To introduce loss-of-function mutations in swine KISS1, a single guide RNA (sgRNA) was designed to target the second exon of this gene, upstream of the sequence encoding the highly conserved KISS1 10-amino acid (aa) core (Figure 1A). The designed sgRNA and Cas9 proteins were co-injected into porcine zygotes with two HDR templates, which separately contained a stop codon (HDR^SC) and a silent blocking mutation (HDR^BM). These single-stranded donor oligonucleotides (ssODNs) were paired to increase the odds of generating pigs with edits capable of disrupting KISS1 (HDR^SC) and/or maintaining KISS1 function (HDR^BM; Figures 1B, C). A total of 684 zygotes were injected and transferred into 19 recipient sows. After full-term pregnancies, 69 liveborn (35 males and 34 females) and five stillborn piglets were farrowed in nine litters (Supplementary Table S1).

Analyses of DNA sequences from tail tissues revealed that 53 (72%) piglets were edited, and 18 of these (24%) bore HDR-mediated edits (Table 1). At least 22 pigs bore multiple (>2) alleles. In pigs with two different alleles, 24 pigs had one wild type (WT) allele while four animals had two different editing-derived alleles. AmpSeq indicated that most of the pigs bearing two alleles were mosaic instead of heterozygous (monoallelic edit) or compound heterozygous (different biallelic edits). Three pigs (4MD, 71MY, and 73SB) presented identical biallelic edits, of which there was one pig for each HDR-intended mutation (Figure 1D).

Detected indels ranged in size from 1 to 68 bp for insertions and 1 to 211 bp for deletions (Supplementary Figure S1), whereas larger indel alleles would not be detected by our AmpSeq assay. Predicted translations indicate that five and 35 alleles contained non-sense and frameshift mutations, respectively. These alleles were predicted to disrupt the KISS1 aa core (Supplementary Figure S2) and were carried by 43 pigs at different frequencies. The percentage of reads attributed to gene-editing-derived alleles in a pig’s genotype was defined as editing percent. In all, 27 animals were generated with KISS1-disruptive alleles between 30% and 90%, with seven pigs having a mosaicism level greater than 90% (Figure 1D).

Four pigs (two boars, two gilts) displayed phenotypic characteristics consistent with hypogonadism. Testicular weight and size were dramatically reduced in two boars (51MY and 54MY) at 8 ± 0.6 months of age compared to control animals (Figures 2A–C; Supplementary Table S2). KISS1 WT alleles were absent in both of these two boars. In all other boars, typical variation in testicular volume throughout development was observed, and the variation was not related to the extent to which pig’s KISS1 was disrupted (Figure 2A). The KISS1 KO gilts had KISS1-disruptive allele frequencies of 96% and 100% (49FY and 32FY, respectively) with considerably smaller ovaries and visual absence of surface follicles. The uteri of these KISS1 KO gilts were underdeveloped when compared to age-matched controls (Figures 2D–F; Supplementary Table S3). These findings alongside behavioral observations and hormonal data (Supplementary Figure S3) suggested that the KISS1 KO pigs failed to become pubertal. The other KISS1-edited and WT pigs were considered sexually mature as they displayed genital sizes and external characteristics typical of boars and gilts that advanced through puberty, which is consistent with the hormonal profiles of these animals. Additionally, mosaic F0 pigs that were mated exhibited sexual behaviors, gamete production, and fertility.

Piglet birth weight was unaffected by sex, KISS1-disruption group, and their interaction (p ≥ .55). The main effect of age was the only fixed effect to influence body weight (p < .001), but average daily gain was influenced by age and piglet birth weight (p < .001). There was no significant interactive effect of age × KISS1-disruption group for body weight or average daily gain (p ≥ .33; Figures 2G, H). Body weight was lower (p = .04) in WT than 5%–90% KISS1-disruptive pigs at 160 days, but no KISS1-disruption groups differed in body weight or average daily gain at any age category evaluated (p ≥ .07). In this study, the pigs considered as WT were those that carried no KISS1-disruptive alleles, see descriptive statistics in Supplementary Table S4. Boar 15MD was classified in the group >90% KISS1-disruption based on initial sequencing, but his phenotype was remarkably similar to an unedited or partially mosaic boar (<90%). Resequencing revealed him to have 88% of KISS1-disruptive allele frequency instead of 95% as originally determined. Reclassification of this boar did not alter statistical significances or interpretation of the results. It cannot be ruled out that similar cases may have occurred. Nevertheless, no further inconsistencies were observed between genotype and expected phenotype aside from pig 67FY. This gilt had a KISS1-disruptive allele frequency of 97% but had no phenotypic abnormalities.

KISS1 KO and WT boars presented an architecture of seminiferous tubules that looked similar. The main difference between these two groups of 8-month-old pigs was related to the size of the observed structures, which were smaller in kisspeptin deficient boars. In the KISS1 KO group, there were no visible lumina and Sertoli cells were arrayed along the parietal edge of these small seminiferous tubules. Germ cells were located both in the adluminal area and toward the edge of the tubules while no spermatids were observed. In the WT boar, the germ cells had all homed to the basement membrane of the germinal epithelium, with none in the seminiferous tubule lumen; germ cells and Sertoli cells were mixed lining the tubule. DAPI staining revealed the presence of cells across the germinal epithelium and adluminal area only in the WT boar, and some sperm tails and elongated spermatids were observed. This indicates that, in this animal, germ cells were undergoing differentiation to eventually become spermatozoa (Figure 3).

Hormone measurements were conducted every 4 weeks beginning when pigs were 40 days old and halved to a fortnight frequency from 130 to 280 days of age. Only follicle-stimulating hormone (FSH) of gilts showed an interaction of age with KISS1-disruptive editing percent (p = .05), but least-squares means of this interactive effect are displayed for all hormones in Figure 4 to describe patterns over time. Gilt FSH was lower for the >90% KISS1-disruption group than all others at 40 and 70 days of age (p ≤ .05; Figure 4B), but no differences between groups were observed thereafter (p ≥ .20). Serum concentrations of boar FSH were not affected by KISS1-disruptive editing percent (p = .39; Figure 4A). All hormones, except for boar FSH (p = .07), were influenced by the main effect of age (p < .001). Overall, boars with KISS1-disruptive allele frequency >90% had less (p < .001) serum luteinizing hormone (LH) and testosterone during all ages of development compared with other groups, which did not differ from one another (p ≥ .91; Figures 4C, E). Concentrations of LH in gilts with KISS1-disruption >90% were numerically less than in other groups, but this did not reach significance (p = .07; Figure 4D). Individual concentrations of serum hormones for some pigs belonging to the group with KISS1-disruptive allele frequency >90% are shown in Supplementary Figure S3.

Intergenerational transmission of the KISS1-mediated castration free trait is required to provide a more effective alternative to end surgical castration. Semen from five F0 boars (13MY, 36MY, 61MY, 53MY, and 15MD) with KISS1-disruptive allele frequencies between 35% and 88%, was collected for analysis (Supplementary Table S5). KISS1 gene edits in sperm DNA were evaluated by NGS and compared to results obtained from the tail samples. Each mutant allele detected in the tail samples could be found in the sperm DNA, confirming germline transmission. However, apart from 13MY and 61MY, there was discordance ranging from 16% to 43% between the allele frequencies observed in the tail versus those in sperm DNA (Figure 5).

Three F0 boars and 10 F0 gilts were selected for mating based highest levels of KISS1 disruption under 90 percent and having the lowest inbreeding coefficient. From the first matings, seven litters were farrowed (Figure 5) with an average of 12.9 ± 1.6 liveborn piglets per litter, for a total of 90 liveborn F1 piglets. The genotype distribution was, 13 KISS1 ^−/−, 29 KISS1 ^+/− and 40 KISS1 ^+/+; 44 males and 38 females. For second matings, five and one of the previously chosen F0 gilts and boars, respectively, were bred again. Eighty liveborn F1 piglets were farrowed through five litters (35 males and 30 females; average of 16 ± 1.7 piglets per litter), from which 21 piglets were KISS1 ^−/−, 31 KISS1 ^+/− and 13 KISS1 ^+/+ (Table 2). In nine piglets farrowed in four litters (from 47FY and 65FY), only the paternal alleles were detected (Figures 1D, 5), suggesting that these two mothers harbored alleles with large deletions that could not be detected with the implemented PCR test. Inherited alleles of genotyped F1 piglets are shown in Figure 5 according to the litter.

All animal procedures followed institutional guidelines for the care and use of animals and were approved by the Recombinetics Institutional Animal Care and Use Committee (IACUC; approval number: RCI-1903-03A-A3).

The F0 pigs were of Duroc or American Yorkshire breeding, derived from commercial herds, and farrowed from first and second parity surrogate sows. The F0 boars and gilts chosen for breeding resulted in both purebred and crossbred animals. An industry standard fortified corn-soybean meal diet formulated to meet and exceed nutrient requirements was provided with ad libitum access to water.

Sus scrofa KISS1 possesses a genomic structure of two exons that encode 138 aa. Amino acids 111–120 form a 10 aa core conserved across vertebrates that is essential for binding and activation of KISS1R. A sgRNA was designed to target a KISS1 exonic region preceding the sequence encoding this completely conserved aa core motif. The online tools Gene Sculpt Suite MENTHU (Ata et al., 2018) and microhomology predictor (Bae et al., 2014) were used to design the sgRNA with the sequence of 5′-GGT​CCC​CCG​AGG​GTT​CGC​CTCGG-3′ (PAM is underlined). crRNA and tracrRNA (IDT, IA, United States) were resuspended in injection buffer (5 mM Tris-HCl, 0.1 mM EDTA, pH 7.5) and incubated with HiFi Cas9 protein (IDT, IA, United States) to form RNP complexes. The assembled RNPs were co-injected in fertilized porcine zygotes with two ssODNs of 90 bp (IDT, IA, United States) that served as templates for HDR-mediated editing (Figure 1B). The first ssODN, HDR^SC, introduces an 8-bp insertion to generate a premature termination codon followed by the HindIII restriction site. The second ssODN, HDR^BM, produces two silent mutations at and close to the PAM site that reduce Cas9 recutting after HDR^BM editing while also introducing the restriction site AcuI.

Single-cell embryos for microinjection were produced in vivo by synchronizing the ovulation of 71 sows with 200 μg of the GnRH agonist triptorelin acetate (OvuGel; JBS United Animal Health, IN, United States) 96 h (h) post-weaning. At 22 ± 2 h after triptorelin acetate treatment, sows were artificially inseminated with extended boar semen. The following day, sows were transferred to the slaughterhouse at the Meat Science Laboratory at the University of Minnesota, where the reproductive tracts were collected. The zygotes were recovered by flushing the oviduct with phosphate-buffered saline (1X PBS; Corning, NY, United States). A total of 836 in vivo fertilized zygotes were obtained. Single-cell embryos were immediately transferred into TCM-199 medium (Gibco, MD, United States) and 3–5 h after zygote recovery, 684 one-cell stage embryos were injected. Zygote microinjection was done into the cytoplasm with sgRNA (25 ng/μl)/Cas9 (50 ng/μl) RNPs and ssODNs (HDR^SC: 33.3 ng/μl; HDR^BM: 66.7 ng/μl) using an inverted microscope (Nikon, Tokyo, Japan), which was equipped with micromanipulators (Narishige, Tokyo, Japan) and an electronic microinjection system (Femtojet; Eppendorf, Hamburg, Germany). Injected zygotes were cultured for 12–16 h in PZM-3 medium (Yoshioka et al., 2002) covered with mineral oil, at 38°C in a humidity-controlled atmosphere containing 5% CO₂ and 5% O₂.

Within approximately 20–24 h from zygote collection, an average of 36 microinjected zygotes were surgically transferred into oviducts of 19 hormonally synchronized surrogate sows. To synchronize the estrous of recipient sows, they were treated with 18 mg/day of altrenogest (MATRIX; Merck & Co., NJ, United States) for 18 days, plus 1,000 IU of human chorionic gonadotropin (hCG; Chorulon, Merck & Co., NJ, United States) on day 19, and embryo transfers were carried out 48 h later. Pregnancies were confirmed and monitored on days 30 and 60 after the embryo transfers using an Aloka 500 Ultrasound Scanner (Aloka Co. Ltd., CT, United States). After 115–118 days of gestation, 10 recipient females farrowed naturally, and piglets were genotyped.

Body weights of all F0 pigs were individually recorded within 48 h after birth (birth weight) and at 40, 70, 100, 130, and 160 days of age ±4 days. The growth rate was calculated as the average daily gain of weight for the time periods birth-40, 40–70, 70–100, 100–130, and 130–160 days of age. The total average daily gain of weight was estimated from birth to 160 days of age. Beginning at 40 days of age and continuing, at each time body weight was collected, the length and width of both testicles were individually measured with a vernier caliper (35-OD8; iGaging, CA, United States).

Testicular volume was estimated using the equation for a prolate spheroidal shape (Young et al., 1986): T V = 4 / 3 ⋅ π ⋅ a ⋅ b 2 ⋅ 2 , where TV is testis volume (cm³), π is 3.14, a is ½ length testis (cm), and b is ½ width testis (cm). This equation was calculated by measuring the width and length of both testes, besides subtracting the skinfold thickness (Sinclair et al., 2001; Jacyno et al., 2015; Han et al., 2017). A Harpenden skinfold caliper was employed for measuring the skinfold thickness (2 layers of scrotal skin) (Jacyno et al., 2015).

Blood samples were collected from F0 pigs monthly starting at 40 days of age and biweekly from 130 to 280 days of age by jugular venipuncture into serum-separating tubes. Serum was obtained by centrifugation at 1,500 × g for 15 min (min) at 4°C, aliquoted, and stored at −80°C until analysis. Serum concentrations of LH (Kesner et al., 1987), FSH (Trout et al., 1992), testosterone (Ford et al., 2001), and progesterone (Calderón et al., 2017) were quantified with validated radioimmunoassay. Testosterone and progesterone assays were obtained from MP Biomedicals (Irvine, CA, United States). The reference standards for LH (AFP-10506A) and FSH (AFP-10640B) were provided by Dr. A. F. Parlow (Scientific Director of the NIH, NIDDK, National Hormone and Peptide Program, Torrance, CA, United States). Pools of porcine serum were included in each assay. For LH, concentrations of pools measured .81, 1.54, and 8.41 ng/ml with an average intra- and inter-assay coefficient of variation (CV) of 6.7% and 10.8%, respectively (n = 8 assays). For FSH, serum concentrations of pools ranged from 2.0 to 6.8 ng/ml with average intra- and inter-assay CV of 6.3% and 7.3%, respectively (n = 6 assays). For progesterone, a serum pool (midluteal phase) measuring 24.2 ng/ml had an intra- and inter-assay CV of 5.7% and 8.6%, respectively (n = 4 assays). Intra- and inter-assay CV were 10.7% and 13.8% (n = 6 assays), respectively, for a pool of boar serum with testosterone concentration of 5.3 pg/ml.

Pigs were humanely euthanized according to established guidelines (American Veterinary Medical Association, 2020) for tissue collection. The length and width of ovaries and testes were individually measured with a vernier caliper, and the number of ovarian follicles was counted. Ovaries and testes were individually weighed. External and internal genitalia were observed for visible abnormalities. The number of pigs that were evaluated postmortem as well as their minimum and maximum ages at this point, for each KISS1-disruption group, are presented in Supplementary Tables S2, S3.

Testes from F0 boars at 255–262 days of age were collected, bivalved, cut through the longest dimension, and 0.5 mm × 0.5 mm pieces containing sections of mediastinum and periphery were collected. These tissues were placed into 15 ml conical tubes with 10% Neutral Buffered Formalin (89370-94; VWR, PA, United States), and maintained at room temperature for 18 h. Tissues were then washed three times in PBS with rocking for 10 min. Samples were placed in 70% ethanol (V1001TP; Decon Laboratories Inc., PA, United States) and stored at room temperature until submission to Scientific Solutions (MN, United States) for paraffin embedding. Tissue sections of 20 microns were mounted on slides (Globe Scientific Inc., NJ, United States) and deparaffinized with Xylene (89370-008; VWR, PA, United States) followed by rehydration with increasing concentrations of 70%–100% ethanol. Antigens were retrieved by boiling the tissue in a microwave for 15 min with a citrate-based unmasking solution (1:100; H-3300; Vector Laboratories Inc., CA, United States). Membrane permeabilization was accomplished by incubating tissue sections in PBS containing 0.1% Triton X-100 (Sigma-Aldrich, MO, United States) for 10 min at room temperature. To block the binding of non-specific proteins, tissue sections were incubated with 10% normal goat serum (Sigma-Aldrich, MO, United States) in PBS containing 0.025% tween-20 (PBST; Bio-Rad, CA, United States) overnight at 4°C. Washes with PBS were done twice for 10 min after blocking, primary antibody stain, secondary stain, and DAPI staining.

Tissue sections were incubated overnight at 4°C with primary antibodies diluted to manufacturer recommendations with 5% normal goat serum-PBST. Vimentin (1:500 dilution; Abcam, Cambridge, UK) and GATA binding protein 4 (GATA4; 1:500 dilution; Cell Signaling Technology, MA, United States) were used as markers for Sertoli cells. Deleted in azoospermia-like protein (DAZL; 1:500 dilution; Abcam, Cambridge, UK) and protein gene product 9.5 (PGP9.5; 1:1,000 dilution; Abcam, Cambridge, UK) were used as germ cells markers. Tissue sections were incubated for 1 h at room temperature in secondary antibodies diluted 1:400 in 5% normal goat serum-PBST. These antibodies were Alexa Fluor 488 (anti-rabbit) or Alexa Fluor 594 (anti-mouse; Invitrogen, MA, United States). Details of antibodies used are in Supplementary Table S6. Nuclei were stained with Hoechst solution (Thermo Fisher Scientific Inc., MA, United States) diluted 1:10,000 in PBS for 15 min. Tissue sections were washed with PBS (5 min, room temperature) and coverslipped using ImmuMount (Fisher Scientific, MA, United States), followed by incubation at room temperature for 30–120 min. A Leica DM6000B microscope equipped with a Leica DFC7000 T camera (Leica Microsystems, Wetzlar, Germany) was used to view and image slides.

Descriptive statistics of all anatomical-related phenotypes and statistical analyses of body weight, average daily gain, and hormone concentration were estimated using SAS 9.4 (SAS Institute Inc., NC, United States). The WT group always consisted of all pigs that harbored 0% of KISS1-disruptive alleles, including individuals carrying KISS1-non-disruptive alleles at different frequencies as these did not differ phenotypically from unedited pigs.

F0 boars that were candidates to be bred were trained on a semen-collection dummy. When mosaic boars were between 272 and 329 days of age, semen was collected via the hand-gloved method. Ejaculates were quantified, samples were sent for NGS-mediated genotyping as outlined above, and replicate semen was extended for evaluation of sperm motility, morphology, and concentration with a light microscope (CxL; Labomed, CA, United States).

In mosaic gilts that were candidates to be mated due to their genotypes, estrus was synchronized using altrenogest for at least 14 days. Gilts that displayed classical signs of estrus were artificially inseminated with semen of F0 boars bearing KISS1-disruptive alleles. First matings by artificial insemination occurred when gilts were between 275 and 355 days of age and boars were 317 and 348 days old. A second mating was performed 9 months later. Semen collections and pregnancy diagnoses were done as previously mentioned.