Efficient production of multi-modified pigs for xenotransplantation by ‘combineering’, gene stacking and gene editing

Xenotransplantation from pigs could alleviate the shortage of human tissues and organs for transplantation. Means have been identified to overcome hyperacute rejection and acute vascular rejection mechanisms mounted by the recipient. The challenge is to combine multiple genetic modifications to enable normal animal breeding and meet the demand for transplants. We used two methods to colocate xenoprotective transgenes at one locus, sequential targeted transgene placement - ‘gene stacking’, and cointegration of multiple engineered large vectors - ‘combineering’, to generate pigs carrying modifications considered necessary to inhibit short to mid-term xenograft rejection. Pigs were generated by serial nuclear transfer and analysed at intermediate stages. Human complement inhibitors CD46, CD55 and CD59 were abundantly expressed in all tissues examined, human HO1 and human A20 were widely expressed. ZFN or CRISPR/Cas9 mediated homozygous GGTA1 and CMAH knockout abolished α-Gal and Neu5Gc epitopes. Cells from multi-transgenic piglets showed complete protection against human complement-mediated lysis, even before GGTA1 knockout. Blockade of endothelial activation reduced TNFα-induced E-selectin expression, IFNγ-induced MHC class-II upregulation and TNFα/cycloheximide caspase induction. Microbial analysis found no PERV-C, PCMV or 13 other infectious agents. These animals are a major advance towards clinical porcine xenotransplantation and demonstrate that livestock engineering has come of age.


Results
We wished to generate cells that support production of animals that ubiquitously express multiple xenoprotective transgenes from a single genomic locus. Two alternative strategies were adopted to assemble an array of colocated transgenes: sequential targeted transgene placement 'gene stacking' at a known permissive locus, and cointegration of engineered high capacity bacterial artificial chromosome (BAC) and phage artificial chromosome (PAC) vectors, 'combineering' . Cell clones with characterised expression were then used to inactivate xenoreactive endogenous genes. Genetic manipulation schemes are summarised in Fig. 1.
Transgene stacking at the porcine ROSA26 locus. As in mice, the porcine ROSA26 locus provides a 'safe harbour' for transgene expression without interrupting the function of essential endogenous genes 25 . We previously demonstrated that porcine ROSA26 could be targeted efficiently and support abundant ubiquitous transgene expression 26 . Here we placed an SV40-driven human HO1 cDNA into porcine ROSA26 essentially as Scientific RepoRts | 6:29081 | DOI: 10.1038/srep29081 described for a reporter gene 26 (Suppl. Fig. 1A). Approximately 5% of mesenchymal stem cell (MSC) cell clones were correctly targeted. Groups of clones were used for nuclear transfer, 177 reconstructed embryos were transferred to two recipient sows, one pregnancy established and one liveborn piglet (pig 74) obtained. This was sacrificed and HO1 detected in all tissues examined (Suppl. Fig. 1B). Porcine kidney fibroblasts (PKF) were derived and used for retargeting.
We generated a CAG promoter-driven CD55 minigene that showed expression levels similar to the average obtained with genomic CD55 in MSCs. A second targeting vector was used to place the human CD55 minigene 5′ to the HO1 cassette and exchange the selectable marker (Suppl. Fig. 1A). Targeted cell clones were again derived with high efficiency (11% of clones analysed) and groups of clones used for nuclear transfer. Nuclear transfer is ongoing, one pregnancy has been established and piglets expected soon.
One step generation of multi-transgenic pigs. We also investigated co-placement of multiple transgenes at random loci because this allowed the use of large genomic constructs. Co-transfection of several DNA constructs commonly results in integration at a single genomic locus. Combining multiple transgenes within the same vector further increases the probability of co-localisation and co-expression. While a permissive integration site could not be predetermined, cell clones could be screened for expression before nuclear transfer, because complement regulators are naturally expressed in MSCs. Serial nuclear transfer was used to check expression in vivo and then regenerate viable animals. Figure 1 outlines the schemes used.
Rounds of construct development were first carried out to maximise transgene expression in porcine MSCs (summarised in Methods section). MSC cell clones were then derived by co-transfection of either two or three constructs: a) human genomic CD46 BAC and human genomic CD55 BAC; b) human genomic CD46 BAC, human genomic CD59 PAC, and a triple-transgene BAC with CAG-driven human genomic CD55, SV40-driven human HO1 cDNA and CAG-driven human A20 cDNA (Suppl. Fig. 2). Several hundred transfected MSC cell clones were screened for transgene expression by RT-PCR, and quantified relative to human MSC line SCP1, which expresses all complement regulators 27 . CD46 mRNA expression was highest in clone 3-6 (> 7 fold higher than SCP1). Clone 4-39 expressed all five transgenes, with especially high levels of CD55 (> 90 fold higher than SCP1) considerably greater than any other clone analysed. These were used for nuclear transfer.
For cell clone 3-6, 253 reconstructed embryos were transferred to three recipients, one pregnancy was established and one stillborn piglet obtained: CD46, CD55 double transgenic animal ID 1107-6. For cell clone 4-39, 119 reconstructed embryos were transferred to one recipient, a pregnancy established and one live-born piglet obtained: CD46, CD55, CD59 A20, HO1 multi-transgenic animal ID 1706. Piglet 1706 was healthy and developed normally, but died due to injury. Organ samples were collected for expression analysis and PKFs cultured for functional analysis, further rounds of genetic manipulation, and nuclear transfer to regenerate each animal.
For re-cloning of piglet 1107-6, 348 reconstructed embryos were transferred to four recipients, two pregnancies were established, and 11 liveborn and one stillborn offspring obtained. One piglet died two weeks post partum. For re-cloning of piglet 1706, 212 reconstructed embryos were transferred to two recipients, one pregnancy established and one liveborn offspring (pig 266) obtained. All surviving animals remain healthy, and we now have F1 generation animals from both lines.
PERV and microbe screening. Xenografts should pose minimal infectious risk to human recipients. We carried out genome and serum analysis of multi-transgenic pig 266 (recloned from pig 1706), which formed the basis for subsequent genetic modifications and found it to be free of PERV-C expression. This animal was also free of other microbial infections. Details are shown in Suppl. Fig. 8.
Multi-transgenes are inserted at a single locus. The integration sites of the two constructs in 1107-6 cells, and the three constructs in 1706 cells were determined by FISH analysis. In each case transgenes were co-located at single genomic loci. The transgene locus in 1706 has been identified as chromosome 6q22. (Suppl. Expression and xenoprotective function of transgenes and genetic knockouts. Lung, liver, spleen, kidney, heart and aorta were analysed for hCD46, hCD55 and hCD59 mRNA expression. Representative RT-PCR results for multi-transgenic animal 1706 are shown in Suppl. Fig. 5A, and Q-RT-PCR of various organs in Suppl. Fig. 5B. Consistent with data from the original MSC cell clones, hCD46 expression was highest in double transgenic animal 1107-6; multi-transgenic animal 1706 showed high levels of hCD55 followed by hCD59 and somewhat less hCD46. Relative protein quantification by FACS analysis of PKFs confirmed these findings ( Fig. 2A).
Protection against human complement-mediated lysis was tested by incubation with concentrations of human serum from 2.5% to 20% (Fig. 2B). Double transgenic 1107-6 PKFs showed markedly less lysis than wild-type, and multi-transgenic 1706 PKFs were completely protected at the highest serum concentration, even though they were still α Gal positive.
Derivation of multi-transgenic and GGTA1 knockout animals. PKFs from piglet 1706 were modified by inactivating GGTA1 using a zinc-finger nuclease targeted to exon 8, as previously described 28 . GGTA1-deficient cells were counter-selected with isolectin B4 (IB4), which specifically binds α Gal, and the resultant population confirmed as α Gal free by flow cytometry. These multi-transgenic, GGTA1-deficient cells were again used for nuclear transfer. 563 reconstructed embryos were transferred to six recipients and three pregnancies established, two of which continued to term. Three liveborn piglets (779, 780, 859) and one stillborn piglet were obtained. Pig 779 was euthanised after three months due to polyarthritis. Pigs 780 and 859 developed normally, continue to thrive and are being used to found a pig line. Multi-transgenic GGTA1 knockout pigs express xenoprotective genes including their isoforms and are resistant to complement-induced lysis. Analysis of animal 779 revealed hCD46, hCD55, hCD59 and hA20 mRNA expression in all organs analysed (Fig. 3A). Human HO1 expression was evident in heart, skin and muscle, with lower levels in liver, kidney, spleen and aortic endothelial cells (PAEC), but not detected in lung.
Human CD46 and CD55 are normally expressed as several RNA splicing variants encoding membrane-bound and soluble protein isoforms that vary by tissue 29 , and thought to be required for full biological activity 30,31 . Figure 3B and C show RT-PCR detection of RNA splicing variants in 1706 PKFs and GGTA1-deficient 779 PKFs. Human CD46 amplified a ~760 bp fragment consistent with expression of splice variants hCD46-002, 004, 005 and 006 (names according to Ensembl database) and a 805 bp fragment consistent with hCD46-001 and hCD46-007. Samples amplified from pig 1706 were isolated, subcloned and the DNA sequence determined, generating sequences consistent with hCD46 splice variants 002/004 and 005/006 from the 760 bp fragment and sequences consistent with hCD46 splice variants 001/007 from the 805 bp fragment (Suppl. Seq. file 1). HCD55 amplified fragments closely similar to splice variants observed in normal human tissue, encoding membrane-bound protein isoforms gDAF, vDAF4 and vDAF5 and soluble isoforms sDAF, vDAF1, vDAF2 and vDAF3 29 . Fragments representing the two major splice variant RNAs encoding membrane-bound (gDAF, hCD55-001) and soluble (sDAF, hCD55-009) CD55 isoforms were subcloned and the sequences determined confirming their identity, (Suppl. Seq. file 2). Most hCD59 splice variants encode the same 128 amino acid protein isoform (Ensembl), so it was not tested. Western analysis of 1706 PKFs revealed expression of hCD46 56 kDa and 66 kDa isoforms, the hCD55 70 kDa isoform and 43 kDa, 46 kDa precursors, and the hCD59 25 kDa protein (Suppl. Fig. 6).
FACS analysis of blood from multi-transgenic GGTA1-deficient piglets 779 and 780 revealed complement regulator expression and the absence of α Gal antigens, consistent with homozygous GGTA1 inactivation (Fig. 4A). Piglet 779 PDKFs showed similar resistance to human complement-mediated lysis as multi-transgenic GGTA1-intact 1706 cells, indicating that abundant complement regulator expression already provides substantial protection (Figs 2B and 4B). A20 and HO1 provide vascular protection. We examined the functional effects of human A20 and HO1 expression. A20 and HO1 both have anti-apopotic effects via inhibition of caspase activity. After treatment with 20 ng/mL human TNFα and 10 μ g/mL cycloheximide, PKFs from multi-transgenic pig 1706 showed a 3% increase of caspase activity while that in wild-type PKFs increased by 124% (Fig. 5C).
We also investigated effects on cytokine-induced upregulation of adhesion molecules. Human A20 also inhibits NFκ B-mediated expression of E-selectin. TNFα -induced upregulation of E-selectin in PAECs from multi-transgenic GGTA1-deficient pig 779 was diminished by 58-69% compared to wild-type cells (Fig. 5A). The ability of human HO1 to inhibit IFNγ mediated upregulation of porcine MHC class-II molecules was tested in 779 PAECs by treatment with 8 ng/ml and 50 ng/ml IFNγ . PAECs of pig 779 showed 42-44% (p < 0.05) lower induction of MHC class-II expression than wild-type cells (Fig. 5B). We cannot exclude that A20 expression did not contribute to the observed results.

CMAH/GGTA1 double knockout.
To generate CMAH/GGTA1 double knockout animals in the 1706 multi-transgenic background we used CRISPR/Cas9 gene editing with serial nuclear transfer. Homozygous CMAH-knockout 1706 PKF cell clones were identified by sequence analysis across the target site. A clone that carried a single base insertion 3bp 5′ of the PAM motif and an 11bp deletion in the other allele was used for   nuclear transfer and two pregnancies established, one was terminated at 27 days, three foetuses explanted and PKFs derived. These were subjected to GGTA1 knockout by CRISPR/Cas9. A cell clone carrying an 11bp deletion in one allele and an 18 bp insertion in the other was identified and used for nuclear transfer. 403 reconstructed embryos were transferred to four recipients, two pregnancies were established, and two liveborn offspring (544 and 545) obtained. One piglet (544) died six days after birth. PKFs were isolated and used for FACS analysis and western blotting, confirming CMAH/GGTA1 double knockout ( Fig. 6 and Suppl. Fig. 7). To facilitate breeding of the multi-transgenic CMAH/GGTA1 double knockout animals once they are available, we have also created female CMAH/GGTA1 double knockout animals using a female PKF clone carrying a homozygous C insertion in GGTA1 exon 8, and a double heterozygous T insertion, 14 bp deletion in CMAH exon 10. Two healthy normal females (488, 490; Suppl. Fig. 7) are currently being raised for mating.

Discussion
Inhibition of hyperacute and acute-vascular rejection are required for long-term porcine xenograft survival in primates. Here we report pigs with a series of genetic modifications designed to address both.
Gene stacking has so far only been performed using site-specific recombinases 32 . We show that it is also possible to add two transgenes (CD55 and HO1) as single copies by successive rounds of homologous recombination at the porcine ROSA26 locus, and obtain high ubiquitous expression of each. Further transgenes can easily be added to the array, and it will be interesting to find out how many can be stacked while maintaining the permissive nature of porcine ROSA26.
The use of E. coli based homologous recombination systems, developed for functional genomics in mice 33 , was key to our engineering of large BAC and PAC constructs incorporating multiple xenoprotective transgenes. To ensure independent expression of each transgene we used separate expression constructs. Alternative means of combining smaller components e.g. cDNAs, such as 2A peptide or IRES-based polycistronic systems, have also been used to produce multiple transgenic xenodonor pigs 34,35 , but these carry a risk of poor or variable expression 36 .
Serial nuclear transfer was key to the strategy we employed as it enabled sequential transgene addition and gene editing steps to generate pigs carrying the battery of modifications necessary to provide robust xenograft protection.
We adopted a multipronged approach to blocking hyperacute rejection. Three important regulators, CD46, CD55 and CD59 were used to inhibit complement activation at the levels of C3 convertase, C5 convertase, and  the membrane attack complex. Such combined inhibition prevents cross-activation of the complement system at stages later than C3 convertase, for example by coagulation factors that interact with the complement system 30 . The transgenes used were genomic sequences, rather than cDNAs or minigenes, to facilitate abundant expression and enable expression of CD46 and CD55 RNA splice variants and the protein isoforms necessary for full biological function 31,37 . Challenge with human complement revealed that a combination of CD46 and CD55 markedly reduced lysis, and that CD46, CD55 and CD59 together provided essentially complete protection in vitro. These results compare very favourably with a previous report of triple complement regulator transgenic pigs that described 20% lysis of PBMCs after one hour incubation 12 , compared to complete protection after four hours in our assay.
Our in vitro lysis assay discerned no additional benefit from GGTA1 gene inactivation in the multi-transgenic background, but the situation in vivo will likely be more rigorous. Pig-to-primate xenotransplantation showed that protection against hyperacute rejection is increased when GGTA1 inactivation is combined with complement regulator expression 38 . GGTA1-KO xenografts are however still subject to acute vascular rejection, characterised by endothelial activation probably by preformed non-Gal antibodies, causing thrombotic microangiopathy 39 . This led us to include the anti-inflammatory and anti-apoptotic genes hHO1 and hA20 as vascular protectors. We also inactivated the CMAH gene responsible for the major non-Gal antigen Neu5Gc (Hanganutziu-Deicher) 40 .
Heme oxygenase-1 has a range of anti-apoptotic and anti-inflammatory effects. It reduces NK cell activity and formation of pro-inflammatory factors such as CCR5 (C-C chemokine receptor type 5), ICAM-1 (intercellular adhesion molecule 1), VCAM-1 (vascular cell adhesion molecule 1) and E-selectin (CD62 antigen-like family member E) 41,42 . It inhibits platelet aggregation, maintaining microcirculation and facilitating angiogenesis 43 . In rodent xenografts HO1 reduces thrombus formation and IgM deposition 44 and prolongs organ survival 45,46 . In pigs HO1 protects endothelial cells from TNF-α -mediated apoptosis, extends ex vivo perfusion of kidneys with human blood 18 , protects fibroblasts from H 2 O 2 damage, and inhibits TNF-α and cycloheximide-mediated apoptosis 47 . Our analysis revealed wide but not ubiquitous HO1 expression, and found that cytokine-induced MHC-class II up-regulation in aortic endothelial cells was reduced by almost one half. This is very encouraging, and could be due to co-expression of hHO1 with hA20.
A20 inhibits NF-κ B activation 48 and thus upregulation of pro-inflammatory and pro-apoptotic cytokines including TNFα and interleukin-1β (IL-1β ) 49 . A20 protects porcine aortic endothelial cells against TNFα -induced apoptosis and confers partial protection against ischemia/reperfusion injury 17 . We found ubiquitous A20 expression, and E-selectin upregulation an indicator of NF-κ B activation was markedly reduced (65% < wild-type) in aortic endothelial cells. Again, this could be due to co-expression of hA20 with hHO1. The anti-apoptotic effects of A20 and HO1 were also confirmed by inhibition of caspase 8 induction.
Safe, effective xenotransplantation requires that donor animals do not transmit zoonotic pathogens. We examined the founder multi-transgenic pig 226 and found it to be to be free of a series of known and potential infectious agents, including expressed PERV-C. Inactivation of all PERV loci in a porcine cell line in vitro has been reported recently 50 , but whether such multiple gene editing is compatible with the production of viable pigs has yet to be established.
We are now breeding multi-modified GGTA1-deficient boars 780 and 859 with homozygous GGTA1-deficient sows to enable ex vivo perfusion of explanted organs, and transplantation into primates. CMAH/GGTA1 double knockout sows 488 and 490 are also available for breeding with multi-transgenic double knockout males.
There have been several recent successes in pig-to-primate xenotransplants using multi-transgenic porcine organs, with record survival times of 125 and 136 days for kidneys 51,52 and a remarkable 945 days for hearts 53 . These donors were all GGTA1-deficient, and expressed one or two complement and one coagulation regulator in the organ used. The combination of genetic modifications we describe is the most extensive reported so far, and should extend xenograft survival further, but might not yet be complete. The full requirements for clinically effective xenotransplantation will become evident as transplant studies continue and salient aspects of delayed xenograft rejection are revealed. Although CD46 affects coagulation and fibrinolytic cascades 54 further transgenes may be required to combat coagulation dysregulation and thrombotic microangiopathy associated with endothelial activation. Candidate transgenes for future inclusion thus include human anticoagulant thrombomodulin 55 , the endothelium protein C receptor 56 and CD39 57 . The adaptive immune response will present the next challenge, and human recipients will probably require immunosuppression. There is encouraging and unanticipated evidence that GGTA1 inactivation and CD46 expression down-regulate the human T-cell response to pig cells in vitro 58,59 , nevertheless T-cell regulators such as CTLA4-Ig/LEA29Y 60 may be necessary. It is eminently feasible to add additional transgenes adjacent to the existing array. Maintaining a single locus is important to simplify breeding of donor animals to meet the demand for xenotransplants.

Methods
Animal experiments were approved by the Government of Upper Bavaria (permit number 55.2-1-54-2532-6-13) or by the Lower Saxony State Office for Consumer Protection and Food Safety (LAVES permit number 33.14-42502-04-12/0891) and performed according to the German Animal Welfare Act and European Union Normative for Care and Use of Experimental Animals.
Reagents. Chemicals were obtained from Sigma-Aldrich Chemie GmbH unless otherwise specified, cell culture media and supplements were obtained from Thermo Fisher Scientific or Invitrogen unless otherwise specified, antibodies were purchased from Santa Cruz unless otherwise specified.
Transgene stacking at the porcine ROSA26 locus. The ROSA26 HO1 targeting vector consisted of: a 2.2 kb short homology arm; splice acceptor, promoterless neo; SV40 driven HO1 cDNA 18 ; and a 4.7 kb long homology arm (Suppl. Fig. 1A). Targeting in MSCs was carried out by standard methods and confirmed in cell clones Scientific RepoRts | 6:29081 | DOI: 10.1038/srep29081 by long-range 5′ and 3′ junction PCR and sequence analysis (PCR primers shown in Suppl. Table 1). The ROSA26 CD55 retargeting vector consisted of the same 2.2 kb short homology arm; splice acceptor; promoterless blasticidin resistance; a CAG-driven CD55 minigene composed of CD55 exon 1, intron 1 and exon2 ligated at the HindIII site within exon 2 to CD55 cDNA (membrane bound form); and a 5 kb long homology arm (Suppl. Fig. 1A). Targeting was carried out in PKF from a ROSA26 HO1 targeted piglet by standard methods, confirmed in cell clones by long range 5′ and 3′ junction PCR and sequence analysis (PCR primers shown in Suppl. Table 1).

Modification of complement regulatory gene constructs. Bacterial artificial chromosome (BAC)
vectors containing genomic sequences of human CD46 and CD55, and a phage artificial chromosome (PAC) vector containing human genomic CD59 were purchased from Source BioScience, UK (CD46 BAC: RP11-99A19; CD55 BAC: RP11-357P18; CD59 PAC: RP4-541C22). Superfluous regions were deleted and drug-selectable cassettes PGK/Em7-driven neomycin resistance (Neo) or SV40/EM7-driven blasticidin resistance (BS) inserted by homologous recombination in E. coli strain SW106 (recombineering) by standard methods 61 . Each molecular clone was then reduced in size by successive restriction digestion and recombineering stages and tested for expression in porcine MSCs to identify a suitable configuration to be introduced into pigs.
The large CD46 BAC construct, which consisted of a 66 kb 5′ flanking/promoter region, 43 kb CD46 gene and 54 kb 3′ flanking region was reduced by deleting distal regions of the 3′ flank to generate 'medium' and 'small' constructs with 34 kb and 7 kb 3′ flanks. The large hCD46 construct provided best expression and was chosen to generate pigs.
The large CD59 PAC construct consisting of 10 kb 5′ flanking/promoter region, 34 kb CD59 gene and 82 kb 3′ flanking region was reduced at the distal 3′ flanking region to 'medium' and 'small' constructs with 57 kb and 37 kb 3′ flanks. The 'small' version provided abundant expression and so was used to generate pigs.
The large CD55 BAC construct consisting of 28 kb 5′ flanking/promoter sequence, 40 kb CD55 gene and 103 kb 3′ flanking region was reduced at the distal 3′ flanking region to 'medium' and 'small' constructs with 33 kb and 6 kb 3′ flanks. The 5′ flanking/promoter region of the 'small' construct was then reduced to 10 kb to generate an 'extra small' CD55 construct. A further construct, termed CAG-CD55, was generated in which the 5′ flanking/ promoter region was replaced by a 1.8 kb CAG synthetic promoter (CMV enhancer, chicken beta-actin promoter, rabbit beta-globin splice acceptor). The 'extra small' CD55 and CAG-CD55 constructs were chosen on the basis of abundant expression in vitro. CAG-driven hA20 cDNA 17 and SV40-driven hHO1 cDNA cassettes 18 were then combined into a single vector and added to the CAG-CD55 BAC at a location 3′ of the CD55 3′ flank by recombineering using two 500bp homologous arms derived from the CD55 3′ flank. Outline structures of the constructs used to generate pigs are shown in Suppl. Fig. 2. Cell isolation, culture and transfection. Mesenchymal stem cells (MSC) were isolated from adipose tissue from German Landrace pigs and cultured as described 62 . Samples of 1 × 10 7 MSCs were co-transfected using Lipofectamine 2000 (Life Tech., USA) with 10-30 μ g DNA composed of either a mixture of CD46 and CD55 (endogenous promoter) constructs in 1:1 molar ratio, or a mixture of hCD46, hCD59 and hCD55-hA20-hHO1 constructs in 1:1:1 molar ratio, cell clones were selected and isolated by standard methods.
Porcine kidney fibroblasts (PKF) were isolated and cultured by standard methods. Porcine aortic endothelial cells (PAEC) were isolated as described 17 and cultured in DMEM, 20% FCS, 4 mM glutamine, 200 U/mL penicillin, 200 μ g/mL streptomycin, 25 mM HEPES, 50 μ g/mL endothelial cell growth supplement (Corning, USA) at 37 °C and 10% CO 2 , adjusted to 5% CO 2 during cytokine stimulation. Porcine peripheral blood mononuclear cells (pPBMC) were isolated from heparinised blood samples by Ficoll density gradient centrifugation. Somatic cell nuclear transfer. Nuclear transfer was performed as described 63,64 . In short, donor cells were arrested at G0/G1 of the cell cycle by serum deprivation, a single cell was inserted into the perivitelline space of enucleated in vitro matured oocytes from prepubertal gilts, then fused and oocytes activated by electric pulse. Reconstructed embryos were transferred into oviducts of hormonally synchronised recipient gilts by mid-ventral laparotomy.
Fluorescence in situ hybridisation (FISH). Prior to hybridization, chromosomes were Q-banded using 0.005% quinacrine mustard solution for 90 s. A standard FISH protocol was applied 65 , using three genomic probes: hCD59 labelled with DEAC-5-dUTP (PerkinElmer, USA), hCD55 with dig-11-dUTP (Roche, Germany), and hCD46 with biotin-16-dUTP (Roche, Germany). In each case mean probe fragment size was Scientific RepoRts | 6:29081 | DOI: 10.1038/srep29081 500 bp. Digoxigenin-labelled probes were detected with anti-digoxigenin-fluorescein Fab fragments (Roche). Chromosome preparations were mounted in Vectashield (Vector Laboratories) with DAPI counterstain and analysed under a Nikon Ni-U fluorescence microscope. International nomenclature for identification of pig chromosomes was applied 66 . GGTA1 and CMAH gene inactivation. Plasmids encoding a zinc-finger nuclease (ZFN) targeted to GGTA1 exon 8 were electroporated by standard methods. α Gal-negative cells were enriched by counter-selection with streptavidin-coated magnetic beads and biotin-conjugated isolectin B4 (Dynabeads, Life Technologies, USA) in a magnetic field 28 . GGTA1 gene editing was analysed by allele-specific sequencing of cloned PCR fragments amplified across the target site.
GGTA1/CMAH double knockout was performed in two steps. A CRISPR/Cas9 enzyme targeted to the sequence 5′ GGAAGAAACTCCTGAACTACA 3′ in CMAH exon 10 was transfected into 1706 cells, and cell clones analysed by allele-specific sequencing of cloned PCR fragments amplified across the target site (Suppl. Table 1). Homozygous knockout clones were used for nuclear transfer, a pregnancy was terminated at day 28, foetal fibroblasts isolated and transfected with a CRISPR/Cas9 enzyme targeted to the sequence 5′ GACGAGTTCACCTACGAG 3′ in GGTA1 exon 8. α Gal-negative cells were counter-selected and verified by sequencing as above.
Immunofluorescence staining and flow cytometry. 0.1-0.5 × 10 6 cells were incubated with one of