Inhibition of ovine in vitro fertilization by anti-Prt antibody: hypothetical model for Prt/ZP interaction

In order to define a role for Prt in ovine fertilization, we hypothesized that the antiserum blockage of ram Prt would limit the fertilizing capacity of ram spermatozoa and subsequent embryo development. As Prt was detected (as described in Pimenta et al. [10]) in the sperm head apical ridge subdomain of ejaculated ram spz (corresponding to the acrosome region) and along in vitro capacitation, it seemed reasonable to investigate its connection to the first steps of sperm-zona binding. To test our hypotheses, different experiments were conceptualized:

Mature ovine oocytes (n = 516) were randomly divided and transferred to fertilization medium alone (control group) or supplemented with 1 μLmL^-1 (v/v) of anti-Prt serum (APPA group), or with 1 μLmL^-1 (v/v) of pre-immune serum (CSerum group) and then co-cultured for 18 h with a mixed pool of frozen/thawed spermatozoa (from three Merino rams) previously submitted to swim up. Afterwards, samples of oocytes were fixed and stained for fertilization assessment (18 h, n = 142). Analysis of data representing 4 replicates of fertilization evaluation was performed.

Presumptive zygotes (from previous steps, n = 374) were denuded and cultured until the stage of 2-4-8 cells embryos. After assessing cleavage, embryo development proceeded until the blastocyst stage, and D6-7 embryo rates were evaluated. Analysis of data representing 4 replicates of embryo production was performed.

The predicted (I-TASSER software [24, 25]) tridimensional structure of ovine Prt was refined (Amber software [26]), and compared to data from CD spectroscopy of Prt in 30% (m/v) and 0.6% (m/v) aqueous TFE. The predicted 3D structures of bovine ZPs (ZP2, ZP3 and ZP4) with homologous ovine fragments were also refined, and used to test for Prt/ZP computational docking, with the HADDOCK software [27].

Unless stated, all reagents used were provided from Sigma-Aldrich (St. Louis, USA). Mouse anti-ovine Prt polyclonal antibody (APPA) which has been shown to be specific for ovine Prt (as described in Pimenta et al. [10]), was used in these studies. Briefly, five to six-week-old female BALB/c mice were injected intraperitoneally with 20 μg of synthetic ovine Prt peptide (obtained from CASLO Laboratory ApS, Denmark; GenBank: ABO86196.1 and emulsified in incomplete Freund’s adjuvant), and boosted monthly (four times) with an equal mass of the referred peptide. Antibody responses generated against ovine Prt were measured by ELISA, and Prt location in semen was demonstrated by Western Blot assay of a protein extract from ram spermatozoa. Membranes were incubated with APPA, with and without previous peptide blocking in order to demonstrate the antibody specificity. Also, mouse serum was collected prior to immunization (pre-immune), and used as a blank (CSerum).

Semen collection was conducted at the experimental farm of INIAV. Semen was collected from native Portuguese Merino rams, already identified by its good in vivo and in vitro fertility results, using an artificial vagina. Immediately after collection, each ejaculate was evaluated for volume, motility and concentration. Good quality ejaculates (mass motility >4; individual motility >60%; concentration >2.5 × 10⁹ spz mL⁻¹) were diluted using an extender containing a solution of 45.0 g L^-1 TRIS, 24.4 gL^-1 citric acid, 5.6 g L^-1 glucose, 15% egg yolk (v/v), 6.6% glycerol (v/v) and antibiotics. The diluted semen was loaded into 0.25 ml mini-straws (300 × 10⁶ spz), refrigerated till 4°C and frozen in nitrogen vapours as described in Valente et al. [21].

Ovine ovaries collected at the local slaughterhouse were transported to the laboratory in Dulbecco’s phosphate buffer saline (PBS, GIBCO) at 37°C. PBS was supplemented with 0.15% (w/v) of bovine serum albumin (BSA, Fraction V) and 0.05 mg mL⁻¹ of kanamycin. At the laboratory, the 2–6 mm follicles were aspirated to obtain cumulus–oocyte complexes. These complexes were incubated in maturation medium (TCM-199, 10 μg mL^-1 FSH, 0.3 mM sodium pyruvate, 100 μM cysteamine, 10 ng mL⁻¹ epidermal growth factor, 10 μg mL⁻¹ estradiol and 10 μL mL⁻¹ gentamicin, as described in Romão et al. [28]), at 38.5°C and 5% CO₂ for 22 h. For oocyte fertilization, a pool of frozen/thawed semen from 3 Merino rams was used in each session. After thawing three straws, pooled sperm motility was immediately examined. Then spermatozoa were submitted to swim-up in capacitation medium (modified Bracket’s medium containing 20% ovine superovulated oestrus serum) at 38.5°C and 5% CO₂ for 1 hour. The upper layer was centrifuged at 225 g for 5 minutes and the supernatant rejected, according to Pereira et al. [22]. The remaining pellet of spermatozoa was evaluated prior to be used to fertilize the oocytes in fertilization medium. Mature ovine oocytes were transferred to in vitro fertilization medium alone (control group) or supplemented with 1 μL mL^-1 (v/v) of anti-Prt serum (APPA group), or with 1 μL mL^-1 (v/v) of pre-immune serum (CSerum group) and then co-cultured with the spermatozoa (1 × 10⁶ spz mL⁻¹) for 18 h. The fertilization medium consisted of synthetic oviductal fluid (SOF) containing glutamine (1.5 μg mL^-1), BME (20 μL mL^-1) and MEM amino acids (10 μL mL^-1), gentamicin (10 μL mL^-1) and 10% ovine superovulated oestrus serum. Samples of oocytes after 18 h of insemination were fixed and stained for fertilization assessment, while the remaining inseminated oocytes proceeded to assess embryo development.

Eighteen hours after oocyte insemination, samples of presumptive zygotes from all groups (control, n = 42; APPA n = 50, CSerum n = 50) were fixed and stained with 1% aceto-lacmoid. Fertilization was considered to occur by the observation in a phase microscope of either a decondensed sperm head or 2 pro-nuclei or zygotes and also when observing more than 2 swollen sperm heads or 2 pronuclei within a single oocyte. Fertilization rate was calculated as the number of fertilized oocytes per number of fertilized and matured oocytes. Immature and unidentified oocytes were not accounted.

After fertilization, presumptive zygotes were denuded and cultured in droplets of SOF enriched with amino acids (20 μL mL⁻¹ BME, 10 μL mL⁻¹ MEM and 6 mg mL⁻¹ bovine serum albumin (BSA,) at 38.5°C, under 5% O₂, 5% CO₂ and 90% N₂ in an humidified atmosphere until the stage of 2-4-8 cell embryos. After assessing cleavage, embryo development proceeded until the blastocyst stage in SOF plus BSA and 10% (v/v) fetal calf serum. Cleavage rate was calculated as the number of cleaved embryos per number of inseminated oocytes and D6-7 embryo rate as the number of morulae and blastocyts at these days per number of cleaved embryos.

Sequence alignment was undertaken with the M-Coffee server, a web server that computes multiple sequence alignments (MSAs) by running several MSA methods and combining their output into one single model [29].

The secondary structure of the Prt peptide previously used to induce APPA antiserum production was studied in two aqueous TFE concentrations (0.6 and 30% v/v) by CD spectroscopy. CD measurements were executed on a Jasco 815 spectrometer between 190 and 260 nm, using of 0.5 cm of path-length cells. The spectra were measured at a peptide concentration of 15 μM m/v (in 30% v/v TFE) and 10.2 μM m/v (in 0.6% v/v TFE). The secondary structure content of Prt peptide was calculated using the CD spectrum deconvolution software CDNN [30] which calculates the secondary structure of this peptide by comparison with a database of CD spectra of known protein structures.

Since the complete ovine ZP sperm-binding proteins sequences are not known, the predicted (with I-TASSER software [24, 25]) tertiary structures of bovine ZP sperm-binding protein sequences [ZP2 (UniProtKB: Q9BH10), ZP3(UniProtKB: P48830) and ZP4(UniProtKB: Q9BH11) [31‐33]] were used, replacing the corresponding bovine sequence, with the homologous ovine fragments [ovine ZP2 (UniProtKB: E5FYD9), and ovine ZP3 (UniProtKB: E5FYG4) [31]].

The ovine Prt (UniProtKB: A4ULE2) and bovine ZPs (with the referred ovine fragments) protein structures predicted by I-TASSER were energy minimized by molecular mechanics in AMBER 12 [26] using the ff12SB force field through 1000 steps of the steepest descent method, followed by the conjugate gradient method until a 0.0001 convergence Kcal.mol^-1 was achieved. The Prt peptide was refined afterwards in explicit solvent using an isometric truncated octahedron TIP3P-water box of 11 Å and the proper number of counter ions was added using tleap as implemented in AMBER package. The simulation was carried out using periodic boundary conditions following a five-step protocol: The first step consisted in a 20000 cycles of minimization to remove any possible unfavorable contacts between the water and the peptide. The first 50 × 10² cycles of the minimization were performed with the steepest descent method, followed by the conjugate gradient method. In this step, the solute was restrained in the cartesian space using a harmonic potential restraint (weight 200 kcal mol^–1.Å^–2). Subsequently, a 10 × 10³ cycles of minimization (30 × 10² steps of steepest descent and 7000 steps of conjugate gradient method) without restraints was performed. The system was then heated up to 298 K for 50 ps using a NVT ensemble and a weak positional restraint (10 mol^–1Å^–2) on the solute, to avoid wild structural fluctuations, using the Langevin thermostat with a collision frequency of 1 ps^-1. The positional restraint was removed and a molecular dynamics run in an isothermal-isobaric (NPT) ensemble at 298 K for 1 ns was performed for equilibration at 1 atm with isotropic scaling and a relaxation time of 2 ps. Finally, NPT data production run was carried out for 10 ns and snapshots of the system were saved to a trajectory file every 0.2 ps. All bonds involving hydrogen atoms were constrained with the SHAKE algorithm [34] allowing the use of a 2 fs time step. The Particle Mesh Ewald method [35] was used to treat the long-range electrostatic interactions and the non-bonded van der Waals interactions were truncated with a 9 Å cut-off. The structural collected data were analyzed with the PTRAJ module as implemented in the AMBER package. The MD trajectory was clustered by RMSD similarity using the average-linkage clustering algorithm [36]. A representative conformation of the cluster with larger population was taken for further docking studies.

The docking simulations were performed with the HADDOCK webserver [37], using the structures previously optimized. The interface prediction in both partners (active and passive residues) were determined through the CPORT [38] prediction method that is optimized for use with HADDOCK. The HADDOCK docking protocol was performed as described elsewhere [39]. The first docking step consisted in a rigid body energy minimization. 10 × 10³ complex conformations were calculated at this stage. The 500 best solutions were then selected for further simulated annealing refinements. Firstly the two proteins were considered as rigid bodies and their respective orientation was optimized, then the side chains at the interface were allowed to move in a second simulated annealing simulation. In the third simulating annealing simulation both backbone and side chains at the interface were allowed to move. The resulting complex structures were energy minimized through 200 steps of the steepest descent method. In the final docking stage, the structures obtained were gently refined (100 MD heating steps at 100, 200 and 300 K followed by 750 sampling steps at 300 K and 500 MD cooling steps at 300, 200 and 100 K all with 2 fs time steps) in an 8 Å shell of TIP3P water molecules. A cluster analysis was performed on the final docking solutions using a minimum cluster size of 6. The root mean square deviation (RMSD) matrix was calculated over the backbone atoms of the interface residues using a 2.0 Å cut off. The resulting clusters were analyzed and ranked according to the HADDOCK score which consists in a weighted sum of intermolecular electrostatic, van de Waals, desolvation and AIR (ambiguous interaction restraints) energies, and a buried surface area term. All molecular diagrams were drawn with PyMOL 1.4.1. [40].

The results were expressed as least squares means ± standard error. Data representing 4 replicates of fertilization stages evaluation and embryo production rates analysis were performed by the MIXED procedure of Statistical Analysis Systems Institute (SAS Inst., Inc., Cary, NC, USA). The mixed linear model included treatment (control, anti-Prt immune serum and pre-immune serum, in fertilization medium) as fixed effect and replicates as random effect. When significant effects were identified, values were compared using the T- or Bonferroni tests. Differences were considered significant when P ≤ 0.05.

A total of 516 ovine mature oocytes were used to assess the fertilizing ability of ovine spermatozoa incubated in the presence and absence of anti-Prt serum and pre-immune serum. The anti-Prt serum clearly interfered (P = 0.006) with the fertilization rate (Table 1). This rate was lower in APPA group than in Control (P = 0.001) and CSerum (P = 0.05) groups. Accordingly the percentage of Metaphase II oocytes was higher in APPA group than in Control (P = 0.001) and Cserum (P = 0.05). CSerum group had more (P = 0.02) zygotes at synkaryosis stage than APPA group No differences (P > 0.05) were identified for sperm head decondensation; two pronuclei, embryo or polyspermic rates among groups (Table 1). In what concerns embryo production rates, only cleavage rates were affected (P < 0.001) by the presence of anti-Prt immune serum in the fertilization medium, being higher (P < 0.0001) in control (44.1 ± 4.15%) than in APPA (19.7 ± 4.22%) group. No differences were identified for D6-7 embryo rates among groups (Table 2).

Figure 1 shows the CD spectrum of Prt peptide dissolved in TFE. In the far UV region (180–240 nm), which corresponds to peptide bond absorption, CD spectrum exhibits two negative bands located at 204 (0.6% TFE) and 207 nm (30% TFE), combined with positives bands near 190 nm, suggesting the presence of a stable α-helical conformation over the TFE range. This is confirmed by secondary structure analysis using the CDNN [30] software that suggests that 17.6% and 22% of the Prt peptide populates an α-helical conformation (in 0,6 and 30% TFE, respectively). In aqueous solution it is mostly a random coil (data not shown).

Biophysical studies of aggregation processes frequently employ chemical cosolvents to reduce experimental variability, simulate cellular conditions, or induce the formation of atypical aggregate structures. The TFE fluorinated alcohol (2,2,2-trifluoroethanol), is one of the most commonly used membrane-mimicking cosolvents in these studies assays [41, 42]. TFE tends to enhance the stability of backbone hydrogen bonds of which the carbonyl groups are from non-polar residues [43]. TFE has been shown to induce and stabilize α-helices in sequences with intrinsic helical propensity [44] without altering helix limits [45] and is a solvent considered to mimic the hydrophobic solvent-excluded environment, which may form early in the protein folding pathway [46]. Moreover, Lehrman et al. [47] suggested that the TFE-enhanced α-helicity, is indicative of the α-helical propensity.

In this colored output (Figure 2), each residue has a color that indicates the agreement of the individual multiple sequence alignments (MSAs) with respect to the alignment of that specific residue. Residues in red are in perfect agreement with every constituting multiple alignments. Previous analysis indicates that 90% of the residues having an individual score of 7 or higher (dark yellow, orange and red) are correctly aligned [29, 48]. Average consistency presents a main score above 70 (Figure 2).

Predicted Prt and ZP3-dimensional protein structures were obtained with the I-TASSER software [24] from amino-acid sequences, using mainly the crystal structures of the full-length chicken sperm receptor ZP3 (PDB: 3NK3 and 3NK4) as templates for bovine ZP2, ZP3 and ZP4 homology modeling, and ab initio structure modeling in the case of ovine Prt, as described in Wu et al. [49]. Compared with other modeling methods, the average performance of I-TASSER is either much better or is similar within a lower computational time [49], and models exhibiting higher C-scores (a confidence score for estimating the quality of predicted models by I-TASSER), were further optimized with the Amber software [26]. The biomolecular modeling program HADDOCK [37, 39] was then used to generate a structural model of the possible interaction between Prt and ZPs. As can be seen from the intermolecular energies in Table 3, the associations are dominated by electrostatic interactions.

Clusters with the lowest not higher scores were selected and the polar contacts mapped, thus giving structural insights of ZP-Prt interaction, as represented in Figures 3 and 4. Images of the three best presenting clusters (Haddock scores), for ZP2, ZP3 and ZP4 are also available as additional material files (Additional file 1: Figure S1). According to the predicted docking results, Prt seems to exhibit a trend towards the ZP domain within the ZP protein chains, that are structural elements predicted to exist as a bipartite structure corresponding to ZP-N and ZP-C regions tethered by a linker [50]. In the case of ZP3, polar contacts are predominantly distributed along the expected bovine ZP-N subdomain (Figure 4). As such, and in order to test if the ZP-N of mouse sperm receptor ZP3, could provide a better template for the homologous bovine subdomain, we specified mouse ZP-N (ZP3) crystal form I (PDB: 3D4C) as a template for bovine ZP-N (ZP3) 3D modeling with I-TASSER. Interestingly, the best C-score was again obtained using the full-length chicken sperm receptor ZP3 as a template (not shown), reinforcing the attained models.