Up-regulated expression of white isthmus genes

The Del-Mar 14 K Chicken Integrated Systems microarray (NCBI GEO Platform # GPL1731) [24] was used for the analysis of over-expressed genes in three segments of the hen oviduct. Our original reports described uterine-specific gene expression relative to other oviduct parts [21], and the validation by qRT-PCR analysis of 16 differentially expressed transcripts detected by microarray analysis of the hen’s oviduct. In the present analysis, we used comparisons of gene expression in the white isthmus (WI) with two other segments (magnum (Ma) or uterus (Ut)) of the hen’s oviduct to predict genes encoding proteins involved in formation of the eggshell membranes.

A general assessment was performed to identify all up-regulated transcripts. A total of 1514 transcripts were over-expressed in the white isthmus compared to the magnum while 422 transcripts were over-expressed in the white isthmus compared to the uterus. However, only 135 clones hybridizing to over-expressed transcripts were common to the two comparisons (Fig. 1); according to our hypothesis, these over-expressed genes are related to white isthmus-specific functions (Additional file 1).

The 135 clones that were common to WI/Ma and WI/Ut comparisons were further assessed. Nine of them could not be further analyzed, since either the clone ID was no longer in the current nucleotide GenBank or in the University of Delaware chick EST database, or Blast searching did not reveal any hit. After removing examples of redundant clones, the over-expressed transcripts corresponding to 107 unique Gene IDs were retained and correspond to over-expressed genes in the white isthmus that possess an NCBI non-redundant Gallus gallus gene ID.

The eggshell membrane fibres are extracellular structures; electron microscopy reveals that they are fabricated in the white isthmus tubular glands and then secreted into the lumen of the oviduct where assembly into the mesh-like membrane structure occurs [25]. The apertures of the tubular glands are more numerous in the white isthmus than in any other oviduct region [26]. Secreted proteins and/or enzymes are candidates to participate structurally or enzymatically in their assembly. The SignalP 4.1 and SecretomeP 2.0 servers were used to interrogate the protein sequences of highly over-expressed genes (Table 3) to search for predictions of protein secretion by classical signal peptide – mediated mechanisms or the unconventional pathway which mediates leaderless secretory proteins. Those with predicted signal peptides included COL10A1 and FCGBP. Similar to spore coat protein SP75, also known as cysteine rich eggshell membrane protein (CREMP), is a potential structural protein that possessed sequence determinants for unconventional secretion. However, it is unclear whether the 5’ sequence of this cDNA (coding for the N-terminus) has been identified since its gene has not yet been cloned. Among the 14 genes that were highly over-expressed, only three have previously been confirmed by proteomics studies. COL10A1, CREMP and hemopexin (HPX) have previously been extracted from eggshell membranes derived from both fertilized and unfertilized eggs [6, 7, 17, 18].

Collagens and related proteins

The eggshell membrane fibres are not exclusively made of collagens as clearly shown by the amino-acid profile of the eggshell membranes, although they contain at least 10 % collagens (types I, V and X) [1, 15, 18, 27–31]. In our study, the microarray contains two independent clones that are specific for different regions of COL10A1 and over-expression levels in the white isthmus were extremely high when compared to both the magnum and the uterus (Table 3). Collagen X is a short chain collagen molecule existing as a homotrimer of the α-1 (X) chains, which contributes to structural integrity. In situ hybridization and immunohistochemistry have previously revealed that collagen X is mainly expressed in the tubular gland cells of the white isthmus segment of the oviduct [32]. Proteomics studies have detected peptides from collagen X in eggshell membrane extracts from both fertilized and unfertilized eggs [7, 17, 18], within the shell matrix corresponding to the mammillary cones [6, 7], and in the uterine fluid during the initial stage of mineralization [33].

Type I and V collagens are known to be less abundant constituents of the eggshell membranes and represent 4 mg/g of whole eggshell membrane, or about 0.6 % of the membrane protein [30, 31]. In the present study, over-expression of COL1A1, COL1A2 and COL3A1 were detected in the WI/Ma contrast only (Table 4). No over-expression of collagens other than COL10A1 was detected in the WI/Ut contrast, suggesting that these genes were expressed at equivalent levels in both white isthmus and uterine segments of the oviduct. This observation is supported by recent proteomics analysis where collagen X (α-1) and collagen II (α-1) were detected in eggshells collected at the early stage of calcification [6]. Proteomics analyses of chicken eggshell membranes have also confirmed the presence of collagens type I (α-2), IV (α-3), V (α-2), X (α-1) and XXIV (α-1) as constituents of this fibrous biomaterial (Table 4) [7, 24, 25, 33]. Additionally, peptides derived from collagens type I (α-1 and α-2), II (α-1), III (α-1), V (α -2), VI (α-1 and α-2), X (α-1), XVII (α-1) and XVIII (α-1) were detected in the matrix of chicken eggshell [6, 34, 35]. In summary, a significant structural fibrillar protein, COL10A1, was found to be the most highly up-regulated gene in the white isthmus, while a number of other collagen genes are also over-expressed in this oviductal segment.

In view of our demonstration that a number of collagens were up-regulated in the white isthmus, we searched for over-expression of genes encoding collagen processing enzymes. Lysyl oxidase is a copper-sensitive enzyme that is associated with the formation of collagen cross-links. Its activity is located in the copper-rich region of the white isthmus of the hen oviduct [36], and can be detected in the eggshell membranes where it participates in the cross-linking of eggshell membrane proteins [37]. A deficiency in hen dietary copper elicited a disruption in eggshell membrane formation [15]. In chicken, there are 5 lysyl oxidase genes (lysyl oxidase: LOX, and lysyl oxidase homologs 1 to 4: LOXL1, LOXL2, LOXL3, LOXL4); the Del Mar 14 K Chicken Integrated Systems microarray contains cDNAs for detection of LOX and LOXL1, LOXL2 and LOXL3.

Our results show that LOXL3 was over-expressed 1.27-fold in the WI/Ma group, but was not over-expressed in the WI/Ut comparison (Table 4). LOXL3 was previously identified as a constituent of eggshell membranes [17, 18, 31] and the shell matrix by proteomics analyses [38]. We also identified LOXL2 as highly over-expressed, especially in the WI/Ut contrast (18.64-fold) (Table 4). Lysyl oxidase homolog 2 was one of the most abundant proteins present at the earliest stage of shell calcification, suggesting a predominant role amongst the lysyl oxidases [6]. Although not over-expressed, LOXL1 was expressed at extremely high (saturating) levels in all three tissues (not shown). Our results indicate that LOXL1, LOXL2 and LOXL3 are likely to participate in the cross-linking of collagen and other fibrillar proteins during formation of the eggshell membranes. While LOXL4 was not identified in this study, recent proteomics analysis confirms its presence in the eggshell membranes [19].

Prolyl 4-hydroxylase subunit α-2 is a collagen-associated protein encoded by the P4HA2 gene that is over-expressed in the white isthmus (1.21-fold and 1.64-fold in the WI/Ut and WI/Ma contrasts, respectively). It is a component of prolyl 4-hydroxylase, a key enzyme required for collagen synthesis which catalyzes the formation of 4-hydroxyproline that is essential to the proper three-dimensional folding of newly synthesized procollagen chains [39, 40]. It is noteworthy that prolyl 4-hydroxylase beta polypeptide (P4HB) was also identified as over-expressed in the WI/Ma contrast. Peptidyl-prolylcis-trans isomerase C (PPIC) is also a collagen-associated protein which catalyzes the cis-trans isomerization of proline imidic peptide bonds. For example, in vitro refolding of denatured type III collagen is rate-limited by cis-trans isomerization of the peptide bond by peptidyl-prolyl cis-trans isomerase [41]. PPIC was found to be over-expressed 1.28 and 1.51-fold in the WI/Ut and WI/Ma contrasts, respectively (Table 4) while proteomics analyses confirm its presence in the shell membranes [17] as well as in the shell matrix during the initial phase of calcification [6]. Therefore, genes encoding collagens as well as collagen-processing enzymes, which are probably involved in the synthesis of additional fibrillar proteins, are upregulated in the white isthmus.

Fibrillin

A structural constituent identified by our analysis for the first time as a potential extracellular component of the eggshell membrane fibres is FBN1. Its over-expression level was 45.89-fold in the WI/Ut contrast and 2.53-fold in the WI/Ma contrast (Table 3); however, fibrillin-1 has only been identified in uterine fluid bathing weak eggs [38] and has yet to be identified in proteomics studies of the shell membranes and shell matrix. This suggests that fibrillin is highly specific to the white isthmus region of the oviduct and possibly is a highly insoluble constituent of the shell membrane fibers. Fibrillin is the major constitutive element of extracellular microfibrils and has widespread distribution in both elastic and non-elastic connective tissues throughout the body [42]. The molecular weight of chicken fibrillin-1 is predicted to be about 335 kDa, with 12 % cysteine residues; similar to human fibrillin (350 kDa; 14 % cysteine, of which one-third appears to be in the free reactive sulfhydryl form) [43]. About 75 % of fibrillin is composed of 46 EGF-like repeats, which are cysteine-rich domains originally found in human epidermal growth factor. Forty-three of these repeats satisfy the consensus for calcium binding, which is a known property of fibrillin [44]. Thus, fibrillin-1 is a likely candidate as a structural protein in the eggshell membranes forming microfibrils that contribute to the elasticity of eggshell membranes.

The cysteine-rich eggshell membrane protein (CREMP)

Another over-expressed gene, CREMP (EST accession BM439825.1; XP_001236415), was strongly expressed in a white isthmus– specific manner (58.08-fold in WI/Ma; 70.03-fold in WI/Ut) (Table 3). The corresponding protein fragment was recently identified in chicken eggshell membranes as potentially significant cysteine-containing constituent [17, 18]. This EST coding for a disulfide-rich sequence was originally annotated as similar to spore coat protein SP75, since a related gene was first identified in slime mold [45]. When calculated on a percent amino acid composition basis, fibrillin-1 (XP_413815.4) and CREMP (BM439825.1) are 12.8 and 13.8 % cysteine, respectively, in contrast to collagen X (α-1) (P08125) which is only 0.2 % cysteine. Thus, fibrillin-1 and CREMP are both in agreement with the literature values for whole eggshell membrane (i.e., 10.1 ± 0.7 % cysteine), as summarized by Kodali et al. [18]. These observations demonstrate that either (or both) CREMP and fibrillin-1 proteins could account for the relatively high cysteine content of eggshell membranes, in contrast to collagen X.

A number of genes encoding enzymes associated with disulfide mediated protein cross-linking were over-expressed in the white isthmus. Thioredoxin is encoded by TXN (Table 3) and participates in many biological processes as an antioxidant by facilitating the reduction of other proteins by cysteine thiol-disulfide exchange [46]; also, protein disulfide isomerase (PDIA5) catalyzes the formation of disulfide bonds (Table S1). Kodali et al. [18] showed that a reduced recombinant cysteine-rich eggshell membrane protein is an efficient thiol substrate of sulfhydryl oxidase (QSOX1) which is detected in the vitelline membrane, egg white, eggshell and hen oviduct tissue [18, 34, 35, 47–52]. Sulfhydryl oxidase 1 is a secreted disulfide catalyst recently shown to control extracellular matrix composition and function [53]. Close inspection of the individual fluorescence intensity data for the 3 cDNA clones annotated as QSOX1 revealed a high and equivalent expression in all three segments of the oviduct (Ma, WI and Ut; data not shown). Recent proteomics studies have confirmed the presence of sulfhydryl oxidase 1 in eggshell membranes, mammillary cones and uterine fluid during the initial phase of eggshell calcification [6, 7, 33]. Therefore, sulfhydryl oxidase is likely to participate in eggshell membrane formation by catalyzing the cross-linking of membrane fibres.

Several other genes are highly expressed in the white isthmus (Table 3). FCGBP interacts with the Fc portion of IgG and with mucin 2 (MUC2) [54]. MAX interactor 1 (MXI1) belongs to the family of proteins that function as potent antagonists of MYC oncoproteins. The antagonism relates to their ability to compete with MYC for the MAX protein and for consensus DNA binding sites [55]. Another highly over-expressed gene was chicken R3H domain and coiled containing 1 (R3HCC1); the function of the protein encoded by this conserved gene is unknown although it is predicted to bind ssDNA or ssRNA in a sequence-specific manner (Entrez Gene annotation). LIM and calponin homology 1 (LIMCH1) is also highly over-expressed in the white isthmus; it contains two domains: the calponin homology domain is actin-binding and the LIM domain is a small protein-protein interaction domain with two zinc fingers [56]. The D-3-phosphoglycerate dehydrogenase (PHGDH) enzyme is involved in the early steps of L-serine synthesis in animal cells [57]. Sorbitol dehydrogenase (SORD) is an enzyme of glucose metabolism which catalyzes the interconversion of polyols and their corresponding ketoses, and produces fructose [58]. HPX encodes a plasma glycoprotein that binds heme with high affinity. The encoded protein transports heme from the plasma to the liver and could be involved in protecting cells from oxidative stress [59]. It is noteworthy that several proteomics analyses have found hemopexin to be associated with the eggshell membranes, mammillary cones and uterine fluid during the early stages of shell mineralization [6, 7, 17, 18, 33]. Tudor domain containing 9 (TDRD9) is a probable ATP-binding RNA helicase which plays a central role during spermatogenesis by repressing transposable elements and preventing their mobilization, which is essential for germline integrity [60]. Adaptins (AP1G1) are important components of clathrin-coated vesicles transporting ligand-receptor complexes from the plasma membrane or trans-Golgi network to lysosomes [61]. TMEM63C is an uncharacterized gene which is conserved from zebrafish to human (Entrez Gene annotation).

Proteoglycans

Proteoglycans with keratan sulfate and dermatan sulfate epitopes have been implicated in the assembly and function of eggshell membranes and shell matrix, respectively [29]. Ovocleidin-116, expressed in uterine cells and found to be most abundant during the rapid growth phase of mineralization, is the core protein of ovoglycan, a variably modified dermatan sulfate proteoglycan detected in SDS-PAGE at 120 and 220 kDa [6, 62]. Previous studies have reported that a keratan sulfate epitope is located at the sites of calcium carbonate nucleation on the eggshell membranes during the initiation of calcification [27]. Secretion of this keratan sulfate epitope (termed “mammillan”) coincides with the formation of the mammillary cones [28, 29]. However, the core protein of this putative keratan sulfate proteoglycan has not yet been cloned or characterized. Therefore, we examined the list of over-expressed genes in the white isthmus carefully to detect possible proteoglycans, proteoglycan-processing enzymes or keratan sulfate annotations. The Del-Mar 14 K Chicken Integrated Systems Microarray contained 23 unique cDNA clones annotated as proteoglycans, none of which was over-expressed in either WI/Ut or WI/Ma contrasts. One over-expressed gene, ADAMTS1 (Table S1), encodes a member of the ADAMTS (a disintegrin and metalloproteinase with thrombospondin motif) protein family, with proteoglycan-degrading activity that has a role in the remodeling of extracellular matrix and is involved in mineralized nodule and bone formation [63]; ADAMTS-1 is also capable of binding to sulfated glycosaminoglycan chains [64].

Keratan sulfate is composed of repeating disaccharide units of Galb1-4GlcNAc (poly-N-acetyllactosamine) with sulfate groups at the C6 position of the Gal and GlcNAc residues [65]. The sulfation of oligosaccharides is carried out in the lumen of the Golgi apparatus by the transfer of a sulfate group from 3’-phosphoadenosine 5’-phosphosulfate (PAPS) to a precursor oligosaccharide. SULT1B1 (sulfotransferase family, cytosolic, 1B, member 1), which utilizes PAPS as sulfonate donor to catalyze the sulfate conjugation of many hormones, neurotransmitters, drugs and xenobiotic compounds, is expressed at very high (even saturating) levels in all three oviduct tissues (not shown). GNS, which codes for N-acetylglucosamine-6-sulfatase precursor, a calcium-dependent lysosomal enzyme associated with keratan breakdown, is specifically upregulated in the white isthmus (1.28-fold and 1.29-fold in the WI/Ut and WI/Ma contrasts, respectively).

Calcium-binding proteins

A number of over-expressed genes are related to cellular calcium handling or could mediate Ca²⁺ signaling in various cellular activities (Table 2). The RCN2 gene encodes reticulocalbin-2, a calcium-binding protein located in the lumen of the endoplasmic reticulum. RCN2 codes for a protein that contains six conserved regions with similarity to a high affinity Ca²⁺ -binding motif, the EF-hand [66]. CHP1 is a calcium-binding protein involved in different processes such as regulation of vesicular trafficking, plasma membrane Na⁺/H⁺ exchanger and gene transcription; it is required for the targeting and fusion of transcytotic vesicles [67]. TMBIM4 is an anti-apoptotic protein which can inhibit apoptosis induced by intrinsic and extrinsic apoptotic stimuli, and modulates both capacitative Ca²⁺ entry and inositol 1,4,5-trisphosphate (IP₃)-mediated Ca²⁺ release [68].

Antimicrobial protection

One over-expressed gene in the white isthmus encodes an antimicrobial protein that could contribute to egg innate defense mechanisms. The antimicrobial peptide, gallinacin-10 [also called avian beta-defensin 10 (AvBD-10)] is encoded by the GAL10 gene [69]. Beta-defensins are a group of cysteine-rich antimicrobial peptides that are effective against Gram-positive and Gram-negative bacteria as well as fungi [70]. Proteomics analyses of egg compartments have detected avian beta-defensins in multiple egg components [35, 49]. This study reports a marginal but significant over-exprresion of GAL10 with a 1.17-fold in the WI/Ut contrast and a 1.15-fold increase in the WI/Ma contrast (Additional file 1). A recent proteomics study identified AvBD-10 associated with eggshell membranes from both fertilized and unfertilized eggs [7]. Bioinformatics indicate that AvBD-10 possesses a signal peptide and is categorized as “defense response to bacterium” by the biological process GO term analysis. Transcriptomic, proteomics and bioinformatics evidence all suggest that AvBD-10 (GAL10) is an antimicrobial defense molecule enriched in the white isthmus region of the oviduct.

Ethics statement

All experiments, including all animal-handling protocols, were carried out in accordance with the European Communities Council Directives of 24 November 1986 (86/609/EEC) concerning the practice for the care and Use of Animals for Scientific purposes and the French ministerial decree 87848 of 19 October 1987 (revised on 31 May 2001) on Animal experimentation under the supervision of authorized scientists (authorization # 7323, delivered by the DDPP, direction départementale de la protection des populations, d’Indre et Loire). The experimental unit UE-PEAT 1295 where the birds were kept has the agreement for rearing birds and for the euthanasia of experimental animals (decree N° B37-175-1 of August 28th 2012 delivered by the Préfecture d’Indre et Loire following the inspection of the Department Direction of Veterinary Services). The protocol was approved by an ethics committee (comité d’éthique de val de Loire, officially registered under number 19 of the French national ethics committee for animal experimentation), under agreement number 00159.02.

Animal handing and tissue collection

Forty-week old brown egg-laying hens (ISA-Hendrix, brown strain) were caged individually, with a light/dark cycle of 14 h light and 10 h darkness (14 L:10D), with controlled humidity and temperature. The chickens were fed a layer mash recommended by the Institut National de la Recherche Agronomique (INRA) and bird cages were equipped with automatic recording devices that monitor the time of oviposition. One sample from each segment (magnum, white isthmus and uterus) of the hen oviduct was collected from eight hens during the phase of eggshell calcification. The “white isthmus” corresponds to the upper part of the isthmus where the eggshell membrane fibres are secreted. This region can be easily distinguished from the magnum due to a translucent ring indicating a change in cell type [71]. The lower part of the isthmus, also known as the “red isthmus”, is very similar to the uterine tissue. Samples were immediately frozen in liquid nitrogen and stored at −80 °C until RNA isolation.

RNA isolation and microarray hybridization

The Del-Mar 14 K Chicken Integrated Systems Microarray (NCBI GEO Platform # GPL1731) was used to analyze gene expression in different parts of the hen oviduct during the formation of the eggshell membranes. This microarray system contains total of 14,053 unique cDNAs with 7,937 unique cDNA clones from the neuroendocrine and reproductive systems as well as 9,833 unique cDNA clones from the metabolic and somatic systems [24].

RNA extraction, quality controls, cDNAs labeling and hybridization were performed as previously reported [21]. Briefly, RNeasy Mini Kit (Qiagen, Courtabeouf, France) was used to extract RNA from frozen tissue samples; the samples were also treated with DNase 1 (Macherey-Nagel EURL, France). RNA concentration was determined at 260 nm and its integrity was analyzed using a 1 % agarose gel with an Agilent 2100 Bioanalyzer (Agilent Technologies, Massy, France). RNA samples with a 28S/18S ratio >1.3 were used for labeling and hybridization. The Superscript Plus Indirect cDNA labeling System (Invitrogen, Cergy Pontoise, France) was used to label total RNA (20 μg). Labeled cDNA was synthesized and purified prior to being assessed with a Nanodrop ND 1000 (Nanodrop, Nyxor Biotech, Palaiseau, France).

Hybridization was carried out using the balanced block design; half of the samples were labeled with Alexa Fluor® 555 and the other half were labeled with Alexa Fluor® 647 dyes (Fisher Scientific BioBLock, Illkirch, France). There were two comparisons: white isthmus vs. uterus (WI/Ut) and white isthmus vs. magnum (WI/Ma), where each comparison utilized 8 microarray slides for hybridization with 16 samples (1 sample from each segment for 8 hens. cDNA probes were used for hybridization only if they had an incorporation efficiency >11.4 dye molecules/1000 bases. Prehybridization was performed with 100 μl DIG Easy Hyb buffer (Roche, Indianapolis, IN) for all microarray slides in a humidified chamber for 1 h at 42 °C. The slides were washed with distilled water for 10 min at room temperature. Equal quantities of Alexa Fluor® 555- and Alexa Fluor® 647- labelled cDNA probes from two samples were combined, added to the hybridization solution, and then denatured at 100 °C for 2 min. The combined solution was placed on the slides, which were then cover slipped and placed in the hybridization chamber for 16 h at 42 °C. The slides were washed with 0.2X saline sodium citrate (SSC) buffer and sodium dodecyl sulfate (SDS) buffer for 15 min at 42 °C, and then washed with 0.2X SSC for 15 min at room temperature. Finally, the slides were washed with distilled water and dried by low speed centrifugation. A GenePix 4000 B scanner (Axon Molecular Devices, California, USA) was used to scan the microarray slides, Alexa Fluor® 555 was scanned at 532 nm and Alexa Fluor® 647 was read at 635 nm. GenePix Pro 6.0 software was used to analyze the spot intensity of microarray data. GenePix report (GPR) files containing spot intensity raw data were stored in the BioArray Software Environment (BASE) of SIGENAE (Système d’Information du projet d’Analyse des Génomes des Animaux d’Elevages) and the microarray data was uploaded in the NCBI Gene Expression Omnibus (GEO) database under the series accession numbers GSE17267 and GSE52491.

Validation of gene expression by qRT-PCR

Sixteen genes, chosen to represent a wide range of gene expression, were selected for verification of transcript abundance using quantitative real time PCR (qRT-PCR), as documented in our previous analysis of this data for uterus-specific over-expression [21].

Statistical data analysis

Gene expression was compared between the white isthmus and the uterus as well as between the white isthmus and the magnum (2 samples/microarray slide, 1 microarray slide per hen, for a total of eight hens for each comparison). The ‘anapuce’ package in R was used to identify differentially expressed genes [72]. Spot intensities were calculated using the median value. These values were transformed to log2 values and normalized using global locally-weighted regression (Lowess) in order to eliminate any bias arising from the efficiency of fluorescent dye incorporation. Subtraction of the median value corrected a block effect and the spot intensities were kept only when present in >50 % of samples. Gene variance was estimated using a mixture model integrated into the VarMixt method [73] followed by a unilateral statistical t-test to identify genes over-expressed in the white isthmus compared to either the magnum or uterus. The Benjamini-Hochberg multiple testing procedure [74] was used to adjust the P-values to control the false discovery rate (FDR < 0.05).

Functional annotation

The clone sequences of the 135 over-expressed transcripts were available from the University of Delaware Chick EST and NCBI databases [75 ]. BlastN analysis of the transcript sequences was performed against the NCBI Gallus gallus Refseq RNA and nr nucleotide databases. BlastP analysis of identified proteins was performed against the NCBI non-redundant Gallus gallus protein databases.

Gene ontology enrichment analysis

Gene Ontology (GO) terms represent the functions of proteins encoded by over-expressed genes that are revealed by microarray analysis. Expression Analysis Systematic Explorer (EASE) software from the Database for Annotation, Visualization and Integrated Discovery (DAVID) Bioinformatics Resources 6.7 was used for functional annotation clustering to identify biological themes for 135 over-represented genes in the white isthmus where the formation of the eggshell membranes occurs [76]. Each GO term corresponded to an EASE score (a modified Fisher Exact P-Value and high enrichment value), using GOTERM_BP_FAT and GOTERM_MF_FAT. Only GO terms with an EASE score ≤ 0.05 were considered to be significantly enriched.

Secreted protein identification

Two protein informatic softwares, SignalP 4.1 [77, 78] and SecretomeP 2.0 [79, 80], were applied to identify the over-expressed genes in the white isthmus that encoded sequences specifying either endoplasmic reticulum – mediated export (SignalP) or protein sequences predictive of unconventional secretion (SecretomeP). For SignalP 4.1 analysis, the proteins were only accepted for secretion if the D-score was above the 0.450 cutoff threshold. However, for the SecretomeP analysis the NN-score cutoff was 0.5 for unconventional secretion.