Generation of cattle knockout for galactose‐α1,3‐galactose and N‐glycolylneuraminic acid antigens

Abstract Two well‐characterized carbohydrate epitopes are absent in humans but present in other mammals. These are galactose‐α1,3‐galactose (αGal) and N‐glycolylneuraminic acid (Neu5Gc) which are introduced by the activities of two enzymes including α(1,3) galactosyltransferase (encoded by the GGTA1 gene) and CMP‐Neu5Gc hydroxylase (encoded by the CMAH gene) that are inactive in humans but present in cattle. Hence, bovine‐derived products are antigenic in humans who receive bioprosthetic heart valves (BHVs) or those that suffer from red meat syndrome. Using programmable nucleases, we disrupted (knockout, KO) GGTA1 and CMAH genes encoding for the enzymes that catalyse the synthesis of αGal and Neu5Gc, respectively, in both male and female bovine fibroblasts. The KO in clonally selected fibroblasts was detected by polymerase chain reaction (PCR) and confirmed by Sanger sequencing. Selected fibroblasts colonies were used for somatic cell nuclear transfer (SCNT) to produce cloned embryos that were implanted in surrogate recipient heifers. Fifty‐three embryos were implanted in 33 recipients heifers; 3 pregnancies were carried to term and delivered 3 live calves. Primary cell cultures were established from the 3 calves and following molecular analyses confirmed the genetic deletions. FACS analysis showed the double‐KO phenotype for both antigens confirming the mutated genotypes. Availability of such cattle double‐KO model lacking both αGal and Neu5Gc offers a unique opportunity to study the functionality of BHV manufactured with tissues of potentially lower immunogenicity, as well as a possible new clinical approaches to help patients with red meat allergy syndrome due to the presence of these xenoantigens in the diet.


| INTRODUC TI ON
Two well-characterized antigens are absent in humans but present in mammals and include galactose-α1,3-galactose (αGal) and N-glycolylneuraminic acid (Neu5Gc) whose synthesis are catalysed by α(1,3) galactosyltransferase (encoded by the GGTA1 gene) 1,2 and CMP-Neu5Gc hydroxylase (encoded by the CMAH gene) 3-5 respectively. These have been identified as major antigens in xenotransplantation studies or retrospective clinical findings 3 . Pigs that carry mutations in both genes, and therefore lack these xenoantigens, have been generated. 6 Moreover, porcine kidneys lacking αGal are not hyperacutely rejected. 7 It is also expected that such tissues will be less immunogenic for patients being implanted with animal-derived tissues engineered to lack both antigens.
One of the major clinical applications of xenogenic tissues is for the manufacturing of bioprosthetic heart valves (BHVs), and it has been shown that such tissues carry the same xenoantigens despite the glutaraldehyde treatments used in the manufacturing process 8,9 .
Almost 300 000 patients are now undergoing BHV replacement each year 10 with a growing demand. The sources of BHV are those manufactured from pig or bovine pericardia as compared to mechanical heart valve (MHV) that require lifelong anticoagulation therapy.
Bovine BHVs suffer however premature structural valve degeneration (SVD). The functionality of BHV is maintained for 10-15 years in older patients. However, in younger (<35 years old) patients, BHVs undergo SVD much earlier. 11 It is hypothesized that among various metabolic causes, SVD is also immune-mediated since both αGal 8,12 and Neu5Gc 9,13 are still present on the BHV used in the clinic.
After BHV replacement, there is an increase of anti-αGal antibodies 14, 15 and it has been reported in an experimental context that implantation of BHV from αGal-knockout pigs into primates is associated with a reduced anti-αGal immune response. 16 Moreover, valves from αGal/Neu5Gc-deficient pigs further reduce human IgM/ IgG binding when compared to BHV from wild-type pigs 17 . A similar situation is likely to occur whether bovine double knockout (DKO) tissue would be used. Seventy per cent of the BHV currently used in the clinic are in fact manufactured with bovine pericardia, that carries non-negligible amounts of αGal 8 and of Neu5Gc 9 even after currently used manufacturing treatments.
Pig-and cattle-derived products are also a major source of proteins for human consumption, and particularly, cattle are the major source of dairy products. Such products can become allergenic for some patients or infants consuming baby milk replacers. This allergy, known as the red meat allergy syndrome, 18,19 generally follows a tick bite inducing an isotype shift for IgE against αGal antigen. Neu5Gc is not synthesized by humans, but it can be incorporated through the diet and found in minute amounts in endothelial or epithelial cells of various tissues, likely contributing to inflammation-related diseases. 20,21 Furthermore, cattle can be used as a "bioreactor" to produce bioactive molecules for nutraceuticals or biomedical use, including r-human lactoferrin 22 in bovine milk. However, the resulting product differs from the human one because of the different glycosylation pattern. 23 Similarly, partially "humanized" antibodies 24 produced in cattle for various purposes still display Neu5Gc epitopes 25,26 that might be the target of an immune response by the host with clinically relevant side effects.
The scope of the present work was to generate cattle KO for both αGal and Neu5Gc antigens using a genome editing approach. 27 A stillborn calf KO for αGal has been reported, 28 but to the best of our knowledge, this work has not progressed further. Availability of DKO cattle line offers the opportunity to explore the potential of such animals to provide low immunogenic cattle-derived products for clinical purposes as well as for the food industry and human consumption.

| Animal experiments and source of animals
All procedures involving the use of animals in this study were approved by the Animal Welfare Committee of Avantea and carried out in accordance with the Italian Law (D.Lgs 26/2014) and EU directive embryos were implanted in 33 recipients heifers; 3 pregnancies were carried to term and delivered 3 live calves. Primary cell cultures were established from the 3 calves and following molecular analyses confirmed the genetic deletions. FACS analysis showed the double-KO phenotype for both antigens confirming the mutated genotypes. Availability of such cattle double-KO model lacking both αGal and Neu5Gc offers a unique opportunity to study the functionality of BHV manufactured with tissues of potentially lower immunogenicity, as well as a possible new clinical approaches to help patients with red meat allergy syndrome due to the presence of these xenoantigens in the diet. Bovine adult fibroblasts (BAFs) were derived from a skin biopsy of a Holstein bull and a cow with previous successful record of somatic cell nuclear transfer (SCNT). Recipient heifers used as surrogate mothers were also of Holstein breed.

| Chemicals
All chemicals were purchased from Sigma-Aldrich (Milano, Italy) unless otherwise stated.

| PCR set-up for identification and validation of target genes
Ensemble database was analysed to obtain the Reference genome sequences for the GGTA1 (ENSBTAG00000012090) and of the CMAH (ENSBTAG00000003892) genes. These sequences were studied in silico to identify possible target sequences, and the selected regions were amplified by PCR and analysed with Sanger sequencing to exclude polymorphisms in male and female fibroblast cell lines selected for the genome editing.
Editing of GGTA1 gene started initially in the male line targeting the exon 9 and because of the paucity of tools available at the time we never found efficient RNA guide. Therefore, we decided to use two guides that targeted the same sequence ( Table 2, btGGTA1cr1 and btGGTA1cr2). Years later, when we targeted the female line, we were able to find an efficient guide for exon 4 ( Table 3, btGG-TA1cr3) used for the pig by Sato et al. 29 Editing of the CMAH gene was achieved efficiently in the exon 2 carrying the ATG codon. For the editing of the male, we used one guide ( Table 2, btCMAHcr1) and subsequently for the female we found a more efficient guide (

| Genomic DNA extraction and PCR conditions
Primary fibroblasts and tissues biopsies were lysed at 55°C for 3 hours using a lysis buffer (100 mmol/L Tris HCl pH 8.  Desired sgRNAs (btGGTA1cr3 and btCMAHcr2) were in vitro synthetized following the CRISPOR guidelines (http://crisp or.org/).
Briefly, oligonucleotides (Table 3) were annealed, amplified and purified before to use the resulting amplification product as template (1μg) for the following transcription step. Single guide RNAs were finally synthetized using the TranscriptAid T7 High Yield Transcription kit (Thermo Fisher Scientific) and purified on silica membranes columns (MEGAclear Transcription clean-up kit, Thermo Fisher Scientific) according to the manufacturer's instructions and stored at −80°C.
We targeted the CMAH gene using as template a synthetized single strand oligonucleotide (ssCMAH-STOP oligo) specific for the exon 2 and symmetric according to the position of the CMAH-START codon.
Its sequence is characterized by the substitution of the START codon (ATG) with a STOP codon (TAA, in bold Figure 1A), generating a new AflII restriction site (CTTAAG, underlined in Figure 1A), useful for the identification of the knock-in colonies (152bp + 73 bp) with the AflII-RFLP analyses (AflII from Thermo Fisher Scientific; 1 hour at 37°C).

| Culture, transfection and selection of adult fibroblasts
Bovine adult fibroblasts (male and female) were cultured in Each colony was analysed for the GGTA1 gene (739 bp) and for the CMAH gene (225 bp). Resulting electrophoretic patterns determined directly that some colonies were characterized by visible Indels, creating bands different from the WT controls. This situation is clear for colonies A1 (double band), A2 (deletion) and A6 (deletion) in PCR analyses for the GGTA1 gene (°) and for colonies A1 (double band) and A5 (deletion) in PCR analyses for the CMAH gene (#). Resulting CMAH-PCR products were also digested with the AflII restriction enzyme, detecting the alleles interested by the targeting event. Due to the introduction of a STOP codon (TAA) in the START position (ATG) of the CMAH gene, only the HDR-CMAH alleles will be cut by the restriction enzyme producing two lower bands (152 + 73 bp). A simple agarose electrophoresis enabled us to identify possible additional edited colonies detecting the STOP codon insertion (**) for colonies A2 and A6 and the single insertion (*) for colonies A3 and A4. In these last ones, the not targeted allele resulted uncut (225 bp) as the WT sample. For this reason, the final determination of the exact Indels, occurred in all the edited colonies, was determined by Sanger sequencing of the resulting TOPO TA E coli clones. 100 = 100 bp ladder (Thermo Fisher Scientific); A1, A2, A3, A4, A5 and A6 = transfected females colonies; WT = wild-type female line; H 2 0 = Nucleases-free water. C, Sequences alignments of colonies used for the SCNT. Sanger sequencing outlining the mutations affecting the GGTA1 and the CMAH genes of colonies selected for the SCNT step. For the GGTA1 gene, the exon 9 was used as reference for the male colonies and a PCR product including the exon 4 was used for the female ones. In both cases, deletions of different lengths were obtained (Table S1). For the CMAH gene, all edited alleles of the edited colonies were aligned using as reference a PCR product including the exon 2 sequence. In this case, in both lines, we were able to determine the TAA substitution, as result of the targeting event mediated by the site-specific cut, produced by the CRISPR/Cas9 system driven by the sgRNA btCMAHcr1 Eight (four male and four female) confirmed DKO colonies, edited for GGTA1 and CMAH genes, were selected for further screening in SCNT to assess developmental potential. Before SCNT embryos were transferred into recipients, at least 10 cloned embryos of each selected colony were analysed for the absence of wild-type genotypes. Genomic DNA extraction procedure, PCR amplification and sequencing reactions were done using the same materials and methods described above.

| Somatic cell nuclear transfer (SCNT)
The

| Recipients synchronization, embryo transfer (ET) and calving
Heifers of 14-16 months of age were used as recipients. Oestrus was synchronized using the Ovsynch protocol with two injections of a

| Genotyping and phenotyping analyses for αGal and Neu5Gc antigens in DKO cattle-derived primary cells
Newborn calves were subjected to ear biopsy to establish a primary cell line, to extract the genomic DNA for genotyping by PCR and

| Disruption of GGTA1 and CMAH genes in primary bovine fibroblast lines
Two millions male fibroblasts were nucleofected and expanded for 7 days to 6.5 × 10 6 . After Dynabeads sorting, 4200 αGal-nega- and AflII-RFLP analyses revealed that 15 appeared to be edited for the CMAH gene and for this reason they were sent for Sanger sequencing analyses of their GGTA1 and CMAH genes ( Figure 1C and Table S1). GGTA1-KO was confirmed in all 15 colonies and 13 (31.7%) resulted also KO for Neu5Gc ( Table 4).
The female colonies were subjected to the same analysis. The PCR analysis ( Figure 1B) followed by Sanger sequencing analyses ( Figure 1C and  (Table S1). These data were finally demonstrated by the deletion (17 bp) generated in the CMAH gene. C, Sequencing results for 9162. The GGTA1 gene sequence presented a 8 bp deletion, and the CMAH gene is characterized by the same 2 different mutations (TAA substitution; del 13 bp) detected in colony E3 (Table S1). D, Sequencing results for 9163. The same Indels, characterizing the GGTA1 (del 54 bp) and the CMAH (TAA substitution) genes of A6 colony (Table S1) (Table 4), confirming the high efficiency of Dynabeads selection for the GGTA1 KO.
Male A4 and E3 and female A6 colonies were used for SCNT based on their morphology and growing characteristics and embryo production after SCNT. Ten SCNT embryos of each colony were sequenced to confirm the purity of the selected colonies for the required mutations ( Figure 1C) to avoid potential contaminations of WT cells.

| Genotyping of cloned calves
Sanger sequencing of TOPO TA-cloned PCR products of DKO calves confirmed the Indels characterizing the colonies used for cloning. In details, in clone 9161, GGTA1 gene is affected by two different mutations in exon 9 (del AGACCCTGGGCGAGTCGGTGG/ del 171bp) and the exon 2 of the CMAH gene carries a deletion (del GGCAGGCAAGTGAGGGA) as it was described for colony A4 (Table S1; Figure 2B). In clone 9162 (Figure 2A), a deletion in GGTA1 gene (del AGTCGGTG) is accompanied by 2 different mutations in the exon 2 of CMAH gene. The first allele was inactivated by the substitution of the ATG codon (START) with the TAA codon (STOP), due to the homology-directed repair (HDR) event driven by the ssCMAH-STOP oligo, as described for colony E3 (Table S1, Figure 2C), and the second allele has 13 bp deletion (del AGGCAAGTGAGGG).
Sanger sequencing results of female clone 9163 (Figure 2A) demonstrated that a deletion in the exon 4 of the GGTA1 gene (del 54bp) and the substitution of the START to a STOP codon (ATG TAA) in the exon 2 of the CMAH gene are identical to the Indels described for the donor female colony A6 (Table S1, Figure 2D). PCR analyses on the male calves (data not shown) demonstrated also that the CRISPR/Cas9 expression vectors were not integrated in the genome of the cloned calves.

| Phenotyping of cloned calves
FACS analysis confirmed the genotyping results of the three calves.
All primary cell lines derived from biopsies of the cloned male calves do not express αGal ( Figure 3A) and Neu5Gc ( Figure 3B) as opposed to WT control cells before genetic engineering. As negative controls, pig cells KO for both antigens were used. The female phenotyping was performed in the same way as for the males but in this case the negative control was the male 9162. In this experiment performed a year later with different experimental context, the αGal was completely negative. In the case of Neu5Gc, both the control (9162) and the female (9163) had some background staining. Since the 9162 pictured in Figure 3 is the same as in Figure 4, we can conclude that the tail of Neu5Gc staining is background staining coming from the different experimental setting and culture conditions.

| D ISCUSS I ON
In this work, we targeted two well-known xenoantigens identified as such from pig xenotransplantation studies that are also expressed in cattle. Here, we show that editing bovine fibroblasts are possible using both CRISPR/Cas9 in plasmid and Cas9-RNP formats and live animals can be generated through SCNT. The advent of programmable nucleases for genome editing in large animals, especially the pig, has greatly increased its efficiency by reducing the number of animals required and the costs involved. The number of genetically modified pigs and the consequent generation of animal models through precise genetic engineering have grown exponentially in the last 10 years. However, genetically modified cattle are still very few due to some constrains for applying this technology to this species such as the long generation interval. Nevertheless, cattle would be more relevant for food production since it is a major source of beef and dairy products. Furthermore, one of the major potential applications for DKO cattle for both GGTA1 and CMAH would be as a source of less immunogenic biological materials (pericadia) to manufacture BHV. In addition, genetically engineered cattle would also allow to produce food to avoid, for example, anaphylactic reaction following the consumption of red meat in some allergic individuals.
Despite the low transfection efficiency in bovine fibroblasts that affected the total number of edited colonies, because of the thorough screening of the few colonies selected and the combination with SCNT, we were able to generate DKO male and female calves.
All the bovine genome editing work was undertaken to disrupt simultaneously the GGTA1 and the CMAH genes without the need of a selectable marker, choosing primary cell lines whose genomic sequences were not affected by polymorphisms. We started the bovine genome editing work in the male line transfecting the plasmid format of the S pyogenes CRISPR/Cas9 system, while later its Cas9-RNP format was tested in the female line.
We selected to target exon 9 of GGTA1 gene in the male line using together two different sgRNAs (btGGTA1cr1 and btGG-TA1cr2- Figure 1A). In contrast to the female line, we targeted the exon 4, using the protein (Cas9-RNP), designing a sgRNA specific for the START codon (btGGTA1cr3 Figure 1A). The use of Dynabeads and IB4 lectins greatly compensated for the low transfection efficiency very effectively since all the analysed colonies derived from cells that did not bind IB4 were all KO for the GGTA1  Table 4). As a consequence, also the KO rate for the CMAH gene was very high indicating that when these nucleases enter the cells, they are very effective on all the targets. This event was also described for the pig by Li et al. 38 The use of ssODN-mediated KI with CRISPR/Cas9 system for KO purposes was also possible in cattle and facilitated the PCR screening because of the insertion of an AflII restriction site. The use of the plasmid to introduce and express all the machinery required was in our experiments more efficient than the use of the protein but because we required only a few cell clones for SCNT, it did not affect the success at the end since we had far more cell clone that we could need for SCNT. The reason for preferring the Cas9 protein to the plasmid is to avoid the risk of integration of the plasmid. Luckily in this case, we did not detect any integration of the CRISPR/Cas9 expressing plasmids in the genome of the male calves. CRISPR/Cas9-mediated genome editing procedures are compatible with SCNT, and the efficiency is comparable when WT cells are used. Three live calves were delivered by caesarean section and the first born (9161) is about to reach puberty while the female is only a few days old. The genotype of the three claves born alive exactly matched the genotype of the three cell clones selected for SCNT (Table S1). To further validate the genotyping findings with the phenotype, FACS analysis was performed on fibroblasts derived from the three newborn animals. The absence of αGal and Neu5Gc was clearly confirmed on the two bull calves. To perform the FACS analysis, primary cells were grown from biopsy taken in the first days of life of the calves that were gestated by WT surrogate mother and fed after birth with milk from WT cows; moreover, the culture of primary cells was performed with FCS supplementation to culture media from WT source. All these conditions favour incorporation of Neu5Gc into the cells that before the analysis requires 2-3 week in culture with serum lacking Neu5Gc. We used human serum in this period to allow the cells to shed the incorporated Neu5Gc but this time is variable depending on culture conditions, not ideal with human serum for bovine fibroblasts and the reagents used. The phenotypical characterization of the female calf by FACS was not yet extensively completed (only one experiment was performed). There is some background noise due to Neu5Gc remnants of the culture conditions; on the other end, also the male cells used as negative control that was completely clear in a previous experiment ( Figure 3B) had the same right shift for Neu5Gc ( Figure 4B). We can conclude that the generation of DKO cattle is possible using the latest genome editing technologies combined with SCNT.
This will offer the opportunity to use novel biological materials of bovine origin for medical and industrial application as well as for human consumption in the form of beef or dairy products for allergic individuals.

ACK N OWLED G EM ENTS
The authors acknowledge the technical support of Gabriella

CO N FLI C T O F I NTE R E S T S
The authors declare no conflict of interests.