Theriogenology 165 (2021) 1e9 Contents lists available at ScienceDirect Theriogenology journal homepage: www.theriojournal.com Effects of intrauterine devices on proteins in the uterine lavage fluid of mares M.M. Rivera del Alamo a, *, T. Katila b, M. Palviainen c, T. Reilas d a Unit of Reproduction, Faculty of Veterinary Medicine, Travessera Dels Turons S/n Autonomous University of Barcelona, 08193 Bellaterra, Spain b Department of Production Animal Medicine, Faculty of Veterinary Medicine, University of Helsinki, Finland c Department of Equine and Small Animal Medicine, Faculty of Veterinary Medicine, University of Helsinki, Finland d Natural Resources Institute Finland (Luke), Jokioinen, Finland a r t i c l e i n f o a b s t r a c t Article history: Intrauterine devices block luteolysis in cyclic mares, but the underlying mechanism is unknown. To Received 13 January 2020 clarify the mechanisms, the protein profile of the endometrial secretome was analyzed using two- Received in revised form dimensional difference gel electrophoresis (2D-DIGE). Twenty-seven mares were classified according 1 February 2021 to whether they were inseminated (AI) or had an intrauterine device (IUD), a water-filled plastic sphere, Accepted 1 February 2021 Available online 6 February 2021 inserted into the uterus on Day 3 after ovulation. Uterine lavage fluids were collected on Day 15 from pregnant inseminated mares (AI-P; n ¼ 8), non-pregnant inseminated mares (AI-N; n ¼ 4), and mares with IUD (n ¼ 15). The IUD group was further divided into prolonged (IUD-P; n ¼ 7) and normal luteal Keywords: Mare phase (IUD-N; n ¼ 8) groups on the basis of ultrasound examinations, serum levels of progesterone and 2D-DIGE PGFM on Days 14 and 15, and COX-2 results on Day 15. Four mares from each group were selected for the Endometrial proteins 2D-DIGE analyses. Ten proteins had significantly different abundance among the groups, nine of the Intrauterine device proteins were identified. Malate dehydrogenase 1, increased sodium tolerance 1, aldehyde dehydroge- Pregnancy nase 1A1, prostaglandin reductase 1, albumin and hemoglobin were highest in pregnant mares; T- complex protein 1 was highest in non-pregnant mares; and annexin A1 and 6-phosphogluconolactonase were highest in IUD mares. The results suggest that the mechanism behind the intrauterine devices is likely related to inflammation. © 2021 Elsevier Inc. All rights reserved. 1. Introduction luteolysis [4] during maternal recognition of pregnancy in the mare. Thirdly, the device is in contact with the endometrium, Intrauterine devices (IUDs) are used as a method to suppress similar to an embryo. estrus behavior in competing mares. The underlying mechanisms It has been shown that embryo contact with the endometrium e of inducing prolonged luteal phases have not been completely which is enhanced by moving around in the uterus e is involved in elucidated, but embryo simulation by contact with the endometrial the inhibition of luteolysis by attenuating the secretion of PGF2a. wall and endometrial inflammation have been suggested [1e3]. Specifically, contact of the embryo with the endometrium triggers The hypothesis of embryo simulation was initially proposed for mechanotransduction mechanisms that induce changes in endo- three reasons. First, the size of the plastic device is identical to that metrial protein abundance during pregnancy [5]. of an embryo during the time frame of maternal recognition of Although inflammation is a logical explanation for the effec- pregnancy. Secondly, the device can move slightly in the uterine tiveness of IUDs, the evidence for this is not consistent. After IUD lumen [2], and embryo movement is mandatory for blocking use for one year in wild horses, all mares had chronic endometritis at the time of IUD removal [6]. In previous IUD experiments in mares, non-echogenic intrauterine fluid was reported during IUD * Corresponding author. Department of Animal Medicine and Surgery, Faculty of treatment [1,2], but it was not associated with inflammation in Veterinary Medicine, Autonomous University of Barcelona, Bellaterra (Cerdanyola endometrial biopsy samples obtained after IUD removal [1,2,7,8]. It del Valles), E-08193, Spain. is possible that the mares had previous inflammation which was E-mail addresses: mariamontserrat.rivera@uab.cat (M.M. Rivera del Alamo), resolved by the time the biopsies were taken, or at least acute terttu.katila@helsinki.fi (T. Katila), mari.palviainen@helsinki.fi (M. Palviainen), tiina.reilas@luke.fi (T. Reilas). inflammation had already been resolved and turned chronic. It is https://doi.org/10.1016/j.theriogenology.2021.02.001 0093-691X/© 2021 Elsevier Inc. All rights reserved. M.M. Rivera del Alamo, T. Katila, M. Palviainen et al. Theriogenology 165 (2021) 1e9 also possible that the diagnostic methods used to detect inflam- mation in the previous studies were not sensitive enough. It has been shown that IUDs suppress cyclooxygenase-2 (COX-2) [3], leading to the inhibition of PGF2a release and maintenance of the corpus luteum (CL) [2]. However, the events preceding COX-2 inhibition remain unknown. We do not know how the presence of an IUD is mediated to the endometrium to suppress COX-2. Likewise, differences between mares in the efficacy of the devices to prolong luteal function have yet to be elucidated. However, adequate perfusion and drainage of the endometrium seem to in- crease the efficacy of the devices to inhibit luteolysis [8]. Uteroferrin staining was increased in endometrial biopsies of mares in which the presence of an IUD resulted in prolonged CL function [8]. This may indicate inflammation, but it also shows that the presence of a device in the uterus induces changes in the composition of endometrial secretions. Other uterine proteins, such as annexin 2, eukaryotic initiation factor 4A1, protein disul- phide isomerase, superoxide dismutase and transketolase have been described to be involved in processes of endometrial inflam- mation in bovine species [9]. During early pregnancy, endometrial secretions e mostly con- sisting of proteins e are essential. They play a crucial role in im- plantation and development of the conceptus [10]. Changes in the composition across the estrus cycle are thought to occur to provide the most appropriate environment in the different events of a successful pregnancy [10]. The aim was to gain new insights into the mechanisms by which IUDs prevent luteolysis. Our hypothesis was that the protein composition in endometrial secretions on Day 15 after ovulation differs between pregnant, non-pregnant and IUD mares. For this purpose, a proteomic study using two-dimensional difference gel electrophoresis (2D-DIGE) analysis [11], which allows to compare simultaneously multiple protein samples in the same gel thus reducing the experimental and analytical time, was performed on uterine lavage fluids collected previously [3] from pregnant and non-pregnant mares, and from mares with an IUD exhibiting a prolonged or normal luteal phase. 2. Material and methods 2.1. Animals Twenty-seven mares (Finnhorses and four warmbloods) from Equine College Yp€ € and MTT Agrifood Research Ypa aja € ja € (Finland) were initially included in the present study. The mean age was 9.6 years (ranging from 4 to 17 years); live weights ranged from 500 to 590 kg. The mares had foaled 0 to 7 times, had no history of reproductive failure, and were clinically normal. They were ranked by age, number of foalings and breed, and then assigned alternately into two groups: the inseminated group (AI) (n ¼ 12; mean age 9 years; mean number of foalings 0.75) and the intrauterine device group (IUD) (n ¼ 15; mean age 10.3 years; mean number of foalings 1). These groups were further divided into four sub-groups depending on the results of serum progesterone, COX-2 and ul- Fig. 1. Experimental protocol for collecting uterine lavage fluid for two-dimensional difference gel electrophoresis (2D-DIGE). AI: artificial insemination; IUD: intrauter- trasound examination [2]: pregnant (AI-P; n ¼ 8; mean age 6.1 ine device; US: ultrasound examination; P4: progesterone; AI-P: pregnant mares; AI- years) and non-pregnant (AI-N; n ¼ 4; mean age 14.8 years); pro- N: non-pregnant mares; PGFM: 15-ketodihydro-PGF2a; COX-2: cyclooxygenase 2; longed (IUD-P; n ¼ 7; mean age 9.3 years) and normal luteal phase IUD-P: prolonged luteal phase; IUD-N: normal luteal phase. (IUD-N; n ¼ 8; mean age 10.8 years). The permission for animal experimentation was granted by the provincial government of Southern Finland (number 1102101). and ultrasonography (SonoSite Vet 180 Plus with a 4 MHz linear array transducer; Sono Site Inc., Bothell, WA, USA). Once in early 2.2. Experimental design estrus, the examinations were performed daily until ovulation was detected. The ovulation day was assigned as Day 0. Experimental protocol is shown in Fig. 1. The mares included in When a 35-mm follicle was observed by ultrasound, an in- the study were examined every other day by transrectal palpation jection of 1500 IU of hCG (Chorulon®, Intervet International B$V., 2 M.M. Rivera del Alamo, T. Katila, M. Palviainen et al. Theriogenology 165 (2021) 1e9 Boxmeer, The Netherlands) was given to time ovulation. The mares following the manufacturer’s instructions. The samples were then in the AI group were inseminated approximately 24 h after hCG labelled with Cy2, Cy3 and Cy5 dyes (CyDye DIGE Fluor minimal administration with a proven stallion. In the IUD group, the device dyes, GE Healthcare) according to the Ettan two-dimensional dif- was inserted in the uterus on Day 3 after ovulation using the ference gel electrophoresis (DIGE) protocol (Supplementary double-glove technique [12]. The device was a water-filled poly- Fig. S1). Briefly, 50 mg of protein from each sample was labelled propylene ball, 20 mm in diameter [2]. with 400 pmol of the Cy3 and Cy5 dyes. An internal standard was Blood samples for progesterone were obtained on Days 0, 3, 5, 7, established by combining 25 mg of each sample and labelled with 9, 11, 13, 14, and 15 after ovulation and analyzed by means of the Cy2 dye. The labelling reaction was incubated for 30 min on ice in Spectria Progesterone radioimmunoassay (RIA) kit (Orion Diag- the dark, and stopped by adding 1 mM lysine to the reaction nostica, Espoo, Finland), using the 1270 Rackgamma counter following a 10-min incubation period. (Wallac Oy, Turku, Finland). The detection limit of the equipment was 0.7 nmol/L. Intra- and inter-assay coefficients of variations for 2.3.3. Two-dimensional gel electrophoresis low, medium and high levels of progesterone were 11.5%, 3.0% and Labelled proteins were analyzed by DIGE as described earlier 3.8% (intra-), and 7.8%, 5.1% and 4.8% (inter-) respectively. In addi- [15]. An immobilized pH gradient (IPG) strip (24 cm, pH 3e10, tion, serial blood samples were obtained to determine serum levels nonlinear, GE Healthcare) was used for isoelectric focusing. IPG of 15-ketodihydro-PGF2a (PGFM) on Days 14 and 15 after ovulation, strips were loaded with 150 mg of protein in total by using the cup- as previously described [3] and analyzed according to Granstro €m loading method. Isoelectric focusing was performed using IPGPhor and Kindahl [13]. The detection limit of the assay was 60 pmol/L. (GE Healthcare) at 20  C as follows: 3 h at 150 V, 3 h at 300 V, linear The inter- and intra-assay coefficients of variation were 11.7% and ramping to 10 000 V and 10 000 V for 50 000 Vh with a maximum 6.6 respectively. current of 75 mA per strip. After focusing, the isoelectric strips were On Day 14, the mares were examined by transrectal palpation prepared for the second-dimension gels by incubation in equilib- and ultrasonography to determine the stage of the estrus cycle, the rium buffer I (50 mM Tris pH 8.8, 6 M urea, 30% glycerol, 2% sodium presence or absence of an embryo in the AI mares, and the location dodecyl sulphate (SDS), 0.2% bromophenol blue, with added 10 mg/ of the device in the IUD mares. ml dithiothreitol (DTT)) solution for 15 min. This was followed by On Day 15, 30 ml of warm (37  C) phosphate buffered saline incubation in equilibrium buffer II (50 mM Tris pH 8.8, 6 M urea, (PBS; pH 7.0) was infused into the uterus via a Foley catheter, and 30% glycerol, 2% SDS, 0.2% bromophenol blue, supplemented with the uterus was briefly massaged per rectum. After 5 min of equil- 25 mg/ml iodoacetamide) for another 15 min. The prepared IPG ibration, the fluid was allowed to drain into a sterile centrifuge tube strips were then placed on 12% sodium dodecyl sulphate (SDS) [13]. The lavage fluid was kept on ice until centrifuged; after polyacrylamide gels (SDS-PAGE) and sealed with overlay agarose centrifugation, the supernatant was collected and stored at 80  C (Bio-Rad). Electrophoresis was initiated at 50 V for 30 min, followed until analyzed. Lavage fluid was used for protein analyses by 2-DIGE by 300 V for 3 h. The gels were scanned between low-fluorescence techniques [14]. glass plates using an FLA-5100 laser scanner (Fujifilm) at wave- lengths 473 (for Cy2), 532 (for Cy3), and 635 nm (Cy5) using volt- 2.3. 2D-DIGE analysis ages of 420, 410 and 400 V, respectively. After scanning, the gels were silver stained as previously described [16]. Each evaluated 2.3.1. Animal selection for 2D-DIGE analysis group had four biological replicates. For 2D-DIGE analysis, four mares from each of the four groups were selected. The selection criterion for AI-N mares was the 2.3.4. Image analysis and statistical processing of the data absence of an embryo, and for AI-P mares the presence of an em- The gel images were analyzed and statistically assessed using bryo and high serum progesterone. In the IUD mares, serum pro- DeCyder 7.0 software (GE Healthcare). First, the gels were auto- gesterone and plasma PGFM concentrations, and endometrial COX- matically analyzed using the batch processor function to normalize 2 expression were used to determine a normal or prolonged luteal the Cy2, Cy3 and Cy5 images from each gel. Spot volumes were phase. PGFM concentrations of 60e100 pmol/L were defined as calculated based on the intensity of the signal of each spot (Cy3 or low, 101e200 pmol/L as intermediate, and >200 pmol/L as high. Cy5), and compared to Cy2 volumes (internal standard) to correct Thus, four mares with progesterone concentrations between 1 and the inter-gel variations. In the biological variation module, the Cy2 5 nmol/L on Day 15, two to five distinctive PGFM pulse releases images of all eight gels were matched and the spot volumes were (368e739 pmol/L), and positive expression for COX-2 in immuno- compared. Approximately 7000 separate spots were detected on blotting and immunohistochemistry were included in the normal each gel. Protein spots demonstrating at least a 1.5-fold difference luteal phase group (IUD-N), while four mares with progesterone in average spot volume ratios between groups in all biological concentrations between 14 and 26 nmol/L, mostly low PGFM values replicates were selected and analyzed with mass spectrometry. (60/64 values low, 3/64 intermediate, and one value of 256 pmol/L), Spot volume ratios were calculated using one-way ANOVA, with a and negative expression for COX-2 were included in the prolonged P-value of less than 0.05 as the selection criteria. luteal phase group (IUD-P) [3]. Mean ages in AI-P, AI-N, IUD-P and IUD-N groups for 2D-DIGE analysis were 5.5, 14.8, 10.5 and 8.8 2.3.5. Protein identification years, respectively. A workflow for the 2D-DIGE analysis of uterine Protein spots of interest were manually cut from the gel and lavage fluid is shown in Supplementary Fig. S1. digested in-gel using trypsin (Trypsin Gold, Mass spectrometry grade, Promega) as earlier described [17,18]. The samples were first 2.3.2. Sample preparation and CyDye labelling concentrated and desalted on a C18 trap column (PROTECOL, SGE Proteins from the uterine lavage samples were first treated with Analytical Science, Griesham, Germany) followed by peptide sep- 0.1 U chondroitinase ABC and 0.1 U betaglucuronidase to remove aration on a PepMap 100 C18 analytical column (LC Packings, Sun- glycans. Proteins were then precipitated with trichloroacetic acid- nyvale, CA). MS/MS of peptides was performed on a hybrid acetone precipitation and solubilized in 50 ml of labelling buffer quadruple/TOF mass spectrometer with Nanospray II source (7 M urea, 2 M thiourea, 4% cholamidopropyl [dimethylammoniol]- (QSTAR Elite, applied Biosystems, Foster City, CA). The identification 1-propanesulfonate hydrate (CHAPS), 30 mM Tris). Protein con- of proteins was performed using the local Mascot version 2.2 centration was measured by using a 2D Quant Kit (GE Healthcare) (matrix science, London, UK) against the in-house database. The 3 M.M. Rivera del Alamo, T. Katila, M. Palviainen et al. Theriogenology 165 (2021) 1e9 search criteria included digestion with one missed cleavage allowed, as a fixed modification carbamidomethyl modification of cysteine and as a variable modification oxidation of methionine. 2.3.6. Verification of identified proteins All individual samples included in 2D-DIGE analysis from each group were pooled and used for validation by Western blot analysis (Fig. 4). SDS-PAGE was carried out in 12% polyacrylamide gels using a vertical slab gel apparatus under non-reducing conditions (Bio- Rad TetraCell) as described previously [19]. A 20-ml aliquot of each pooled sample was loaded to the gel, and proteins were separated by electrophoresis at 100 V for 3 h. Proteins were transferred to a PVDF membrane (Immobilon, Amersham) using a semidry blotting apparatus (Bio-Rad). Membranes were blocked for 1 h using 5% bovine serum albumin in Tris-buffered saline containing 0.1% Tween-20 (TBST), and probed using polyclonal anti-6- phosphogluconolactonase (6PGL C14200, Assay Biotech, 1:1000), monoclonal anti-increased sodium tolerance 1 (IST1, clone C7, antibodies-online, 1:1000), polyclonal anti-prostaglandin reduc- Fig. 2. A representative 2D-DIGE gel image of the endometrial secretome of mares. tase 1 (PTGR1, C16862, Assay Biotech, 1:1000), polyclonal anti- Proteins with different abundance are marked with red circles. 1: non-identified protein; 2: T-complex protein (TCP1); 5: hemoglobin; 8: 6-phosphogluconolactonase aldehyde dehydrogenase 1A1 (ALDH1A1, bs-6509R, antibodies- (6PGL); 11: increased sodium tolerance 1 (IST1); 12: prostaglandin reductase 1 online, 1:500), polyclonal anti-malate dehydrogenase 1 (MDH1, (PTGR1); 13: aldehyde dehydrogenase 1A1 (ALDH1A1); 16: malate dehydrogenase 1 ABIN2783316, antibodies-online, 1:500) or polyclonal anti-annexin (MDH1); 17: annexin A1 (ANXA1). (For interpretation of the references to colour in this A1 (ANXA1, bs-1562R, antibodies-online, 1:1000), in TBST at 4  C, figure legend, the reader is referred to the Web version of this article.) respectively. The signals were detected using horseradish peroxidase-conjugated secondary anti-rabbit (sc-2004, Santa Cruz, 1:2000) or anti-mouse (P0447, Daco, 1:1000) and SupserSignal different in TCP-1, ANXA1 and 6PGL. Finally, the average ratio be- West Dura Chemiluminescent Substrate (Thermo Scientific), and tween the IUD-P and pregnant mares was significantly different in the intensities were quantified with a chemiluminescent image TCP-1, IST1, MDH1, and ANXA1 (Table 2). analyzer LAS3000 (Fujifilm). There was no commercial antibody All identified proteins were within the expected size and pI available against the T-complex protein 1 subunit theta (TCP1) and ranges in the 2D analysis. Western blot results verified the identi- therefore the biological verification of the protein was not fication of the proteins (Fig. 4). conducted. 3.2. GO annotations 2.3.7. Gene ontology analysis To gain further insight into the role of these proteins, gene The top ten GO terms enriched for the cellular component, ontology (GO) analysis was conducted with the FunRich analysis molecular function and biological process of all identified proteins tool [20]. Proteins were mapped for molecular function, cellular are shown in Fig. 5. Statistically significant GO terms are listed in component and biological process of GO terms. Supplementary Table S1. The most prominent cellular component GO terms were extrinsic component of endosome membrane 3. Results (GO:0031 313), extrinsic component of external side of plasma membrane (GO:0031 232) and phagocytic cup (GO:0001891). For 3.1. 2D-DIGE the molecular function, the most enriched GO terms were phos- pholipase A2 inhibitor activity (GO:0019 834), retinal dehydroge- In the 2D-DIGE analysis, approximately 7000 different spots nase activity (GO:0001758), and calcium-dependent phospholipid were visualized in gels. Ten of these spots had different (P < 0.05) binding (GO:0005544). The proteins were enriched in 56 biological abundance in the uterine lavage fluids between AI-P mares and the process GO terms, of which the most prominent ones were regu- other groups (Fig. 2). Nine of the ten spots were identified by LC- lation of interleukin-1 production (GO:0032 612), negative regu- MS/MS: in addition to albumin and hemoglobin, TCP1, 6PGL, IST1, lation of T-helper 2 cell differentiation (GO:0045 629), and positive PTGR1, ALDH1A1, MDH1, and ANXA1 (Table 1). Of the total ten regulation of T-helper 1 cell differentiation (GO:0045 627). proteins, seven were highest in the AI-P group, one in the AI-N group, one in the IUD-N group, and one in the IUD-P group 4. Discussion (Fig. 3; Table 2). Annexin A1 was up-regulated in IUD mares, with the IUD-P Pregnant mares had the highest endometrial levels of MDH1, mares showing the highest values, while it was down-regulated IST1, ALDH1A1 and PTGR1. GO analysis of molecular function of in pregnant mares (Table 2; Fig. 3). In contrast, IST1, ALDH1A1, these proteins revealed significant enrichment in oxidoreductases: PTGR1, MDH1, albumin and hemoglobin, as well as the non- oxidases and dehydrogenases. identified protein, were up-regulated in pregnant mares, while MDH1 is a cytosolic protein involved in the Krebs cycle and down-regulated in non-pregnant mares. In addition to ANXA1, NADH metabolic processes that transform malate into oxaloacetate. TCP1 and 6PGL were down-regulated in pregnant mares (Fig. 3). Literature on the role of MDH1 in female reproduction is scarce. The average ratio of 2D-DIGE expression of the different iden- Higher endometrial levels of MDH1 have been observed in preg- tified proteins was calculated by using pregnant mares as the nant mares compared with cyclic mares on day 13 after ovulation control group. The average ratio between non-pregnant and preg- [21]. This protein is crucial for the proper development of the nant mares was significantly different in all identified proteins. The embryo in mice [22] and sows [23]. Considering that the embryo is average ratio between IUD-N and pregnant mares was significantly a highly replicating structure, it can be hypothesized that MDH1 4 M.M. Rivera del Alamo, T. Katila, M. Palviainen et al. Theriogenology 165 (2021) 1e9 Table 1 Proteins identified by LC-MS/MS following 2D-DIGE analysis. Spot no Identified protein Accession number Theoretical pI/MW (Da) Matched peptides Sequence coverage (%) MASCOT score 2 TCP1 Q4R5J0 5.5/59.7 4 8 43 8 6PGL OP5RR6 5.6/27.5 6 27.1 38 11 IST1 Q3ZBV1 5.23/39.7 1 2.4 41 12 PTGR1 Q29073 8.39/35.7 3 10.3 52 13 ALHD1A1 P15437 6.43/54.7 1 2.2 60 16 MDH1 P40925 6.91/36.4 4 11.9 93 17 ANXA1 Q8HZM6 5.67/38.7 6 16.7 111 pI: isoelectric point; MW: molecular weight; TCP1: T-complex protein 1; 6PGL: 6-phosphogluconolactonase; IST1: increased sodium tolerance 1; PTGR1: prostaglandin reductase 1; ALDH1A1: aldehyde dehydrogenase 1A1; MDH1: malate dehydrogenase 1; ANXA1: annexin A1. might be involved in fulfilling the glucose, glutamine and oxidative GO analysis of these proteins shows that they are involved in phosphorylation demands of the equine embryo. phospholipase A2 (PLA2) inhibitor activity, calcium-dependent IST1 is a member of the endosomal sorting complexes required phospholipid binding, protein binding, bridging, calcium ion for transport (ESCRT), specifically ESCRT-III [24,25]. The GO analysis binding, and structural molecule activity. All the ten mostly revealed that the microtubule organizing center was the most enriched biological processes and cellular components were asso- enriched component for IST1. It has been shown that the depletion ciated with annexin A1, as well as the most important molecular of IST1 inhibits cellular division [26], suggesting that IST1 is function, PLA2 inhibitor activity. Hence, annexin A1 appears to be involved in the cytokinesis phase of cellular reproduction. It is the most significant component of endometrial secretomics at the hence possible that the presence of an embryo induces the pro- time of sampling in our study. duction and/or release of IST1 in the uterine lumen. Furthermore, it Annexin A1 is a calcium/phospholipid-binding protein that has can be hypothesized that the higher levels of IST1 in the uterine several functions including promotion of membrane fusion, lumen may be involved in the cellular multiplication of the growing involvement in exocytosis, and regulation of PLA2 activity [36]. embryo. According to the literature, this is the first time that this Phospholipase A2 induces the release of arachidonic acid, which is protein is described in equine uterine fluid. the precursor molecule of prostaglandins, through the action of ALDH1A1 is a cytosolic enzyme that metabolizes retinal to ret- cyclooxygenases. In the present study, Annexin A1 was down- inoic acid [27]. Its molecular functions include retinal, aldehyde, regulated in pregnant mares. However, previous studies have and benzaldehyde dehydrogenase activity, and it participates in shown that this specific protein is up-regulated in the intrauterine retinoid and retinol metabolic processes. Retinoid acid is crucial for fluid during early pregnancy in sows [37], sheep [38] and mares the proper development of the embryo. Both deficiency and excess [21], linking its action to maternal recognition of pregnancy. The of this molecule are linked with teratogenic abnormalities during main difference between the studies performed in mares is the embryogenesis [28e30]. Retinoic acid is involved in the processes different sampling time. In the present study, the mares were of neurogenesis, cardiogenesis and body axis extension, as well as sampled on Day 15 after ovulation, while in the study of Smits et al. in the development of the forelimb buds, foregut and eye [27]. Thus, [21] they were sampled on Day 13. Since maternal recognition of the high expression of ALDH1A1 in the uterine lumen of pregnant pregnancy in the mare occurs between days 12 and 14 after mares can be explained by the presence of the developing embryo. ovulation [21], it is possible that ANXA1 might be involved in the Since these proteins, MDH1, IST1, and ALDH1A1, are all involved process of preventing luteolysis by inhibition of PLA2. Once in either cellular replication, embryogenesis or metabolic path- maternal recognition has taken place, ANXA1 levels might decrease ways, it seems logical to assume they also play a role in conceptus again similar to the present results. This hypothesis is sustained by development. an earlier study [39], which demonstrated that proteomic expres- PTGR1 is a cytosolic prostaglandin reductase involved in PG- sion of uterine fluid in pregnant heifers varies depending on the day metabolism. As it has been stated above, PTGR1 was up-regulated of pregnancy. in pregnant mares, which is in agreement with a previous study Annexin A1, an anti-inflammatory mediator, may be induced by [21]. Its molecular functions include 2-alkenal reductase [NAD(P)] glucocorticoids in inflammatory cells, and shares with these drugs with 15-oxo -PGE1, -PGE2 or -PGE2-a as substrates, 13- many anti-inflammatory effects. It is important in the resolution of prostaglandin reductase, which produces the transient PGF inflammation and is therefore induced in inflammatory conditions metabolite, and 15-oxoprostaglandin 13-oxidase activity. The latter [40]. Annexin A1 inhibits inducible nitric oxide synthetase (iNOS) in enzyme initiates metabolic inactivation of leukotriene B4 (LTB4) macrophages and COX-2 in activated microglia. The inhibition of [31], which is known to be involved in many inflammatory pro- iNOS expression may be caused by the stimulation of IL-10 release cesses [32] including neutrophil recruitment [33]. During the induced by annexin A1 in macrophages [40]. Like glucocorticoids, period of implantation, the human endometrium shows signs of annexin A1 exerts profound inhibitory effects on both neutrophil inflammatory processes, such as the presence of interleukin 1 (IL1), and monocyte migration in inflammation. Annexin A1 has been IL6, IL8, leukemia inhibitory factor and tumor necrosis factor [34], recently identified as one of the signals on apoptotic cells to be demonstrating that some steps of inflammatory pathways are recognized and ingested by phagocytes, thus it may contribute to necessary for the proper establishment of pregnancy. However, the safe post-apoptotic clearance of dead cells [41]. Since ANXA1 some important differences have been observed between implan- had the highest abundance in IUD mares, it seems likely that the tation inflammation and inflammation in response to disease or presence of IUDs in the uterine lumen provokes inflammation, injury. The most outlined difference is the lack of neutrophils which in turn induces annexin A1 release contributing to the in- during implantation [35]. The present results suggest that the in- hibition of luteolysis. The highest abundance of annexin A1 was hibition of LTB4 by PTGR1 might be involved either in the maternal detected in IUD-P mares, which suggests that an intense inflam- recognition of pregnancy or pregnancy maintenance. mation may increase the efficacy of the device. ANXA1 and 6PGL were up-regulated in IUD-P and IUD-N groups. 6PGL is an intermediate enzyme in the pentose phosphate 5 M.M. Rivera del Alamo, T. Katila, M. Palviainen et al. Theriogenology 165 (2021) 1e9 Fig. 3. Graphical representation of the standardized log abundance of the identified proteins in the endometrial secretome from each group of mares. TCP1: T-complex protein 1; PTGR1: prostaglandin reductase 1; 6PGL: 6-phosphogluconolactonase MDH1: malate dehydrogenase 1; IST1: increased sodium tolerance 1; ANXA1: annexin A1; ALDH1A1: aldehyde dehydrogenase 1A1. AI-P: pregnant mares; AI-N: non-pregnant mares; IUD-N: device mares with normal luteal phase; IUD-P: device mares with prolonged luteal phase. 6 M.M. Rivera del Alamo, T. Katila, M. Palviainen et al. Theriogenology 165 (2021) 1e9 Fig. 4. Western blotting confirmation of 2D-DIGE results for proteins 6PGL, IST1, PTGR1, ALDH1A, MDH1 and Annexin A1. A 20-ml aliquot of pooled samples were used for each verification. AI-P: pregnant mares; AI-N: non-pregnant mares; IUD-P: device mares with prolonged luteal phase; IUD-N: device mares with normal luteal phase. pathway (PPP) that transforms 6-phosphogluconolactone to 6- did not induce any change in the expression of 6PGL compatible phosphogluconate [42]. During pregnancy e mainly during im- with a pregnancy. plantation e glucose consumption is increased to meet ATP re- T-complex protein 1 subunit theta is a cytosolic protein that quirements [43], and the PPP plays a key role to cover these belongs to the T-complex protein 1 (TCP-1) chaperone family. It necessities. In the present study, 6PGL showed a lower expression participates in the folding of protein complexes [44,45], and is in pregnant mares (AI-P) compared to non-pregnant mares (AI-N). involved in cell growth, proliferation and apoptosis [46] and This result disagrees with that of Smits et al. [21] who observed inflammation [47]. In the present study, pregnant mares showed higher concentrations of 6PGL in uterine fluid on day 13 of preg- the lowest expression of TCP-1, whereas AI-N mares had the nancy in comparison with cyclic mares on the same day after highest expression. ovulation. Once again, a possible explanation for this difference Finally, hemoglobin and albumin were also present in higher could be the fact that the mares were sampled on different days of abundance in intrauterine fluid samples from pregnant mares. The pregnancy. Expression of 6PGL was similar in IUD and non- role of hemoglobin and albumin in intrauterine fluid is controver- pregnant mares. Thus, the presence of the device in the uterus sial. Since intrauterine fluid contains abundant blood proteins, some studies consider their presence as contamination [48]. However, some other studies have demonstrated the presence of hemoglobin in human endometrium and suggested a role during Table 2 Average ratios of 2D-DIGE analyses between pregnant mares (AI-P) and either non- implantation [49,50]. Higher concentrations of hemoglobin in in- pregnant (AI-N) or device (IUD) mares. trauterine fluid from pregnant mares have been previously described [21], which is in accordance with our results. However, Spot nº Identified protein AI-N vs AI-P IUD-N vs AI-P IUD-P vs AI-P further research is warranted to establish the possible role of he- Average ratio, one-way ANOVA (p-value) moglobin in the pregnant mare. Uterine concentration of albumin 2 TCP-1 2.46 (0.0064) 1.29 (0.016) 1.88 (0.0025) is increased in acute iatrogenic endometritis in mares [51], and its 8 6PGL 2.08 (0.0064) 3.60 (0.00036) 1.68 (0.27) presence has been reported both in pregnant and non-pregnant 11 IST1 4.21 (0.015) 1.69 (0.35) 3.51 (0.037) mares [52], but no role in normal early pregnancy has been 12 PTGR1 2.49 (0.03) 2.05 (0.11) 1.27 (0.6) 13 ALDH1A1 4.31 (0.023) 1.96 (0.26) 4.07 (0.33) described to date. 16 MDH1 3.376 (0.021) 1.85 (0.21) 3.09 (0.029) Given these results, it appears that the presence of an embryo or 17 ANXA1 1.40 (0.041) 1.67 (0.012) 1.84 (0.0014) IUD induces changes in the protein composition of endometrial AI-P: pregnant mares; AI-N: non-pregnant mares IUD-N: device mares with normal secretions on Day 15 after ovulation. In the study by Klohonatz et al. luteal phase; IUD-P: device mares with prolonged luteal phase; TCP1: T-complex [5], the contact of the embryo/IUD with the endometrial wall protein 1; 6PGL: 6-phosphogluconolactonase; IST1: increased sodium tolerance 1; induced the formation of adhesion molecules that are involved in PTGR1: prostaglandin reductase 1; ALDH1A1: aldehyde dehydrogenase 1A1; mechanisms of mechano-transduction. The released adhesion MDH1: malate dehydrogenase 1; ANXA1: annexin A1. 7 M.M. Rivera del Alamo, T. Katila, M. Palviainen et al. Theriogenology 165 (2021) 1e9 Fig. 5. Gene ontology (GO) classification of the identified proteins. Proteins were grouped into three main GO categories: biological process (BP), molecular function (MF) and cellular component (CC). Bonferroni corrected p-values were transformed by log 10. molecules differed between embryos and IUDs However, unlike our Declaration of competing interest results, Klohonatz et al. [5] found that the IUD did not block PGF2a release. A possible explanation for this disagreement is the fact that The authors declare that no conflict of interest exists. the former study was performed in vitro, while ours was done in vivo. Funding Annexin A1 proved to be the most important protein in the endometrial secretomics in this study, as all major biological pro- This work was supported by Finnish Veterinary Research cesses were due to annexin A1. Moreover, they were related to the ncia de Gestio Foundation and the Age  d’Ajuts Universitaris i de regulation of inflammation or immune reactions: differentiation of Recerca (AGAUR). T-cells, leukocyte migration, granulocyte chemotaxis, neutrophil apoptosis and interleukin production. The high levels of annexin A1 in IUD mares suggest that the IUD caused inflammation, which CRediT authorship contribution statement subsequently induced annexin A1 to down-regulate the inflam- mation. Since annexin A1 is a PLA2 inhibitor, it also prevented COX- M.M. Rivera del Alamo: All authors contributed to the planning 2 and subsequent PGF2a release. This is likely the mechanism by of the study, carried out the animal experiments, Writing - original which IUDs block the luteolysis. draftwrote the draft of the manuscript. T. Katila: All authors The present study has some limitations, such as the low number contributed to the planning of the study. M. Palviainen: All authors of mares in all groups, as well as the subgroup heterogeneity in contributed to the planning of the study, performed the DIGE terms of age. Another point to consider is the possibility of a analysis and analyzed the data. all authors contributed to the delayed [53] or partial luteolysis [54], making Day 15 maybe too writing of the manuscript and approved it. T. Reilas: All authors early for luteolysis to be completed in some non-pregnant mares. contributed to the planning of the study, carried out the animal The protein composition of endometrial secretions differed be- experiments. tween pregnant and IUD-P mares, and Annexin A1, an inflamma- tory mediator, was up-regulated in IUD-mares. In conclusion, the Appendix A. Supplementary data results of the present study suggest that intrauterine devices cause endometrial inflammation which contributes to the inhibition of Supplementary data to this article can be found online at luteolysis. https://doi.org/10.1016/j.theriogenology.2021.02.001. 8 M.M. Rivera del Alamo, T. Katila, M. Palviainen et al. Theriogenology 165 (2021) 1e9 References stage-specific repatterning of the rostral central nervous system. Dev Biol 1996;175:347e57. [29] Gale E, Zile M, Maden M. Hindbrain specification in the retinoid-deficient [1] Nie GJ, Johnson KE, Braden TD, Wenzel JGW. Use of an intra-uterine glass ball quail. Mech Dev 1999;89:43e54. protocol to extend luteal function in mares. J Equine Vet Sci 2003;23:266e73. [30] White JC, Highland M, Kaiser M, Clagett-Dame M. Vitamin A deficiency results [2] Rivera del Alamo MM, Reilas T, Kindahl H, Katila T. Mechanisms behind in- in the dose-dependent acquisition of anterior character and shortening of the trauterine device-induced luteal persistence in mares. Anim Reprod Sci caudal hindbrain of the rat embryo. Dev Biol 2000;220:263e84. 2008;107:94e106. [31] Nordling E, Jornvall H, Persson B. Medium-chain dehydrogenases/reductases [3] Rivera del Alamo MM, Reilas T, Galv~ ao A, Yeste M, Katila T. Cyclooxygenase-2 (MDR). Family characterization including genome comparisons and active site is inhibited in prolonged luteal maintenance induced by intrauterine devices modeling. Eur J Biochem 2002;269:267e76. in mares. Anim Reprod Sci 2018;199:93e103. [32] Yokomizo T, Izumi T, Shimizu T. Leukotriene B4: metabolism and signal [4] Ginther OJ. Mobility of the early equine conceptus. Theriogenology 1983;19: transduction. Arch Biochem Biophys 2001;15:231e41. 603e11. [33] Lee EKS, Gillrie MR, Li L, Arnason JW, Kim JH, Babes L, et al. Leukotriene B4- [5] Klohonatz KM, Nulton LC, Hess AM, Bouma GJ, Bruemmer JE. The role of mediated neutrophil recruitment causes pulmonary capillaritis during lethal embryo contact and focal adhesions during maternal recognition of preg- fungal sepsis. Cell Host Microbe 2018;23:121e33. nancy. PloS One 2019;14:e0213322. [34] Dekel N, Gnainsky Y, Granot K, Mor G. Review article: inflammation and [6] Daels PF, Hughes JP. Fertility control using intrauterine devices: an alternative implantation. Am J Reprod Immunol 2010;63:17e21. for population control in wild horses. Theriogenology 1995;44:629e39. [35] Chavan AR, Griffith OW, Wagner GP. The inflammation paradox in the evo- [7] Argo CM, Turnbull EB. The effect of intra-uterine devices on the reproductive lution of mammalian pregnancy: turning a foe into a friend. Curr Opin Genet physiology and behavior of pony mares. Vet J 2010;186:39e46. Dev 2017;47:24e32. [8] Klein V, Müller K, Schoon HA, Reilas T, Rivera del Alamo MM, Katila T. Effects [36] Oliani SM, Paul-Clark MJ, Christian HC, Flower RJ, Perretti M. Neutrophil of intrauterine devices in mares: a histomorphological and immunohisto- interaction with inflamed postcapillary venule endothelium alters annexin 1 chemical evaluation of the endometrium. Reprod Domest Anim 2016;51: expression. Am J Pathol 2001;158:603e15. 98e104. [37] Jalali BM, Bogacki M, Dietrich M, Likszo P, Wasielak M. Proteomic analysis of [9] Piras C, Guo Y, Soggiu A, Chanrot M, Greco V, Urbani A, Charpigny G, et al. porcine endometrial tissue during peri-implantation period reveals altered Changes in protein expression profiles in bovine endometrial epithelial cells protein abundance. J Proteomics 2015;125:76e88. exposed to E. coli LPS challenge. Mol Biosyst 2017;13:392e405.  MJ, Barros E, Vital CE, et al. [38] Romero JJ, Liebig BE, Broeckling CD, Prenni JE, Hansen TR. Pregnancy-induced [10] Bastos HBA, Martínez MN, Camozzato GC, Estrade changes in metabolome and proteome in ovine uterine flushings. Biol Reprod Proteomic profile of histotroph during early embryo development in mares. 2017;97:273e87. Theriogenology 2019;125:224e35. [39] Forde N, Carter F, Spencer TE, Bazer FW, Sandra O, Mansouri-Attia N, et al. [11] Arentz G, Weiland F, Oehler MK, Hoffmann P. State of the art of 2D DIGE. Conceptus-induced changes in the endometrial transcriptome: how soon Proteonomics Clin Appl 2015;9:277e88. does the cow know she is pregnant? Biol Reprod 2011;85:144e56. [12] Portus BJ, Reilas T, Katila T. Effect of seminal plasma on uterine inflammation, [40] Perretti M, D’Acquisto F. Annexin A1 and glucocorticoids as effectors of the contractility and pregnancy rates in mares. Equine Vet J 2005;37:515e9. resolution of inflammation. Nat Rev Immunol 2009;9:62e70. [13] Reilas T, Katila T. Proteins and enzymes in uterine lavage fluid of postpartum [41] Parente L, Solito E. Annexin 1: more than an anti-phospholipase protein. and nonparturient mares. Reprod Domest Anim 2002;37:261e8. Inflamm Res 2004;53:125e32. [14] Piras C, Soggiu A, Greco V, Martino PA, Del Chierico F, Putignani L, Urbani A, [42] Collard F, Collet JF, Gerin I, Veiga-da-Cunha M, Van Schaftingen E. Identifi- Nally J, Bonizzi L, Roncada P. Mechanisms of antibiotic resistance of enro- cation of the cDNA encoding human 6-phosphogluconolactonase, the enzyme floxacin in uropathogenic Escherichia coli in dog. J Proteom 2015;127:365e76. catalyzing the second step of the pentose phosphate pathway. FEBS (Fed Eur [15] Ünlü M, Morgan ME, Minden JS. Difference gel electrophoresis: a single gel Biochem Soc) Lett 1999;459:223e6. method for detecting changes in protein extracts. Electrophoresis 1997;18: [43] Gardner DK, Harvey AJ. Blastocyst metabolism. Reprod Fertil Dev 2015;27: 2071e7. 638e54. [16] O’Connell KL, Stults J. Identification of mouse liver proteins on two- [44] Liu X, Lin CY, Lei M, Yan S, Zhou T, Erikson RL. CCT chaperonin complex is dimensional electrophoresis gels by matrix-assisted laser desorption/ioniza- required for the biogenesis of functional Plk1. Mol Cell Biol 2005;25: tion mass spectrometry of in situ enzymatic digests. Electrophoresis 1997;18: 4993e5010. 349e59. [45] Zebol JR, Hewitt NM, Moretti PA, Lynn HE, Lake JA, Li P, et al. The CCT/TRiC [17] Shevchenko A, Jensen ON, Podtelejnikov AV, Sagliocco F, Wilm M, Vorm O, chaperonin is required for maturation of sphingosine kinase 1. Int J Biochem et al. Linking genome and proteome by mass spectrometry: large-scale Cell Biol 2009;41:822e7. identification of yeast proteins from two dimensional gels. Proc Natl Acad [46] Granthman J, Brackley KI, Willison KR. Substantial CCT activity is required for Sci USA 1996;93:14440e5. cell cycle progression and cytoskeletal organization in mammalian cells. Exp [18] Jensen ON, Larsen MR, Roepstorff P. Mass spectrometric identification and Cell Res 2006;312:2309e24. microcharacterization of proteins from electrophoretic gels: strategies and [47] Pejanovic N, Hochrainer K, Liu T, Aerne BL, Soares MP, Anrather J. Regulation applications. Proteins 1998;Suppl 2:74e89. of nuclear factor kB (NF-kB) transcriptional activity via p65 acetylation by the [19] Laemmli UK. Cleavage of structural proteins during the assembly of the head chaperonin containing TCP1 (CCT). PloS One 2012;7:e42020. of bacteriophage T4. Nature 1970;227:680e5. [48] Hannan NJ, Stoikos CJ, Stephens AN, Salamonsen LA. Depletion of high- [20] Pathan M, Keerthikumar S, Chisanga D, Alessandro R, Ang CS, Askenase P, et al. abundance serum proteins from human uterine lavages enhances detection A novel community driven software for functional enrichment analysis of of lower-abundance proteins. J Proteome Res 2009;8:1099e103. extracellular vesicles data. J Extracell Vesicles 2017;26:1321455. [49] Borthwick JM, Charnok-Jones DS, Tom BD, Hull ML, Teirney R, Phillips SC, [21] Smits K, Willems S, Van Steendam K, Van De Velde M, De Lange V, Ververs C, Smith SK. Determination of the transcript profile of human endometrium. Mol et al. Proteins involved in embryo-maternal interaction around the signaling Hum Reprod 2003;9:19e33. of maternal recognition of pregnancy in the horse. Sci Rep 2018;8:5249. [50] Ponnampalam AP, Weston GC, Trajstman AC, Susil B, Rogers PA. Molecular [22] Yoon SJ, Koo DB, Park JS, Choi KH, Han YM, Lee KA. Role of cytosolic malate classification of human endometrial cycle stages by transcriptional profiling. dehydrogenase in oocyte maturation. Fertil Steril 2006;86:1129e36. Mol Hum Reprod 2004;10:879e93. [23] Breininger E, Vecchi Galenda BE, Alvarez GM, Gutnisky C, Cetica PD. Phos- [51] Arlas TR, Wolf CA, Petrucci BPL, Estanislau JF, Gregory RM, Jobim MIM, phofructokinase and malate dehydrogenase participate in the in vitro matu- Mattos RC. Proteomics of endometrial fluid after dexamethasone treatment in ration of porcine oocytes. Reprod Domest Anim 2014;49:1068e73. mares susceptible to endometritis. Theriogenology 2015;84:617e23. [24] Dimaano C, Jones CB, Hanono A, Curtiss M, Babst M. Ist1 regulates Vps4 [52] Zavy MT, Sharp DC, Bazer FW, Fazleabas A, Sessions F, Roberts RM. Identifi- localization and assembly. Mol Biol Cell 2008;19:465e74. cation of stage-specific and hormonally induced polypeptides in the uterine [25] Rue SM, Mattei S, Saksena S, Emr SD. Novel Ist1-Did2 complex functions at a protein secretions of the mare during the oestrous cycle and pregnancy. late step in multivesicular sorting. Mol Biol Cell 2008;19:475e84. J Reprod Fertil 1982;64:199e207. [26] Bajorek M, Morita E, Skalicky JJ, Morham SG, Babst M, Sundquist WI. [53] Ginther OJ, Gastal EL, Gastal MO, Utt MD, Beg MA. Luteal blood flow and Biochemical analyses of human IST1 and its function in cytokinesis. Mol Biol progesterone production in mares. Anim Reprod Sci 2007;99:213e20. Cell 2009;20:1360e73. [54] Ginther OJ, Castro T, Baldrighi JM, Wolf CA, Santos VG. Defective secretion of [27] Duester G. Retinoic acid synthesis and signaling during early organogenesis. prostaglandin F2a during development of idiopathic persistent corpus luteum Cell 2008;134:921e31. in mares. Domest Anim Endocrinol 2016;55:60e5. [28] Avantaggiato V, Acampora D, Tuorto F, Simeone A. Retinoic acid induces 9