell proliferation, adhesion and migration processes (Fig. 5). These functions were also significantly highlighted in the analysis of human AF proteins (9, 11). However, the biological significance of these functions in the chicken and human AF are likely to be overestimated since most proteins annotated with this function are very-low-abundance proteins and may reflect some embryonic or extraembryonic tissues desquamation (Supplemental data S2 and S3, sheet #2).

Overall, it is noteworthy that of the ten most abundant proteins in the chicken AF, four of them are major egg white proteins (OVAL, SPINK7, LYZ, TF) (42) whereas ALB, APOA1, GC, TTR, APOC3 are egg yolk abundant proteins (43). Gene expression of OVAL, SPINK7, LYZ and TF, which together account for about 80% of the egg white proteins, is under hormonal control in the oviduct and will not be expressed until the onset of sexual maturity in hens (44). OVAL, SPINK7 and TF are thus likely to flow from egg white into the amniotic sac, rather than being expressed by embryonic tissues. As for LYZ, the ongoing hypothesis is that it also originates from egg white although it is also expressed by a wide range of tissues (45). On the other hand, APOA1, ALB, GC, TTR and APOC3 represent high-abundance proteins of egg yolk (43), which also suggests that they are somehow transferred to the chicken AF from the egg yolk, or from the yolk sac where they are highly expressed (4). The molecular mechanisms by which egg white and egg yolk proteins enter the AF during the first half of incubation are still not completely understood (16, 18). In contrast, the embryonic origin of AFP is unequivocal since it is a fetus-specific protein present in a range of (extra)embryonic tissues, highly expressed in the yolk sac and in the embryonic liver (46). Although all these proteins are the most abundant proteins identified in the chicken AF at ED11, their protein concentration in the AF at this stage (0.01 mg/ml) is not comparable to those recovered in the egg yolk (>10 mg/ml, 22% of dry matter (47)) and egg white (380 mg/ml (30)). The low protein concentration of chicken AF at ED11 suggests that the presence of proteins in this fluid may reflect some passive transfer from these egg compartments toward AF, rather than an active mechanism. This concentration is also lower as compared with that of human AF: it is about 4 mg/ml at 15 weeks of gestation (10), it reaches its maximum (6 mg/ml) at 22–27 weeks of gestation and starts to decrease to about 2 mg/ml at 37–40 weeks of gestation (48). It also highlights that chicken AF cannot meet embryo's requirements in energy up to ED11, suggesting that during the first half of incubation, the egg yolk is the major source of energy/lipoproteins for the embryo. Some proteolytic activities (18) have been however detected in AF at ED11, but, because of the low abundance of the corresponding proteins, the biological significance of these activities in AF remains questionable as compared with the egg yolk, where hydrolytic enzymes are concentrated (43, 49). The same observation can be made for the faint antibacterial activities detected at ED11 in chicken AF, compared with egg white, which concentrates high amounts of antibacterial proteins such as active lysozyme (30, 42).

These considerations along with the detection of antibacterial activities, corroborates the role of chicken AF in protecting the embryo against a range of potential physical, physicochemical and antibacterial threats. Moreover, it also emphasizes some particularities such as the presence in this fluid of bird-specific proteins involved in embryonic nutrition, eggshell biomineralization and egg defense (OVAL, OVALY, OVALX, OC-116, AvBD11) and likely SPINK7/ovomucoid and SPINK5/ovoinhibitor, recently renamed SPIK7 and SPIK5 in the NCBI Gene database (May 2017) and whose functions are still speculative. We also identified some oviparous-specific proteins (VTG1, VTG2, AVD, APOV1) (Fig. 9B,), that appear concomitantly with the appearance of the egg-laying type of reproduction. Additionally, this analysis revealed the presence of 43 proteins that have been uniquely found in the chicken AF as compared with human AF (Fig. 6). Nevertheless, the high number of proteins reported for human AF as compared with chicken AF, merits further studies. A comprehensive study of the proteome profiling of the allantoic fluid would probably answer some of these questions, as we might find allantoic fluid-specific proteins whose orthologues have been identified in human AF. Future studies of this specific fluid would confirm or not whether proteins from both chicken amniotic and allantoic fluids together encompass most proteins identified in the human AF.

To conclude, the story of the chicken AF remains quite simple up to ED11, with a protein composition that is comparable to that of human AF or at least with similar general functions. However, the protein composition of AF is deeply revised after egg white transfer, which occurs between ED11 and ED12 (18). The AF protein profile then completely merges with the egg white protein profile (Fig. 3A,) (16, 18, 50), with ten major proteins consisting of OVAL, LYZ, TF, OVALY, SPINK7, VMO1, SPINK5, AVD, OVALX, and CST3 (42). Despite the presence of AF hydrolytic enzymes (18), no major proteolytic degradation of egg white proteins was detected after its inflow into the amniotic cavity, up to ED19. These data suggest that abundant egg white antiproteases, namely SPINK7, SPINK5, cystatin (CST3) and ovostatin (OVST), remain active during the entire duration of incubation. It is rather interesting to note that the inflow of egg white reinforces some proteins that were already present as major proteins in ED11-AF (OVAL, LYZ, TF and SPINK7, Table I). More than half of these major egg white proteins are related to innate defense (LYZ, TF, VMO1, SPINK5, AVD, OVALX) while three of them are protease inhibitors (SPINK5, SPINK7, CST3). Therefore, we suspected that chicken AF enriched in egg white proteins from ED12 onwards, acquires an increased antibacterial potential. An in-gel antibacterial assay was developed to better appreciate the relevance and effective activity of these so-called antibacterial proteins in the AF-egg white mixture. We showed that ED16-AF displays higher antibacterial potential against the two bacterial strains tested, namely Listeria monocytogenes, (Gram-positive strain), and Salmonella enterica, Enteritidis (Gram-negative strain) compared with ED11-AF. A total of 29 proteins were identified in the corresponding zones lacking bacterial growth (Table II), including LYZ, AvBD11, VMO1, AVD, LOC101750704 (BMSP), OVALX, TF, OC-17, BPIFB2 and MDK that were all previously reported to exhibit antibacterial activities (27, 32–34, 41, 51–56). These results demonstrate that the antibacterial potential of egg white proteins is not impaired by the incubation temperature, nor by proteolytic degradation (thanks to egg white antiproteases), and that its antimicrobial potential remains effective after transfer into the AF. However, subtle differences are detected in AF at ED16 after completion of egg white transfer, but also in remnant egg white at ED8–14. Indeed, additional antibacterial zones were visualized on anti-Listeria monocytogenes, assays performed with ED16-AF and ED8-14-EW (Fig. 7 and 8, respectively). Unexpectedly, they correspond to proteins or protein complexes/associations of very high molecular mass (>250 kDa). These high molecular mass complexes were identified as a mixture of egg white proteins: clusterin (CLU, 62 kDa), SPINK5 (52 kDa), OVALX (44 kDa), AVD (tetramer of 17 kDa units), OVAL (43 kDa), TF (78 kDa), ovomucin (234 kDa), LYZ (14 kDa), BPIFB2 (47 kDa), etc. The progressive appearance of these complexes suggests some specific regulation occurring during incubation. Indeed, the re-distribution of the egg white water toward extraembryonic sacs during the first half of the development, concentrates egg white proteins, thus promoting protein-protein interactions between antibacterial proteins/peptides of low molecular masses (LYZ, LOC101750704 (BMSP), AvBD11, etc.) and proteins of higher molecular masses (ovomucin). These protein complexes do not seem to affect the antibacterial protein/peptide activities, which are still effective even before oral absorption by the embryo (Fig. 8), thus forming another barrier against bacteria around the embryo's body. As the embryo moves and develops in the AF until hatching, it can be hypothesized that this mixture of antimicrobial molecules and mucins may also constitute a protective biofilm deposited on the embryonic feathers and skin to protect the embryo and the newborn chick. This proposal is partly supported by the presence of lysozyme and defensin-like molecules in human vernix caseoa, (57, 58), a protective lipid-rich substance that covers the skin of the human fetus during the last trimester of gestation and the newborn. Whether this enhanced antimicrobial potential of AF prior to its swallowing has some relevant biological significance with respect to the protection of the embryonic digestive tract is also of interest. Indeed, the fate of the AF-egg white protein mixture that is ingested by the embryo is still poorly understood. These proteins remain very stable in egg white during incubation but also in the AF (16, 18) and all along the length of the digestive tract of the embryo, up to ED19 (50). Similarly, some egg white typical profiles can be visualized in the yolk sac contents at ED20 just before emergence of the chick, with OVAL, LYZ and TF as major proteins (13, 50). These data together with the fact that OVAL was recovered as a native form in the central nervous system and in other embryonic organs (16), suggest that this protein and possibly related proteins OVALX and OVALY may have other functions than nutrition (31). From these data, we infer that intrinsic egg yolk proteases exhibit limited proteolytic activities within the egg white contents at this stage. This can be partly explained by the gradual increase of the yolk pH, which impacts the activity of aspartic proteases (59), such as chicken cathepsin D, a key enzyme in yolk processing (49). It seems that everything converges to protect egg white-AF protein content from uncontrolled proteolytic and thermal degradation (presence of numerous proteases inhibitors together with the progressive increase in thermostable S-ovalbumin during incubation (60, 61)), to be utilized by the embryo and/or the chick at a very precise moment of its development/growth.

Besides its major interest for comparative biologic approaches and for deciphering the functions and specificities of the avian extraembryonic compartments, these results (protein profile and composition) constitute a reference starting point for analyses of pathological conditions due to infection or impaired development occurring during the chicken embryonic development. As examples, in humans, differences in concentration of AFP, defensins, vitamin-binding proteins, APOA1, and TTR in AF (also identified in chicken AF) have been associated with many metabolic and developmental disorders such as trisomy, developmental delays, preterm-labor, inflammation or infections (62–64). These data combined with other high throughput methods such as metabolomics or epigenetics, and general egg quality traits, may constitute useful tools for poultry production. Indeed, these approaches may help to further investigate the impact of chicken lines, housing systems and nutrition of hens (65, 66) or of more subtle egg manipulations including conditions of egg storage prior to the incubation (67) or thermal manipulation of eggs (68), on many aspects of chicken embryo development and health. We believe that the identification of a set of molecular markers for abnormal embryonic development in chicken extraembryonic fluids will contribute to the development of tools to improve poultry managements and, to increase the robustness of chicks and chickens to help them face contrasted and changing environments.

DATA AVAILABILITY

Data are available via, the PRIDE partner repository with the data set identifiers PXD008046 and 10.6019/PXD008046.