Follicular size predicts success in artificial insemination with frozen-thawed sperm in donkeys

In asses, semen collection, cryopreservation, and artificial insemination (AI) with frozen-thawed semen have been scarcely described and success rate, particularly following AI, is reportedly low. In the absence of reliable protocols, assisted reproductive technologies cannot support the conservation efforts aimed at endangered wild ass species and domestic donkey breeds. Two experiments were conducted in this study. In experiment 1 we evaluated freezing Abyssinian donkey (N = 5, 4 ejaculates each) spermatozoa using three freezing extenders (Berliner Cryomedium + glycerol, BC+G; BotuCrio, BOTU; INRAFreeze, INRA) and two cryopreservation techniques (liquid nitrogen vapour, LNV; directional freezing, DF). Post-thaw evaluation indicated that BOTU and INRA were similar and both superior to BC+G (P ≤ 0.004 for all motility tests), and that DF was superior to LNV (P < 0.002 for all evaluation parameters). In experiment 2, relying on these results, we used Abyssinian donkey sperm frozen in BOTU and INRA by DF for AI (N = 20). Prior to AI, thawed samples were diluted in corresponding centrifugation media or autologous seminal fluids at 1:1 ratio. No difference was found between BOTU and INRA or between the addition of seminal fluids or media, all resulting in ~50% pregnancy, and no differences were noted between males (N = 4). The size of pre-ovulatory follicle was a significant (P = 0.001) predictor for AI success with 9/10 pregnancies occurring when follicular size ranged between 33.1–37.4 mm, no pregnancy when it was smaller, and only one when larger. A number of ass species face the risk of extinction. Knowledge gained in this study on the Abyssinian donkey can be customised and transferred to its closely related endangered species and breeds.


Ethical considerations
The College of Veterinary Medicine and Agriculture (CVMA) of Addis Ababa University, Ethiopia, as a teaching, research, and community service rendering institution, has the mandate to house, breed, and experiment with animals. All activities involving animals (semen collection, artificial insemination, pregnancy evaluation) were performed by and under the strict supervision of veterinarians with long-standing experience, following approval of the project by the CVMA. These activities are non-invasive procedures performed routinely in domestic and often also in non-domestic species, are integrated into husbandry and breeding guidelines, and require no specific ethical permit. Privately owned jennies participated in the study only if their owners gave their full consent after being briefed in detail about the study, the specific procedures to be performed on their animals and the potential risks and benefits.

Animals
A total of 25 domestic Abyssinian donkeys participated in this study. Five jacks, aged between 7 and 10 years and with at least one foaling experience based on information provided by their seller, acted as semen donors. All jacks were housed at the stable of the CVMA. Twenty parous jennies, aged 3.5 to 20 years, were recruited for the study. Of these, three were owned by the CVMA and the rest privately owned by local farmers. The breeding history, reproductive soundness, general health and body condition of each jenny were taken into consideration during recruitment into this study. Age of the selected animals, when not known to the owner, was determined by dentition [45]. All animals participating in this study were of body condition 4.0-5.0 according to the established Pearson and Ouassat scoring system [46]. Animals at the CVMA were free to graze during the day and their diet was supplemented by hay and wheat bran. They had access to water ad libitum. Privately owned animals were kept at their owners' home, without any contact with males for the duration of the study. Animals had free access to water, were free to graze and their diet supplemented with straw and cereal by-products. All animals were in good body condition throughout the study.

Semen collection
Semen was collected by letting the jacks mount on an oestrous female and using Colorado model equine artificial vagina (Agtech, Inc., Manhattan, KS, USA) with disposable liner (Minitube, Tiefenbach, Germany) connected to a collection bottle lined with sterile Whirl-Pak 1 sampling bag (Minitube) and disposable filter (Minitube). Each jack was collected five times with 3-4 days rest between consecutive collections. The jacks had no previous training for the procedure. All jacks responded well and semen was normally collected within 20-30 min.

Semen evaluation
Samples were evaluated for total and progressive motility, vigour, volume, concentration, viability, and morphology. Evaluations were performed after collection, before centrifugation, and again after thawing. Motility was also assessed 3 h after thawing and incubation at 37˚C. Total ejaculate volume was assessed using a graduated tube. Sperm concentration was estimated using Neubauer haemocytometer. Total and progressive motility were evaluated by phase contrast microscopy at ×100 and × 200 magnifications. A drop of 10 μL of the sample was placed on a pre-warmed (37˚C) slide, covered with a pre-warmed cover slip and place on MiniTherm Stage Warmer (Hamilton Thorne, Beverly, MA, USA) set at 37˚C on the microscope for evaluation. The same experienced spermatologist evaluated all samples. Vigour was estimated on a scale of 0 to 5 (0 = immotile, 5 = highly active and vigorously motile). Sperm viability was assessed by staining 10 μL of suspended sperm with eosin-nigrosin (eosin Y yellow CI 45380, nigrosin CI 50420 dissolved in 0.9% NaCl; VWR International, Darmstadt, Germany) at a 1:1 ratio, incubating at room temperature for 2 min and then smears were prepared and air-dried before evaluation, which took place within two hours. At least 100 spermatozoa of each sample were evaluated with a light microscope (oil immersion; ×1000). White (unstained) sperm were classified as live and those that showed pink or red coloration were classified as dead. The same slides were also used to assess sperm morphology. Cells were classified as normal, or as having defect in the head, neck, midpiece, or endpiece. Evaluation of sperm morphology included search for a wide range of abnormalities as previously described [39].

Experimental design
The study was divided into two experiments that were implemented in a step-wise manner to ensure that the second builds on the previous step's results (Fig 1).

Experiment 1: Donkey sperm cryopreservation
This experiment was designed to evaluate the suitability of three freezing extender and two freezing techniques for cryopreservation of jack sperm.
Sperm cryopreservation. Of the 20 samples collected, one was deemed unsuitable for freezing (only 10% motility). Good quality samples were processed for freezing as follow: each sample was split into three parts, each diluted 1:1 in one of three different centrifugation media: Berliner Cryomedium (BC) [47], BotuSemen (Nidacon, Mölndal, Sweden), and INRA96 (IMV Technologies, L'Aigle cedex, France). Samples were then underlain with 2 mL of pre-warmed (37˚C) 60% iodixanol (OptiPrep™; Sigma Aldrich Chemie GmbH, Taufkirchen, Germany), an inert material used as a protective cushion against centrifugation damage [48], and centrifuged at 1000 g for 10 min. Following centrifugation, the supernatant and iodixanol were discarded and the pellet was re-suspended in one of three freezing extenders that were matched to the three centrifugation media: BC with glycerol (BC+G,~4.5% glycerol final concentration), BotuCrio (BOTU; Nidacon), or INRAFreeze (INRA; IMV Technologies) to a final concentration of 200 x 10 6 spermatozoa/mL. After dilution with the freezing extender, samples were cooled to 4˚C over 2 h at~0.3˚C/min by placing the tubes in an isothermal water bath inside a refrigerator. Samples from each extender were split into two parts and packaged into 0.5 mL plastic straws and HollowTubes™ (IMT Ltd., Ness Ziona, Israel). Two sizes of Hollow-Tubes™ were used: 8 mL for storage as insemination doses, and 2.5 mL for post-thaw evaluation. Straws were frozen by sustaining them at four cm above liquid nitrogen inside a freezing unit's metal box (Minitube) for 15 min. Straws were then plunged into the liquid nitrogen. HollowTubes™ were frozen in a multi-thermal gradient DF machine (MTG-550; IMT Ltd.) as described previously [34,38,49]. All frozen samples were stored under liquid nitrogen pending evaluation and use.
Thawing and post-thaw evaluation. Straws were thawed by submerging them in a water bath at 37˚C for 30 s. HollowTubes™ were thawed by first holding them in the air, at room temperature (24˚C) for 90 s and then plunging them into a water bath at 37˚C for 60 s. Samples were evaluated immediately after thawing for total and progressive motility, vigour, viability, and morphology, as described above. Remainder of the samples was split into two. One part was kept as is. The other part was diluted 1:1 in the respective centrifugation diluent (BC, INRA96, BotuSemen) to dilute out the cryoprotective agent(s). All samples were incubated at 37˚C and evaluated again for motility variables 3h after thawing.

Experiment 2: Artificial insemination
Based on the results of Experiment 1, only samples frozen by the DF technique with INRA and BOTU as the freezing extender were used in experiment 2. Jennies were randomly split into four equally sized groups for insemination with one of four combinations: i) INRA as is (mean age 11.8±5.8 y); ii) INRA diluted 1:1 with autologous seminal plasma (mean age: 8.8±1.9 y); iii) BOTU diluted 1:1 with BotuSemen (mean age 9.0±3.3 y); iv) BOTU diluted 1:1 with autologous seminal plasma (mean age: 8.3±7.1 y).
Seminal plasma collection. Semen was collected once from each of the five jacks for seminal plasma stockpiling. Collected semen was evaluated as described above and then centrifuged without iodixanol at 3000 g for 30 min. The supernatant was alliquoted and stored in insemination portions (8 mL) at -21˚C till use.
Oestrous induction. Each jenny was evaluated using an ultrasound scanner with a 5MHz linear array transducer (Aloka SSD-500V, Hitachi Aloka Medical, Ltd., Tokyo, Japan) to determine the ovarian status and ascertain they were not pregnant. Jennies were then given a 1.0-mL intramuscular injection of cloprostenol sodium (250 μg/mL Estrumate 1 , Schering-Plough Animal Health Corp, Germany). The jennies were then monitored by ultrasonography for follicular growth and development starting two days after injection and then on a daily basis when the largest follicle has reached the size of 20 mm until ovulation, recording the size of the pre-ovulatory follicle. Endometrial oedema was also monitored as a proxy of estimating approaching ovulation.
Artificial insemination. Cryopreserved sperm of four jacks from experiment 1 was used randomly to inseminate 20 jennies. Frozen quantity from the fifth male was insufficient for the multiple inseminations indicated by experimental needs. Only sperm with acceptable motility ! 30% was used for the AI procedures. A total of five jennies were assigned to each insemination group as described above. Samples frozen in 8-mL HollowTubes™ were thawed, evaluated for motility, and, if suitable, used for insemination. AI was performed only once ovulation was anticipated to occur within the next 24 h during ultrasonographic examination based on pre-ovulatory follicular morphology and endometrial oedema. Each jenny was inseminated twice. Once when the shape and size of the pre-ovulatory follicle, as viewed by ultrasonography, suggested approaching ovulation, and a second after ovulation, which was determined by the disappearance of the preovulatory follicle and the appearance of corpus haemorrhagicum in its place during ultrasonographic examination. In two cases ovulation took an extra day to occur so these jennies were inseminated a third time to make sure all jennies were inseminated twice in proximity to ovulation. Prior to each AI procedure, one insemination dose of 8-mL was thawed, evaluated, prepared with or without the isothermal diluents or seminal fluid, put into a pre-warmed 12 or 24-mL syringe (without plastic plunger), which was then fit onto the insemination catheter (Minitube). The insemination catheter was inserted deep into the uterus where the sperm was deposited. A small amount of air was then pushed through to expulse remaining sperm and to prevent retrograde flux.
Pregnancy evaluation. Pregnancy was diagnosed between days 16 and 19 after the second (or third) AI procedure, using ultrasound scanner with a 5MHz linear array transducer. Females were considered pregnant when an embryonic vesicle or an early conceptus was demonstrated.

Statistical analysis
Statistical analysis was performed using PASW Statistics software for Windows, version 18.0.0 (Formerly SPSS Statistics; IBM Inc., Chicago, IL, USA). Differences in sperm quality parameters were tested using the Multivariate test within the General Linear Model, where the extender was considered fixed factor. For comparison between the two freezing techniques the paired-sample t-test was used. Comparison between jacks in fresh semen parameters was done using One Way ANOVA. When values were not normally distributed based on Levene Test of Homogeneity of Variances, values were SQRT transformed. The Fisher exact test and the Wilcoxon's Signed Rank test were used to evaluate differences between groups in pregnancy rate. Differences were considered significant when P < 0.05.

Results
A total of 25 semen collection procedures were performed, five per jack. Despite the fact that the jacks were not collected before the start of the study to deplete old semen from the system, only one sample was deemed unsuitable for cryopreservation (only 10% motility). One sample per jack (last collection) was used for seminal plasma extraction. All other samples were of very good quality ( Table 1, S1 Data) and were used in the experiments.

Experiment 1: Donkey sperm cryopreservation
Semen was collected successfully from all males on all occasions. One sample was of too low motility to justify freezing. Nineteen samples were processed and frozen. Generally speaking, freezing outcome was fair to good with all samples having post thaw motility of 30% or more in at least one extender/freezing technique.
Comparison between the two freezing techniques revealed that freezing in large-volume HollowTubes 1 by the directional freezing technique was superior to freezing in straws over liquid nitrogen (Table 2). This was true when the two freezing techniques were compared for each evaluation parameter within each freezing extender and it held true also when data of all freezing extenders were combined ( Table 2, S2 Data). The exception to this was the percentage of spermatozoa with normal morphology in which samples frozen in straws were superior to those frozen in tubes when compared for BOTU, INRA, or all samples combined, but not for BC+G.
Comparison (Table 3, S2 Data) showed that there was a statistically significant difference between the three freezing extenders based on post-thaw evaluations [F (12, 194) = 27.074, P < 0.0005; Wilk's Λ = 0.140]. Comparison between individual extenders by Post Hoc tests indicated that INRA and BOTU were superior to BC+G in total, progressive and 3 h motility and in vigour (P < 0.004 for all). In vigour, INRA was also superior to BOTU (P < 0.0005). INRA and BOTU did not differ in total and progressive motility and all three extenders showed similar results when viability and morphology were compared.
As results indicated that BC+G was inferior to the other two extenders, in three ejaculates an attempt was made to reduce the final concentration of glycerol in BC+G from~4.5% to

Experiment 2: Artificial insemination
Of the 20 jennies inseminated in this study, nine were confirmed pregnant and in one the ultrasonographic examination suggested lost pregnancy (S3 Data). This last female had matured corpus luteum and uterine fluids characteristic for lost pregnancy. We can thus say that all over conception rate was 50%. Of these, five jennies gave birth to five healthy foals (four males and a female; Fig 2). Three females were lost to follow-up (owners moved away or jennies sold), and one has aborted after about two months of pregnancy. Results were analysed in search for possible combinations that resulted in higher pregnancy rate. However, we found no difference between INRA and BOTU, or between insemination doses that were diluted with seminal plasma and those that were not. There was also no effect to the level of sperm motility after thawing (range between 30 and 50%) and no effect to the jennie's age. When comparing between males, one male stood out as contributing to pregnancy rate lower than the total average (2/7 = 28.6%). Further analysis of the data found that when the pre-ovulatory follicular size was between 33.1 and 37.4 mm, nine of the ten females in this sub-group conceived (90%) whereas only one in the remaining ten jennies conceived. The difference between these two sub-populations was significant (two-tailed Fisher's exact text, P = 0.001). When splitting the jennies based on preovulatory follicular size, we found that the male with the low pregnancy rate happened to have inseminated mainly (5/7) females with follicles larger than 38.5 mm.

Discussion
"Sancho hastened to his Dapple, and embracing him he said, 'How hast thou fared, my blessing, Dapple of my eyes, my comrade?' all the while kissing him and caressing him as if he were a human being. The ass held his peace, and let himself be kissed and caressed by Sancho without answering a single word" [50].
Donkeys have accompanied humans for thousands of years, providing them with a variety of services. Similar to dogs, this human-donkey interaction resulted in a very wide variety of breeds. In recent decades, however, the use of donkeys has drastically dropped and with it came the dwindling of populations, bringing many of these breeds to the verge of extinction or beyond. Protecting this cultural heritage is of paramount importance. Furthermore, in the wild, of the three extant donkey species, two are also under severe threat of extinction. This dire state of donkeys needs to be halted and hopefully reversed. To this end, the present study was devised, targeting a long-standing hurdle faced by assisted reproduction in donkeys. Whereas freezing donkey sperm seem to be working well, using frozen-thawed donkey sperm to inseminate jennies result in consistently low conception rates, ranging between zero and about 36% [12,13,19,51] with just one report showing higher conception rate of about 62% [26]. Despite a number of studies on the topic over the last 20 years or so, it remains unknown why frozen donkey sperm does not succeed in fertilizing the jennies. It has been hypothesised that the culprit is with the glycerol in the freezing extender. One line of thought suggested that glycerol is toxic to donkeys' sperm [12] while another suggested that it is the jennie's uterine environment that is sensitive to the compound [13]. The latter study even showed that pregnancy rate was higher in mares inseminated with frozen-thawed donkey sperm than it was in jennies, further supporting the theory that the jennies' sensitivity to glycerol stand behind the low pregnancy rates observed in donkeys. If this explanation was correct, however, one would have expected proper conception rates when other cryoprotectants were used. Yet, this was not really the case [19]. Freezing with glycerol but removing it prior to insemination did improve things in a study on the Poitou donkey but pregnancy rate was zero in its presence making comparison questionable [12]. In our study we found no effect to the dilution out of the glycerol with autologous seminal plasma (INRA) or using glycerol-free extender (BOTU). All these suggest that glycerol might not be the factor affecting conception rate in jennies inseminated with frozen-thawed donkey sperm. An alternative explanation for the better success rate achieved thus far when frozen-thawed donkey sperm was used to inseminate mares is because the knowledge of the mare reproductive biology, cycle and follicular growth dynamics, and when is the ideal insemination window are better known compared to those of the jenny and so inseminations in those studies possibly took place at a more suitable time. It was previously reported in other donkey breeds (Nordestina or other undetermined breed) that seasonality had no effect on the jack's sexual behaviour [52] or ejaculate characteristics [53]. It would help to find out if the same applies to the Abyssinian donkey jacks in Ethiopia, as this information will indicate if there is optimal time of the year for semen collection and preservation. The same studies, while noting difference between jacks, found no difference between the two weekly semen collections. In our study we noted variations between jacks only in ejaculate concentration. In some jacks, between-ejaculates differences were apparent but we noted no relation between collection sequence and ejaculate parameters. In some species, e.g. llama [54] or pig [55], addition of seminal plasma to the thawed sperm before insemination seems to improve fertilisation rate. In donkeys, the addition of seminal plasma to post-thaw sperm showed no improvement in in vitro sperm characteristics [27] and, as shown in our study, had no effect on pregnancy rate, when it was added to the thawed sample before insemination. This is yet another demonstration of the fact that species are different from each other and thus require species-specific customisation of protocols.
In an attempt to resolve the issue at hand, we tried to use different freezing extenders, containing different cryoprotectants, and to use the directional freezing-a freezing technique never tested in donkeys. Looking at the results, however, nothing of all these, or the males themselves, seems to have any effect on conception rate. With 50%, conception rate was at the higher end of what thus far has been achieved in donkeys and on par with what is customary in horses [56,57] or the highly-managed dairy and beef cattle industry (e.g. [58]). Considering the fact that most of the jennies in this study were working animals with less-than-ideal nutrition, 50% pregnancy rate is a great outcome. We did, however, hope for more. It was only when we looked at the association between conception rate and pre-ovulatory follicular size that we got a clue as to how pregnancy rates in donkeys inseminated with frozen-thawed donkey sperm may be improved. The association between pre-ovulatory follicular size and conception rate has been reported before in other species [59,60] and it was suggested that when ovulation is induced, some immature follicles are ovulated, leading to lower pregnancy rates [59]. Still, the general association seems to be one that binds between follicular size, follicular maturation, and pregnancy rate. In other words, the larger the follicle, the higher pregnancy rate in induced ovulation cycles [60,61]. In our study ovulation was not induced. All jennies ovulated naturally. It could certainly be a matter of numbers as 20 jennies are not a large enough sample to provide a full picture of the population, but it could also be a species-specific character. Follicles smaller than 33.1 mm may have ovulated before they had properly matured and were thus less likely to result in pregnancy. Why follicles larger than the higher end of the range (37.4 mm), with the exception of the largest follicle observed in this study (41.3 mm), did not result in pregnancy is not known. Based on our previous [62,63] and the present study's results, the Ethiopian long (June to September) and short (March to May) rainy seasons seem to be the right mating seasons for these jennies when average pre-ovulatory follicles were measured at 33.2±1.2 mm and 37.8±2.1 mm, respectively. During the dry season (October to February), when conditions are tough, the pre-ovulatory follicles are smaller (31.0±1.8) thus appear to be rendering the jennies with lower probability of getting pregnant.
It has been suggested that when ovulation is induced, supporting hormones, such as estradiol, do not reach their optimal levels [59]. This, however, was not the case in our study, as we relied on natural ovulation. Inducing the cycle may have its affect on the hormonal status of the females and thus on follicular development and maturation. A future study, evaluating the hormonal status of the jennies before, at, and after ovulation and insemination, may provide the answer. To make sure that all jennies were inseminated in proximity of ovulation, two jennies in our study were inseminated three times-twice,~24h apart, before ovulation and once after ovulation as opposed to once before and once after in all other jennies. If this had an effect on the success rate we do not know but one should keep this in mind when interpreting the results of our study.
This study provides a further confirmation to the superiority of the directional freezing technique over freezing in straws in nitrogen vapour. We and others have shown this to be the case in a wide variety of species including domestic species such as bovine bull [32] and stallion [34], and non-domestic species such as rhinoceros [44], dolphin [42], killer whale [64], and most recently also in the onager [29]. The most plausible explanation for this outcome is the different way ice crystals grow within the container in these two techniques-stochastic with possible recrystallisation in straws over liquid nitrogen vs. highly organised and controlled ice crystal morphology and efficient heat dissipation in tubes frozen by the directional freezing technique [65]. The result is a much better protection of the delicate cells inside the solution when directional freezing is used.
In conclusion, in this study we have demonstrated that when freezing donkey sperm the directional freezing was superior to freezing in straws over liquid nitrogen, and that INRA and BOTU were the freezing extenders of choice. Pre-ovulatory follicular size was found to be a reliable predictor for successful AI. The protocols developed through this project will contribute to a long-standing deficiency in wildlife management, namely the ability to artificially inseminate endangered donkey species and breeds, including the Senar donkey in Amhara Regional State in Ethiopia. With this knowledge in hand, the captive populations and isolated wild populations will gain access to unrepresented genetic material, thus improving their resilience and sustainability in face of inbreeding and its negative consequences. It would, however, be beneficial to conduct a large-scale study to reconfirm the preliminary findings of this study and their applicability to the donkey population.