A piggyBac-based platform for genome editing and clonal rhesus macaque iPSC line derivation

Non-human primates (NHPs) are, due to their close phylogenetic relationship to humans, excellent animal models to study clinically relevant mutations. However, the toolbox for the genetic modification of NHPs is less developed than those for other species like mice. Therefore, it is necessary to further develop and refine genome editing approaches in NHPs. NHP pluripotent stem cells (PSCs) share key molecular signatures with the early embryo, which is an important target for genomic modification. Therefore, PSCs are a valuable test system for the validation of embryonic genome editing approaches. In the present study, we made use of the versatility of the piggyBac transposon system for different purposes in the context of NHP stem cell technology and genome editing. These include (1) Robust reprogramming of rhesus macaque fibroblasts to induced pluripotent stem cells (iPSCs); (2) Culture of the iPSCs under feeder-free conditions even after removal of the transgene resulting in transgene-free iPSCs; (3) Development of a CRISPR/Cas-based work-flow to edit the genome of rhesus macaque PSCs with high efficiency; (4) Establishment of a novel protocol for the derivation of gene-edited monoclonal NHP-iPSC lines. These findings facilitate efficient testing of genome editing approaches in NHP-PSC before their in vivo application.

piggyBac transposition using our previously published 6-factors-in-one-vector transposon system (Fig. 1a) 27 . The reprogramming transposon encodes the marmoset factors SOX2, OCT4A, KLF4, c-MYC, NANOG, and LIN28. Marmoset and rhesus macaque pluripotency factors show a very high degree of conservation on the cDNA and the protein level ( Table 1). The iPSC lines were generated from skin fibroblasts from 2 adult macaques (DPZ_iRhpb#1-3, male) (DPZ_iRhpb#4, female). Approximately 20 days after transfection and selection of the fibroblasts, the first colonies emerged; new colonies appeared at least until day 60. Between 100 and 150 colonies were identified per reprogramming experiment (100-150 primary colonies/1 × 10 6 transfected cells), resulting in a reprogramming efficiency of approximately 0.08-0.12% (~ 16-25 colonies per primary plate; 0.2 × 10 5 cells per primary plate). Primary colonies showed the typical morphology of human iPSCs in feeder cell culture (not shown). During the first passages, the best colonies were selected by manual picking according to their morphology. Four of the colonies with good morphology were selected in passage 5 to proceed with further passaging and characterization (Fig. 1a). Between passages 10 and 20, the lines became stable, showing almost no differentiation. The first assessment of the pluripotent state of the cells was done by alkaline phosphatase (AP) staining. All generated lines showed AP activity (Fig. 1b).
Three iPSC lines, DPZ_iRhpb#2, DPZ_iRhpb#3, DPZ_iRhpb#4, and a rhesus control ESC line (Rh_ESC), were adapted to feeder-free conditions (DPZ_iRhpb#4 and Rh_ESC, Fig. 1c) (Supplementary Fig. 1) using Stem Max iPS-Brew supplemented with 1 µM IWRI and 0.5 µM CHIR. We recently named this formulation universal primate pluripotent stem cell (UPPS) medium 33 . DPZ_iRhpb#4 was adapted at early passage (passage 10) and DPZ_iRhpb#2 and 3 were maintained in feeder conditions for expansion and characterization and adapted to feeder-free conditions at late passages (between passage 50 and 60). Colony morphology was very similar before and after adaptation (Fig. 1c, and Supplementary Fig. 1a,b, compare with Fig. 1a). The colonies of the iPSC and of the ESC lines present a typical compact structure and a high nucleus/cytoplasm ratio. DPZ_iRhpb#4 was cultured for more than 60 passages in feeder-free conditions, DPZ_iRhpb#2 and 3 for more than 10 passages. All cell lines showed undifferentiated morphology and AP activity during adaptation and maintenance in UPPS medium (DPZ_iRhpb#4 and Rh_ESC, Fig. 1d) (DPZ_iRhpb#2 and 3 Supplementary Fig. 1c,d).
In order to evaluate if the expression of the pluripotency factors was downregulated during differentiation, teratomas were stained for pluripotency markers. OCT4A, NANOG, TRA-1-60, and LIN28 expression was absent from most cells of the teratomas and was limited to a few isolated clusters found in all teratomas (DPZ_iRhpb#2, Fig. 2c) (DPZ_iRhpb#1, DPZ_iRhpb#3, data not shown). This finding shows that endogenous (OCT4A, NANOG, TRA-1-60, and LIN28) as well as piggyBac-dependent (OCT4A, NANOG, and LIN28) pluripotency factor expression is generally downregulated on the protein level during differentiation (Fig. 2c). Isotype controls were performed as negative control ( Supplementary Fig. 5b).
Teratoma formation assay was performed with DPZ_iRhpb#1, DPZ_iRhpb#2, DPZ_iRhpb#3 cultured on feeder cells and DPZ_iRhpb#4 in feeder-free conditions. In order to confirm that all iPSC lines maintain their differentiation potential in feeder-free conditions, we performed the embryoid body (EB) formation assay. DPZ_iRhpb#2, DPZ_iRhpb#3, and DPZ_iRhpb#4 were able to differentiate into representative cell types of the three germ layers. Successful differentiation was evaluated by immunostaining for β-III tubulin, SMA and AFP (DPZ_iRhpb#3 and DPZ_iRhpb#4, Supplementary Fig. 5c,d) (DPZ_iRhpb#2 data not shown). Finally, we also evaluated the presence of cells positive for pluripotency markers after in vitro differentiation. All cells were negative for OCT4A, and only two isolated clusters of TRA-1-60 positive cells were found in the culture, confirming our findings in the teratomas (DPZ_iRhpb#3, Supplementary Fig. 5e). piggyBac silencing during reprogramming and differentiation. The histological analysis of the teratomas generally suggested silencing of the piggyBac transposon used for reprogramming; most of the cells neither expressed OCT4A nor NANOG. For both proteins, we have established very specific and sensitive IHC detection protocols 34 . In order to check if silencing occurs during reprogramming or differentiation, we performed comparative methylation analysis of the reprogramming construct in iPSCs and teratomas ( Fig. 3) (Supplementary Fig. 6).
The reprogramming construct contains two separate CAG promoters; one driving the reprogramming factor cassette and one driving the puromycin resistance gene (Fig. 3a). We aimed at getting an overview of the global methylation of the construct in the cultures and gaining insights into possible differential methylation patterns between both promoters (CAG reprogramming vs. CAG puromycin). DNA methylation analysis was performed by bisulfite conversion using primers designed to specifically amplify the promoter driving the expression of the reprogramming 6-factor cassette (CAG reprogramming). Another primer pair was designed to amplify the puromycin resistance promoter (CAG puromycin). Additionally, primers non-selectively amplifying both promoters were used (CAG) (Fig. 3a). The three amplicons were sequenced with two different oligonucleotides, i.e., S2 (Fig. 3) and S1 ( Supplementary Fig. 6). Fibroblasts and Rh_ESCs were included as negative controls.
Three teratomas and three iPSC lines were included in the analysis. The methylation of the reprogramming cassette promoter (CAG reprogramming, 60-95% methylation) was higher than the methylation of the puromycin promoter (CAG puromycin, 17-30% methylation) in all samples (Fig. 3b,c) ( Supplementary Fig. 6a,b). As expected, the non-selective analysis of both promoters (CAG) showed intermediate values. No significant differences were found between teratomas and iPSC for any of the two promoters. These findings suggest that epigenetic modifications are mainly triggered already during reprogramming of fibroblasts into iPSCs and not during differentiation of the iPSC during the teratoma formation. GpC island-specific evaluation of the bulk analysis of the samples shows homogenous methylation levels at all CpG sites in all amplicons (Fig. 3e). This advocates against high heterogeneity in the epigenetic modifications in the different GpC sites contained in each promoter. The two different primers used for sequencing show consistent results (Fig. 3) (Supplementary Fig. 6). This data shows that already iPSCs have close to maximum methylation levels of the CAG reprogramming similar to the teratomas. This may explain the down-regulation of the expression of the exogenous pluripotency factors as shown by IHC of the teratomas. However, the CAG puromycin promoter is significantly less methylated than www.nature.com/scientificreports/  27 . Stop codons were substituted by 2A peptide sequences (F2A, T2A, and E2A). Expression of the reprogramming transcript is driven by a CAG promoter. An independent CAG promoter controls the expression of the puromycin resistance gene (P). Below the vector: Schematic representation of the amplicons used for pyrosequencing. Three different primer combinations were used, one to amplify both CAG promoters (CAG, blue) simultaneously, one to amplify specifically the CAG promoter driving reprogramming cassette expression (CAG reprogramming, red), and one specific for the promoter associated with the puromycin resistance gene (CAG puromycin, green www.nature.com/scientificreports/ the CAG reprogramming promoter demonstrating differential methylation of the two sequence-wise identical CAG promoters present in the piggyBac construct. Transposon removal by re-expression of pBase-tdTomato in iPSCs. We have shown high methylation of the CAG reprogramming promoter (Fig. 3) (Supplementary Fig. 6). However, RT-PCR analysis still detected the transposon-encoded transcripts in the iPSC lines ( Supplementary Fig. 3). In order to generate a transgene-free iPSC line and demonstrate that the stability of the iPSC line is based exclusively on endogenous gene expression, we exemplarily tried to remove the reprogramming transposon from DPZ_iRhpb#2 and 4. Cells were re-transfected with the transposase vector pBase-tdTomato ( Fig. 4a) ( Supplementary Fig. 7) 27 . Two days after transfection, clusters of Tomato-positive cells were identified. Tomato-positive cells were sorted by FACS to evaluate transfection efficiency and for selection of the potential transgene-free clones. Transfection efficiency was evaluated for DPZ_iRhpb# 2 and 4. The average percentage of tdTomato positive cells for the two iPSC lines was 11% (DPZ_iRhpb# 4) and 15% (DPZ_iRhpb# 2) ( Supplementary Fig. 7b). For the selection of transgene-free clonal lines cells, different clones were sorted and analyzed for the presence/absence of the transposon in gDNA by PCR (Fig. 4b,c). Five different primer pairs were used to detect different fragments of the transposon (Fig. 4c). Successful removal of the transposon was only achieved for DPZ_iRhpb# 4. In this line, one out of 28 clones showed no presence of the reprogramming cassette (Fig. 4b).
In summary, iRhpb#4∆pb cells remain pluripotent under feeder-free conditions and have a normal karyotype (Fig. 4e) after removal of the transposon.

Derivation of clonal rhesus macaque pluripotent stem cell lines. To generate clonal iPSC lines
with defined mutations, single-cell isolation and propagation protocols need to be established. Two cell lines, Rh_ESC and DPZ_iRhpb#4 were selected to develop a work-flow for clonal rhesus PSC line generation, based partially on the work-flow published by Chen and colleagues for human iPSC (Fig. 1c) 29 .
The first step towards single-cell cloning is reliable and complete dissociation of cell clusters to a single cell suspension. In order to find the most suitable dissociation reagent, we tested three reagents commonly used in human PSC protocols: versene, accutase, and TrypLE Select (Fig. 5a). After dissociation with the different reagents, we analyzed proliferation and differentiation of the cells. Five days after splitting, spontaneous differentiation of Rh_ESC and DPZ_iRhpb#4 was assessed by alkaline phosphatase staining and proliferation by cell counting. The starting cell number was 200.000 cells. No significant differentiation was found for any of the three tested reagents (Fig. 5b). Proliferation analysis of DPZ_iRhpb#4 revealed no significant differences between the dissociation reagents. In contrast, Rh_ESCs dissociated with accutase or versene showed higher proliferation recovery in comparison with TrypLE Select (Fig. 5c). These data suggest that accutase and versene may be more suitable reagents for single-cell dissociation compared to TrypLE for Rh_ESCs.
We then transfected Rh_ESC and DPZ_iRhpb#4 with the piggyBac vector pTT-PB-pCAG-eCas9-GFP-U6-gRNA-Neo together with pBase-tdTomato encoding the transposase, for constitutive expression of Cas9 in the PSC (Fig. 6a). In order to analyze the transfection efficiency, replicates for each one of the two PSC lines were sorted 5 days after transfection. The overall transfection efficiency for the combination of transposon/transposase CRISPR/Cas vector system was 49,03% for iRhpb#4 and 12% for Rh_ESC ( Supplementary Fig. 9). For the derivation of the monoclonal lines containing the mutations of interest, we decided to passage the cells three times before sorting in order to increase the likelihood that the GFP-positive population integrated the transposon. Single cells were sorted into a 96-well plate by FACS. For control purposes, one or three cells were sorted into each well. This approach allowed us to expand clones and to obtain a representative number for analysis. We tested three different conditions to improve clonal expansion (Fig. 6b). Firstly, culture with and without prosurvival compound, secondly, with and without Activin A and, thirdly, under normoxia and hypoxia (5% O 2 ). Only pro-survival compound significantly increased cell survival and/or proliferation (Fig. 5d), while neither Activin supplementation for two days after sorting (Fig. 5e) nor hypoxia (Fig. 5f) improved clonal expansion. In summary, the use of accutase in combination with medium supplemented with pro-survival compound for clonal expansion provides optimized conditions for the establishment of clonal rhesus PSC lines.
Genome editing of rhesus macaque stem cells using the piggyBac transposon Cas9/GFP vector. In order to validate the clonal expansion protocol, we designed and tested CRISPR/Cas9 guides to target clinically relevant mutation sites in the sarcomeric gene Titin (TTN). Two sites were targeted, one encoding the N-terminus and another one encoding the C-terminal part of the protein (rhesus exons homologous to human TTN exons 38 and 280, respectively; ENST00000589042.5). Guides were inserted in the pTT-PB-pCAG-eCas9-GFP-U6-gRNA-Neo vector and the construct was transfected together with the transposase-expressing vector following a DNA-integrative approach. After transfection, PSC were passaged three times before the generation of the monoclonal lines. Increasing the time span (8-10 days/three passages) between transfections and sorting www.nature.com/scientificreports/ we expect that the GFP signal comes from constitutive and not transient expression of the construct to ensure the integration of the transposon. Even though transfection efficiency following the protocol detailed in the methods part is usually high, variation between different cell lines was observed ( Supplementary Fig. 9b). The pTT-PB-pCAG-eCas9-GFP-U6-gRNA-Neo vector contains a neomycin resistance cassette that can be used to enrich the transfected population before sorting (Fig. 6a). In parallel to the transfection of the PSC for sorting, we performed bulk population analysis of the targeted TTN sites in fibroblasts and iPSC. We performed Sanger www.nature.com/scientificreports/ sequencing (data not shown) and T7 endonuclease I assay 35 in order to gain insights into the functionality of the gRNAs plus vector before starting the laborious single-cell cloning process ( Supplementary Fig. 10). Both gRNAs showed evidence of on-target editing activity ( Supplementary Fig. 10). While the targeted site encoding the N-terminal sequence showed only faint digestion products in the T7 endonuclease I assay, the targeted site encoding the C-terminal sequence showed more intense bands. After transfection and expansion of the transfected cells, the single-cell cloning protocol developed above was followed, and monoclonal lines containing frameshift mutations (fxs) in both loci were expanded. Eight clones with truncating N-terminal mutations were obtained: three for Rh_ESC and five for DPZ_iRhpb#4. Mutations in the different clones were analyzed by PCR amplification of the targeted genomic loci and subsequent sequencing of the products (data not shown). The efficiency of the process was also analysed for the C-terminus, and independent biological replicas for each cell line were performed (transfection, sorting, and expansion). For the C-terminal mutation, 37 mutated clones were obtained from 42 clones analyzed in total (DPZ_iRhpb#4: 27 mutated/28 analyzed; Rh_ESC: 10 mutated/14 www.nature.com/scientificreports/ analyzed). Exemplary chromatograms are shown in Fig. 6c for DPZ_iRhpb#4::eCas9-GFP C-terminus fsx TTN and Rh_ESC::eCas9-GFP C-terminus fsx TTN. The overall efficiency of the process for this gRNA was 85% (mutated clones/analyzed clones; Fig. 6d,e). The N-terminal gRNA worked less efficiently than the C-terminal gRNA. From 49 clones analyzed 5 mutated were identified, with an overall efficiency of 11.53% (Fig. 6d,e). However, each experiment performed with any of the gRNAs led to at least one mutated clone (number of clones analyzed per experiment ranges between 9 and 12, Fig. 6e). In conclusion, we established tools and a protocol for the robust and reliable generation of genetically modified clonal rhesus macaque ESC and iPSC lines carrying clinically relevant mutations.
Isogenic control cell line generation and characterization. The next step after the generation of the mutated clones is to assess the phenotypic alterations in differentiated cells generated from the pluripotent stem cells lines, e.g., iPSC-derived cardiomyocytes 33 . Downstream analysis of the mutated cells requires the generation of proper isogenic controls, to exclude possible effects of the cloning process and the genome editing construct expression. Therefore, isogenic controls were generated for both lines. Transfection of the cells with the piggyBac Cas9 vector without guides (pTT-PB-pCAG-eCas9-GFP-U6-gRNA-Neo), but with transposase, was performed (DPZ_iRhpb#4::eCas9-GFP and Rh_ESC::eCas9-GFP). Isolation of appropriate isogenic clonal control populations expressing the Cas9-GFP construct was achieved for both Rh-ESC and DPZ_iRhpb#4 (Supplementary Fig. 11a). Expression of Cas9-GFP was evaluated every passage by fluorescent microscopy and at passage 5 by immunostaining ( Supplementary Fig. 11b). All cells in the populations show Cas9-GFP expression ( Supplementary Fig. 11a,b). In order to show the functionality of the clonal lines (mutated and isogenic controls) derived from the genome editing process, basic characterization of Rh_ESC and DPZ_iRhpb#4 sublines was performed. Isogenic controls ( Supplementary Fig. 11c) and TTN mutated lines still express key pluripotency markers ( Supplementary  Figs. 11c, 12a). Additionally, all lines present a normal chromosome number ( Supplementary Fig. 12b, shown for DPZ_iRhpb#4::eCas9-GFP C-terminus fsx TTN). Finally, in order to ensure that the cloning process does not affect the differentiation potential of the iPSCs we performed spontaneous in vitro differentiation (shown for DPZ_iRhpb#4::eCas9-GFP and Rh_ESC::eCas9-GFP). During the embryoid body formation, no downregulation of the GFP expression in the cell aggregates was observed ( Supplementary Fig. 11d). Additionally, immunostaining of both cell lines analyzed showed differentiation into representative cell types of the three germ layers ( Supplementary Fig. 11e,f). These results indicate that the cloning protocol neither changed the cells' differentiation potential nor caused major chromosomal abnormalities in the genome-edited PSCs.

Discussion
NHPs as our closest phylogenetic relatives are excellent animal models to study human diseases. The phylogenetic relationship is reflected in similar physiology, genetics, life span, and relatively equal size 5,6,[8][9][10]13 . In order to model human pathologies with genetic origin in NHPs, one of the main strategies is to edit the embryonic genome. Therefore, genome editing tools for these applications need to be validated in vitro before in vivo translation to guarantee the efficiency, accuracy and hence safety of the process. NHP-PSCs share significant parts of their molecular signatures with those of the pluripotent cells present in the early embryo. Therefore, testing of novel editing tools in PSCs will help to assess their performance in vivo 7,36,37 .
Four novel rhesus macaque iPSC lines were generated using our previously published 6-reprogrammingfactor transposon 27 . The generated iPSCs were pluripotent as indicated by pluripotency marker expression and teratoma formation. RT-PCR was performed with primer pairs able to discriminate between endogenous and exogenous origin of pluripotency-related genes. Robust expression of endogenous OCT4A, SOX2, NANOG, and c-MYC transcripts was detected in the iPSCs as judged by comparison with Rh_ESCs. However, the analysis also revealed the presence of piggyBac-derived exogenous transcripts showing that the expression cassette was not fully silenced in iPSCs. The potency of the novel NHP-iPSC lines was also evaluated by teratoma formation. All lines analyzed formed teratomas. We were wondering whether the piggyBac cassette would be silenced during iPSC generation or upon iPSC differentiation over several weeks in the context of teratoma development. Interestingly, most teratoma sections were negative for the pluripotency markers OCT4A, LIN28, TRA-1-60 and NANOG, for which we have established specific and sensitive immunohistochemical detection protocols 34,38 . We found only scattered and small clusters of pluripotency factor-positive cells. This indicates down-regulation of endogenous pluripotency genes and, even more important, also down-regulation of the reprogramming factor expression from the reprogramming construct in the differentiated cells of the teratoma.
In order to get first insights into piggyBac silencing in the teratoma, we performed comparative methylation analysis of the two CAG promoters present in the piggyBac construct isolated from iPSCs and the teratomas. There was no clear difference between the methylation patterns found in iPSCs and in teratomas. However, between the two promoters, i.e., the CAG puromycin and the CAG reprogramming, we found differential methylation patterns. Surprisingly, the CAG reprogramming was highly methylated (in most samples close to 100%) in comparison with the CAG puromycin promoter, which shows around 30% methylation in teratomas and around 20% in iPSCs. These findings were generally reproduced in all three independent iPSC lines. This suggests that the cells can discriminate between the two CAG promoters in the construct, which are sequence-wise identical. Either the gene products (indirectly) regulate the methylation of the promoters or the broader DNA sequence context in which the actual CAG promoters are embedded. Targeted methylation has been previously described for this and other reprogramming approaches 39 . Moreover, the high methylation of the CAG reprogramming was not only found in the teratomas, but also in the iPSC. This indicates that methylation of the reprogramming construct occurs mainly during fibroblast reprogramming and/or early passages of the iPSCs, but not during differentiation, as we initially hypothesized. Constitutive reprogramming factor expression could lead to www.nature.com/scientificreports/ differentiation failure and tumorigenicity 40,41 . However, in our experiments, we observed mainly differentiated cells within the teratomas. Even more important, also the iPSC line after deletion of the piggyBac cassette (see below) shows some remaining OCT4A-positive cells in the teratoma, indicating that not (only) the exogenous OCT4A expression was still detectable in the teratomas, but also the endogenous protein.
The piggyBac transposon can be removed from the genome of the cells without leaving a footprint 31,32,42,43 . Even though the teratoma showed silencing of the exogenous reprogramming factors in most of the cells, this would not be a safe condition regarding cell transplantation, and exogenous genetic material present in the iPSC genome is not acceptable for transplantation purposes. In order to generate transgene-free macaque iPSC, we exemplarily excised the reprogramming transposon from the genome of one iPSC line. The resulting transgenefree clone was then fully re-characterized. Pluripotency factor expression and potency of the cell line remained unchanged after transposon removal. In addition, karyotyping showed that the excision-ligation process did not generate any detectable chromosomal abnormalities. We are aware of reprogramming approaches based on non-integrating vectors like Sendai viruses and episomes 33 . However, we have shown that rhesus monkey iPSC generation is significantly less efficient than human iPSC generation 33 . Considering this, the integrating, yet reversible piggyBac approach for NHP-iPSC generation is still useful because of its robustness 43 . In fact, piggyBac-based reprogramming proved useful for the generation of marmoset monkey 27 , baboon 44 and rhesus monkey (present study) iPSCs. From the latter two species we obtained iPSCs also from adult and aged animals.
We adapted three of the generated NHP-iPSC lines to feeder-free conditions. Feeder-free culture is essential to obtain pure iPSC populations without the presence of other (feeder) cells. Furthermore, this allows upscaled production and facilitates more efficient use of biotechnological tools, like CRISPR/Cas9 45,46 . The cell lines remained undifferentiated under the new conditions. Importantly, also the transgene-free iPSC line was cultured in feeder-free conditions. Hence, feeder-free and transgene-free culture of piggyBac-derived rhesus iPSCs is possible.
Genome editing and single-cell cloning of PSCs is challenging due to the low efficiency of the process and stress-induced death of the stem cells forced to single-cell separation 28,29 . Fine-tuning of the protocols developed for human 29,47 and mouse PSCs 48 was required in order to generate a robust protocol for rhesus macaque PSCs 49 . We have demonstrated that accutase and versene are more suitable dissociation reagents for the generation of rhesus ESC single-cell suspensions than TrypLE. Furthermore, we tested different compounds and conditions to increase the number of surviving single cells/clones after sorting. ROCK inhibitor (pro-survival compound, PSC) shows a beneficial effect in single-cell survival in the two lines analyzed, while hypoxia and Activin showed no effect on the cloning efficiency in our experimental setup. Taking these results together, we developed a robust single-cell cloning protocol for rhesus PSC. Finally, we used a piggyBac vector containing eCas9-GFP plus guide RNAs to target two clinically relevant loci in the TTN gene. The piggyBac transposon combining all required components of the CRISPR/Cas9 editing machinery in one vector in combination with the fine-tuned protocol for single-cell cloning allowed us to reach high efficiency in the generation of mutated clones, e.g. over 85% for the C-terminal TTN mutation.
Altogether, we demonstrated the suitability of the piggyBac system to reprogram and gene edit rhesus macaque iPSCs. We developed an efficient platform to evaluate CRISPR/Cas-based genome editing approaches in PSCs before in vivo application. In combination with the new protocol for clonal expansion of gene-edited rhesus PSCs, this represents a useful in vitro screening platform for gene editing and thereby contributes to the 3Rs (reduce, replace, and refine) in animal experimentation. Differential methylation of the two identical CAG promoters present in the vector was encountered, with high methylation of the CAG reprogramming and low methylation of the CAG puromycin promoter. This may suggest a specific methylation response to the gene product controlled by the respective promoters and deserves further investigations. Finally, we have demonstrated removal of the reprogramming transposon by re-expression of the transposase, resulting in transgene-free clonal iPSCs cultured under feeder-free conditions. However, a higher efficiency of the removal procedure of the piggyBac transposon would be desirably.

Materials and methods
Animals. The German Primate Center (DPZ) is registered and authorized by the local and regional veterinary governmental authorities. Rhesus macaque skin samples were made available during necropsy from animals kept in the context of an unrelated project (Approval number of the Niedersächsisches Landesamt für Lebenmittelsicherehit und Verbraucherschutz 33.42502-04-16/2370). All animal experiments and methods were performed in accordance with relevant guidelines and regulations for animal use, and the institutional guidelines of the DPZ for the care and use of rhesus macaques were followed. All methods were carried out in accordance with the arrive guidelines. The cell lines DPZ_iRhpb#1-3 iPSC were derived from an adult male macaque (16 years) and the cell line DPZ_iRhpb#4 iPSC from a female (8 years). Regarding the use of mice in the context of the mouse embryonic fibroblast generation and the teratoma formation experiments, please see "Mouse embryonic fibroblasts (MEFs)" and "Teratoma formation and histological analysis", respectively.
In vitro differentiation. Rhesus  For PCR, Fast Start Taq polymerase (Roche) was used to amplify the bisulfite-converted DNA. PCR conditions and primers were designed with the Pyrosequencing Assay Design Software (Qiagen). The piggyBac 6-factor-reprogramming construct contains two CAG promoters, one driving the expression of the reprogramming cassette, and the other one the puromycin resistance gene. In order to analyze both promoters separately, one assay for each CAG region was designed. Both assays contained the same forward primer but a different reverse primer located in the unique sequence of the reprogramming cassette (CAG Reprog) and the puromycin resistance gene (CAG Puro). In addition, a third assay was designed to evaluate both CAG promoters (CAG). For reducing the amplification bias, all three measurements were arranged in a nested/semi-nested approach consisting of two following PCRs for each assay. The first PCR round generates three different outer amplicons related to the three different assays (CAG Reprog, CAG Puro and CAG). In the second PCR round, one primer combination of two primers located within the CAG promoter sequence was applied to all three different outer amplicons (CAG Reprog, CAG Puro and CAG) and generate the same inner amplicon for all three assays. This inner amplicon was sequenced with two sequencing primers, which together covered 15 CpGs in total (S1: 7 CpGs; S2: 8 CpGs). As cycler conditions of the outer-PCR reactions, the standard PCR procedure according to manufacturer's instruction was applied with a cycle number of 30 (annealing CAG Reprog: 60 °C; CAG Puro: 64 °C; CAG: 52 °C). The inner-PCR contained for all three assays the same combination of two primers. The cycler conditions of the inner-PCR corresponded to the standard PCR procedure according to manufacturer's instruction (annealing 53 °C and 35 cycles). Finally, the products of the inner-PCR were sequenced in triplicates by pyrosequencing following the manufacturer's instruction. 10 µL of the generated inner-PCR products were used for the immobilization to Streptavidin Sepharose HP beads (GE Healthcare Life Sciences). The pyrosequencing was performed on a Pyromark Q96MD system (Qiagen) and analyzed with the Pyromark Q-CpG software.
PiggyBac transposon removal/rhesus iPSC transfection. iRhpb#4 in feeder-free conditions was nucleofected using 4D-nucleofector device (Lonza). 1 × 10 6 cells were transfected with pBase-tdTomato transposase vector (6 µg), using P3 nucleofection solution and program CA-137 (Lonza). UPPS medium was supplemented for 2 days with PSF (5 µM) after transfection. After 3 days Tomato positive clones were sorted using SH800S Cell Sorter (Sony Biotechnology), sorting 10 cells per well in a 24 well plate. Potential transgene-free colonies were picked and expanded for 2-5 passages. Colonies were then pelleted and gDNA was extracted. Detection PCR was performed using primers specifically designed to detect different regions of the transposon (Supplementary Table 1). PCR was performed using Taq DNA polymerase with Standard Taq Buffer (New England BioLabs) according to manufacturer instructions. Original gel electrophoresis pictures of PCR analysis can be found in Supplementary Fig. 14. Transgene-free clones were checked for the absence of the reprogramming construct every 10 passages.
Transfection efficiency was evaluated 3 days after transfection by FACS sorting. The analysis was performed in triplicates for two iPSC lines in feeder-free conditions (iRhpb#2 and 4). 10 × 10 3 events were recorded for the analysis of each sample.
Dissociation reagent testing. Two stem cell lines, iRhpb#4 and Rh_ESC, were dissociated with three different reagents. Dissociation conditions are listed in detail in Fig. 6A. After dissociation, cells were collected and centrifuged (180g, 5 min, RT). After centrifugation 200,000 cells were transferred into a well of a 6-well plate and cultured for 5 days. Alkaline phosphatase staining and cell counting were subsequently performed. Statistical analysis of the cell numbers was performed using GraphPad PRISM and one way ANOVA test [(*) p < 0.05, (**) p < 0.01.].
T7 endonuclease I assay. T7 endonuclease I assay 35 was performed according to manufactures instructions (New England Biolabs, M0302L). In brief, CRISPR target sites were amplified from the genomic DNA of the transfected cells. PCR products were purified, heated to 95 °C and re-hybridized. Finally, T7 endonuclease I was added to the re-annealed PCR products. Cleavage of the PCR product by the endonuclease due to the presence of DNA mismatches was evaluated in 1.5% agarose gels. Original gel electrophoresis pictures of T7 endonuclease I assay can be found in Supplementary Fig. 15.
Clonal expansion analysis. iRhpb#4 and Rh_ESC were used for the testing of the clonal expansion potential after single-cell dissociation and FAC-sorting. The transfection efficiency of pTT-PB-pCAG-eCas9-GFP and pBase-tdTomato was evaluated three days after nucleofection according to the protocol detailed in "Promoter methylation assay" ("Materials and methods"). For single cell cloning, cells transfected with pTT-PB-pCAG-eCas9-GFP-U6-gRNA-Neo and pBase-tdTomato were cultured for three passages and digested with accutase (37 °C, 5 min). The SH800S Cell Sorter (Sony Biotechnology) was used to sort Cas9-GFP positive cells. To evaluate the effect of PSF during clonal expansion, a control group was sorted with no PSF and compared to a test group with PSF. For the evaluation of the effect of Activin A (50 µg/µL; Miltenyi), or hypoxia (5% O 2 ), the different experimental groups were compared to each other. Since PSF turned out in the first experiment to support clonal cell line derivation, it was the standard based on which the remaining variables (Activin and hypoxia) were tested. All experiments were performed in triplicates. For the significance analysis, paired T-test was performed using GraphPad PRISM software.

Induction of truncation mutations in rhesus macaque stem cells. Two different guide RNAs for
the CRISPR/Cas9 system were cloned and validated in the pTT-PB-pCAG-eCas9-GFP-U6-gRNA-Neo vector (Supplementary Table 1). Transfection of the iPSC was described above (piggyBac transposon removal/rhesus iPSC transfection). Transfection efficiency of the pTT-PB-pCAG-eCas9-GFP-U6-gRNA-Neo vector was evaluated 5 days after transfection by FACS sorting (according to 4.12 PiggyBac transposon removal/rhesus iPSC transfection). After transfection rhesus iPSCs and ESCs were culture for three passages with/without Geneticin (Gibco) (200 µg/mL). Antibiotic selection was used depending on the proportion of GFP-positive cells in the transfected population. If the percentage of GFP positive cells was estimated (by fluorescence microscopy) lower than 10%, Geneticin was added to the medium. After expansion the cells were FACS-sorted in order to obtain clonal sublines, expanded, and pelleted for gDNA extraction and analysis (Macherey-Nagel™ NucleoSpin™ Gel and PCR Clean-up Kit). Primers for the mutation site amplification and PCR conditions are described in Supplementary Table 1. PCR products of the different clones were purified and sent for sequencing (LGC Genomics). Sequence analysis was performed using Serial Cloner, Geneious and Chromas software. Mutations in the two different alleles of the target gene of each clone were predicted using Poly Peak Parser 51 .