Primer lists and cycling programs can be found in Supplementary Data 4 and Supplementary Table 1.
Target site identification
In order to nominate a suitable target site, we first identified a sequence of the inverted repeat int1h, by aligning the first intron of the F8 gene against a reverse-complement sequence of 200 kb DNA fragment located upstream of the F8 transcription start site (reference genome assembly hg38), using EMBOSS Water as an alignment tool42. In the following step, using a python script, the inverted repeat (1041 bp) was scanned for all occurrences of potential target sites, defined as palindromic repeats of 13 bp (half-sites) separated by 8 bp spacer sequence. The half-sites were allowed to have up to 7 positions of asymmetry. A set of 82 target sites fulfilling this criterion was sorted by a larger count of mismatches between one of the half-sites (left or right) and any of the half-sites recognized by previously evolved Cre-type recombinase libraries16,17. As the final target site, we picked the sequence with the highest score from the sorted list and named it loxF8.
In order to avoid potential off-targeting, we scanned the human genome for any occurrences of the selected target site, allowing up to one mismatch in each half-site and any sequence in the spacer region. A search for genomic sequences with the highest resemblance toward loxF8 was performed using an exhaustive short sequence aligner PatMaN, allowing up to 8 mismatches in both half-sites in total43.
Plasmid construction
Evolution vectors with the different target sites for SLiDE were cloned as described previously21. In short, primers were designed to carry the desired target sites and an overlap with the pEVO vector. The PCR fragment generated when using the pEVO vector as template was then cloned via Cold Fusion into a BglII digested pEVO backbone (System Biosciences).
A DNA fragment (synthesized by Twist Bioscience) coding for TRE3G-EF1a-Tet-ON® 3G-P2A-eGFP was inserted into the lentiviral backbone of the plentiSAMv2, a gift from Feng Zhang, (Addgene plasmid #75112; http://n2t.net/addgene:75112; RRID:Addgene_75112)44 utilizing NheI and KpnI restriction enzymes (NEB). The resulting plasmid (plentiX) can be used to clone recombinases under the control of a doxycycline inducible promoter employing BsrGI and XbaI restriction enzymes (NEB). In order to fuse the recombinases to EGFP the plasmid was modified and a TRE3G-NLS-eGFP-EF1a-Tet-ON®3G-P2A-PURO cassette (synthesized by Twist Bioscience) was inserted between the LTRs. The resulting plasmid was used to exchange EGFP with mCherry or tagBFP.
The plentiCRISPR v2, a gift from Feng Zhang, (Addgene plasmid #52961; http://n2t.net/addgene:52961; RRID:Addgene_52961)45 was used to generate the loxF8 reporter plasmid containing a SFFVpromoter-loxF8-PURO-loxF8-mCherry cassette between the LTRs.
Cloning and expression of recombinases
Single recombinases are cloned into the pEVO vector utilizing BsrGI and XbaI restriction enzymes (NEB). Dimer recombinases were cloned into the pEVO vector employing BsrGI and XbaI (second position) and SacI and XhoI restriction enzymes (first position). A Shine-Dalgarno (SD) sequence was located in front of each recombinase gene. In the pEVO vector carrying two recombinases, the SD sequence allows bicistronic expression of both recombinases. Expression of recombinases was controlled by an L-arabinose inducible promoter system (araBAD).
One or two recombinases were cloned into the mammalian plentiX recombinase expression vector under the control of a doxycycline inducible promoter. The first recombinase is cloned employing BsrGI and XbaI restriction enzymes (NEB). If the expression of two recombinases is intended, the stop codon of the first recombinase is removed and the second recombinase is cloned utilizing EcoRI and XhoI restriction enzymes. The plentiX vector harbors a SV40 NLS before each recombinase cloning site. A T2A site situated between the first and second recombinase gene enforced expression of both recombinases.
Transient expression of recombinases was achieved by cloning the recombinase gene in a mammalian expression vector (EF1a-Rec-P2A-EGFP) utilizing BsrGI and XbaI restriction enzymes (NEB). The STOP codon of the recombinase was removed to allow translational linking with EGFP using a P2A self-cleaving peptide sequence. Recombinase expression was driven by an EF1a promoter.
Substrate-linked directed evolution (SLiDE)
Recombinases were evolved using substrate-linked directed evolution as described previously14,16,17,21. A schematic overview of the experimental procedure is depicted in Fig. 1b. Recombinase libraries were evolved stepwise on different subsites to obtain active variants on the final target sites loxF8, loxF8-1, loxF8-2, loxF8-L, and loxF8-R.
In short, two parallel evolutions were set up to evolve either a recombinase monomer or a recombinase dimer being able to recombine the final asymmetric loxF8 target site. In the first step different pEVO vectors were generated containing the evolution sites (Supplementary Figs. 2 and 3) as a recombinase excision substrate. Diversified Cre-type recombinase libraries from previous evolutions were used to initiate the evolution process14,16,17,21. These libraries were cloned via XbaI and BsrGI in the pEVO vectors containing the first subsites (loxF8-1a, loxF8-2a, loxF8-L2 and loxF8-R2) and recombinase expression was induced at high levels (200 μg/ml L-arabinose). Recombinases active on the target sites will excise a part of the pEVO vector containing two unique restrictions sites (AvrII and NdeI) making this vector ‘immune’ to a subsequent digest with AvrII and NdeI-HF (NEB). These active recombinases were then retrieved using a PCR designed to only amplify from a circular template (Fig. 1b). This PCR step was performed with a low-fidelity DNA polymerase (MyTaq, Bioline, primers 79 + 80) to incorporate random mutations in the recombinase gene. The amplified recombinases were then cloned into a non-recombined pEVO vector and recombinase expression was induced again. This cycling process was repeated with lowering the recombinase expression by reducing the L-arabinose concentration (from 200 μg/ml down to 1 μg/ml) to select for the most active recombinase variants. SLiDE evolution was stepwise guided through the different evolution-target sites to obtain either an active monomer or an active dimer library for the final loxF8 target site (Supplementary Figs. 2 and 3).
Clonal analysis of recombinases
Recombinase activity was either analyzed with a plasmid-based assay or a PCR-based assay as previously described21. Schematics of both assays are depicted in Fig. 1b and Supplementary Fig. 6a. Briefly, recombination of the respective target sites on the evolution plasmid leads to the excision of a fragment from the plasmid. The resulting size difference is an indication for recombinase activity and can be detected by a restriction digest or by PCR. In order to detect the recombination by PCR a three-primer assay was used. The amplicon derived using primers 1 + 2 (Supplementary Fig. 6) detects the non-recombined product as the primer 2 binds in between the lox-sites. The resulting PCR product is 475 bp. Another primer set (1 + 3, Supplementary Fig. 6) will generate an amplicon (400 bp) from the recombined plasmid. The elongation time of the PCR was chosen so that primer 1 + 3 will not generate any amplicon on non-recombined templates.
Recombination quantification
Recombination efficiency of candidate recombinases was quantified using a plasmid-based assay. The recombined pEVO is smaller in size compared to the non-recombined version. After linearization, this size difference can be visualized on a standard agarose gel by gel electrophoresis. The bigger band (~5 kb) shows non-recombined substrate, while the smaller band (~4.3 kb) shows recombined substrate. Recombination efficiencies were calculated by measuring the bands intensities using Fiji. The average recombination efficiency was calculated from gel images of three independent biological experiments.
Cell culture of HEK293T and HeLa cells
HEK293T (ATCC) and HeLa (MPI-CBG, Dresden) were cultured in DMEM, Dulbecco’s modified Eagle’s medium (Gibco) with 10% fetal bovine (tetracycline-free) serum and 1% Penicillin- Streptomycin (10,000 U/ml, ThermoFisher).
Fluorescent activated cell analysis
HEK293T, HeLa, or iPSCs were washed once with PBS and then detached using Trypsin (Gibco) or Accutase (Sigma). Cells were resuspended in FACS buffer (PBS with 2.5 mM EDTA and 1% BSA) and analyzed with the MACSQuant® VYB Flow Cytometer (Miltenyi) or the BD FACSCanto™ II Cell Analyzer (BD). Analysis of the data was performed using FlowJo™ 10 (BD). Gating strategies are displayed in Supplementary Fig. 21.
Long term expression of recombinase libraries
Recombinase libraries were cut out of the evolution vectors via BsrGI and XbaI restriction sites, purified from an agarose gel and subsequently ligated to the plentiX viral vector. HeLa (MPI-CBG Dresden) cells were transduced with lentiviral particles generated from the lentiX-loxF8-L and lentiX-loxF8-R libraries. Transduced HeLa cells will express EGFP and recombinase expression can be induced upon administration of doxycycline. Cells were transduced with a MOI of one to enrich for clones harboring only one recombinase integration per cell. Recombinase expression was induced for 21 days with 100 ng/ml doxyclcline and the medium was renewed every other day. Next, genomic DNA was isolated using the QIAamp DNA Blood Mini Kit (Qiagen) and recombinases were retrieved by PCR using the high-fidelity Herculase II Phusion DNA polymerase (Agilent). Retrieved recombinase libraries were ligated into the evolution vectors and the activity of single clones and libraries was assessed by PCR or by a digest as previously described earlier.
Generation of the HEK293TloxF8 and HEK293TloxP cell lines
HEK293T (ATCC) cells were transduced with lentiviral particles generated from the SFFVpromoter-loxF8-PURO-loxF8-mCherry or SFFVpromoter-loxP-PURO-loxP-mCherry plasmids. The cells were exposed to 2 µg/ml puromycin selection 48 h after transduction for 7 days. Different concentrations of viral particles were used for transduction to estimate the MOI. The viral load that resulted in a transduction efficiency of ~5% (95% of cells were depleted during the puromycin selection) was used to establish reporter cell lines with an estimated MOI of 1. Genomic DNA was isolated from the surviving cells and a reporter-specific PCR was performed. The PCR fragment was sequenced to confirm that the reporter construct was integrated in the genome.
In vitro transcription
Recombinase, eGFP, tagBFP, and mCherry mRNA was produced by in vitro transcription (IVT) using the HiScribeTM T7 ARCA mRNA Kit (NEB) and purified using the Monarch© RNA Cleanup Kit (NEB). The DNA templates for the IVT were generated by PCR using the lentiviral plasmids with eGFP (primers 1 + 2), mCherry (primers 3 + 4), tagBFP (primers 5 + 6). Recombinase DNA templates for IVT were generated by PCR using the pEVO vectors as templates with D7 (primers 7 + 8 and 9 + 10) or RecF8 (primers 11 + 12). Cycling program 3 was used for the mRNA transcription and poly(A) tailing reaction.
mRNA transfection
IVT produced mRNA was transfected using Lipofectamine™ MessengerMAX™ Transfection Reagent (ThermoFisher). HEK293TloxF8 cells were transfected in a 12-well format, seeded at a density of 250,000 cells/well and transfected 24 h after seeding. For each well 250 ng of mRNA (200 ng recombinase mRNA and 50 ng tagBFP mRNA) and 2 μl Lipofectamine™ MessengerMAX™ was mixed with 50 μl Opti-MEM I Reduced Serum Medium (ThermoFisher, prewarmed at RT). After 5 min of incubation at RT the mRNA and Lipofectamine™ MessengerMAX™ mixtures were combined, shortly vortexed and incubated for 10 min. The transfection mixture was directly added to the cells without changing the medium. Cells were analyzed 48 h post transfection by FACS and fluorescent microscopy.
iPSCs were cultured using StemFit basic 02 (AJINOMOTO) and iMatrix-511 silk laminin coating (NIPPI) as described elsewhere46. In short, iPSCs were maintained in 6-well plates using StemFit 02 (Reprocell/AJINOMOTO) supplemented with 100 ng/ml FGF2 (R&D) and iMatrix-511 silk laminin coating (NIPPI). 10 μl laminin in 2 ml PBS for each well was used for coating and incubated 1 h at 37 °C. Confluent iPSCs were washed once with 2 ml PBS and detached using 1 ml Accutase (TheromFisher) for 1 min at 37 °C. Cells were collected in 10 ml DMEM F12 (ThermoFisher) and pelleted for 3 min at 300 × g. The pellet was dissolved in 5 ml DMEM F12. Depending on the cell line 1 × 105–5 × 105 cells/well were plated on coated plates. The first 24 h after splitting the medium was supplemented with 10 uM Rock-inhibitor (Y-27632, Tocris). Medium change was performed daily. iPSCs were transfected as previously described47.
Detection of the int1h inversion by PCR on genomic DNA
Genomic DNA of HEK293TloxF8, iPSCs and ECs was isolated 48 h post transfection using the QIAamp DNA Blood Mini Kit (Qiagen). Two sets of primers were designed to detect the orientation of the 140 kb DNA fragment between the two loxF8 target sites. Primers (15 and 18) are located outside the fragment and do not change upon recombination. Primers (16 and 17) bind inside the 140 kb fragment and will change their orientation upon recombination. Primer pairs 15 + 16 and 17 + 18 were used to detect the WT orientation of the 140 kb fragment. Primer pairs 15 + 17 and 16 + 18 were used to detect the inverted variant. Independent of the orientation and the primer combinations, cycling program 1 was used for the PCR.
Linker selection
The linkers of eight GGS repeats were synthesized by Sigma-Aldrich as annealing compatible oligonucleotides. The designed linker library contained the core 12 amino acids coded by the degenerate codon RVM that were flanked by two GGS repeats from each side, and the sequence was synthesized. In both cases, the linker sequences were inserted in the pEVO plasmid via XhoI and BsrGI, between the two recombinase monomers. The linker selection was performed following the SLiDE protocol, with the exception of using the high-fidelity Herculase II Phusion DNA polymerase (Agilent) in order to select the heterodimer fused with a linker with the best properties without introducing new mutations in the recombinase sequences. By varying selection of active and inactive recombinase heterodimers on the loxF8 and symmetric sites (loxF8-L and loxF8-R), respectively, the counter selection was performed. Active recombinases on the loxF8 site were retrieved the same way as described previously in the SLiDE procedure. Inactive variants for loxF8-L and loxF8-R were retrieved by PCR using a primer binding between two recombinase target sites and another primer binding before the recombinase dimer (primers 79 + 81). Only recombinase coding sequences carrying an inactive recombinase for loxF8-L or loxF8-R will be amplified. This simple assay selection of linked heterodimers that are able to recombine the final asymmetric loxF8 site, but do not recombine the symmetric loxF8-L and loxF8-R target sites.
Inversion quantification
Quantification of the inversion efficiencies was performed with a qPCR-based assay. In order to detect the WT orientation primer pair 17 + 18 was used together with a TaqMan specific probe for this amplicon. To detect the inversion orientation primer pair 16 + 18 was used. For both reactions cycling program 2 was used. A standard curve of 1%, 5%, 10%, 25%, 50%, and 100% inversion was generated by mixing genomic DNA of WT iPSCs and F8 iPSCs at appropriate rations. The standard curve was used to extrapolate the inversion efficiency of the genomic DNA samples. As the standard curve was generated using genomic DNA of male iPSCs (one X-chromosome), the calculated inversion efficiencies for the HEK293T cells (female, two X-chromosomes) were divided by two.
ChIP-Seq and qPCR validation of potential binding sites for RecF8
RecF8 recombinase was fused with EGFP and cloned in a modified version of the tetracycline-inducible plentiX vector. Hela cells were infected with the lentivirus and selected with 2 µg/ml puromycin 48 h after transduction for 7 days.
Cells were grown in 10 cm dishes and the expression of RecF8 or EGFP (control) was induced for 24 h with 100 ng/mL doxycycline. The cell lines were crosslinked with 1% formaldehyde for 10 min at room temperature and further processed following the manufacturer’s protocol for High Cell number using the kit TruChip Chromatin Shearing Kit (Covaris). Chromatin shearing was performed using a Covaris M220 sonicator. 1% of the sheared chromatin was separated for qPCR validation (input sample) and the rest was used for immunoprecipitation.
Sonicated chromatin was immunoprecipitated using a goat GFP-antibody (MPI-CBG antibody facility, 1:5000) and Protein G sepharose beads (Protein G Sepharose® 4 Fast Flow, GE Healthcare). Eluates were reverse crosslinked followed by RNA and protein digestion.
Sequencing libraries were prepared using NEBNext® Ultra™ DNA Library Prep Kit for Illumina® from 17 to 75 ng of ChIP DNA with 15 PCR amplification cycle and size selection using AMPure XP beads. Paired-end sequencing was performed on an Illumina HiSeq 2000, aiming for approximately 25 million pairs of sequencing reads per sample, with each read being 76 bp long. Sequencing reads were aligned to a human reference genome assembly GRCh38.p1248 using STAR aligner49 tuned for ChIP-Seq analysis pipeline (disabled intron detection, 400 bp of maximum gap between read mates, up to 50 reported alignments of multi-mapper reads). Peak calling was performed with Genrich (https://github.com/jsh58/Genrich), using the ENCODE blacklist (v2)50 for filtering out problematic regions. All steps involving manipulations and comparisons of genomic intervals were done using BEDTools51. Visualizations of the ChIP-Seq pile-up signals were generated with the USCS Genome Browser52 (USCS genome browser tracks are available upon request), directly from BAM files sorted by samtools command line tool53.
Ten out of twelve peaks that were identified by ChIP-Seq were additionally tested by qPCR for recombinase binding (Primers 49-78). qPCR analysis comparing immunoprecipitated samples and input samples was performed using SYBR green MasterMix (Thermo Scientific ABsolute qPCR SYBR Green Mix).
De-novo motif discovery was performed with MEME-ChIP script from the MEME suite54. It was executed with the following set of arguments: “-order 2 -seed 0 -meme-mod oops -meme-minw 13 -meme-maxw 34 -meme-nmotifs 3 -centrimo-local” and a target database of loxP and loxF8 sequences used at a motif comparison stage. To generate an input file, BEDTools51 were used to extract 330 bp-long sequences centered at positions of 85 peak summits reported by the ChIP-Seq analysis.
Recombination assays of ChIP-seq peaks in bacteria and human cells
Twelve peak sequences were tested in a plasmid-based assay for recombination in bacteria and human cells. A DNA insert (95 bp) around each peak was generated by PCR (primer 23–46) using cycling program 4 and cloned into the pEVO vector via BglII digestion and ligation. After expression of the RecF8 recombinase at 100 µg/ml L-arabinose in E. coli, plasmid extraction was performed and recombination of the peak sequence was analyzed by gel electrophoresis.
In order to test the putative off-target sequences in HEK293T cells, reporter plasmids were generated based on the pCAG-loxPSTOPloxP-ZsGreen (Addgene plasmid # 51269; http://n2t.net/addgene:51269; RRID:Addgene_51269)55. The plasmid was modified as described by Lansing et al.21. In short, the ZsGreen was exchanged for mCherry and the loxP sites were exchanged with the 95 bp fragment of the 12 different peak hits. Successful recombination will remove a stop cassette, allowing for the expression of mCherry. RecF8 or inactive Cre324 were cloned in the transient mammalian expression vector (EF1a-Rec-P2A-EGFP) and co-transfected to HEK293T cells together with the different peak reporter vectors. Recombination was measured via FACS 48 h after transfection.
HG2 off-target translocation detection
Primers were designed around the HG2 off-target (chromosome 15, primers 19 and 20) site and its potential recombination-target site HG2-1 (chromosome 7, primers 21 and 22). PCR products generated from these combinations were sequenced and the presence of the off-targets was validated. Next, a combination of primes as depicted in Supplementary Fig. 16 was used for PCR. In order to validate the sensitivity of the PCR and to have a positive control, a DNA fragment that resembles the translocation product if HG2 is recombined was synthesized. The sensitivity of the PCR was tested by a serial dilution of the DNA fragment (1:102–1:107). PCR products of primer combinations 21 + 19 and 20 + 22 should reveal a translocation event. PCR program 5 was used for the amplification. The PCR reaction was then carried out on genomic DNA from human cells treated with and without D7 and RecF8. For a positive control genomic DNA of untreated HEK293T cells was mixed with the synthesized translocation fragment for TL7/15 and TL15/7 (Supplementary Fig. 16) in a ratio that would equal 1:200 genome equivalence.
IPSCs and endothelial cell culturing, differentiation and transfection
Reprogramming, culturing, and characterization of patient-derived iPSCs (F8 iPSCs) were performed at the Stem Cell Engineering Facility of the Center for Molecular and Cellular Bioengineering (CMCB) at TU Dresden using the CytoTune-iPS 2.0 Sendai Reprogramming Kit (Thermo Fisher Scientific, A16517, Waltham, MA, USA) according to the manufacturer’s guidelines. Human iPSCs were cultured in StemFit02 and laminin coating as described above. Endothelial cells were differentiated from F8 iPSCs and maintained in a 6-well dish as described elsewhere46. In short, 3 × 105 iPSCs were seeded for each well on a laminin coated 6-well plate (day 0) using 2 ml of StemFit 02 medium. The first 3 days cells were cultured in the following medium: DMEM F12 with 1%B27, 1% Glutamax, 8 μM CHIR and 25 ng/ml BMP4. The next 3 days cells were cultured in following medium: StemPro34 with 200 ng/ml VEGF and 2 μM Forskolin. Cells were split in on day 7 after seeding. 6-well plates were coated for 1 h at 37 °C with 2 ml PBS containing 10 μg/cm2 fibronectin. ECs were washed using 2 ml PBS per well and detached for 2 min at 37 °C using 0.3 ml Trypsin. ECs were collected in DMEM F12 and pelleted for 3 min at 300 × g. The pellet was resuspended in EC maintenance medium (StemPro34 with 100 ng/ml VEGF). The cells were split in a ration of 1:4 and seeded on the coated 6-well plates. For each well 700 ng mRNA (250 ng mRNA of each recombinase monomer or 500 ng of the linked RecF8 mRNA and 200 ng mCherry mRNA) and 4 μl Lipofectamine™ MessengerMAX™ was mixed with 100 μl Opti-MEM I Reduced Serum Medium (ThermoFisher, prewarmed at RT) and 4 μl Lipofectamine™ MessengerMAX™. The mRNA and Lipofectamine™ MessengerMAX™ mixtures were combined and incubated 15 min at RT. Afterward the transfection mixture was added to the ECs. Four hours after transfection the EC maintenance medium was renewed and the cells were analyzed 48 h post-transfection.
Factor VIII qPCR
RNA was isolated using the RNeasy mini kit (Qiagen). Reverse transcription was carried out using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturer’s protocol. qPCR was carried out using the ABsolute QPCR Mix, SYBR Green, no ROX (Thermo Fisher) on the C1000 Touch CFX96 Real-Time System (BioRad). In order to detect the right boundary of exon 1 and exon 2 after recombinase mediated correction of the genomic inversion, specific primers (47 and 48) in exon 1 and exon 3 were used for quantification. F8 mRNA expression was normalized to TATA-Box Binding Protein (TBP).
Whole-genome sequencing of ECs
Genomic DNA was extracted from mRNA transfected patient-specific ECs 72 h post transfection. The average size of extracted DNA was determined using a fragment analyzer (average size 10–50 kb). Libraries were prepared following the manufacturer’s instructions with 1.25–1.5 μg using the Ligation Sequencing Kit SQK-LSK109 (Oxford Nanopore Technologies). 1 μl of DNA Control sequence (DCS) was included in the library preparation. Samples were sequenced using the PromethION (Oxford Nanopore Technologies). Runs were set to 72 h, the “non-treated” sample was washed and reloaded after 44 h. “Cre-treated” and “Rec8-treated” flow cells were refueled after 44 h. Sequencing reads were aligned to the reference genome using minimap2 (ver. 2.17)56 with the following command-line arguments: “–MD -a -x map-ont”. The reference sequence was modified from GENCODE Release 26 (GRCh38.p10)57 to accommodate for the patient-specific chrX inversion between positions 155,007,147 and 155,147,780. The minimap2 output was converted into BAM format with samtools58 Variant calling was performed with two software packages: Sniffles (ver. 1.0.12)59 and npInv (ver. 1.28)60, where the latter was exclusively used for calling inversions. Each sample was processed independently and variants predicted from each sample were merged into a single vcf file using SURVIVOR (ver. 1.0.7)61 with arguments set to “500 1 1 -1 -1 -1”. In case of the Sniffles pipeline, the variant calling step was repeated in a forced genotyping mode, using the merged file as a fixed reference, as described on the Sniffles web page (https://github.com/fritzsedlazeck/Sniffles/wiki/SV-calling-for-a-population). Sniffles was executed with “−max_distance 500” parameter, and a “−min_support” parameter set to 3, in case of the first run, or -1 in case of the latter genotyping mode. Inversions were detected by npInv executed with “−min 150 −max 500000 −threshold 1” arguments and merged with SURVIVOR. In case of npInv, the forced genotyping mode is not available. Using bcftools58, resulting vcf files were converted into a tab-separated format containing positions of variants and counts of reads supporting either a reference or a variant allele.
Counting and filtering of variants was performed in R using the GenomicRanges package (ver. 1.44.0)62. In order to exclude patient-specific variants, all variants detected in recombinase-treated samples having their rearrangement breakpoints overlapping (±500 bp) with breakpoints detected in the non-treated sample were filtered out. To nominate variants potentially caused by recombinase activity, the following additional criteria were applied: a variant has to be either an inversion, a translocation or a deletion; in case of deletions and inversion, a distance between breakpoints has to be at least 150 bp; the total number of reads supporting either a reference or a variant allele has to be more than 10 and a fraction of variant-supporting reads cannot exceed 20%. The WGS data is deposited at the European Genome-phenome Archive (EGA) under the ID EGAS00001005496.
Factor VIII immunohistochemistry
iPSCs were cultured in 6-well plates (Corning) and were differentiated into endothelial cells as described earlier. At day 3 cells were fixed in 4% paraformaldehyde (PFA) at room temperature for 20 min, permeabilized and immunostained following standard protocol using anti-Factor VIII antibody (Abcam, ab236284, 1:200) and fluorescently labeled secondary antibody (Invitrogen, Alexa fluor 546 goat anti-rabbit, 1:500). Additionally, nuclei were labeled with Hoechst 33342 (Invitrogen, H3570, 1:5000). Images of cells were automatically captured using an automated Operetta CLS confocal microscope (PerkinElmer) at 20× magnification. Subsequent image capturing was performed using the PerkinElmer Columbus Image Analysis System as previously described63.
CRISPR/Cas9 and RecF8 fidelity
In order to compare the accuracy of recombinase-based inversion with nuclease-based inversion, we tested these two different approaches in patient-specific endothelial cells (ECs). Transfections were performed in a 6-well format as described earlier using mRNAs together with Lipofectamine™ MessengerMAX™. One sample was co-transfected with 500 ng of RecF8 mRNA and 200 ng of mCherry mRNA. The other sample was transfected with 1000 ng Cas9 mRNA (TriLink® BioTechnologies) and 10 pmol of gRNA (5′-GGUCCCCGGGGUUGUGCCCC-3′) targeting the inverted repeat as published by Park et al.7 Two days after transfection genomic DNA was isolated and analyzed for the inversion accuracy. The recombinase or CRISPR/Cas9 mediated inversion product was amplified using primers 16 + 18 spanning over the loxF8 target site and the gRNA binding site. The recombinase or CRISPR/Cas9 mediated inversion products were amplified using primers 15 + 16 or 17 + 18 spanning over each of the loxF8 target sites and the gRNA binding site. The PCR products were purified using ISOLATE II PCR and Gel Kit (Bioline, Meridian Bioscience), cloned into pMiniT 2.0 plasmid, and transformed into NEB 10-beta E. coli using the NEB PCR Cloning Kit (NEB) according to the manufacturer’s instructions. For each sample, twenty-two colonies were picked and sent for E. coli overnight Sanger sequencing (Microsynth) with the cloning analysis forward primer provided with the kit.
Statistics and reproducibility
Statistical analysis was performed using GraphPad Prism 8. Relevant details of the statistical test are provided in the figure legends. Experiments that were further statistically analyzed or quantified were replicated and reproduced independently. Representative gel pictures were not replicated for restriction digests (e.g., activity of recombinase libraries during SLiDE or single clones) or PCR reactions (e.g., PCR-based activity of single clones or orientation-PCR of the loxF8 locus in human cells) if no comparative quantification was performed.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
