Porcine sperm binding to oviduct cells and glycans as supplements to traditional laboratory semen analysis

Abstract

Accurate semen evaluation is necessary to maintain high reproductive efficiency but difficult to accomplish. The objective was to determine if the ability to bind oviduct cells or oviduct glycans are useful supplements to traditional semen analyses. Measuring binding to specific soluble glycans is less laborious than assessing binding to oviduct cell aggregates and more suitable for routine use. Previous work has shown that sperm binding to oviduct cells improves fertility prediction, possibly by estimating the ability of sperm to form an oviduct reservoir. The two oviduct glycan motifs, biantennary 6-sialylated N-acetyllactosamine (bi-SiaLN) and Lewis^X trisaccharide (Le^X), that bind boar spermatozoa with high affinity and specificity were tested. Semen from 30 boars was shipped overnight for laboratory analysis and for inseminations to determine fertility (n = 3 replicates). Oviduct cell binding and traditional sperm analyses including motility and morphology were completed. Additionally, binding to soluble fluoresceinated glycans bi-SiaLN, sulfated Le^X (suLe^X), and the control lactosamine disaccharide (LacNAc) was measured. Inseminations were at 15 farms (>50 matings per boar) in the Midwest and farrowing data from all matings were used. Pregnancy rate (PR) and litter size (LS) were adjusted to account for different farms, number of services, number of doses inseminated, and sow parity, using the MIXED procedure in SAS 9.4. A fertility index (FI) was generated, consisting of PR × LS, to estimate boar overall fertility. Finally, the GLMSELECT procedure was used to select variables having a significant impact on PR, LS, and FI. The predictive models constructed were further analyzed using the REG procedure and accounted for 58% or more of the variation in PR, LS, and FI [PR (P < 0.001, r² = 0.60), LS (P < 0.001, r² = 0.58), and FI (P < 0.001, r² = 0.63)]. The final model for PR includes oviduct cell binding as well as boar age, % distal droplets, head morphology, tail morphology, beat/cross frequency, and curvilinear velocity. The final model for LS includes boar age, % distal droplets, tail morphology, and overall morphology. Finally, the FI model included boar age, % distal droplets, head morphology, tail morphology, curvilinear velocity, and semen volume per ejaculate. Although binding to intact oviduct cells was impactful as a means to predict PR, binding to specific soluble oviduct glycans was not a useful supplement to traditional semen analysis.

INTRODUCTION

To maintain high reproductive efficiency, boars with low fertility need to be identified accurately and quickly so that they can be removed from routine service. Selecting for boars with high reproduction can increase the number of animals produced using the same number of females and reduce the number of nonproductive days. This will lower the cost per pound of pork produced and reduce the environmental impact.

One limitation in the selection of boars based on reproduction is that there is no single laboratory assay that can predict an individual boar’s fertility. Traditional semen evaluation includes motility and morphology but these alone are not sufficient (Gil et al., 2005). Using a combination of traditional and functional assays can allow for fertility prediction with greater accuracy. Previous work has identified a correlation between sperm-oviduct epithelial cell binding and fertility (Waberski et al., 2005; Daigneault et al., 2015). This correlation is likely due to the function of the oviduct isthmic epithelium in binding and retaining sperm to form a reservoir (Gualtieri and Talevi, 2003; Rodríguez-Martínez et al., 2005; Töpfer-Petersen et al., 2008). Retention of sperm in the reservoir is critical in swine due to the large variation in time between insemination and ovulation (Soede and Kemp, 1997). Formation of the sperm reservoir in the isthmic portion of the oviduct maintains sperm viability so that oocytes ovulated later after ovulation might still be fertilized (Gualtieri and Talevi, 2003; Rodríguez-Martínez et al., 2005; Töfer-Petersen et al., 2008). An inability of sperm to be retained in the reservoir could result in higher return-to-estrus rates and reduced LS.

The oviduct is made up of a complex matrix of cells where sperm interact and bind. The carbohydrate-mediated binding within the oviduct to create the sperm reservoir has been studied previously in multiple species (hamster: DeMott et al., 1995; horse: Dobrinski et al., 1996; bull: Lefebvre et al., 1997; pig: Green et al., 2001; Ignotz et al., 2001; Wagner et al., 2002; Kadirvel et al., 2012). This interaction is partially controlled by glycans present on the oviduct as well as receptors on the sperm head. A study completed by Kadirvel et al. identified the main carbohydrate structures that noncapacitated boar sperm bind. A glycan array was used, consisting of 377 glycans. The glycan array identified the glycans to which noncapacitated porcine sperm bound. Common among those structures were the glycan motifs biantennary 6-sialylated N-acetyllactosamine on a mannose core (bi-SiaLN) and Lewis^X trisaccharide (Le^X) (Kadirvel et al., 2012). Both bi-SiaLN and Le^X motifs bind with high affinity to the sperm head prior to capacitation and are necessary for sperm binding to the porcine oviduct (Kadirvel et al., 2012; Machado et al., 2014).

We hypothesized that combining functional assays like oviduct epithelial cell binding with traditional assays would improve the accuracy of fertility prediction. Herein, we determined if the ability of sperm to bind to soluble oviduct glycans was a suitable replacement for the more laborious sperm-oviduct cell-binding assay. Binding to oviduct cells and specific soluble oviduct glycans was tested as a supplement to routine semen analysis. Regression models were tested for their ability to predict pregnancy rate (PR), litter size (LS), and the product of the two, PR × LS, defined as fertility index (FI).

MATERIALS AND METHODS

Materials

Chemicals were purchased from Sigma-Aldrich Co. (St. Louis, MO) unless stated otherwise. Semen was provided by Pig Improvement Company (PIC). Reproductive tracts were provided by Rantoul Foods (Rantoul, IL).

Boars and Semen Collection

Boars were all large white males selected for maternal purposes with some carcass traits, all of PIC genetics. All boars were born on the same farm. Boar age varied from 9 to 35 mo. Boars were housed in barns kept at 61–64 °F (16 to 18 °C).

Semen was collected by the gloved hand method. Criteria for acceptable semen quality included ≥80% motility, ≥70% normal sperm, <15% cytoplasmic droplets, and <30% agglutination. After assessment, semen was extended and dispensed into doses. Doses ranged from 3 to 4 billion sperm cells. These doses were cooled and packaged for shipment. They were shipped overnight to our lab as well as to the farms for insemination. Semen was inseminated within 96 h of collection. Spermatozoa from 30 boars were collected, three ejaculates per each boar. Semen was collected from January 4–July 11, 2016. There was a minimum of 3 wk between collections analyzed for each boar.

Birchwood Genetics compiled various collection evaluation parameters including boar rest interval, sperm volume per ejaculate, sperm concentration per ejaculate, total sperm per ejaculate, number of doses poured, and total sperm per dose. Overall motility and progressive motility, as well as various computer-assisted sperm analysis parameters were measured with PRISM software. For morphology, percentages of sperm possessing normal heads, normal tails, proximal droplets and distal droplets were collected.

Females and Breeding Trial

Gilts were exposed to a boar daily to stimulate puberty and estrus. Gilt boar exposure lasted approximately 10 to 15 min. Gilts were inseminated at the second observed estrus. Boar exposure was also used in sows to stimulate them for the duration of their insemination. Doses were administered by traditional AI. Catheters were bent and left in the female for 5 to 7 min after insemination to prevent backflow and allow uterine contractions to pull the remainder of the extended semen into the female. Pregnancy checks were conducted on all females starting at 3 wk of gestation. Those that were considered open by boar exposure–heat check were considered returns. Those females that were classified as abortions were either identified by ultrasound at week 5 or visually by weeks 8 to 10. Some females that were classified as pregnant were culled for unknown reasons. Data collected by PIC included: number of services (times an individual female had an estrus event), number of matings (doses inseminated per estrus), parity (how many litters a female has had), and outcome (return-to-estrus, abortion, died, farrowing/pregnant, or pregnant-at-slaughter). The farrowing data collected include: total born alive, total stillborn, and total mummified piglets. Females were all large whites of PIC genetics with parities ranging from one to 10 (P1–P10). Parity 1 females were gilts at the beginning of the trial. These animals were housed at 15 farms in the greater Midwest. The farms varied in size from 400 to 4,000 females. Each boar was mated to >50 females over the course of the trial (Supplementary Table S1).

Sperm Motility Assessment

Samples were held in NUTRIXcell+ extender (IMV Technologies, Maple Grove, MN) and shipped overnight to the University of Illinois at Urbana-Champaign. Motility was assessed at time of oviduct-binding assay ~36 h after shipping. Motility analyses were conducted using computer-assisted sperm analysis (Hamilton Thorne, Beverly, MA). Sperm were warmed at 37 °C for 15 min. Sperm were centrifuged at 2,400 × g for 1 min, extender was removed and the pellet was resuspended in dmTALPC (2.1 mM CaCl₂, 3.1 mM KCl, 1.5 mM MgCl₂, 100 mM NaCl, 0.29 mM KH₂PO₄, 0.36% lactic acid, 0.6% fraction V BSA, 1 mM pyruvic acid, 25 mM HEPES [pH 7.3], sterile filtered). The sperm suspension was kept at 37 °C and 3 µL of each sample were loaded onto a pre-warmed four chamber Leja4 slide (IMV Technologies). A minimum of eight fields and 450 cells were observed at 37 °C and 45 frames were acquired at a rate of 60 Hz with a minimum contrast of 55, minimum cell size of five pixels, average path velocity (VAP) cutoff of 20 µm/s, progressive minimum VAP cutoff 45 µm/s, straight-line velocity (VSL) cutoff of 5 µm/s, static head size of 0.53 to 4.45 and magnification of 1.73. Total motility, progressive motility, VAP, VSL, VCL (curvilinear velocity), straightness = VSL × 100/VAP, linearity = VSL × 100/VCL, BCF (beat cross frequency), amplitude of lateral head displacement, elongation, percent of head width to head length, and area of sperm head were measured.

Acrosomal and Morphological Analyses of Spermatozoa

Coomassie blue staining was done to determine acrosome integrity and morphology for each sample, modified as follows (Larson and Miller, 1999). Spermatozoa (10 µL) from each group were fixed with 4% paraformaldehyde at 24 h after shipping to fix cells. Cells were then centrifuged at 2,000 × g for 1 min, the supernatant was discarded and the pellet was resuspended in 50 µL of 0.1 M ammonium acetate. The suspension was centrifuged again at 2,000 × g for 1 min, the supernatant was discarded and the pellet was resuspended in 25 µL of ammonium acetate. The sperm suspension (10 µL) was placed onto a slide and allowed to air dry. Dried slides were stained with Coomassie Blue G-250 solution (0.22% Coomassie Blue G-250, 50% methanol, 10% glacial acetic acid, and 40% water) for 4 min. Slides were rinsed with distilled water and allowed to air dry. A drop of 1× PBS was added and topped with a cover glass. A minimum of 300 sperm per sample was evaluated using Zeiss Axio-CamHc (Zeiss Microscopy, LLC, Thornwood, NY) at 400× magnification. Spermatozoa with a stained acrosome were considered to be acrosome intact. While assessing the acrosome status, morphology was also analyzed. The percentage of sperm with normal heads, normal tails, proximal droplets, and distal droplets was recorded. Sperm possessing multiple abnormalities were only considered once when calculating the total number of normal and abnormal sperm.

Sperm-Oviduct Cell-Binding Assay

For each experiment, entire female reproductive tracts were collected from pre- and post-pubertal females from an abattoir. Tracts were brought back to the lab and 20 to 30 oviducts were removed and held in 1× PBS on ice before being processed the same day. Oviducts were transferred to a 100-mm petri dish containing 1× PBS and the ampulla was removed from the isthmus portion using a glass slide. Epithelial sheets from the isthmus were collected by gently pressing down at a 45° angle and squeezing the sheets out. The epithelial sheets were transferred to a 15-mL conical tube and centrifuged at 100 × g for 30 s. After the supernatant was removed, cells were re-suspended in 1 mL of dmTALPC. Cells were dis-aggregated by passage though a 1-mL pipette tip 10 times. Partially disaggregated cells were washed with 5 mL of dmTALPC and centrifuged at 100 × g for 1 min. The supernatant was removed and cells were re-suspended in 1 mL of dmTALPC. The cells were passed through a 23-gauge needle 10 times for further disaggregation, the volume of the cell suspension was adjusted to 9 mL with dmTALPC and cells were divided evenly into three 100-mm petri dishes. The cells were allowed to re-aggregate for 1.5 to 2 h at 37 °C. Spherical aggregates (150 to 200 µm in diameter) were selected for the experiments.

About 30 spherical oviduct cell aggregates were selected and washed twice in 100 µL drops containing fresh dmTALPC. A Stripper Pipette (MidAtlantic Diagnostics, Inc. Marlton, NJ) with a 250-µm internal diameter tip was used to collect oviduct epithelial cell aggregates and wash them. Oviduct cell aggregates were moved to a drop containing a final volume of 50 µL with 1 × 10⁶ sperm/mL for co-incubation at 37 °C for 15 min to allow sperm to bind to the aggregates. Each group contained three 50 µL drops with 10 oviduct cell aggregates each. After incubation, the sperm-bound aggregates were transferred in a volume of 3 µL onto a microscope slide. Each slide corresponded to one boar. Each droplet with 10 sperm-bound aggregates was considered an experimental unit for statistical analysis. Images were captured using a Zeiss Axioskop and AxioCam Hc digital camera (Carl Zeiss, Thornwood, NY). The number of sperm bound to the periphery of each aggregate was counted and the circumference for each aggregate was calculated using ZEN AxioVision V 4.5 software (Carl Zeiss). The number of sperm bound was divided by the linear periphery of the aggregate to normalize for differences in aggregate size. Three boars were tested on an experimental day.

Fluorescent Glycan-Binding Assay

Localization of glycan-binding receptors in live spermatozoa was determined by binding between spermatozoa and three fluoresceinated glycans. Sperm from each boar were held overnight at 17 °C in the original NUTRIXcell+ extender. Sperm were warmed at 37 °C for 15 min. Sperm were centrifuged at 2,400 × g for 1 min, extender was removed and the pellet resuspended in dmTALPC. The concentration was brought to 10 × 10⁶ sperm/mL. An aliquot of 38 µL of the sperm suspension was transferred to an Eppendorf tube and 2 µL of fluorescent glycan was added (final glycan concentration of 50 µg/mL). The glycans used were biantennary 6-sialylated N-acetyllactosamine (bi-SiaLN), sulfated Lewis^X (suLe^X), and N-acetyllactosamine (LacNAc). N-Acetyllactosamine is a disaccharide present in both motifs and was used as the control for nonspecific binding. All glycans were covalently attached to a 30-kDa polyacrylamide chain modified by fluorescein addition. The sperm suspension was incubated with glycans for 30 min at 37 °C. A small drop of 6 µL from the suspension was placed on a slide and covered with a coverslip. A minimum of 200 sperm were counted for each glycan, using a 63× oil immersion objective (630× magnification). Positive binding was considered when green fluorescence was detected. Each of the three glycans was tested for each of the ejaculates for every boar.

Statistical Analyses

All analyses were completed using statistical software from SAS 9.4. Farrowing data were collected by PIC. Females that died during the trial were removed from the data set because there was no way to identify their pregnancy status. Females that were classified as abortion, farrowing/pregnant or pregnant-at-slaughter (those females that were classified as pregnant but were culled before farrowing) were given an outcome value of 1, as they were pregnant for a period of time after insemination. Those females that had been marked as having returned to estrus were given an outcome value of 0 because pregnancy was not established. These two outcomes were used to determine the average PR for each boar. Construction of the second response variable, LS, resulted from those females that had farrowed. The raw LS was calculated by taking the sum of total born alive, total stillborn, and total mummified piglets. Through the MIXED procedure, PR and LS for each boar were adjusted by farm, number of services, number of matings, and sow parity. The third response variable was an FI created by multiplying PR × LS.

After semen shipment, morphology results from four boars during one replicate were not collected. The average of the other two replicates was used to impute values for boars 8, 10, 11, and 12 to allow for complete analysis. Oviduct cell aggregates were normalized using a log transformation in SAS. The values were then back-transformed in Excel using the exponentiation function.

Using the GLMSELECT procedure, the variables that had a significant impact on PR, LS, and FI were selected. This procedure created the models for PR, LS, and FI. The models were further analyzed using PROC REG with the stepwise selection function resulting in the R² values, partial R², and P-values for each independent variable.

RESULTS

Boar Fertility Overview

Boar fertility was based on the three response variables, PR, LS, and FI. There was consistent variation among males based on PR and LS. Although boars were screened by PIC and met the standard minimums of sperm motility and morphology, there were 12 out of the 30 boars used (Figure 1A) that had an adjusted PR lower than the desired PR in females of 90% (Knox, 2016). Eleven boars had an average LS of ~14 piglets per litter and an additional seven were above 13 piglets per litter. The remainder were below 13 piglets per litter with the lowest adjusted average LS of five piglets (Figure 1B). There were seven boars with the lowest PR and LS, yielding low FI values (Figure 1C).

Figure 1.

Boar fertility performance based on their pregnancy rate (PR) (A), litter size (LS) (B), and fertility index (FI) (C) after insemination. Semen from 30 boars was inseminated into at least 50 females. FI was established by multiplying the adjusted PR by the adjusted LS. The mean ± standard error is presented.

Sperm-Oviduct and Glycan-Binding Assays

The 30 boars used in this study varied in the ability of their sperm to bind to oviduct cells in vitro (Figure 2A). Binding to the oviductal epithelium also varied among ejaculate collections from the same boar (data not shown). Oviduct binding was correlated to PR, LS, and FI each individually; however, when the other possible predictors were tested in the model, the only final model that included oviduct binding was the model to predict PR (Table 1). To identify the abundance of glycan receptors, sperm binding to specific soluble fluoresceinated glycans, bi-sialylated lactosamine (bi-SiaLN), sulfated Lewis^X (suLe^X), and lactosamine disaccharide (LacNAc) as a control were assessed. Among the 30 boars, there were no significant differences in the percentage of sperm that bound any of the three glycans analyzed (Supplementary Table S2). The glycan-binding patterns on the sperm head were also identified. The three patterns present were binding to the apical ridge of the sperm head, binding to both the apical ridge and the post-acrosomal area, and last, the post-acrosomal area only. The percent of sperm binding to the individual patterns showed no differences among boars or ejaculates (Supplementary Tables S3A–C). For further analysis, sperm-binding patterns were grouped to create two new variables of total apical binding (apical ridge only + apical ridge and postacrosomal) and total postacrosomal binding (apical ridge and postacrosomal + postacrosomal only). Creating these two variables was done to determine if some boars produced sperm with preferential glycan binding in the post-acrosomal area, due to a leaky membrane. However, there were no differences in the percentage of sperm that bound the targeted glycans (bi-SiaLN and suLe^X) in the apical region (total apical binding) (Table 2) or the postacrosomal binding (total) among the boars used (Supplementary Table S4). Further, binding to soluble fluoresceinated glycans was not correlated with oviduct cell binding (Supplementary Table S5). Finally, binding of soluble fluoresceinated glycans to sperm from the 30 boars was not correlated with either PR or LS (Supplementary Table S6).

Figure 2.

Boar average semen performance in oviduct epithelial cell binding (A), motility (B), and morphology (C). (A) Sperm bound to the periphery of aggregates (n = 30) were counted and normalized to the size of the aggregate. (B) Sperm motility before (light gray bars) and after shipment (black line) were analyzed. (C) Sperm were fixed and stained. Sperm were classified as normal if they presented no defects mean ± standard error of three semen collections from 30 boars is presented.

Table 1.

Binding to intact oviduct cells regressed against pregnancy rate (PR), litter size (LS), fertility index (FI) value, motility, and normal morphology

Oviduct cell-binding correlations
Variable	Estimate	Standard error	t-Value	P-Value	Adj R2
PR	0.07	0.02	4.27	0.17	0.16
LS	0.02	0.01	2.57	0.07	0.06
FI	2.87	0.92	3.13	0.10	0.09
% Motile cells (A)	0.31	0.06	5.03	0.22	0.21
% Total normal morphology (A)	0.24	0.06	3.75	0.14	0.13

Variables were regressed using PROC REG in SAS 9.4. Motility and morphology assessments were performed before and after (A) shipment.

Table 2.

Specific binding to the apical ridge of the sperm head by the three soluble fluoresceinated glycans

Apical ridge glycan binding (%)
Boar	bi-SiaLN	suLeX	LacNAc
1	81.85 ± 3.72	32.91 ± 2.46	12.53 ± 2.84
2	77.24 ± 9.14	45.27 ± 10.27	17.28 ± 6.93
3	80.43 ± 3.00	31.13 ± 14.69	12.23 ± 2.95
4	68.01 ± 8.85	22.51 ± 3.48	29.94 ± 9.79
5	69.15 ± 13.93	37.75 ± 15.38	22.36 ± 1.94
6	76.57 ± 8.63	29.97 ± 5.58	25.43 ± 14.84
7	74.86 ± 3.69	39.15 ± 11.18	23.01 ± 6.43
8	74.86 ± 3.69	39.15 ± 11.18	23.01 ± 6.43
9	78.13 ± 8.59	47.98 ± 36.13	37.86 ± 32.74
10	77.85 ± 9.02	34.33 ± 7.94	25.58 ± 6.56
11	69.84 ± 19.02	36.11 ± 17.47	14.10 ± 7.44
12	77.64 ± 11.44	32.52 ± 9.62	17.49 ± 10.42
13	70.48 ± 3.95	26.90 ± 7.30	17.93 ± 3.67
14	71.70 ± 2.62	48.50 ± 19.89	27.03 ± 13.22
15	77.29 ± 6.13	32.65 ± 15.91	11.06 ± 4.75
16	78.24 ± 11.47	30.08 ± 24.10	26.78 ± 22.76
17	76.40 ± 2.96	20.80 ± 3.48	23.24 ± 8.10
18	84.72 ± 3.86	22.80 ± 11.07	10.71 ± 2.18
19	72.63 ± 15.13	33.34 ± 13.63	20.98 ± 5.82
20	84.51 ± 2.17	36.32 ± 7.59	27.83 ± 9.14
21	81.04 ± 14.10	40.61 ± 17.93	21.78 ± 11.98
22	75.43 ± 10.29	43.79 ± 3.45	13.03 ± 6.62
23	80.55 ± 9.59	37.73 ± 14.39	11.56 ± 3.48
24	78.61 ± 8.44	34.12 ± 11.03	20.96 ± 2.42
25	66.55 ± 3.52	26.42 ± 9.49	21.05 ± 3.09
26	84.65 ± 4.92	34.91 ± 25.51	26.17 ± 12.34
27	69.54 ± 20.87	47.48 ± 23.91	20.38 ± 1.69
28	73.59 ± 4.89	17.58 ± 6.23	22.96 ± 17.12
29	78.52 ± 11.75	31.65 ± 13.23	17.68 ± 6.24
30	68.55 ± 9.79	26.06 ± 13.37	22.55 ± 11.22

The three patterns on the sperm head were apical ridge only, apical ridge and post-acrosomal area, and post-acrosomal only. Patterns were re-grouped as apical ridge + apical ridge and post-acrosomal to have a total percent of sperm binding to the edge of the head which would be in contact with oviduct cells. Three ejaculates were analyzed for each of the 30 boars. Fluorescence detected on the sperm head was considered positive binding. The mean ± standard deviation is presented. No significant differences were detected.

Sperm Motility and Morphology

Sperm from samples of fresh extended boar semen were assessed based on a variety of characteristics before and after shipment. All 30 boars had ≥90% motile cells prior to shipment (Figure 2B). However, after overnight shipping, the average semen motility had decreased from as little as 7% to as much as 60% among the different boars. Boars showed a decrease in the percentage of motile cells an average of 28% at the time of the sperm-oviduct-binding assay (Figure 2B).

The percentage of morphologically normal sperm ranged from over 90% to below 50% among the boars (Figure 2C). The most frequent defects observed were proximal and distal droplets, head and tail problems, and those that were acrosome reacted. The types of morphological problems varied between boars and replicates (data not shown).

Final Model Variables

Variables selected through GLMSELECT were subjected to regression analysis resulting in the final models represented in Tables 3 to 5. Table 3 presents the estimate values in addition to the partial R² values for PR. Binding to oviduct cells had a significant impact on the PR. Two CASA motility parameters observed after semen shipment were included in the final model, BCF and VCL. Various morphology problems were correlated with PR. The percent of head defects before shipment and the percent of sperm with distal droplets and tail defects after shipment were useful in predicting PR (Table 3). The amount of time a boar has been at the stud was included in the model for PR as well (Table 3). Boars that were regularly collected over a longer time had higher PR than those that entered the stud at a later date.

Table 3.

Multiple linear regression analysis to assess the relationship of sperm traits to the adjusted PR

PR model
Variable	Estimate	Standard error	t-Value	P-Value
Intercept	96.31	37.82	2.55	0.01
% Distal droplets (B)	0.37	0.09	4.26	<0.01
Boar stud entry date	−0.008	<0.01	−5.68	<0.01
Oviduct binding	0.03	<0.01	3.32	<0.01
% Normal head morphology (A)	0.47	0.11	4.20	<0.01
% Normal tail morphology (B)	0.47	0.17	2.79	<0.01
CASA–BCF (A)	0.50	0.14	3.49	<0.01
CASA–VCL (A)	0.04	0.01	2.62	0.01
Partial values for PR model
Variable	Partial R2	Model R2	Partial P-value
% Distal droplets (B)	0.15	0.15	<0.01
Boar stud entry date	0.15	0.3	<0.01
Oviduct binding	0.10	0.39	<0.01
% Normal head morphology (A)	0.08	0.47	<0.01
% Normal tail morphology (B)	0.05	0.52	<0.01
CASA–BCF (A)	0.05	0.57	<0.01
CASA–VCL (A)	0.03	0.60	0.01
PR model summary
R 2	Adj R2		P-Value
0.6	0.57		<0.001

Three ejaculates from 30 boars were used in homospermic matings and laboratory assays. PR was adjusted by PROC MIXED for farm, parity, number of doses, and the number of inseminations. Variables were selected using PROC GLMSELECT. Boar and sperm traits were regressed against the adjusted mean PR using PROC REG. Motility and morphology assessments were performed before (B) and after (A) shipments.

Those variables included in the final model to predict LS are in Table 4. Boars that entered the stud later had reduced average LS. Semen with a higher percentage of sperm with distal droplets before shipment and tail defects observed after shipment had reduced LS. Additionally, the percent of normal sperm after shipment had a significant impact on the LS model.

Table 4.

Multiple linear regression analysis to assess the relationship of sperm traits to the adjusted LS

LS Model
Variable	Estimate	Standard error	t-Value	P-Value
Intercept	28.08	8.38	3.35	<0.01
% Normal tail morphology (A)	0.04	0.01	2.84	<0.01
% Distal droplets (B)	0.11	0.03	4.37	<0.01
Boar stud entry date	−0.002	<0.01	−4.16	<0.01
% Total normal morphology (A)	0.02	<0.01	3.02	<0.01
[Total sperm per ejaculate]	0.005	<0.01	2.29	0.02
Partial values for LS model
Variable	Partial R2	Model R2	Partial P-Value
% Normal tail morphology (A)	0.29	0.29	<0.01
% Distal droplets (B)	0.14	0.43	<0.01
Boar stud entry date	0.08	0.50	<0.01
% Total normal morphology (A)	0.05	0.55	<0.01
[Total sperm per ejaculate]	0.03	0.58	0.02
LS model summary
R 2	Adj R2		P–Value
0.58	0.55		<0.001

Three ejaculates from 30 boars were used in homospermic matings laboratory assays. LS was adjusted by PROC MIXED for farm, parity, number of doses, and the number of inseminations. Variables were selected using PROC GLMSELECT. Boar and sperm traits were regressed against the adjusted mean LS using PROC REG. Motility and morphology assessments were performed before (B) and after (A) shipments.

Finally, the sperm assessments that had a significant impact on the FI, the product of PR and LS, are shown in Table 5. The percentage of sperm with tail problems and distal droplets before shipment and the percentage of normal cells were correlated with FI. The CASA motility trait VCL was also included in the final model to predict FI. Additionally, boars that entered the stud later also had reduced FI.

Table 5.

Multiple linear regression analysis to assess the relationship of sperm traits to the FI

FI model
Variable	Estimate	Standard error	t–Value	P-Value
Intercept	3,771.6	1,795.92	2.10	0.04
% Normal head morphology (A)	0.54	1.05	5.14	<0.01
Boar stud entry date	−0.36	0.07	−5.38	<0.01
% Distal droplets (B)	21.89	4.54	4.82	<0.01
CASA–VCL (A)	1.94	0.70	2.77	<0.01
Sperm volume/ejaculate	0.51	0.23	2.19	0.03
% Normal tail morphology (B)	17.34	8.33	2.08	0.04
Partial values for FI model
Variable	Partial R2	Model R2	Partial P-value
% Normal head morphology (A)	0.28	0.28	<0.01
Boar stud entry date	0.18	0.45	<0.01
% Distal droplets (B)	0.09	0.54	<0.01
CASA–VCL (A)	0.05	0.59	<0.01
Sperm volume per ejaculate	0.02	0.61	0.03
% Normal tail morphology (B)	0.02	0.63	0.04
FI model summary
R 2	Adj R2		P-Value
0.63	0.61		<0.001

Three ejaculates from 30 boars were used in homospermic matings and laboratory assays. FI was developed by multiplying the adjusted PR by the adjusted LS. Variables were selected using PROC GLMSELECT. Boar and sperm traits were regressed against the FI value using PROC REG. Motility and morphology assessments were performed before (B) and after (A) shipments.

Aside from laboratory assays, some information collected at the boar stud was correlated with fertility outcomes. Those variables with significance were the time at which boars entered the stud (in reality the age of the boar), the total number of sperm in the ejaculate, and the total volume of the ejaculate impacted the models. Semen from boars that were younger resulted in a lower FI value, a result of lower PR and smaller LS (Tables 3 to 5). Total sperm in the ejaculate was related to LS (Table 4). Total semen volume of the ejaculate influenced the FI (Table 5).

DISCUSSION

Predicting the outcome of an individual artificial insemination is a challenge for multiple reasons. First, males make up only half of the equation; successful fertilization also requires a suitable female at the proper stage of estrus (Amann et al., 2018). Second, there is significant variation among individual males and individual ejaculates from the same male (Windsor, 1997). Predicting fertility using laboratory analyses is challenging; commonly used observations like sperm motility and morphology do not identify lower fertility males consistently. Using a combination of different traditional semen analyses in conjunction with laboratory assessments of sperm functions should improve the ability to identify males with superior or poor fertility. An important functional assessment to consider is sperm interaction within the female reproductive tract, specifically the oviduct, where they are stored until fertilization (Holt and Fazeli, 2016).

Sperm interact with the extracellular matrix of the oviduct epithelium. Exposed glycans on this epithelium bind receptors on sperm to create a sperm reservoir that has been studied previously in multiple species (hamster: DeMott et al., 1995; horse: Dobrinski et al., 1996; bull: Lefebvre et al., 1997; pig: Green et al., 2001; Ignotz et al., 2001; Wagner et al., 2002; Kadirvel et al., 2012). A previous study (Kadirvel et al., 2012) screened 377 glycans and identified two glycan motifs that were common to all glycans that bound porcine sperm. The first was a biantennary 6-sialylated N-acetyllactosamine on a mannose core (bi-SiaLN) and the second was Lewis^X trisaccharide (Le^X). Both bi-SiaLN and Le^X motifs bound with high affinity to the sperm head prior to capacitation and were necessary for sperm binding to the porcine oviduct (Kadirvel et al., 2012; Machado et al., 2014). We hypothesized that inadequate sperm reservoir formation due to poorly functional glycan receptors on sperm could result in higher return-to-estrus rates and reduced LS.

To assess sperm adhesion to the oviduct reservoir, we allowed sperm to bind to oviduct cell aggregates and to soluble fluoresceinated glycans in vitro. Although time-consuming, sperm binding to oviduct cell aggregates offers insight into how sperm are interacting with intact oviduct cells. This in vitro assay is likely a good measure of the ability of sperm to form a reservoir in vivo because the oviductal cells maintain their polarity, resembling an epithelium (Thomas et al., 1997). We found that samples with few sperm that bound to oviduct cells in vitro had reduced PR, confirming previous studies (Waberski et al., 2005; Daigneault et al., 2015). Furthermore, the ability of sperm to bind to oviduct cells was retained in the final model to predict pregnancy and had the third highest partial R² value (0.10). Thus, oviduct cell binding provided additional ability to predict fertility beyond sperm motility and morphology.

Though the sperm-oviduct-binding assay was correlated with PR in this study, it is a time-consuming assay and is not feasible in most production settings. Assessing sperm binding to each of the specific oviduct glycans presented in a soluble fluorescent form is much simpler. Incubating sperm with individual glycans can identify whether the glycan receptors are present on the sperm head. These receptors could indicate the amount of success sperm would have forming the oviduct reservoir. However, binding to soluble glycans showed no correlation with oviduct cell binding. Furthermore, sperm binding to oviduct glycans had no correlation with PR, LS, or FI. Additionally, there were no significant differences among the individual boars in binding to the fluoresceinated glycans. This lack of variation in glycan binding could be due to the soluble nature of the glycans. The binding affinity required to tether a highly motile cell to an insoluble matrix found on an oviduct cell is higher than that to soluble glycans. These two assays have different affinity thresholds required to produce a signal.

Motility characteristics assessed by computerized semen analysis after semen shipments were related to PR and FI but not LS. Parameters collected after semen shipment included in the proposed model to predict PR were BCF and VCL. Curvilinear velocity is the average velocity of the sperm head through its actual path. BCF is the number of times the tail of an individual sperm crosses the average directional path. Overall, these parameters indicate how fast the sperm tail is beating, BCF, and how fast a sperm is traveling, VCL. The sperm that pass through barriers in the female reproductive tract to reach the oocytes are selected for, among other things, adequate motility (Suarez et al., 1991; Petrunkina et al., 2001). It has been proposed that fertilizing sperm requires adequate BCF and VCL to move through the oviduct mucus and barriers such as the utero-tubal junction (Broekhuijse et al., 2012), offering an explanation for the relationship to fertility. Although not related to fertility assessments, the percentage of motile sperm decreased in some samples considerably during shipment but changed little in others. The percentage of motile cells after shipment was not included in the final predictive model.

The percentage of abnormal sperm was included in the model to predict all three assessments of fertility. Our results are in agreement with previous work that samples with higher percentages of abnormal sperm had decreased binding to oviduct cells in vitro (Petrunkina et al., 2001). We observed a two-fold variation among the boars in the percentage of morphologically normal cells after shipment, illustrating the variation among the boars used in this experiment. The morphological abnormalities that were observed varied between boars and the standard errors for the boars were relatively small, indicating that, among the three replicates, the sperm consistently had poor morphology during the period of the experiment. Samples with a higher percentage of abnormal sperm are expected to have fewer sperm capable of being retained by the reservoir, resulting in lower PR and smaller LS.

Examination of data from individual boars was interesting. The oviduct selects for morphologically normal spermatozoa (Suarez et al., 1991; Petrunkina et al., 2001) and sperm interaction with oviduct cells is correlated with fertility (Waberski et al., 2005; Daigneault et al., 2015). Boar #25 had the lowest FI, which was a result of a combined lower PR and a small LS. Additionally, boar #25 had the lowest percentage of morphologically normal sperm and the second lowest percentage of motile cells after shipment, which together suggests that he should have the lowest oviduct cell-binding assay results. However, that was not the case. Sperm from boar #25 did not have lower binding of oviduct glycans and were average in their ability to bind to oviduct epithelial cells. This example supports the idea that other components have an important influence on sperm binding to oviduct cells. The identity of these components is unclear.

Laboratory assays that most closely mimic the challenges sperm encounter to fertilize oocytes are expected to be the most predictive of fertility. Assays using more viscous medium similar to what is in female genital tract fluid and requiring that sperm travel a longer distance are a couple of additional possibilities. Regardless, our results demonstrate that sperm binding to oviduct cells, sperm motility characteristics, and sperm morphology are related to boar fertility. The predictive models developed in this research can improve quality control at boar studs and reduce losses due to poor semen quality.