Identification of 56 Proteins Involved in Embryo–Maternal Interactions in the Bovine Oviduct

The bovine embryo develops in contact with the oviductal fluid (OF) during the first 4–5 days of pregnancy. The aim of this study was to decipher the protein interactions occurring between the developing embryo and surrounding OF. In-vitro produced 4–6 cell and morula embryos were incubated or not (controls) in post-ovulatory OF (OF-treated embryos) and proteins were then analyzed and quantified by high resolution mass spectrometry (MS) in both embryo groups and in OF. A comparative analysis of MS data allowed the identification and quantification of 56 embryo-interacting proteins originated from the OF, including oviductin (OVGP1) and several annexins (ANXA1, ANXA2, ANXA4) as the most abundant ones. Some embryo-interacting proteins were developmental stage-specific, showing a modulating role of the embryo in protein interactions. Three interacting proteins (OVGP1, ANXA1 and PYGL) were immunolocalized in the perivitelline space and in blastomeres, showing that OF proteins were able to cross the zona pellucida and be taken up by the embryo. Interacting proteins were involved in a wide range of functions, among which metabolism and cellular processes were predominant. This study identified for the first time a high number of oviductal embryo-interacting proteins, paving the way for further targeted studies of proteins potentially involved in the establishment of pregnancy in cattle.


Introduction
In mammals, embryo development starts in the oviduct, a tubular organ connecting the ovary to the uterus. The bovine embryo develops up to the 16-cell or early morula stage in the oviduct, in close contact with the oviductal epithelial cells and their secretions, the oviductal fluid (OF) [1]. Important embryonic changes including the first mitotic cleavages and the embryonic genome activation, at 8-cell, occur in this oviductal micro-environment [2,3]. The OF is a dynamic and complex fluid composed

New Embryo-Interacting Proteins Were Identified by NanoLC-MS/MS and Changed According to the Embryonic Stage
Among the proteins identified in the OF, 56 were classified as interacting with embryos (i.e., detected in OF-treated embryos but not detected in controls, or detected at significantly higher abundance in OF-treated than in control embryos). The 56 embryo-interacting proteins accounted for only 0.03% of the 1707 proteins identified in the OF (see all OF proteins identified in Table S1).
In total, 4-6 cells and morulas interacted with 37 and 43 proteins, respectively (Tables 1-3), Interacting proteins accounted for 0.02% of the 1616 and 1765 proteins identified in OF-treated 4-6 cells and morulas, respectively (see all proteins identified in embryos in Table S2). The embryo-interacting proteins changed according to the developmental stage: • 13 proteins interacted exclusively with 4-6 cells (Table 1). • 19 proteins interacted exclusively with morulas (Table 2). • 24 proteins interacted with both embryonic stages (Table 3).    There was no correlation between the initial abundance of the embryo-interacting proteins in the OF and their abundance in OF-treated embryos ( Figure S1). Nineteen 4-6 cell-interacting proteins were detected only in OF-treated embryos and 18 were measured at a higher abundance in the OF-treated than in the control embryos (mean ± SEM of treated:control ratio = 8.1 ± 2.1; range: 2.1-37; Tables 1  and 3). For morulas, 11 proteins were detected only in OF-treated embryos and 32 measured at higher abundance in OF-treated than in controls (mean ± SEM of treated:control ratio = 9.3 ± 2.7; range: 2-71; Tables 1 and 2).

Figure 1.
Mean abundance of embryo-interacting proteins in OF-treated embryos at the 4-6 cell and morula stages. Blue bars, proteins interacting exclusively with 4-6 cell embryos. Red bars, proteins interacting exclusively with morulas.

Most Embryo-Interacting Proteins Were Presumed to be Exosomal and Secreted via Non-Conventional Pathways
Five embryo-interacting proteins (9%) contained a signal peptide and were predicted to be

Most Embryo-Interacting Proteins Were Presumed to be Exosomal and Secreted via Non-Conventional Pathways
Five embryo-interacting proteins (9%) contained a signal peptide and were predicted to be secreted in a conventional way. Furthermore, 20 (36%) were predicted to be secreted by non-classical pathways. In addition, 33 embryo-interacting proteins (59%) were reported previously in oviductal exosomes ( Figure 2) [9,30]. Green, proteins possessing a peptide signal and presumed to be conventionally secreted; blue, proteins predicted to be non-conventionally secreted; orange, proteins reported in bovine [9] and feline [30] oviductal exosomes.

Embryo-Interacting Proteins Were Mainly Involved in Metabolism and Cellular Processes
Functional annotation clustering of embryo-interacting proteins resulted in six enriched clusters, among which 'Metabolism' and 'Cellular processes' were the most significant. These findings are visualized using Proteomaps in Figure 3. Proteins such as annexins and alpha-2 macroglobulin (A2M) were assigned to the 'Exosome' category while proteins such as epoxide hydrolase 2 (EPHX2), retinal deshydrogenase 1 (ALDH1A1) and PYGL were assigned to various metabolic processes. Oviductin did not have any functional category annotation and therefore was not included in the proteomap. Green, proteins possessing a peptide signal and presumed to be conventionally secreted; blue, proteins predicted to be non-conventionally secreted; orange, proteins reported in bovine [9] and feline [30] oviductal exosomes.

Embryo-Interacting Proteins Were Mainly Involved in Metabolism and Cellular Processes
Functional annotation clustering of embryo-interacting proteins resulted in six enriched clusters, among which 'Metabolism' and 'Cellular processes' were the most significant. These findings are visualized using Proteomaps in Figure 3. Proteins such as annexins and alpha-2 macroglobulin (A2M) were assigned to the 'Exosome' category while proteins such as epoxide hydrolase 2 (EPHX2), retinal deshydrogenase 1 (ALDH1A1) and PYGL were assigned to various metabolic processes. Oviductin did not have any functional category annotation and therefore was not included in the proteomap.

Protein Interactions Were Localized in Different Embryo Subcompartments
Two highly abundant interacting proteins (OVGP1 and ANXA1) and one among the least abundant ones (PYGL) were chosen to visualize protein interactions in 4-8 cell embryos by immunohistochemistry. The signal for ANXA1 was recorded in the zona pellucida, perivitelline space and into blastomeres of OF-treated embryos (Figure 4a). The pattern of interactions was slightly different for OVGP1 and PYGL, which was localized in the perivitelline space and in blastomeres but not in the zona pellucida of OF-treated embryos (Figure 4c,e, respectively). A negligible diffuse signal was observed in blastomeres of control embryos incubated with primary antibodies against ANXA1, OVGP1 and PYGL (Figure 4b,d,f). No signal was detected in OF-treated and control embryos incubated with IgG isotypes (frames in Figure 4). proteins (down panel) are shown by polygons. Areas of polygons illustrate protein abundance, weighted by protein size. Functionally related function/protein are arranged in common regions and coded using similar colors.

Protein Interactions Were Localized in Different Embryo Subcompartments
Two highly abundant interacting proteins (OVGP1 and ANXA1) and one among the least abundant ones (PYGL) were chosen to visualize protein interactions in 4-8 cell embryos by immunohistochemistry. The signal for ANXA1 was recorded in the zona pellucida, perivitelline space and into blastomeres of OF-treated embryos (Figure 4a). The pattern of interactions was slightly different for OVGP1 and PYGL, which was localized in the perivitelline space and in blastomeres but not in the zona pellucida of OF-treated embryos (Figure 4c,e, respectively). A negligible diffuse signal was observed in blastomeres of control embryos incubated with primary antibodies against ANXA1, OVGP1 and PYGL (Figure 4b,d,f). No signal was detected in OF-treated and control embryos incubated with IgG isotypes (frames in Figure 4).

Discussion
The bovine embryo develops in contact with the OF for the first 4-5 days of its life. To date, there is limited information regarding the molecular interactions occurring between the embryo and its maternal microenvironment. In this study, using a high-resolution MS technique, a number of new embryo-interacting proteins originated in the OF were identified. To our knowledge, this is the first study providing a significant list of proteins interacting with the early embryo in mammals.
The first criterion retained for the definition of embryo-interacting proteins was their detection in the post-ovulatory OF ipsilateral to ovulation. A total of 1707 proteins were identified by nanoLC-MS/MS in the OF used to produce OF-treated embryos. The most abundant proteins in the OF included serum albumin, heat shock proteins (HSP90AA1, HSP90B1, HSP90AB1, HSPA1B, HSPA5), oviductin (OVGP1), annexin A4 (ANXA4), complement C3 (C3), myosin 9 (MYH9) and numerous tubulin subunits. This is in agreement with previous proteomic analyses of post-ovulatory OF collected from cows at the slaughterhouse [5] or by transvaginal endoscopy [7]. Based on our

Discussion
The bovine embryo develops in contact with the OF for the first 4-5 days of its life. To date, there is limited information regarding the molecular interactions occurring between the embryo and its maternal microenvironment. In this study, using a high-resolution MS technique, a number of new embryo-interacting proteins originated in the OF were identified. To our knowledge, this is the first study providing a significant list of proteins interacting with the early embryo in mammals.
The first criterion retained for the definition of embryo-interacting proteins was their detection in the post-ovulatory OF ipsilateral to ovulation. A total of 1707 proteins were identified by nanoLC-MS/MS in the OF used to produce OF-treated embryos. The most abundant proteins in the OF included serum albumin, heat shock proteins (HSP90AA1, HSP90B1, HSP90AB1, HSPA1B, HSPA5), oviductin (OVGP1), annexin A4 (ANXA4), complement C3 (C3), myosin 9 (MYH9) and numerous tubulin subunits. This is in agreement with previous proteomic analyses of post-ovulatory OF collected from cows at the slaughterhouse [5] or by transvaginal endoscopy [7]. Based on our previous work on the regulation of the bovine OF proteome across the estrous cycle [5], some proteins more abundant in the OF around the time of ovulation compared with the luteal phase were identified as embryo-interacting proteins. This is the case, among others, for OVGP1, CD109 and PFKL. Some of these proteins and others were also at higher levels on the side of ovulation, i.e., the side of embryo development, than on the contralateral side at the post-ovulatory stage-this is the case for A2M, CD109 and PFKL [5]. However, at the same stage, EPHX2 was less abundant in the ipsilateral than in the contralateral OF, showing that the secretion of some but not all embryo-interacting proteins may be upregulated at the time and place of embryo presence in the oviduct.
We hypothesized that numerous proteins present in the OF could interact with in vitro-produced early embryos. However, the 56 embryo-interacting proteins accounted for only 0.03% of the identified OF proteins. To our knowledge, this is the first study deciphering OF embryo-interacting proteins using a MS-based approach with no a priori. In an earlier study using the same methodology on bovine spermatozoa, we identified 27 oviductal proteins that interacted with bovine sperm cells [31]. Similarly, sperm-interacting proteins accounted for less than 0.06% of proteins previously identified by MS in the OF [5,31]. In addition to the low proportion of embryo-interacting proteins among OF proteins, there was no relationship between the initial abundance of the embryo-interacting proteins in the OF and their abundance in OF-treated embryos. To illustrate this, galectin-3 (LGALS3) was the fifth most abundant embryo-interacting protein in morulas but was detected with low abundance (12 normalized weighted spectra (NWS)) in the OF. On the other hand, PYGL was among the top-50 most abundant proteins (83 NWS) in the OF but one of the least abundant embryo-interacting proteins. Thus, it seems that very few OF proteins interacted with embryos and that these interactions were not related to their initial abundance around embryos, suggesting highly selective and specific embryo-OF interactions. However, we cannot exclude that longer incubation times (>6 h) may enable more OF proteins to interact. Moreover, this study was carried out on in-vitro produced embryos for obvious economical and ethical reasons (more than 800 embryos were used). Although oviductin is known to interact with bovine embryos in vivo [24] and was identified as interacting proteins under our conditions, it cannot be ruled out that embryo-protein interactions differ in vivo.
Oviductin was identified by nanoLC-MS/MS as interacting in high abundance with both 4-6 cells and morulas and was immunolocalized in the perivitelline space and blastomeres of OF-treated embryos. This is in agreement with previous studies in which OVGP1 was identified by immunostaining and/or western blot in bovine oocytes [28,29] and embryos [19,24] exposed to OF in vivo or in vitro. However, we did not observe a strong signal for OVGP1 in the zona pellucida of OF-treated embryos. Bovine embryos collected in vivo [24] or produced in vitro in the presence of recombinant OVGP1 [19] displayed high immunostaining in the zona pellucida. ANXA1 was observed in the zona pellucida of OF-treated embryos, showing that our conditions did not prevent protein interactions with the zona pellucida. These differences in OVGP1 localization may be due to differences in the origin of embryos used and to the duration of contact with OF or recombinant OVGP1 before immunostaining. A 6-h incubation was used in the present study, whereas embryos were retrieved in vivo after approximately 2 days within the oviduct in the study from Boice et al. [24] or incubated in vitro with recombinant OVGP1 for 3.5 days in the study from Algarra et al. [19]. These differences may also be due to the different antibodies used: a monoclonal antibody raised against the C-terminus of mouse oviductin in the present study vs. a home-made polyclonal antiserum directed against bovine oviduct glycoproteins [24], or a home-made monoclonal antibody directed against purified recombinant porcine OVGP1 [19].
In addition to oviductin, osteopontin (SPP1) and L-PGDS have been reported earlier as oviductal proteins interacting with the zona pellucida of bovine oocytes [29]. However, in this study, SPP1 and l-PGDS were not identified in the OF used for embryo incubation, and therefore could not be identified as interacting proteins. In line with our results, SPP1 and L-PGDS were not identified in previous MS-based analyses of bovine OF, either throughout the estrous cycle [5] or at Days 1 and 3 of the estrous cycle [7], both studies identifying more than 3000 proteins. Thus, the presence of SPP1 and L-PGDS in the bovine OF and their potential interaction with the bovine embryo cannot be confirmed. Moreover, inactivated complement-3b (iC3b), a derivative of the human complement protein C3 (C3), was shown to be taken up by mouse embryos, resulting in an increase in embryo development up to the blastocyst stage [32]. In the present study, C3 was identified at high abundance in the post-ovulatory OF (113 NWS, Table S1). However, C3 was not identified as interacting with cattle embryos. Therefore, some oviductal protein interactions with embryo are likely to be species-specific.
There is some evidence that the developing embryo interacts with its oviductal microenvironment [13,33,34]. However, little is known about the modulating role played by the embryo in these interactions. In order to address this question, the same OF and conditions of embryo incubation were used for both 4-6 cells and morulas. The results showed that 13 OF proteins interacted exclusively with 4-6 cell embryos, while 19 interacted only with morulas. Furthermore, for some proteins interacting at both stages, their abundance in OF-treated embryos differed between stages. For instance, the fold-change between OF-treated and control embryos for ANXA1 was approximately twice higher in 4-6 cells than in morulas (37 vs. 15). This suggests a modulating role played by the zona pellucida and/or embryonic cells in the process of protein interaction. The zona pellucida surrounding all mammalian embryos constitutes the first barrier for interactions between OF protein and embryonic cells. Several studies indicated that the zona pellucida is a dynamic envelope that changes in structure and properties depending on its environment [35][36][37]. The zona pellucida of mouse oocytes were shown to be permeable to macromolecules at molecular weights up to 170 kDa, while zygotes showed a decreased permeability at around 110 kDa [37]. Using colored molecular probes, it was shown that the size and hydrophilic-lipophilic balance of the probe were important in determining its interaction with the mouse embryo [36]. Scanning electron microscopy observation of bovine in vitro-produced embryos showed that the outer zona pellucida surface typically forms a spongy network with a rough surface containing numerous pores; however, the mean number of pores doubled from the 8-cell to the morula stages (1658 vs. 3259 per 5000 µm 2 ) and their mean diameter decreased in parallel (203 vs. 155 nm) [35]. These changes may contribute to the observed stage-specific embryo-protein interactions. Furthermore, the 8-cell stage was identified as the period of major embryonic genome activation in the bovine embryo [3,38]. At this time, maternal RNAs and proteins stored in the oocyte are gradually degraded and actively replaced by embryonic transcripts and proteins [3]. Therefore, molecules and interaction processes at the embryonic cell surface are likely to change from 4-6 cell to morula and to contribute, in association with changes in the zona pellucida permeability, to the differences in OF protein interactions between embryonic stages.
Oviductal EVs comprise exosomes, which are small 30-150 nm vesicles endocytic in origin and released upon fusion of multi-vesicular bodies with the membrane of oviduct epithelial cells, and microvesicles, which are larger vesicles (100-1000 nm) budding directly from the cell membrane [8]. The proteomic contents of oviductal EVs were recently published in the bovine [9] and feline [30], with many more proteins identified in the latter (1511 vs. 319 protein groups). Therefore, both species were considered to analyze potential secretion pathways of interacting proteins. In the present study, only five embryo-interacting proteins (9%) contained a signal peptide and appeared likely to be secreted in a classical way, whereas the majority of proteins were presumed to be secreted by non-conventional pathways and/or previously reported in oviductal EVs, including OVGP1, several annexins (A1, A2, A4, A5) and the liver form of glycogen phosphorylase (PYGL). It is important to note that these secretion pathways are not exclusive-interacting proteins like OVGP1, CD109 and alpha-2 macroglobuline (A2M) possess a signal peptide but were also identified in oviductal EVs. Furthermore, in the present study, OVGP1, ANXA1 and PYGL were immunolocalized in the perivitelline space but also in blastomeres of OF-treated embryos, showing that these molecules crossed the zona pellucida and were internalized by embryonic cells. To our knowledge, this is the first report of immunolocalization of ANXA1 and PYGL in mammalian embryos. Bovine embryos were shown to be able to internalize PKH67-labelled in vivo-derived oviductal EVs during in vitro development [9]. Cloned and parthenogenic porcine embryos were also able to uptake embryo-derived membrane-labelled EVs from the culture medium [39]. In both studies, oviductal EVs were observed in the whole cytoplasm of blastomeres [9,39]. Using electron microscopy on 2-to 8-cell embryos collected from hamster oviducts, endocytic structures, many endosomes and multivesicular bodies associated with OVGP1 immunolabeling were observed in the blastomeres [26]. Taken together, these results strongly suggest that oviductal proteins previously reported as exosomal were internalized into OF-treated embryos via exosomal cargos. However, the exact mechanisms by which OF proteins interacted with embryonic cells were beyond the scope of this study and remain to be determined.
Cattle embryos enter the uterus at the early morula stage and have a long pre-implantation period during which blastocyst hatching (Days 9-10), trophoblast elongation and intense production of interferon-tau (IFNT) occur before implantation begins, around Day 19 of pregnancy [2]. The possible functions of oviductal interacting proteins on pre-implantation steps are poorly understood. Some roles played by oviductin on early embryo development were reported from in vitro studies. Consistent positive effects of purified oviductin were observed on blastocyst yield in goat [17], sheep [40], pig [41] and cattle [18]. Antibodies directed against the C-terminal peptide of rabbit oviductin were shown to inhibit mouse embryo development at the 2-cell stage, suggesting that in this species, oviductin has a function in overcoming the development block at this stage [42]. In bovine, the addition of porcine recombinant oviductin during in vitro fertilization, in vitro development or both, increased the relative abundance in the embryo of mRNA of DSC2, ATF4, AQP3 and DNMT3A, genes involved in cell proliferation, cell adhesion, cellular homeostasis and epigenetics [19].
Four annexins, namely ANXA1, ANXA2, ANXA4 and ANXA5, were identified as embryo-interacting proteins, only at the 4-6-cell stage for ANXA5 and at both stages for ANXA1, ANXA2 and ANXA4. It is well established that some annexins, including ANXA1 and ANXA2, can be secreted out of the cell through unconventional secretory mechanisms, with implications in many functions such as the endocrine regulation, inflammatory response and cancer [43]. Several studies have associated annexins with early embryo-maternal interactions. A greater abundance of ANXA4 was reported in the OF of pregnant mares compared with cyclic mares four days after ovulation [44] and both ANXA1 and ANXA2 were increased in the uterine fluid around the signaling of maternal recognition in this species [45]. Similarly, increasing amounts of ANXA1, ANXA2 and ANXA5 were reported in the uterine fluid of pregnant ewes in the pre-implantation period [45]. Annexin A1 knock-out female mice displayed numerous changes in early gestation, including increased sites of implantation, increased inflammatory reaction in the uterine fluid during implantation, reduced pre-and post-implantation losses and enhanced plasma progesterone [46]. Furthermore, ANXA2 was shown to be crucial for embryo adhesiveness to the endometrium, a critical step for implantation, in humans [47] and mice [48].
Galectin-3 and -9 interacted only with morulas and galectin-3 was one of the most abundant interacting proteins at this stage. Galectins have a varied array of activities both inside and outside cells [49]. Galectin-3 and -9 are members of the lectin family and contain carbohydrate recognition domains [49]. Galectin-3 is expressed in several parts of the female genital tract, including the uterine endometrium and oviduct [50,51]. When galectin-3 was knocked down in the mouse endometrium, the number of embryos implanted decreased substantially [51], showing that, like annexins, galectins have important roles in the establishment of pregnancy in mice.
In conclusion, proteins in the post-ovulatory OF that interact with the early bovine embryo before and after the embryonic genome activation were identified and quantified on a large scale for the first time. Some protein interactions were developmental stage-specific, revealing new roles of the embryo in modulating early maternal interactions. These data provide new protein candidates potentially involved in pre-implantation development and establishment of pregnancy in cattle. Targeted studies are required to go further in the search for underlying mechanisms and functions.

Bovine Oviductal Fluid (OF) Collection
Oviducts connected to ovaries from adult Bos taurus cows were collected at a slaughterhouse and transported to the laboratory on ice within 2 h after the death of the animal. According to the morphology of the ovary and corpus luteum, only oviducts ipsilateral to the side of ovulation at the post-ovulatory phase of the estrous cycle (Days 1-5, i.e., at the expected time and place of embryo development) were used. Mixtures of OF and epithelial cells were collected from the whole oviducts by gentle squeezing, then the OF was isolated by two centrifugations (2000× g, 15 min then 12,000× g, 10 min) at 4 • C. The OF from 22 cows were pooled, assayed for protein concentration, divided into 15-µL aliquots and stored at −80 • C before used for incubation with embryos. The same pool of post-ovulatory OF was used for all embryo co-incubations.

Nanoliquid Chromatography Coupled with Tandem Mass Spectrometry (NanoLC-MS/MS)
For proteomics, pools of 25 embryos (OF-treated and controls) and one aliquot of OF were lysed in a 10 mM Trizma ® base supplemented with 4% sodium dodecyl sulfate (SDS) and 0.05% of a protease inhibitor cocktail for 15 min at ambient temperature. Protein concentrations were determined using a Nanodrop 2000/2000c ® (Thermo Scientific, Waltham, MA, USA). The samples were incubated for 5 min at 95 • C in Laemmli Buffer then protein lysates (4 µg for 4-6 cells, 6 µg for morulas, 15 µg for the OF) were briefly migrated on a home-made 0.75-mm thick 10% SDS-PAGE (50 V, 15 min) to get one band per sample. The gel bands were stained with BluePage ® overnight then proteins were in-gel digested with bovine trypsin, as previously described [5]. Salts were removed from samples using C18 SpinColumns (Harvard Apparatus, Les Ulis, France). The resulting peptide mixtures were separated on a 75 µm × 250 mm IonOpticks Aurora 2 C18 column (Ion Opticks Pty Ltd., Bundoora, Australia). A gradient of basic reversed-phase buffers (Buffer A: 0.1% formic acid, 98% H 2 O MilliQ, 2% acetonitrile; Buffer B: 0.1% formic acid, 100% acetonitrile) was run on a NanoElute HPLC System (Bruker Daltonik GmbH, Bremen, Germany) at a flow rate of 400 nL/min at 50 • C. The liquid chromatography (LC) run lasted for 120 min (2% to 15% of buffer B during 60 min; up to 25% at 90 min; up to 37% at 100 min; up to 95% at 110 min and finally 95% for 10 min to wash the column). The column was coupled online to a TIMS TOF Pro (Bruker Daltonik GmbH, Bremen, Germany) with a CaptiveSpray ion source (Bruker Daltonik). The temperature of the ion transfer capillary was set at 180 • C. Ions were accumulated for 114 ms, and mobility separation was achieved by ramping the entrance potential from −160 V to −20 V within 114 ms. The acquisition of the MS and MS/MS mass spectra was done with average resolutions of 60,000 and 50,000 full width at half maximum (mass range 100-1700 m/z), respectively. To enable the PASEF method, precursor m/z and mobility information was first derived from full scan TIMS-MS experiments (with a mass range of m/z 100-1700). The quadrupole isolation width was set to 2 and 3 Th and, for fragmentation, the collision energies varied between 31 and 52 eV depending on the precursor mass and charge. TIMS, MS operation and PASEF were controlled and synchronized using the control instrument software OtofControl 5.1 (Bruker Daltonik). LC-MS/MS data were acquired using the PASEF method with a total cycle time of 1.31 s, including 1 TIMS MS scan and 10 PASEF MS/MA scans. The 10 PASEF scans (100 ms each) containing, on average, 12 MS/MS scans per PASEF scan. Ion mobility-resolved mass spectra, nested ion mobility vs. m/z distributions, as well as summed fragment ion intensities were extracted from the raw data file with DataAnalysis 5.1 (Bruker Daltonik GmbH, Bremen, Germany).

Quantification of Proteins, Identification of Embryo-Interacting Proteins and Statistical Analysis
All proteins with more than two peptides identified were considered for protein quantification. Protein quantification was based on a label-free approach using spectral counting, as previously described [31]. Scaffold Q+ software (version 4.9, Proteome Software; www.proteomesoftware.com) was used using the Spectral Count quantitative module. Peptide identifications were accepted if they could be established with greater than 95.0% probability as specified by the Peptide Prophet algorithm [53]. Peptides were considered distinct if they differed in sequence. Protein identifications were accepted if they could be established with greater than 95.0% probability as specified by the Protein Prophet algorithm [54] and contained at least two identified peptides (false discovery rate (FDR) < 0.01%).
The normalization of spectra among the samples was realized in Scaffold by adjusting the sum of the selected quantitative values for all proteins within each MS sample to a common value, which was the average of the sums of all MS samples present in the experiment. This was achieved by applying a scaling factor for each sample to each protein or protein group. Thus, the numbers of the normalized weighted spectra (NWS) were tabulated using experiment-wide protein clusters.
Proteins were defined as embryo-interacting proteins originating in the OF if they met the following conditions: (i) detection at a minimum level of 5 NWS in the OF and (ii) detection at a minimum level of 5 NWS in OF-treated embryos with no detection in controls or significantly higher detection in OF-treated embryos than in controls after Student's t-test with Benjamini-Hochberg correction (p-value < 0.05; fold-change > 2).

Functional Analyses of Interacting Proteins
To predict the secretion ways of embryo-interacting proteins, the online tools SignalP 5.0 (http: //www.cbs.dtu.dk/services/SignalP/) and SecretomeP 2.0 (http://www.cbs.dtu.dk/services/SecretomeP/) were used. Proteins predicted as possessing a standard secretory signal peptide and a peptide cleavage site in their N-terminal sequence were considered as secreted in a conventional way. In addition, proteins predicted as being targeted to a non-classical pathway in the mammalian dataset of SecretomeP were considered as potentially undergoing non-conventional secretion. A cutoff NN-score value of 0.6 was applied. To go further in the non-conventional secretion pathway, we looked for the presence of the protein among those reported in bovine [9] and feline [30] oviductal EVs. The online tool Proteomaps 2.0 (https://bionic-vis.biologie.uni-greifswald.de) was used for the functional annotation of embryo-interacting proteins using NWS values and Gene IDs of interacting proteins, and the Bos taurus dataset. 4.6. Immunolocalization of ANXA1, OVGP1 and PYGL By western blotting, the primary antibodies used gave one band at the expected molecular weight in bovine post-ovulatory oviduct epithelial cells and OF ( Figure S2). For immunostaining, embryos of normal morphology at Day 3 were used. Embryos were incubated or not (controls) in OF and washed in Tris-sucrose as described above. Embryos were then fixed for 30 min in 4% paraformaldehyde at 35 • C then washed three times in PBS supplemented with 0.1% (w/v) BSA (PBS-BSA). For blocking, embryos were incubated for 40 min at ambient temperature in PBS-BSA supplemented with 10% (v/v) serum from the same host species as the secondary antibody (donkey for OVGP1 and PYGL; goat for ANXA1). After three washings in PBS-BSA, the embryos were incubated overnight at 4 • C with either anti-ANXA1, anti-OVGP1 (both at 1 µg/mL) or anti-PYGL (at 0.5 µg/mL) diluted in PBS-PSA. Isotypes at the same concentrations were used as negative controls. After two washings in PBS-BSA, embryos were incubated overnight at 4 • C in the secondary antibody diluted at 1/1000 in PBS-BSA. At the end of incubation, Hoechst 33,342 was added (10 µg/mL, 10 min) for nucleus staining. After three washings in PBS-BSA, embryos were mounted in PBS-BSA and immediately observed under confocal microscopy (Zeiss LSM 700, Carl Zeiss, Oberkochen, Germany). Two biological replicates were made for each antibody.