Purification of pregnancy-associated glycoproteins from late-pregnancy Bubalus bubalis placentas and development of a radioimmunoassay for pregnancy diagnosis in water buffalo females

Background Pregnancy-associated glycoproteins (PAGs) were first described as placental antigens present in the blood serum of the mother soon after implantation. Here, we describe the purification of several pregnancy-associated glycoproteins from water buffalo placenta (wbPAGs). A specific radioimmunoassay (RIA) was developed for early pregnancy diagnosis in buffalo species. Results Amino-terminal microsequencing of immunoreactive placental proteins allowed the identification of eleven wbPAGs sequences [Swiss-Prot accession numbers: P86369 to P86379]. Three polyclonal antisera (AS#858, AS#859 and AS#860) were raised in rabbits against distinct wbPAG fractions. A new RIA (RIA-860) was developed and used to distinguish between pregnant (n = 33) and non-pregnant (n = 26) water buffalo females. Conclusions Our results confirmed the multiplicity of PAG expression in buffalo placenta. In addition, the RIA-860 system was shown to be sensitive, linear, reproducible, accurate and specific in measuring PAG concentrations in buffalo plasma samples from Day 37 of gestation onwards.


Background
Domestic buffalo are considered to have low reproductive efficiency, characterised by late attainment of puberty and maturity, seasonality of calving, long postpartum anoestrus, poor expression of oestrus signs, low conception rate and long calving intervals [1,2]. Furthermore, there is some evidence of a high rate of embryonic loss, in particular during the critical phase of embryonic attachment [3,4]. Thus, an accurate early distinction of pregnant and non-pregnant animals is essential for improvement of reproductive efficiency in buffalo, particularly when breeding techniques such as "out of breeding season mating" or artificial insemination (AI) are applied. Moreover, the study of embryonic mortality through the detection of pregnancy markers could support researchers aiming to improve oestrus synchronisation and fixed time AI programs in buffaloes.
Since the early 1980s, a large family of placental proteins without known biological activity (pregnancy-associated or pregnancy-specific proteins, namely pregnancy-associated glycoprotein (PAG) or pregnancy-specific protein B (PSPB)) has been purified from ruminant placenta [5,6]. Some of them are detected in the peripheral circulation of pregnant females, and are used as a tool to investigate placental function in ongoing or endangered pregnancies [7][8][9].
PAGs are expressed in mono-and binucleate trophoblastic cells of the outer epithelial layer in the synepitheliochorial cotyledonary placenta [10][11][12][13]. Using an antiserum (AS) raised against bovine PAG (AS#PAG-F4), Carvalho et al. [14] confirmed that there is a strong homology between water buffalo and ruminant binucleate cells concerning cell morphology, protein expression, glycosylation pattern and characteristics of cell migration and fusion. Buffalo binucleate cells migrate toward the maternal epithelium and fuse with a uterine epithelial cell to form a trinucleate cells [14]. Maternal hybrid trinucleate cells can also further fuse with adjacent cells, resulting in the formation of a multinuclear syncytium [15]. However, larger syncytia, with more than three nuclei, are much less frequent than trinucleate cells in buffalo placentas [14].
Molecular biology techniques have allowed for huge progress in understanding the phylogenetic diversity of PAG molecules. However, these techniques are not adequate to obtain purified and semi-purified PAG preparations necessary for the development of radioimmunoassay (RIA) and enzyme-linked immunosorbent assay (ELISA) protocols. Fortunately, since the 1990s, the selection of the most convenient chromatography procedures allowed for the isolation of an important number of native purified PAG. At present, forty PAG isoforms have been isolated from cotyledons of the cow [6,23,24], ewe [20,25,26], goat [27], buffalo [28], bison [29,30], moose and elk [31]. Some of them were used to immunise rabbits and the antisera obtained allowed the immunolocalisation of PAG in placental tissue [10,13,14,32] and the determination of secretory pattern measured in peripheral maternal blood (reviewed by Sousa et al. [33]).
In buffalo species, despite efforts made on the purification of PAG molecules, the quantity of purified PAG has not been sufficient to raise new PAG antisera. So far, PAG concentrations in buffalo species were determined by heterologous PAG-RIA systems based on antisera raised against bovine or caprine PAG [34,35]. Interestingly, by using these systems, concentrations of PAG were remarkably distinct from those measured in cattle, increasing gradually from the 6 th week of gestation to parturition (28 th week of pregnancy), and reaching relatively low peripartum levels [35].
Here, we describe the successful isolation and characterisation of new buffalo PAG molecules. These newly purified proteins were used for the production of three specific antisera to develop a new PAG-RIA system (RIA-860). Finally, its ability to discriminate between pregnant and non-pregnant females was evaluated in Italian Mediterranean buffalo cows.

Results
Isolation and characterisation of water buffalo PAG (wbPAG) Figure 1 schematically shows the protocol used to isolate wbPAG from fetal cotyledons. Amounts of total protein (TP) and equivalent immunoreactive PAG contents (equivPAG; determined by RIA-708) in the different steps of wbPAG purification are summarised in Table 1. Highest immunoreactivity was observed at the DEAE 80 mM NaCl (D80) when compared to DEAE 40 mM NaCl (D40) and DEAE 160 mM NaCl (D160). Sephadex G75 peaks exhibiting the highest PAG/TP ratios ( Figure 2) were loaded onto VVA lectin-affinity chromatography. The VVA chromatographies of Sephadex G75-D40 Peak III, G75-D80 Peak II and G75-D160 Peak I and II resulted in PAG/TP ratios higher than 100% (Table 2). Table 2 summarises the different purification steps used to obtain the eleven N-terminal sequenced water buffalo PAG molecules (wbPAG), as well as their apparent molecular masses as determined after Coomassie staining of the PVDF membrane ( Figure 3A) and Western blot ( Figure 3B). The apparent molecular masses of immunoreactive PAG isoforms from VVA peaks (revealed by AS#708) ranged from 55 to 70 kDa. These molecular masses were systematically slightly lower than those revealed after Coomassie staining of the PVDF membrane (58 to 76 kDa). Proteins were submitted to Edman degradation based on their availability ( Figure 3A). The protein sequence data described are available in the UniProt knowledgebase under the accession numbers P86369 to P86379. Isolated wbPAG showed highly conserved amino acid residues at the beginning of the N-terminal extremity (RGS-), with Edman sequencing failing to give any signal on cycle 4 ( Table 2). Buffalo sequenced N-termini contained the consensus PLR (residues 9 to 11).
Screening of the EMBL and Swiss-Prot data banks revealed 100% identity between four PAG N-termini characterised here (wbPAG 73kDa _B, wbPAG 61kDa _C, wbPAG 73kDa _I and wb 63kDa _J) and those deduced from cDNA from buffalo species ( Figure 4). Finally, the micro-sequence comparison also revealed that some wbPAG forms (wbPAG 73kDa _B, wbPAG 61kDa _C and wbPAG 63kDa _J) are unique compared to other purified proteins forms isolated from bovine, ovine, caprine and bison species ( Figure 5). Figure 6 shows displacement of standard (0.2 to 25 ng/ mL) inhibition curves (B/B 0 ) tested using three different antisera raised against buffalo PAG (AS#858, AS#859 and AS#860). All tested antisera gave very similar slopes. Highest dilutions of primary antisera were obtained with AS#860 (1:840,000). Therefore, this system (RIA-860) was chosen for measuring concentrations of PAG in water buffalo cows.

Development and validation of a new PAG RIA
Concerning RIA-860 validation, MDL was 0.1 ng/mL. Parallelism between standard curve and serial dilutions from a pregnant buffalo female is demonstrated in Figure 7. Reproducibility measured as the coefficients of variation intra-and inter-assay was 6.7% (2.3 ± 0.2 ng/ mL) and 8.0% (3.1 ± 0.2 ng/mL), respectively. Regarding specificity, the presence of different concentrations (0.19 until 1,000 ng/mL) of placental proteins, sugars and other plasmatic compounds did not alter the binding of radio-labelled PAG. Finally, regarding recovery, it ranged from 101 to 110% when concentrations of 10.0 and 2.0 ng/mL were added to a sample containing low PAG concentrations.

Concentrations of PAG in plasma samples of pregnant and non-pregnant buffalo cows
From a total of 59 buffalo females used, 33 became pregnant after AI, as detected by both rectal palpation and RIA-860. As shown in Table 3, in non-pregnant animals, mean PAG concentrations remained under 0.5 ng/mL at Days 0, 30 and 37 after AI. Two non-pregnant females exhibited PAG concentrations of 1.0 ng/mL: one female at Day 0, and the other at Day 30 after AI. In the pregnant group, two females exhibited PAG concentrations under 1.0 ng/mL at Day 30 after AI. In this group, concentrations increased significantly from Day 0 to 30 and from Day 30 to 37, when the best threshold for   discrimination between pregnant and non-pregnant animals was observed.

Discussion
Purified native PAG preparations are required for the development of specific and/or more sensitive immunoassay techniques that are currently used for pregnancy diagnosis and physiopathological investigations in ruminant species [6,27,33]. Our work describes the isolation and characterisation of different PAGs from buffalo placentas. Some proteins were used for production of polyclonal antisera, allowing the development of new radioimmunoassay systems. In parallel, our work confirmed the large heterogeneity of PAG molecules for both molecular mass and N-terminal amino acid sequence. Heterogeneity of the PAG subfamily was evoked as early as 1982, when Butler et al. [5] estimated molecular mass of bovine PSPB to range from 47 to 53 kDa. However, after several years, it was not clear if such diversity was due to the expression of distinct PSPB/PAG forms in ruminant placenta, to an extensive post-translational processing in placental tissue (glycosylation, phosphorylation or others), or to both phenomena. The identification of a high number of N-terminal sequences and cDNA corresponding to different ruminant PAG and the demonstration of an extensive glycosylation mechanism in bovine and ovine placenta [36][37][38] confirmed that both factors contribute to the PAG diversity described here and in previous works [24][25][26][27][28][29][30]. For instance, different numbers of N-glycosylation sites of asparagines have been observed in PAG isoforms identified in placentas of cattle, sheep and goats [6,11,12,32,37]. In cattle, the dominant boPAG 67kDa form has been shown to present 4 potential glycosylation sites and 10% of oligosaccharides content [6]. Multi-antennary oligosaccharides have been shown to represent 17.83% of the relative molecular mass of ovPAG [37]. The number of potential sites of glycosylation of buffalo PAG had not yet been described in the literature. However, as previously reported, N-terminal sequences of wbPAG failed to give any signal on cycle 4, this blank cycle being followed by the consensus sequence L-T that is a characteristic of the N-glycosylation site.
Our work describes for the first time the production of polyclonal antisera raised against PAG molecules isolated from buffalo placenta. Proteins issued from three different fractions (DEAE 40 mM NaCl, Sephadex G75-Peak III; DEAE 80 mM NaCl, Sephadex G75-Peak I and DEAE 80 mM NaCl, Sephadex G75-Peak II) were used to immunise rabbits resulting in the production of three distinct antibodies: AS#858, AS#859 and AS#860, respectively. These three obtained antisera gave very similar displacement of standard inhibition curves. This finding was not surprising because PAG molecules expressed in the same species can exhibit high sequence identities [20,25,26,30], and thus probably share common epitopes [11,33].
The highest dilution of primary antiserum (1:840,000) was obtained with AS#860. The remarkable parallelism between serial dilutions of pregnant buffalo samples and the standard curve indicates that RIA-860 is a good immunoassay allowing distinguishing subtle quantitative differences in wbPAG concentrations. RIA-860 also proved to be very repeatable for measurement of PAG concentrations (intra-and inter-assay CV lower than 8%). Finally, it was observed that MDL of RIA-860 was equivalent (0.1 ng/mL) to that previously described for RIA-497 [39], RIA-706 [39][40][41] and RIA-Pool [41].
The use of PAG/PSPB RIA systems are considered as reliable method for early pregnancy diagnosis and follow-up of trophoblastic function in ruminant species (reviewed by Sousa et al. [33]). Measurements of PAG/ PSPB concentrations in the peripheral circulation of pregnant and non-pregnant buffalo cows have been performed using different RIA methods [35,42]. Recently, Barbato et al. [35] compared RIA-497, RIA-706 and RIA-708 (antisera raised against boPAG 67kDa , caPAG 55+59kDa and caPAG 55+62kDa , respectively) for detecting PAG molecules in pregnant buffalo females. They reported that analogous PAG antigens were better recognised using RIA-706 from week 6 of gestation onwards. By using the same PAG-RIA system, Karen et al.   Comparison of the N-terminal amino acid sequences of the newly isolated water buffalo PAG (wbPAG) with ruminant PAG exhibiting the highest sequence identity. Peptide and DNA sequence databases were screened, and the identities (id) were determined by the EBI (European Bioinformatics Institute) using the FASTA3 network service. Grey square represent radical substitutions. [34] described that pregnancy diagnosis reached 100% between days 31 and 35.
An additional feature of the present study was the observation of a rapid increase in PAG concentrations from Day 30 (2.9 ± 0.4 ng/mL) to Day 37 of pregnancy (11.6 ± 1.6 ng/mL). Interestingly, a rapid increase in PAG concentration during pregnancy is characteristic of ovine [43] and caprine species [44]. However, it differs largely from those described in bovine species, in which PAG concentrations increase slowly during the first trimester of gestation [39,45].
In Egyptian buffalo cows, Karen et al. [34] and El-Battawy et al. [42] described mean PAG concentrations ranging from 6.4 to 9.8 ng/mL (15 and 5 pregnant females, respectively) at around Day 30 of pregnancy. Those values were higher than mean concentrations (2.9 ng/mL) measured by RIA-860 at Day 30 in Italian Mediterranean females (n = 33). Egyptian and Mediterranean buffalo cows are quite different in several aspects. Mediterranean females have a much higher milk yield than Egyptian females (12 vs. 5 kg/day) [46,47]. This observation of low PAG levels in high producing milk females agrees with previous findings described by Lopez-Gatius et al. [48] who reported that there is a strong negative correlation between milk production and PAG concentrations during the first trimester of pregnancy. Differences in PAG concentrations could also be partially due to other factors such as breed, placental mass, age of female or even farm (genetic selection and nutritional management) [49]. Moreover, as previously described in bovine and ovine species, PAG expression is complex throughout pregnancy. Some PAG molecules are expressed early, while others only as pregnancy progresses. Other PAG are also expressed throughout the whole pregnancy period [11].
PAG concentrations were detectable in peripheral plasma of all Italian Mediterranean pregnant buffalo cows at Day 37. At Day 30, 31 out of 33 pregnant females were already positive by RIA-860. In dairy and beef cows, PAG concentrations can be detected as early as at Day 21 after AI [39]. However, pregnancy diagnosis is only recommended from Day 28 to 30 because of important individual variation on PAG appearance in peripheral blood [50]. According to different authors, expression of PAG family members as early as Day 7 after fertilisation suggests their potential role in cellular growth and differentiation, elongation, apposition, attachment and placentogenesis processes [51][52][53][54].
On the other hand, in the present study, PAG concentrations were higher than 1.0 ng/mL in one non-pregnant female at Days 0 and 30. Recently, Karen et al. [34] described pregnancy diagnosis in Egyptian buffalo cows by using RIA-706 (antiserum raised against caPAG 55+62kDa ). The authors also reported 3/39 and 2/35 incorrect pregnancy diagnoses between days 31-35 and 36-40, respectively. The presence of detectable concentrations in non-pregnant females can be explained by the existence of extra-placental sources of PAG, as suggested by Zoli et al. [50]. Indeed, the presence of antigens immunologically related to PAG has been demonstrated in testicular and in ovarian extracts, justifying, also, the adjective "associated" and not "specific" given to this family of placental glycoproteins [55].

Conclusion
We have described for the first time the use of antisera raised against buffalo PAG for RIA development and pregnancy detection in buffalo cows. Moreover, we showed that RIA-860 was quantitative, precise, accurate and sensitive in measuring PAG concentrations.

Purification of pregnancy-associated glycoproteins from buffalo placenta
Throughout the procedure, the presence of immunoreactive proteins was screened by PAG RIA-708 and Western blot techniques, as previously described [28]. Total protein (TP) concentrations were determined by the Lowry method [56] with bovine serum albumin (BSA; ICN Biochemicals Inc., Aurora, OH, USA) as the standard. After Sephadex G-75 and VVA chromatographies, protein contents were monitored by measuring the UV absorption at 280 nm.
Two placentas were collected from pregnant buffalo cows (Bubalus bubalis) immediately after slaughter, washed with 0.9% NaCl and frozen at -20°C. The stage of pregnancy was 8 months (determination based on the day of artificial insemination). Approximately 4 kg of fetal cotyledons were thawed, finely minced, mixed and homogenised (2 h, 4°C) in 10 mM potassium phosphate buffer containing 100 mM KCl (pH 7.6; ratio buffer to tissue 4:1, v:v) in the presence of protease inhibitors (PMSF, EDTA, NaN 3 ). After centrifugation (16,000 × g, 1 h, 4°C), a second extraction followed by homogenisation (2 h, 4°C) was performed. The pellet obtained after centrifugation was thawed and frozen five times, before a third extraction in phosphate buffer.
Supernatants issued from the three extractions were pooled, adjusted to pH 4.5 with 0.5 M H 3 PO 4 and allowed to precipitate at 4°C overnight. After centrifugation, the pH of the supernatant was readjusted to 7.6 with 0.5 M KOH solution. Extracted proteins were then precipitated by ammonium sulphate ((NH 4 ) 2 SO 4 ) at 40% saturation (16 h, 4°C) and centrifuged. Additional ammonium sulfate was added to achieve 80% of saturation (40-80% (NH 4 ) 2 SO 4 fraction). Following 4 h precipitation, the solution was centrifuged (16,000 × g) and the pellet was suspended in 10 mM Tris-HCl buffer (pH 7.6), extensively dialysed against the same buffer (48 h) and centrifuged (16,000 × g, 1 h) to eliminate the insoluble proteins.
Batches of 800 mg of proteins issued from D40, D80 and D160 were subjected to gel filtration on a Sephadex G75 column (5 × 100 cm; Amersham Biosciences) equilibrated in 5 mM ammonium bicarbonate buffer. After each gel filtration, proteins from the tube presenting the highest immunoreactivity were lyophilised and used to immunise rabbits ( Figure 2). The other tubes corresponding to rich fractions were pooled together and dialysed against 10 mM HEPES buffer (pH 7.5). After dialysis, each rich fraction was run through an 8 mL agarose-bound Vicia villosa lectin (VVA) column (0.7 × 20 cm; Vector Laboratories, Burlingame, CA, USA). Proteins were eluted with 80 mL HEPES buffer containing 50 mM N-acetyl-galactosamine (GalNAc; Acros Organics, Morris Plains, NJ, USA). All VVA eluted fractions were pooled, dialysed against 5 mM ammonium bicarbonate (pH 8), centrifuged (27,000 × g, 15 min) and lyophilised.
The VVA-eluted proteins were separated by onedimensional SDS-PAGE. They were either visualised by Coomassie Brilliant Blue R250 staining, transferred to PVDF membranes for NH 2 -terminal microsequencing or transferred to nitrocellulose membranes for Western blot analysis as described elsewhere [25]. Amino acid (aa) micro-sequencing analyses were performed by Edman degradation on a pulsed liquid-phase protein sequencer (Procise 492; Applied Biosystems Inc., Foster City, CA, USA). The N-terminal sequences obtained in Values with similar superscripts in the same column are statistically different ( a P < 0.01; b P < 0.05). Significant differences between PAG concentrations from pregnant and non-pregnant females are indicated by asterisks (*P < 0.05; **P < 0.0001).
water buffalo placentas were deposited in the SwissProt database (access numbers P86369 to P86379). The NH 2terminal aa sequences of isolated PAG were compared with previously deposited full-length sequences of polypeptide PAG precursors identified from cloned cDNAs (GenBank) and to the micro-sequences of identified native PAG forms (EMBL-EBI). The comparison between N-terminal amino acid microsequence and those deduced from cDNA was performed using Blast program from NCBI. Identities were determined by the EBI (European Bioinformatics Institute) using the Fasta 3 network service [57]. Since it is known that the X in position 4 is part of a N-glycosylation site in PAG, it was substituted by asparagines (N) for database searches [18].

Antisera production and determination of their dilutions for use in routine RIA
Three mature New Zealand white rabbits (AS#858, AS#859 and AS#860) were immunised with distinct purified PAG preparations ( Figure 1) by intradermal route [58]. For the first immunisation, 300 μg of proteins were dissolved in 1.0 mL phosphate buffer 0.5 M (pH 7.5) and emulsified with Freund complete adjuvant (Difco Labs, Detroit, MI, USA). Booster doses (300 μg) were injected at 3-4 week intervals (Freund incomplete adjuvant). Blood was collected from the marginal ear vein starting one month after the second injection and then once a month. Rabbit blood samples were allowed to clot overnight at room temperature. Thereafter, they were centrifuged at 1,500 × g for 20 min, and the sera were stored at -20°C until used. The immunisation protocol was approved by the Animal Ethics Committee of the University of Liege (Dossier number 95).

PAG radioimmunoassay procedure
The PAG measurements were performed according to the method described by Zoli et al. [50] with some modifications. All assays were performed in Tris-HCl buffer (adjusted to pH 7.6) containing 0.1% BSA (Fraction V; Sigma-Aldrich Co., St Louis, MO, USA). Measurements were performed in duplicate in polystyrene tubes and incubations were performed at room temperature (20 to 22°C). Bovine PAG 67 kDa preparation (boPAG 67kDa , accession number Q29432) was used as standard and tracer for all assays. Pure stock boPAG 67kDa (lyophilised powder) was diluted with assay buffer to give standard curves ranging from 0.2 to 25 ng/mL (preincubated system). Iodination (Na-I 125 , Amersham Biosciences) was carried out according to the Chloramine T method [59]. The double antibody precipitation system was composed of a mixture of sheep anti-rabbit immunoglobulin (0.83% v:v), normal rabbit serum (0.17% v:v), polyethylene glycol 6000 (20 mg/mL; Fluka Biochemika, Buchs, Switzerland), cellulose microcrystalline (0.05 mg/mL; Merck, Darmstad, Germany) and BSA (2 mg/mL) diluted in Tris buffer (25 mM Tris, 10 mM MgCl 2 and 0.02% w/v NaN 3 ; pH 7.5).
Briefly, standard and plasma samples (0.1 mL) were diluted in 0.1 mL and 0.2 mL of Tris-BSA buffer, respectively. Virgin buffalo heifer serum (PAG-free serum; 0.1 mL) was added to each tube of the standard curve. The maximum binding (B 0 ) was determined by replacing standard preparations by 0.1 mL of assay buffer. The NSB tubes contained 0.3 mL of buffer and 0.1 mL of PAG-free serum. After the addition of an appropriate dilution of antiserum (AS#860 at 1:840,000; 0.1 mL), the serum samples and the standard tubes were incubated overnight at room temperature (20 to 22°C). The following day, 0.1 mL of I 125 -PAG (≈ 25,000 cpm) was added and the tubes were incubated for 4 h at room temperature. For separation of bound and free fractions, 1 mL of second antibody polyethylene glycol (PEG) solution was added to all the tubes and a further incubation (30 min) at room temperature was performed. The tubes were then washed with 2 mL of assay buffer and centrifuged at 2,500 × g at 4°C for 30 min. The supernatant was discarded and the pellet was washed again and counted in a gammacounter (Packard Cobra II AutoGamma, Milan, Italy). The same person handled the entire experimental protocol. The results were expressed as the ratio (%) between the amounts of tracer bound to antibody in the presence (B) and in the absence (B 0 ) of unlabelled PAG.

Validation of PAG radioimmunoassay
The minimum detection limit (MDL) was defined as the minimum amount of unlabelled PAG that caused a reduction in the percentage of tracer bound to the antibody greater than twice the standard deviation of 20 determinations of B 0 .
Parallelism was assessed by serially diluting pregnant buffalo serum containing relatively high PAG concentrations with PAG-free serum. Parallelism for each RIA system was determined by evaluating a sample at its initial strength (1/1), and at dilutions of 1/2, 1/4, 1/8 and 1/16.
Reproducibility was determined by calculating the intra-and inter-assay coefficients of variation (CV) as follow: [%CV = (SD/mean)*100]. For intra-assay CV, the same serum was assayed 10 times within the same assay. The inter-assay reproducibility was assessed by analysing each serum in four consecutive assays.
Accuracy was determined by adding increasing concentrations of purified boPAG 67 (1.0, 2.0, 4.0 and 10.0 ng) to buffalo sera containing known PAG concentrations. These amounts were chosen to be in the range of PAG concentrations generally found during early pregnancy. The percentage of recovery was calculated as follows: [observed value (ng/mL) / expected value (ng/mL)] × 100.

Collection of buffalo plasma samples and experimental design
The trial was carried out at the experimental farm of the Animal Breeding Research Center of Monterotondo (Rome, Italy, 42°N parallel). Fifty-nine Italian Mediterranean buffalo cows of different ages and parity were used for determination of PAG concentrations. The animals were housed in an open paddock, fed ad libitum on total mixed ration based on maize silage, alfalfa hay, soya bean meal, maize meal and barley meal (containing 0.90 UFL/Kg of dry matter (DM) and 15% crude protein on DM) and milked twice daily.
The buffalo cows were treated with a progesteronereleasing intravaginal device (PRID®; Sanofi, France), containing 1.55 g natural progesterone and a gelatine capsule with 10 mg oestradiol benzoate (which was included within the device). PRID® was kept in place for 10 days. On the 7 th day after PRID® insertion, an i.m. injection of 1000 IU of PMSG (Ciclogonina®, Fort Dodge, Italy) and 0.15 mg of cloprostenol (PGF2alpha analogue) (Dalmazin®, FATRO, Italy) were given. Buffaloes were artificially inseminated using frozen-thawed semen at 72 and 96 h after PRID® removal. Pregnancy status was confirmed by rectal palpation 40 days after AI.
Blood was withdrawn at Days 0, 30 and 37 after artificial insemination (AI). Approximately 10 mL were collected from the jugular vein into EDTA coated tubes. Plasma was separated by centrifugation at 2,500 × g for 10 min, and stored at -20°C until assayed. Blood sampling was performed in accordance with good veterinary practices and approved by the Animal Ethics Committee of the University of Perugia.
Statistical analysis was carried out by using Student t-test. Concentrations of PAG in pregnant and non-pregnant buffalo females were expressed as the means ± standard error of mean (SEM). Statistical significance was considered at the P < 0.05 level.