1. Stem Cells and Regenerative Medicine
  2. Chromosomes and Gene Expression
Download icon

Targeted, homology-driven gene insertion in stem cells by ZFN-loaded ‘all-in-one’ lentiviral vectors

  1. Yujia Cai
  2. Anders Laustsen
  3. Yan Zhou
  4. Chenglong Sun
  5. Mads Valdemar Anderson
  6. Shengting Li
  7. Niels Uldbjerg
  8. Yonglun Luo
  9. Martin R Jakobsen
  10. Jacob Giehm Mikkelsen  Is a corresponding author
  1. Aarhus University, Denmark
  2. Technical University of Denmark, Denmark
  3. Beijing Genomics Institute, China
Research Advance
  • Cited 8
  • Views 1,089
  • Annotations
Cite as: eLife 2016;5:e12213 doi: 10.7554/eLife.12213

Abstract

Biased integration remains a key challenge for gene therapy based on lentiviral vector technologies. Engineering of next-generation lentiviral vectors targeting safe genomic harbors for insertion is therefore of high relevance. In a previous paper (Cai et al., 2014a), we showed the use of integrase-defective lentiviral vectors (IDLVs) as carriers of complete gene repair kits consisting of zinc-finger nuclease (ZFN) proteins and repair sequences, allowing gene correction by homologous recombination (HR). Here, we follow this strategy to engineer ZFN-loaded IDLVs that insert transgenes by a homology-driven mechanism into safe loci. This insertion mechanism is driven by time-restricted exposure of treated cells to ZFNs. We show targeted gene integration in human stem cells, including CD34+ hematopoietic progenitors and induced pluripotent stem cells (iPSCs). Notably, targeted insertions are identified in 89% of transduced iPSCs. Our findings demonstrate the applicability of nuclease-loaded ‘all-in-one’ IDLVs for site-directed gene insertion in stem cell-based gene therapies.

https://doi.org/10.7554/eLife.12213.001

Introduction

Uncontrolled genomic insertion of transgenes, leading to insertional mutagenesis, remains a primary challenge for safe genetic therapies. Lentiviral vectors offer an effective gene delivery route, but also suffer from prevalent gene insertion into transcriptionally active regions (Schröder et al., 2002; Modlich et al., 2009). Novel strategies to attain targeted gene integration (TGI) into predetermined genomic loci are highly warranted. TGI using lentiviral vectors can be achieved by delivering donor sequences for insertion by homologous recombination (HR) into double-strand breaks (DSBs) generated by tailored endonucleases (Lombardo et al., 2011). Development of programmable nucleases, ZFNs, TALENs, and now CRISPR/Cas9, has made effective gene disruption feasible, even in stem cells, but genetic operations involving TGI in stem cells are still challenged by low membrane permeability, low HR efficiency, and high sensitivity to external manipulation (Genovese et al., 2014Hockemeyer et al., 2011; Hendel et al., 2015Tay et al., 2013Tiyaboonchai et al., 2014).

We previously established genomic editing in cells treated with ‘all-in-one’ lentiviral particles loaded with locus-targeted ZFN proteins and a donor sequence flanked by homology arms (Cai et al., 2014a). Here, building on these findings, we report a simple method for establishing TGI by treating cells with integrase-defective lentiviral vectors (IDLVs) packaged with both the transgene-containing donor template and ZFN proteins fused to the N-terminus of lentiviral Gag and GagPol polypeptides. Locus-directed cleavage and TGI is evident in cell lines, CD34+ hematopoietic stem cells, and induced pluripotent stem cells (iPSCs) upon transduction with ZFN-loaded vectors. Notably, the prevalence of TGI in two safe harbors in the human genome, the CCR5 and AAVS1 loci, is high in iPSCs treated with ZFN-loaded lentiviral vectors, resulting in non-mosaic clones harboring a site-directed gene insertion and no additional cutting at top-ranked off-target sites.

Results

IDLVs harboring both ZFN protein and vector RNA with the transgene flanked by homology arms are here designated ‘IDLV-ZFN(locus)/gene’, where ‘gene’ refers to the transgene cassette flanked by homology arms and transferred by the vector. Principles of production, transduction, and integration of such vectors are illustrated in Figure 1A. Together with a dimer of lentiviral vector RNAs, approximately 5000 copies of Gag and GagPol polypeptides (of which about half carry a ZFN fused to the matrix protein) form the capsid, which is wrapped in a segment of the membrane during budding from the producer cells. ZFNs are released from Gag and GagPol upon cleavage by the viral protease during maturation of the virus particle. The mature virus particles contain two types of ZFNs, left and right, which upon DNA binding in the target cell form a catalytically active dimer. After viral entry in the target cells and endosomal escape, vector RNA is reverse transcribed to DNA, which can serve as donor for homology-directed insertion at targeted breakage sites generated by the co-delivered ZFNs (Figure 1A, lower panel).

Figure 1 with 1 supplement see all
Proof-of-efficacy studies and characterization of ZFN-loaded IDLVs carrying a transgene flanked by homology arms facilitating insertion by homologous recombination.

(A) Schematic illustration of the production, transduction and integration of IDLVs carrying ZFNs and transgene. Enrichment of ZFN-fused viral polypeptides and donor-carried viral RNAs at the cell membrane leads to assembly of ZFN-containing virions. During virus maturation, the polypeptides are cleaved into enzymatic and structural proteins by HIV protease. In the transduced cells, the viral RNAs are reverse transcribed to DNA that serves as donor for repair of ZFN-induced DSB by HR, leading to site-specific insertion of a gene expression cassette (here containing the puro gene). RT, reverse transcription; HR, homologous recombination. (BD) Immunostaining and confocal microscopy of producer cells, virions, and transduced cells. Green, HA-tagged ZFN-GagPol or ZFN; blue, nuclei; red, p24; scale bar indicates 5 μm. (B) HEK293T cells co-transfected by pMD.2G, pHA-ZFNL(gfp)-PH-gagpol-D64V and pHA-ZFNR(gfp)-PH-gagpol-D64V. (C) Incorporation of ZFNs into lentiviral particles was confirmed by co-localization of HA-ZFNs and p24 (see also Figure 1—figure supplement 1B). (D) Recording of the ZFN lifespan in HEK293 cells treated with LP-HA-ZFN(gfp) by immunostaining at different time points after treatment. (EG) TGI by IDLV-ZFN(AAVS1)/egfp in HEK293T cells. (E) IDLV-ZFN(AAVS1)/egfp transduction resulted in higher percentage of eGFP+ cells relative to the IDLV/egfp counterpart after 10 and 18 days. (F) PCR analyses of TGI at the AAVS1 locus. Illustration shows primer designs with one primer matching inside the insert and one in the flanking region allowing amplification of 5’ and 3’ junction sites (5’ and 3’ JS). Amplification of the CCR5 locus included as a control. (G) Rate of imprecise gene repair by NHEJ in target cells. (H-J) TGI by IDLV-ZFN(AAVS1)/fluc in HEK293T cells. (H) Higher levels of Fluc expression in IDLV-ZFN(AAVS1)/fluc-treated HEK293T cells relative to IDLV/fluc-treated cells 10 and 18 days after transduction. (I) PCR analyses of TGI at the AAVS1 locus by IDLV-ZFN(AAVS1)/fluc. (J) Rates of imprecise NHEJ in target cells induced by IDLV-ZFN(AAVS1)/fluc. (K) Schematic representation of Southern blot for analysis of TGI at AAVS1 locus using IDLV-ZFN(AAVS1)/puro. The probe used is indicated by the dark blue line; cleavage site for the used restriction enzyme (BstXI) is also indicated. (LM) Southern blot analyses of puromycin-resistant HEK293T clones resulting from treatment with IDLV-ZFN(AAVS1)/puro (L) and IDLV/puro (M). Fragments indicated with * are results of NHEJ leading to imprecise transgene insertion in the AAVS1 locus.

https://doi.org/10.7554/eLife.12213.002

To visualize the enrichment of ZFN-containing lentiviral polypeptides at the membrane of the producer cells, we transfected HEK293T cells with GagPol plasmid harboring either the coding sequence for a HA-tagged ZFN (HA-ZFN(gfp)) recognizing the egfp gene or the egfp gene itself fused to the 5’-end of gag gene and analyzed the virus-producing cells using confocal microscopy (Figure 1B and Figure 1—figure supplement 1A). By staining with an antibody specific for the tag, we found that the ZFNs were highly enriched at the cell membrane (Figure 1B). We then visualized the lentiviral particles (designated LPs since the vector was not included) and found for both LP-HA-ZFN(gfp) and LP-eGFP co-localization of the fusion protein with viral p24 protein (Figure 1C and Figure 1—figure supplement 1B and C), suggesting that ZFN (or eGFP) proteins were indeed packaged into the LPs. To measure the lifespan of LP-delivered ZFNs in transduced cells, HEK293 cells were synchronized at 4°C for 60 min before fixation at different time points (1, 12, 24, and 48 hr, respectively) after transduction with ZFN-loaded LPs. Already within the first hour after exposure to LP-HA-ZFN(gfp), the ZFNs were easily detectable inside the cells, whereas after 24 hr essentially all ZFNs had been degraded or diluted (Figure 1D). These findings showed that nucleases delivered by LPs were immediately available after transduction and that the time of action was restricted due to decay of the proteins.

We then performed proof-of-efficacy studies by transducing HEK293T cells with IDLVs (500 ng p24), with or without AAVS1-directed ZFNs, carrying an eGFP expression cassette (IDLV-ZFN(AAVS1)/egfp). Initially, three days post-transduction, a higher frequency of eGFP+ cells was observed among cells transduced with IDLV/egfp than for cells transduced with IDLV-ZFN(AAVS1)/egfp (59.1% versus 44.7%), indicating that the overall transduction potential was higher for IDLVs that did not carry ZFNs (Figure 1E and Figure 1—figure supplement 1D and 1E). Nevertheless, at later time points 2.5% of the cells were eGFP+ after treatment with IDLV-ZFN(AAVS1)/egfp, whereas the percentage dropped to below 1% in IDLV/egfp-treated cells (Figure 1E and Figure 1—figure supplement 1E). PCR amplification and sequencing of the 5’ and 3’ junction sites including part of the donor sequence and the regions flanking the inserted transgene cassette showed that events of correct integration into AAVS1 locus were evident after treatment with IDLV-ZFN(AAVS1)/egfp, whereas PCR products indicative of site-directed insertion did not appear for IDLV/egfp (Figure 1F). However, DSBs were not repaired by HR only, and mismatches introduced by non-homologous end joining (NHEJ) were detected in 5% of the AAVS1 alleles, as measured by Surveyor nuclease assay (Figure 1G). To evaluate the levels of TGI with another reporter gene, we treated HEK293T cells with IDLV-ZFN(AAVS1)/fluc carrying AAVS1-targeted ZFNs and a firefly luciferase (fluc) expression cassette. Indeed, fluc gene expression was maintained for up to 18 days after transduction with IDLV-ZFN(AAVS1)/fluc (Figure 1H), although the levels dropped from the initial 3-day time point, where episomal forms were still likely to support expression. For IDLV/fluc, in contrast, luminescence dropped to almost undetectable levels. TGI of the fluc gene into the AAVS1 locus was verified by PCR (Figure 1I), and the rate of NHEJ estimated to be 8% (Figure 1J).

Next, we created ZFN-loaded IDLVs carrying the puromycin resistance gene (IDLV-ZFN(AAVS1)/puro) allowing us to easily isolate single puromycin-resistant clones and further scrutinize AAVS1-targeted integration events by Southern blot analysis (Figure 1K). Among 16 clones generated with IDLV-ZFN(AAVS1)/puro, a 2.2-kb fragment, indicative of precise TGI by HR, was detected in 11 clones (Figure 1L). Other types of integrations, resulting in lower-mobility bands, were evident in the five remaining clones. Notably, two clones (clone 5 and clone 13) showed a 3.7-kb band, and these clones therefore were among clones that were analyzed by Southern blotting using another restriction enzyme and an AAVS1-specific probe (Figure 1—figure supplement 1F and 1G). A common 6.4-kb band was observed for clones 5 and 13 (Figure 1—figure supplement 1G), suggesting that the full-length lentiviral DNA in these two cases was captured and inserted by chance into the ZFN-induced DSB at the AAVS1 locus. These findings demonstrated that site-directed gene insertion into ZFN-generated DSBs also occurred, although less frequently, in an HR-independent fashion.

We analyzed 15 additional puromycin-resistant clones by PCR and found that 13 out of these clones yielded PCR products indicative of TGI (Figure 1—figure supplement 1H). In total, 84% (26/31) of the analyzed clones contained a targeted insertion of the puro gene in the AAVS1 locus. In contrast, the fragments detected by Southern blot analysis of IDLV/puro-treated clones had variable sizes (Figure 1M), indicating that the integrations were random events that occurred at different genomic locations.

To demonstrate the feasibility of this approach in stem cells, we isolated CD34+ hematopoietic stem cells from human cord blood and focused initially on optimizing lentiviral transduction rates for optimal ZFN delivery. We first verified that ZFN-loaded VSV-G-pseudotyped LPs loaded with AAVS1-directed ZFNs resulted in targeted cleavage upon transduction of CD34+ cells. Hence, indels introduced through repair by NHEJ were found in 5% of the alleles (Figure 2—figure supplement 1A). Next, we compared the capacity of recombinant human fibronectin (RetroNectin) and EF–C peptides, the latter recently discovered artificial peptide nanofibrils, to support virus entry to CD34+ cells (Yolamanova et al., 2013Lump et al., 2015). In our setting, the presence of EF-C allowed transduction of 85.1% of the cells with a standard eGFP-encoding lentiviral vector, whereas between 27.5 and 30.8% of the cells were transduced in the presence of RetroNectin (Figure 2—figure supplement 1B). Therefore, we chose to include EF-C for targeted integration in CD34+ cells.

In the first of two protocols (Figure 2A, orange labeling), we transduced CD34+ cells derived from two different donors with IDLV-ZFN(AAVS1)/egfp or IDLV/egfp (MOI = 5) and analyzed for eGFP+ cells 9 days post-transduction (Figure 2B). On average, treatment with IDLV-ZFN(AAVS1)/egfp resulted in (0.49 ± 0.14)% eGFP+ cells, which was significantly higher than the percentage of eGFP+ cells (0.07 ± 0.03)% measured after transduction with the IDLV/egfp control without ZFNs (Figure 2B). PCR analyses confirmed events of TGI by IDLV-ZFN(AAVS1)/egfp, but not by IDLV/egfp (Figure 2C). Notably, IDLV-ZFN(AAVS1)/egfp transduction only slightly affected the cell viability (Figure 2D), and cell proliferation was not disturbed by the virus treatment (Figure 2E). Encouraged by this, we reasoned that the efficiency could be further increased (i) by using higher MOIs, (ii) by pretreating CD34+ cells with Vpx-loaded lentiviral particles (to degrade the reverse transcription inhibitor SAMHD1 [Laguette et al., 2011]), and (iii) by treating the cells with an extra dose of ZFNs delivered by LP-ZFNs. Indeed, using this improved protocol 2 (Figure 2A, green labeling), we were able to increase the percentage of eGFP+ cells to (0.98 ± 0.13)% after transduction with IDLV-ZFN(AAVS1)/egfp (Figure 2F). Also based on this protocol, TGI in the bulk was confirmed by PCR analysis (Figure 2G). Notably, IDLV-ZFN(AAVS1)/egfp treatment did not compromise the colony-forming capacity of CD34+ cells (Figure 2H). In accordance with the flow cytometry data, three eGFP+ colonies (example shown in Figure 2—figure supplement 1C) were identified out of a total of 302 analyzed colonies.

Figure 2 with 1 supplement see all
Targeted gene integration mediated by IDLV-ZFN(AAVS1)/egfp in CD34+ cells.

(A) Schematic illustration of protocols used for IDLV-ZFN(AAVS1)/egfp transduction in CD34+ cells. (B) Flow cytometry analysis showing the percentage of eGFP+ cells after transduction using protocol 1 (Figure 2A, orange labeling). Experiments were performed in triplicates using CD34+ cells from two different donors (donor 1 and donor 2 marked by black and red color, respectively). Representative dot plot is shown on the top right. (C) PCR analyses of targeted gene integration. The primer sets used were same as in Figure 1F. (D) Cell death was measured 4 days after transduction. IDLV-ZFN(AAVS1)/egfp treatment was performed in triplicates. (E) Growth curve of transduced cells and non-transduced cells. (F) Flow cytometry analysis showing the percentage of eGFP+ cells after transduction using protocol 2 (Figure 2A, green labeling). Experiments were performed in triplicates using CD34+ cells from donor 3. Representative dot plot is shown on the top right. (G) PCR analyses of targeted gene integration. The primer sets used were the same as in Figure 1F. (H) The number of colonies formed from IDLV-ZFN(AAVS1)/egfp-transduced and non-transduced CD34+ cells. The experiments were performed in triplicates, and the numbers of eGFP+ colonies are provided above each bar.

https://doi.org/10.7554/eLife.12213.004

Ways of engineering cells by TGI may be of particular interest for therapeutic modification and use of iPSCs. To establish TGI in iPSCs using lentiviral protein transduction, we first evaluated the permissiveness of iPSCs to LPs loaded with ZFNs targeting the AAVS1 and CCR5 loci, respectively. LP-ZFN(AAVS1) and LP-ZFN(CCR5) disrupted 4% and 8%, respectively, of the targeted alleles in iPSCs (Figure 3A).

We then transduced iPSCs with IDLV-ZFN(AAVS1)/puro. Of the approximately 500 puromycin-resistant clones formed from 50,000 transduced iPSCs (Figure 3B), we initially randomly picked, isolated and expanded eight clones (clones 1–8) for further analysis. From a subsequent experiment we picked 15 additional clones (clones 9–23). PCR amplification of the 5’ and 3’ junction sites showed that TGI had occurred in 19 out of the 23 clones (Figure 3C and D). Findings based on Southern blot analysis of clones 9–23 were consistent with the PCR analysis and confirmed that 11 of these 15 clones contained an AAVS1-directed insertion (Figure 3D). Eight out of the 11 clones harbored a precise, homology-directed transgene insertion, whereas only one of the two junction sites in the remaining 3 clones (clones 9, 17, and 18) was intact. Hence, the 5’ junction site was intact in clone 17, whereas clones 9 and 18 contained an intact 3’ junction site, indicating that site-directed insertion in these cases involved both HR and NHEJ. Sequence analysis of the 5’ and 3’ junction sites amplified from clones 1–8 showed junctions sequences indicative of precise, homology-driven insertion (Figure 3E and Supplementary file 1A). For one particular clone, clone 7, we were not able to amplify the 469-bp AAVS1 fragment, suggesting that both alleles in this clone contained a TGI (Figure 3C). Interestingly, sequencing of the AAVS1 locus provided evidence of biallelic cleavage in at least one of the clones (clone 5), resulting in a 1-bp insertion in the allele that did not contain the transgene (Supplementary file 1A). Stemness of TGI-modified iPSCs was verified by characterization of a single representative clone using alkaline phosphatase (AP) and the TRA-1-60 pluripotency marker (Figure 3F). Also, genomic integrity was confirmed by karyotyping (Figure 3G).

To detect potential off-target cleavage by the virus-delivered ZFNs, we first sequenced the three top-ranked off-target sites in clones 1–8, but did not identify indels in any of these loci (Supplementary file 1B). We then isolated genomic DNA from a pool of more than 100 puromycin-resistant iPSC clones and confirmed by PCR analysis the presence of TGI events in the pool (Figure 3H, insert). Then, next-generation sequencing was performed on PCR amplicons representing the on-target AAVS1 locus as well as seven top-ranked off-target sites. Among this pool of clones, most of which carry an AAVS1-directed gene insertion, we were not able to detect indels at off-target positions, whereas 3% of the AAVS1 alleles that were not already targeted by transgene insertion contained indels (Figure 3H and Figure 3—figure supplement 1A). Altogether, our findings demonstrated high levels of ZFN specificity and precise TGI in the AAVS1 locus in iPSCs treated with IDLV-ZFN(AAVS1)/puro vectors, resulting in at least one TGI event in 19 of 23 analyzed iPSC clones.

To demonstrate that such features of the ‘all-in-one’ gene delivery system was not limited to insertion into the AAVS1 locus, we produced IDLVs (designated IDLV-ZFN(CCR5)/puro) carrying ZFNs targeting the human CCR5 locus as well as vector RNA carrying the PGK-puro cassette flanked by homology arms matching the CCR5 locus. Transduction of iPSCs with IDLV-ZFN(CCR5)/puro resulted in effective colony formation (Figure 3I). Among the puromycin-resistant clones, we randomly picked, isolated and expanded 15 clones (clones 1–15) and showed by PCR analysis of each of these colonies that the transgene cassette in all 15 clones (100%) had been inserted site-specifically into the CCR5 locus (Figure 3J). For a single representative clone, stemness was verified using alkaline phosphatase (AP) and the TRA-1-60 pluripotency marker (Figure 3K). Once again, we pooled more than 100 puromycin-resistant iPSC clones and performed next-generation sequencing on PCR amplicons amplified from the on-target locus and nine high-ranked off-target sites in the pooled DNA sample (Figure 3L and Figure 3—figure supplement 1B). Notably, we identified an indel in approximately 19% of the CCR5 alleles that did not already carry a TGI, demonstrating that the second CCR5 allele was cleaved by ZFNs in approximately one of every five clones harboring a CCR5-directed insertion. Importantly, off-target cleavage and indel formation by NHEJ was not evident in any of the nine analyzed off-target candidate loci. Hence, not even the CCR2 locus (off-target 3), which is highly homologous to the CCR5 site and a known off-target cleavage site for CCR5-targeted ZFNs, harbored detectable levels of indels.

Figure 3 with 1 supplement see all
Targeted gene integration in iPSCs by IDLV-ZFN(AAVS1)/puro and IDLV-ZFN(CCR5)/puro.

(A) Gene disruption mediated by transduction of LP-ZFN(AAVS1) and LP-ZFN(CCR5) in iPSCs. (B) Puromycin-resistant iPSC clones appearing after transduction with IDLV-ZFN(AAVS1)/puro. Clones were stained with methylene blue. (C) PCR analyses of targeted gene integration by IDLV-ZFN(AAVS1)/puro in clones 1–8. Primer sets used for amplification of 5’ and 3’ junction sites were the same as in Figure 1F, except that amplification of the AAVS1 locus without the insertion was included as a control. (D) Southern blot (top) and PCR analyses (bottom) of targeted gene integration in clones 9–23 resulting from transduction with IDLV-ZFN(AAVS1)/puro. Schematic illustration of Southern blot mimics that in Figure 1K, except that a longer puro probe was used. Fragments indicated with * are indicative of imprecise TGI involving NHEJ at either the 5’ or 3’ junction site. (E) Representative sequence chromatograms of amplified 5’ and 3’ junctions (see also Supplementary file 1A). (F) Alkaline phosphatase (left) and TRA-1-60 (right) staining of a representative iPSC clone derived from IDLV-ZFN(AAVS1)/puro treatment. (G) Karyotyping analysis of a representative iPSC clone generated after transduction with IDLV-ZFN(AAVS1)/puro. (H) Indel frequencies obtained by deep sequencing of on- and top-ranked off-target sites in pools of iPSC clones generated by transduction with IDLV-ZFN(AAVS1)/puro. Inserted PCR analysis confirms the presence of TGI in a pool of iPSCs consisting of approximately 110 iPSC clones. (I) Puromycin-resistant iPSC clones appearing after transduction with IDLV-ZFN(CCR5)/puro. Clones were stained with methylene blue. (J) PCR analyses of targeted gene integration in iPSCs (clones 1–15) obtained by transduction with IDLV-ZFN(CCR5)/puro. (J) Alkaline phosphatase (left) and TRA-1-60 (right) staining of a representative iPSC clone derived from IDLV-ZFN(CCR5)/puro treatment. (L) Indel frequencies obtained by deep sequencing of on- and top-ranked off-target sites in pools of iPSC clones generated by transduction with IDLV-ZFN(CCR5)/puro. Inserted PCR analysis confirms the presence of TGI in a pool of iPSCs consisting of approximately 105 iPSC clones.

https://doi.org/10.7554/eLife.12213.006

In conclusion, lentivirus-based co-delivery of nuclease proteins and a HR donor represents a powerful and safe approach for inserting transgenes into predetermined ‘safe harbors’ in the human genome.

Discussion

On average, four out of five lentiviral integrations land in genes (Mitchell et al., 2004). It is still unclear to which extent this biased integration pattern induces toxicity due to insertional mutagenesis, but new approaches to alter the integration properties of lentiviral vectors remain attractive. Different strategies have been employed to reprogram the integration machinery of lentiviral vectors including (i) addition of a DNA-binding zinc-finger protein to the integrase (Bushman and Miller, 1997), (ii) fusion of a chromosomal binding domain to LEDGF, a cellular co-factor of the integrase (Gijsbers et al., 2010), and (iii) insertion of lentiviral DNA intermediates by transposases and recombinases (Vink et al., 2009Staunstrup et al., 2009Moldt et al., 2011Cai et al., 2014b). These approaches successfully alter the integration preference, but do not allow site-specific insertion into a preferred locus. Here, we present an alternative homology-directed insertion strategy facilitated by nuclease-directed DNA cleavage. By co-packaging designer nucleases and donor sequences into lentiviral vectors and utilizing cellular HR repair pathways to integrate the transgene, we can guide lentiviral cargo toward a safe harbor in up to 84% of HEK293T cells and 89% of iPSCs that were transduced with the ‘all-in-one’ vector system. The efficiency may likely be further improved by exploiting heterodimer-forming rather than homodimer-forming FokI domains (Doyon et al., 2011).

TGI into the AAVS1 and CCR5 loci of stem cells has previously been achieved using HR-based strategies exploiting two- or three-component systems based on delivery of plasmids, IDLVs, and baculovirus vectors as well as combinations of mRNA encoding ZFNs and IDLVs serving as the HR donor (Lombardo et al., 2011Genovese et al., 2014Tay et al., 2013Tiyaboonchai et al., 2014Hockemeyer et al., 2009Lei et al., 2011). Despite the potential of genetic therapies based on TGI in stem cells, like CD34+ cells and iPSCs, existing technologies remain insufficient and highly toxic (Genovese et al., 2014Hoban et al., 2015Merling et al., 2015). Exploiting the capacity of lentiviruses to co-deliver nucleases and a transgene cassette, we showed TGI in CD34+ cells transduced with ZFN-loaded IDLVs without affecting the viability, proliferation, or colony-forming capacity of the cells. However, the efficiency was only modest, suggesting that a selective growth advantage of the TGI-engineered cells, as in SCID-X1, would be required for successful engraftment. Possibly, efficacy could be boosted by combining virus treatment with exposure to chemicals enhancing stem cell self-renewal, by supplying NHEJ inhibitors (Chu et al., 2015), or by manipulating cell cycle conditions in favor of HR (Marniemi et al., 1975).

Stable transduction rates with ZFN-loaded IDLVs leading to TGI are undoubtedly reduced relative to conventional lentiviral gene delivery strategies. In iPSCs, however, transduction with IDLVs carrying ZFNs targeting either the AAVS1 or the CCR5 locus generated a high number of puromycin-resistant clones. Remarkably, 34 out of a total of 38 analyzed clones contained the transgene expression cassette inserted precisely into the locus targeted by the ZFNs. Also, as typically observed in relation to the standard use of IDLVs, we found that a few clones contained an additional, randomly inserted transgene cassette, which could be of relevance for potential future therapeutic use. Importantly, however, we found that lentiviral ZFN protein delivery did not lead to detectable off-target cleavage. Hence, cleavage in a total of 16 predicted off-target loci (7 for AAVS1 and 9 for CCR5) did not appear as analyzed by deep sequencing of large pools of clones carrying the transgene cassette. If desired, this approach can be combined with removal of the selection marker by seamless excision, exploiting for example lentiviral protein transduction of piggyBac transposase (Cai et al., 2014b). Remarkably, in a large pool of iPSCs, most if not all carrying a site-targeted gene insertion in the CCR5 locus, off-target cleavage and indel formation at the neighboring CCR2 locus was not evident after deep sequencing analysis. This is in contrast to higher levels of CCR2 targeting reported in previous studies using other delivery strategies based on adenoviral vectors, plasmid transfection, or administration of recombinant proteins (Perez et al., 2008Gaj et al., 2012), suggesting that lentiviral delivery of site-targeted endonucleases supports safer engineering of the genome. Together, our findings demonstrate that lentivirus particles can ferry programmable nucleases, facilitating homology-driven, site-directed genomic insertion of a co-delivered transgene. Such co-delivery ensures time-limited exposure of the treated cell to nucleases and, yet, results in frequent TGI with particular applicability in iPSCs.

Materials and methods

Plasmid construction

To construct pLV/AAVS1-donor-fluc-PGK and pLV/AAVS1-donor-egfp-PGK, inserts were amplified from pLV/fluc-PGK-PBT (Cai et al., 2014b) and pLV/egfp-PGK-PBT (cloned similarly as pLV/fluc-PGK-PBT), respectively, using primer sets YJ219F-YJ210R (primers are listed in Supplementary file 2A). The inserts were subsequently digested by BsiWI/SalI and cloned into pLV/AAVS1-donor-LS (Cai et al., 2014a) that was digested with the same pair of enzymes. pLV/AAVS1-donor-puro-PGK and pLV/CCR5-donor-puro-PGK were cloned similarly by amplifying inserts from pLV/puro-PGK-PBT using primer set YJ262F-YJ210R and ligated with pLV/AAVS1-donor-LS and pLV/CCR5-donor-LS (Cai et al., 2014b), respectively. Restriction enzymes used for inserts and vectors were BsrGI/SalI and BsiWI/SalI, respectively. All donor sequences are provided in Supplementary file 3. The packaging construct, pMDlg/pRRE-D64V-218, for production of IDLVs was constructed by introducing a D64V mutation into the integrase of pMDlg/pRRE by overlap PCR. For the first round PCR, primers AgeI-int(s)-int-D64V(as) and int-D64V(s)-AccIII-int(as) were used, respectively, with pMDlg/pRRE as the templates. For the second round PCR, primers AgeI-int(s) and AccIII-int(as) were used with the PCR product from the first round as template. The inserts and vector pMDlg/pRRE were then both digested by AgeI and AccIII before ligation.

Cells and culture conditions

HEK293 and HEK293T were cultured in Dulbecco’s Modified Eagle’s Medium (Lonza, Basel, Switzerland). Culture medium was supplemented with 10% fetal bovine serum, 100 U/ml penicillin, 100 μg/ml streptomycin and 250 μg/ml L-glutamine. Human cord blood was obtained from Aarhus University Hospital. CD34+ cells were purified from umbilical cord blood using EasySep Human Cord Blood CD34 Positive Selection Kit (STEMCELL Technologies, Grenoble, France) and cultured in StemSpan serum-free expansion medium (STEMCELL Technologies) supplemented with human early-acting cytokines (PeproTech, Rocky Hill, NJ) including stem cell factor (SCF) 100 ng/ml, Flt3 ligand (Flt3-L) 100 ng/ml, thrombopoietin (TPO) 50 ng/ml, and interleukin 3 (IL-3) 20 ng/ml and enhancer of self-renewal UM171 50 nM (STEMCELL Technologies). Induced pluripotent stem cells, described in a previous publication (Kang et al., 2015), were cultured using TeSR-E8 medium (STEMCELL Technologies) on Vitronectin XF (STEMCELL Technologies) coated culture dish. All cells were cultured at 37°C and 5% (vol/vol) CO2.

Lentiviral vector production and titration

In this work, ‘LP’ (as in LP-ZFN(AAVS1) and ‘LP-ZFN(CCR5)) was used to designate lentiviral particles which did not carry a vector genome but contained a pair of ZFN proteins targeting the AAVS1 and CCR5 locus, respectively. When a lentiviral genome was included, the vectors were referred to as ‘IDLVs’. IDLV-ZFN(AAVS1)/egfp, IDLV-ZFN(AAVS1)/puro and IDLV-ZFN(AAVS1)/fluc all contained a pair of AAVS1-targeting ZFNs as well as RNAs, which upon reverse transcription formed DNA donors for homologous recombination at the AAVS1 locus. The ‘donor’ genes carried by these vectors were egfp, puro and firefly luciferase, respectively, all driven by the PGK promoter. Preparations of LPs and IDLVs were produced as previously described (Cai et al., 2014a). Briefly, LP-ZFN(AAVS1) and LP-ZFN(CCR5) were produced by co-transfection of pMD.2G, pZFNL(AAVS1 or CCR5)-PH-gagpol-D64V and pZFNR(AAVS1 or CCR5)-PH-gagpol-D64V. IDLV/egfp, IDLV/puro and IDLV/fluc were produced by co-transfection pMD.2G, pRSV-Rev, pMDlg/pRRE-D64V-218, and donor-containing vector (either pLV/AAVS1-donor-egfp-PGK, pLV/AAVS1-donor-puro-PGK or pLV/AAVS1-donor-fluc-PGK). IDLV-ZFN(AAVS1)/egfp, IDLV-ZFN(AAVS1)/puro, IDLV-ZFN(AAVS1)/fluc and IDLV-ZFN(CCR5)/puro were produced by co-transfection pMD.2G, pRSV-Rev, pMDlg/pRRE-D64V-218, pZFNL(AAVS1)-PH-gagpol-D64V (or pZFNL(CCR5)-PH-gagpol-D64V), pZFNR(AAVS1)-PH-gagpol-D64V (or pZFNR(CCR5)-PH-gagpol-D64V), and donor-containing vector (either pLV/AAVS1-donor-egfp-PGK, pLV/AAVS1-donor-puro-PGK, pLV/AAVS1-donor-fluc-PGK, or pLV/CCR5-donor-puro-PGK). Concentrations of HIV-1 p24 were measured by ELISA assay (XpressBio, Frederick, MD) according to the manufacturer’s protocol. The MOIs of IDLV-ZFN(AAVS1)/egfp and IDLV/egfp were determined by virus transduction to HEK293T cells followed by flow cytometry analysis allowing quantification of the percentage of eGFP+ cells three days after transduction.

Confocal microscopy

For confocal microscopy, HEK293T cells or HEK293 cells were seeded at a concentration of 2×105 cells/well 1 day before transfection or transduction. HEK293T cells were co-transfected with 0.75 μg pMD.2G and 1.3 μg pHA-ZFNL(gfp)-PH-gagpol-D64V (Cai et al., 2014a) and 1.3 μg pHA-ZFNR(gfp)-PH-gagpol-D64V (Cai et al., 2014a) or with 0.75 μg pMD.2G and 2.6 μg pGFP-PH-gagpol-D64V (Cai et al., 2014a). Cells were fixed after 24 hr. HEK293 cells were transduced with LP-HA-ZFN(gfp) and incubated at 4°C for 1 hr to synchronize virus entry. The cells were then incubated at 37°C for 1 hr, 12 hr, 24 hr and 48 hr, respectively. To detect lentiviral particles, LP-HA-ZFN(gfp) and LP-eGFP were attached to poly-L-lysine coated coverslips by spinoculation at 1200 g for 99 min. In all the experiments, cells or lentiviral particles were washed three times with PBS before fixation with 4% paraformaldehyde. Coverslips were stored in 70% ethanol for at least 15 min. To visualize HA-tagged ZFNs, fixed cells or lentiviral particles were incubated with mouse-derived primary HA monoclonal antibody (Covance, Princeton, NJ) and secondary Alexa Fluor 488 Donkey anti-Mouse IgG (LifeTechnologies). To visualize p24, the lentiviral particles were stained with rabbit-derived primary p24 polyclonal antibody (ThermoFisher Scientific, Waltman, MA) and secondary Alexa Fluor 568 Donkey-anti-Rabbit IgG (LifeTechnologies). The nuclei were stained by DAPI. Sequential imaging was done by using a 488 nm line of a multiline argon laser (detection of Alexa-488), the 405 nm diode laser (detection of DAPI) and 561 nm DPSS laser (detection of Alexa-568) on a confocal microscope (LSM 710, ZEISS) using 63 × oil immersion objective with a numerical aperture of 1.4.

Lentiviral vector transduction in CD34+ cells and iPSCs

In protocol 1, CD34+ cells were transduced with IDLV-ZFN(AAVS1)/egfp or IDLV/egfp at MOI = 5. In protocol 2, CD34+ cells were pretreated with 300 ng p24 Vpx-loaded lentiviral particles for 6 hr and then transduced with IDLV-ZFN(AAVS1)/egfp at MOI = 20. On the next day, cells were transduced with 300 ng p24 LP-ZFN(AAVS1). Both protocols used EF-C nanofibrils to boost virus transduction to CD34+ cells. The iPSCs were pretreated with 2 μM ROCK inhibitor (Sigma-Aldrich, St. Louis, MO) for 1 hr and dissociated into single cells in the presence of ROCK inhibitor before transduced with LP-ZFN(CCR5), LP-ZFN(AAVS1), IDLV-ZFN(AAVS1)/puro and IDLV-ZFN(CCR5)/puro at a dose of 500 ng p24 per 0.5×105 cells. Five days after transduction, iPSCs treated with IDLV-ZFN(AAVS1)/puro and IDLV-ZFN(CCR5)/puro were selected with puromycin (0.5 μg/ml). Representative clones were further stained using alkaline phosphatase and TRA-1-60 live stain dyes (Life Technologies). For colony-forming unit assays, after transduction with and incubation for 4 days, 1200 CD34+ cells were mixed with Human Methylcellulose Complete Media (R&D Systems Inc., Minneapolis, MN) and transferred to a 35-mm plate. The plates were incubated in a humidified atmosphere with 5% (vol/vol) CO2 at 37°C for two weeks.

Flow cytometry

HEK293T cells were seeded in 6-well plates (1×105 cells/well) one day before transduction with 500 ng p24 IDLV-ZFN(AAVS1)/egfp or IDLV/egfp. Cells were fixed with 4% paraformaldehyde. Data were collected on a FACSCalibur (BD Biosciences, Franklin Lakes, NJ). CD34+ Cells were analyzed on LSRFortessa cell analyzer (BD Biosciences) for viability using the LIVE/DEAD fixable near-IR dead cell stain kit (Life Technologies) 4 and 9 days after transduction and for the presence of eGFP+ cells 9 days after transduction.

Luciferase assay

HEK293T cells were seeded in 6-well plates (1×105 cells/well) one day before transduction with 500 ng p24 IDLV-ZFN(AAVS1)/fluc or IDLV/fluc. Luminescence analysis was carried out 3, 10 and 18 days after transduction, respectively. 0.5×105 cells were transferred to 96-well plates 6 hr before luciferase activity analysis using ONE-Glo luciferase assay system (Promega, Madison, WI). Luminescence analysis was performed on a multi-sample plate-reading luminometer (Berthold Technologies, Bad Wildbad, Germany).

Surveyor nuclease assay

Surveyor nuclease assay (Integrated DNA Technologies, Leuven, Belgium) was used to determine the frequencies of gene disruption as described previously (Cai et al., 2014a). Briefly, iPSCs or CD34+ cells were incubated with ZFN-loaded LPs for 48 hr and collected for DNA extraction. HEK293T cells transduced by IDLV-ZFN(AAVS1)/egfp or IDLV-ZFN(AAVS1)/fluc were also subjected to Surveyor nuclease assay. CCR5 and AAVS1 fragments were amplified using primer sets YJ224F-YJ225R and YJ222F-YJ223R, respectively. The PCR products were incubated with 1 μl Surveyor nuclease plus 1 μl enhancer at 42°C for 1 hr after denaturation and re-annealing. The cleavage products were separated on a 1.5% agarose gel and stained with ethidium bromide. The percentage of indels was determined by the formula 100 × (1−(1−fraction cleaved)1/2), wherein the fraction cleaved is the sum of the cleavage product peaks divided by the sum of the cleavage product and parent peaks.

Southern blot analysis

HEK293T cells were transduced with IDLV-ZFN(AAVS1)/puro or IDLV/puro at a dose of 500 ng p24 per 10cells and selected by puromycin. The puromycin-resistant clones were isolated and expanded for Southern blot analysis. Fifteen micrograms of genomic DNA from each clone was digested overnight with BstXI or EcoNI (for subsequent hybridization with a puromycin and an AAVS1 probe, respectively) before gel electrophoresis and vacuum blotting. The puromycin fragment was derived from pLV/AAVS1-donor-puro-PGK using BstXI and SalI, whereas the AAVS1 probe was amplified from genomic DNA by nested PCR using primer sets YJ599F-YJ600R and YJ601F-YJ602R, successively. The iPSC clones derived from treatment with IDLV-ZFN(AAVS1)/puro were prepared and analyzed in the same way, except for the use of a puromycin probe amplified from pLV/AAVS1-donor-puro-PGK using primers Puro-1 and Puro-2. The probes were randomly labeled using the Prime-It random primer labeling kit (Agilent Technologies, Santa Clara, CA) according to the manufacturer’s instructions.

PCR analyses of targeted gene integration

Nested PCR reactions were performed to confirm homology-directed integration at the AAVS1 and CCR5 loci. In the case of the AAVS1 locus, primer sets YJ706F-0279 and YJ707F-YJ738R were used as first round and second round PCRs, respectively, to detect 5’ junctions. YJ251F-YJ205R and YJ252F-YJ666R primer sets were used as first round and second round primer set, respectively, to detect 3’ junctions. For insertion into the CCR5 locus, YJ825F-0279 and YJ833F-YJ738R were used for amplification of 5’ junctions. Primer sets YJ251F-YJ253R and YJ252F-YJ208R were used to detect 3’ junctions. The CCR5 locus was amplified using primer set YJ224F-YJ225R, whereas the AAVS1 locus was amplified using primer set YJ222F-YJ223R.

Karyotyping

For karyotyping analysis, actively growing cells on 6-well plate were treated with KaryoMAX ColcemidSolution (0.1 μg/ml, ThermoFisher Scientific) for 2 hr at 37°C. Cells were then swollen with warm 0.56% KCL for 10–15 min at 37°C. After centrifugation, the cell pellets were fixed with fresh cold fixative solution (methanol: acetic acid, 3:1) for 30 min on ice. Cells were gently washed twice with fixative solution. A drop of fixed cell suspension was dropped on a slide and then dried in a humid condition. Metaphase chromosomes were imaged by microscope and analyzed by Quips CGH software.

Next generation sequencing

The on-target and top ranking off-target loci of AAVS1 and CCR5 predicted by PROGNOS were amplified from genomic DNA of iPSCs treated with IDLV-ZFN(AAVS1)/puro and IDLV-ZFN(CCR5)/puro as well as from non-treated iPSCs using Phusion High-Fidelity PCR Master Mix (ThermoFisher Scientific). Primers and predicted off-target loci are listed in Supplementary file 2B. The purified PCR fragments amplified from each locus were mixed in equimolar amounts. Adaptors were ligated to the pools, which were then deep sequenced with an Illumina HiSeq2500, HiSeq Rapid Run 125 bp PE mode (GATC Biotech, Constance, Germany). The indel frequencies were analyzed using CRISPResso (Pinello, 2015) using the CRISPRessoPooled amplicons mode, while ignoring substitutions. Alternatively, and for further validation, the HiSeq reads were trimmed with Sickle (Joshi and Fass, 2011), merged with FLASH (Magoč and Salzberg, 2011), and mapped to the amplicons using bowtie2 (Langmead and Salzberg, 2012). Reads with an average Phred score below 23 were discarded. The reads were then aligned using EMBOSS water (Rice, 2000) and analyzed for indels. The maximum-likelihood estimate of the true-indel fraction was calculated as previously described (Hsu et al., 2013).

References

  1. 1
    Tethering human immunodeficiency virus type 1 preintegration complexes to target DNA promotes integration at nearby sites
    1. FD Bushman
    2. MD Miller
    (1997)
    Journal of Virology 71:458–464.
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
    Chemically modified guide rnas enhance crispr-cas genome editing in human primary cells
    1. A Hendel
    2. RO Bak
    3. JT Clark
    4. AB Kennedy
    5. DE Ryan
    6. S Roy
    7. I Steinfeld
    8. BD Lunstad
    9. RJ Kaiser
    10. AB Wilkens
    11. R Bacchetta
    12. A Tsalenko
    13. D Dellinger
    14. L Bruhn
    15. MH Porteus
    (2015)
    Nature Biotechnology, 33, 10.1038/nbt.3290.
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
    Sickle: A sliding-window, adaptive, quality-based trimming tool for fastq files, version Version 1.33
    1. N Joshi
    2. J Fass
    (2011)
    Sickle: A sliding-window, adaptive, quality-based trimming tool for fastq files.
  15. 15
    Mesenchymal stem cells derived from human induced pluripotent stem cells retain adequate osteogenicity and chondrogenicity but less adipogenicity
    1. R Kang
    2. Y Zhou
    3. S Tan
    4. G Zhou
    5. L Aagaard
    6. L Xie
    7. C Bünger
    8. L Bolund
    9. Y Luo
    (2015)
    Stem Cell Research & Therapy, 6, 10.1186/s13287-015-0137-7.
  16. 16
  17. 17
  18. 18
  19. 19
    Radiochemical assay of glutathione s-epoxide transferase and its enhancement by phenobarbital in rat liver in vivo
    1. J Marniemi
    2. MG Parkki
    3. S Lin
    4. BT Staahl
    5. RK Alla
    6. JA Doudna
    (1975)
    Biochemical Pharmacology, 24, 10.1016/0006-2952(75)90080-5.
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
    Crispresso: Sequencing analysis toolbox for crispr-cas9 genome editing
    1. L Pinello
    (2015)
    bioRxiv.
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35

Decision letter

  1. Charles L Sawyers
    Reviewing Editor; Memorial Sloan-Kettering Cancer Center, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your work entitled "Targeted, homology-driven gene insertion in stem cells by ZFN-loaded 'all-in-one' lentiviral vectors" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by Charles Sawyers as the Senior Editor.

The reviewers have discussed the reviews with one another and the Reviewing editor has drafted this decision to help you prepare a revised submission.

Your manuscript presents a method for targeted gene integration (TGI) into human cells via an all-in-one integration defective lentiviral system coupled with ZFN protein and donor template delivery. Previously, your group demonstrated the efficacy of this approach in human cell lines, and has now extended this work into human cord blood and human iPS cells.

Overall, the manuscript is well written and clearly presented. The experimental objectives and methods are well described, and the key findings should benefit those interested in similar therapeutic gene-delivery strategies. However, there are a few key points that impact the novelty of the results. Furthermore, some of the finding implications would challenge ZFN-loaded 'all-in-one' vectors as an efficient method for targeted gene integration (TGI). These concepts are addressed below.

It is essential that your revision, if you choose to resubmit, address the following two points, which are elaborated in further detail in the specific comments by the referees.

1) Do you have evidence supporting the use of the all-in-one system at a second locus, beyond AAVS?

2) Can you elaborate more on the issue of non-specific integration, either by inclusion of additional data to address this point or by more detailed acknowledgement of the issue?

We also request that you address the additional issues surrounding methodology and data presentation that are raised by both referees.

Reviewer #1:

The manuscript by Cai and colleagues presents a method for targeted gene integration (TGI) into human cells via an all-in-one integration defective lentiviral system coupled with ZFN protein and donor template delivery. Previously, this group demonstrated the efficacy of this approach in human cell lines, and has now extended this work into human cord blood and human iPS cells. In general, I found the study very well described, however there is some lack of clarity in experimental descriptions, figure legends and methods.

1) It is unclear to me whether the study demonstrates replication at a second independent target site from AAVS1. I cannot find in the text nor the legend which site they are targeting in Figure 2 They compared EF-C and RetroNectin using the AAVS1 target, but the PCR in Figure 2C and 2G shows CCR5 this the target? I see in Figure 1 that CCR5 is amplified in 1F and 1I, why would you not assay AAVS1 to see if the target site is intact? This lack of clarity and description is a problem throughout the text and figures as the authors have oversimplified the reagent descriptions. For example, instead of just saying IDLV-ZFN/egfp, it should read IDLV-CCR5/egfp or IDLV-AAVS1/egfp in each part, or something similar. The manuscript needs a thorough review in this regard and Figures and legends should be described and labeled accordingly. If Figure 2 is also showing targeting AAVS1, then it seems they have the tools available to attempt the TGI into the CCR5 locus in either the CD34+ or iPS cells.

2) I am not familiar with RetroNectin or EF-C. Please define and reference these approaches.

3) In the last paragraph of the Results, it is unclear what is meant by sequencing the AAVS1 locus. Do the authors mean sequencing the non-modified allele?

4) Please describe the methods used to identify the off target sites reported. Was off target analysis performed with CCR5, such as looking at CCR2 in targeted clones?

5) The authors should describe and provide the full sequences (or reference) the donor vector homology arms. How long are they? I also cannot seem to discern the actual target sequences for the ZFN reagents for AAVS1 or CCR5. Are the constructs used in this study available upon request?

6) Figure 1L – is there some explanation of the varying intensity of the TGI band on the Southern? Is it possible that some TGI clones have more than 1 copy inserted? I am not clear if the restriction site chosen would distinguish whether multiple tandem head-to-tail copies are inserted in some clones.

Reviewer #1 (Additional data files and statistical comments):

It would be great if the authors could provide full sequence maps of these vectors, they are likely to be widely sought after and it would facilitate others trying to replicate this and similar approaches.

Reviewer #2:

The manuscript entitled, "Targeted, homology-driven gene insertion into stem cells by ZFN-loaded 'all-in-one' lentiviral vectors" by Cai et al. describes the use of homology-directed integration of transgenes into the 'safe-harbor' AAVSI locus by integrase-defective lentiviral vectors (IDLVs) that are co-packaged with zinc-finger nuclease (ZFN) proteins and a transgene-containing donor template. The authors demonstrate the specificity of targeting to the AAVSI locus in HEK293 cells by conventional PCR and Southern blot analyses. They show that greater long-term transgene expression (eGFP, Luciferase, or puromycin) is achieved when cells are transduced with IDLV packaged with ZFN (targeted integration, 2.5%) than without (random integration, >1%). Interestingly, the authors note that integrations were not wholly by homologous recombination (HR), but a few events were due to non-homologous end-joining (NHEJ). They further apply the strategy in CD34+ hematopoietic stem cells and iPSCs, and show successful transgene integration into the AAVS1 locus. Finally, the authors argue that treatment with IDLV-ZFN/puro vectors do not alter the integrity of iPSC stemness evidenced by stem cell marker expression and karyotyping.

Overall, the manuscript is well written and clearly presented. The experimental objectives and methods are well described, and the key findings should benefit those interested in similar therapeutic gene-delivery strategies. However, there are a few key points that impact the novelty of the results. Furthermore, some of the finding implications would challenge ZFN-loaded 'all-in-one' vectors as an efficient method for targeted gene integration (TGI). These concepts are addressed below.

As pointed out in the text, the work described in the manuscript is a follow-up study to the paper in eLife (2014) entitled: "Targeted genome editing by lentiviral protein transduction of zinc-finger and TAL-effector nucleases" by the same group. In this previous work, the authors described the use of similar 'all-in-one' lentiviral vectors co-packaged with ZFN or TALEN proteins to promote HR-mediated gene editing at the CCN5 and AAVS1 loci. The two main differences between the previous work, and the current one is that, 1) the authors attempt to explore in depth, the specificity and efficacy by which 'all-in-one' lentiviral vectors can promote TGI; and 2), the authors describe a protocol for TGI in hematopoietic stem cells and iPSCs.

1) There are a handful of published reports that have explored the insertion of transgenes (including ZNFs, albeit in a two-vector strategy) into the AAVS1 locus in ES cells and iPSCs and these should be cited:

Select publications:

Tay FC, et al., Targeted transgene insertion into the AAVS1 locus driven by baculoviral vector-mediated zinc finger nuclease expression in human-induced pluripotent stem cells. J Gene Med. 2013 Oct; 15(10):384-95. doi: 10.1002/jgm.2745.

Tiyaboonchai, A et al., Utilization of the AAVS1 safe harbor locus for hematopoietic specific transgene expression and gene knockdown in human ES cells. Stem Cell Res. 2014 May; 12(3):630-7. doi: 10.1016/j.scr.2014.02.004. Epub 2014 Feb 21.

2) The authors do a satisfactory job describing the insertion events occurring at the AAVS1 site. However, their evidence of site-specific integration falls short. As the authors have illustrated in Figures 1L and Figure 1—figure supplement 1F, low frequency, off-target integration events do occur. However, it is in this reviewer's opinion that conventional PCR and Southern blot analysis are limited in their ability to detect the prevalence of additional off-target events. Although the authors do query 5 top off-target sites, these are only predicted sites that are based on the designed ZFN's position weight matrix using the PROGNOS tool. They do not consider off-target integration events at hotspots that are independent of ZFN activity. For instance, the use of genome-wide approaches have found that translocations via non-specific NHEJ events due to CRISPR/Cas9- or TALEN-mediated gene editing are numerous:

Frock RL et al., Genome-wide detection of DNA double-stranded breaks induced by engineered nucleases. Nat Biotechnol. 2015 Feb;33(2):179-86. doi: 10.1038/nbt.3101. Epub 2014 Dec 15.

This is especially critical when the authors demonstrate that 5-8% of the insertions are due to NHEJ (Figures 1G and 1J). Therefore, the frequency by which delivered ZFNs will promote non-specific NHEJ events on a genome-wide scale is unclear. The authors do attempt to address whether lentiviral treatment can cause genomic instability by demonstrating stem-cell integrity and karyotyping in Figure 3. However, only a single-representative clone was analyzed as described in the last paragraph of the Results (Figures 3E and 3F). Furthermore, Figure 3F is of low quality and karyotyping to assess translocation events should be performed using spectral-karyotyping, on more than one clone.

3) In addition, Figure 1L, lane 9 suggests that the AAVS1 locus has been successfully targeted by IDLV-ZFN/puro, however a second integration event is detectable. This observation suggests that the methodology can lead to multiple integration events in the genome, within single clones. If this is indeed the case, the authors need to discuss how this can impact the efficacy of TGI via IDLV-ZFN/transgene vectors in therapeutic settings.

In conclusion, since the study appears to parallel many aspects of the research group's previous work, the unique findings are limited in range, and may not be highly influential to the field, this reviewer cannot recommend the manuscript by Cai et al. to be accepted for publication in eLife.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Targeted, homology-driven gene insertion in stem cells by ZFN-loaded 'all-in-one' lentiviral vectors" for further consideration at eLife. Your revised article has been favorably evaluated by Charles Sawyers as the Senior editor and two reviewers.

The manuscript has been improved but there are two remaining issues that need to be addressed before acceptance, as outlined below:

1) Please acknowledge any published reports that have explored the insertion of transgenes into the AAVS1 locus in ES cells and iPSCs using ZNFs using 2-vector systems, per the comment below:

"It should again be noted that the study's results expand upon the group's previous innovation, and a handful of published reports have already explored the insertion of transgenes into the AAVS1 locus in ES cells and iPSCs using ZNFs (but using 2-vector systems)."

2) Please address the concern raised about the quality of the blot in Figure 1L, as discussed below:

"Figure 1L blot is now of very low quality. It seems like the same figure used in the previous version of the manuscript. Please ensure that this is fixed."

https://doi.org/10.7554/eLife.12213.012

Author response

[…] Overall, the manuscript is well written and clearly presented. The experimental objectives and methods are well described, and the key findings should benefit those interested in similar therapeutic gene-delivery strategies. However, there are a few key points that impact the novelty of the results. Furthermore, some of the finding implications would challenge ZFN-loaded 'all-in-one' vectors as an efficient method for targeted gene integration (TGI). These concepts are addressed below.

It is essential that your revision, if you choose to resubmit, address the following two points, which are elaborated in further detail in the specific comments by the referees.

1) Do you have evidence supporting the use of the all-in-one system at a second locus, beyond AAVS?

2) Can you elaborate more on the issue of non-specific integration, either by inclusion of additional data to address this point or by more detailed acknowledgement of the issue?

To address the two major points listed in the decision letter, we have carried out substantial additional experimentation. Hence, focusing on targeted gene insertion in iPSCs, we generated ZFN-loaded ‘all-in-one’ lentiviral vectors targeting a second locus (the human CCR5 locus) and verified site-directed gene insertion also into this locus. Notably, in all of 15 analyzed iPSC clones the transgene cassette was inserted into the CCR5 locus. Also, in the revision we are the including the analysis of 15 additional iPSC clones treated with AAVS1-directed ‘all-in-one’ vectors bringing the total number of analyzed AAVS1 clones up to 23. Of these 23 clones, 19 harbored an inserted transgene in the AAVS1 locus. Moreover, to address the second point, we have now markedly improved our analysis of specificity and potential off-target cleavage. Hence, we performed next-generation sequencing on pools of iPSC clones appearing after treatment with AAVS1 and CCR5-directed ‘all-in-one’ vectors. Importantly, deep sequencing of panels of top-ranked off-target gave no indication of cleavage at off-target sites. Encouragingly, our analysis also showed that targeted gene insertion into the CCR5 locus by lentiviral ZFN protein delivery was not accompanied by cleavage and indel formation in the CCR2 locus, a well-known off-target locus, which is known from the literature to be frequently targeted when other delivery approaches are utilized.

We also request that you address the additional issues surrounding methodology and data presentation that are raised by both referees.

Reviewer #1:

The manuscript by Cai and colleagues presents a method for targeted gene integration (TGI) into human cells via an all-in-one integration defective lentiviral system coupled with ZFN protein and donor template delivery. Previously, this group demonstrated the efficacy of this approach in human cell lines, and have now extended this work into human cord blood and human iPS cells. In general, I found the study very well described, however there is some lack of clarity in experimental descriptions, figure legends and methods.

1) It is unclear to me whether the study demonstrates replication at a second independent target site from AAVS1. I cannot find in the text nor the legend which site they are targeting in Figure 2 They compared EF-C and RetroNectin using the AAVS1 target, but the PCR in Figure 2C and 2G shows CCR5 this the target? I see in Figure 1 that CCR5 is amplified in 1F and 1I, why would you not assay AAVS1 to see if the target site is intact? This lack of clarity and description is a problem throughout the text and figures as the authors have oversimplified the reagent descriptions. For example, instead of just saying IDLV-ZFN/egfp, it should read IDLV-CCR5/egfp or IDLV-AAVS1/egfp in each part, or something similar. The manuscript needs a thorough review in this regard and Figures and legends should be described and labeled accordingly. If Figure 2 is also showing targeting AAVS1, then it seems they have the tools available to attempt the TGI into the CCR5 locus in either the CD34+ or iPS cells.

We agree with the reviewer that the names and descriptions of the reagents were perhaps oversimplified in the original submission. In the original manuscript we only targeted theAAVS1 locus, and this is why we decided to simplify names of lentiviral particles and viral vectors. PCR amplification of the CCR5 locus was only occasionally used as a control, and this clearly led to some confusion. Things are different in the revised manuscript, as we have now included vectors that target the human CCR5 locus. This also means that the nomenclature has been changed (and improved). Throughout the text, we now use the format ‘LP-ZFN(locus)’ to describe lentiviral particles loaded with ZFNs targeting the mentioned locus and ‘IDLV-ZFN(locus)/gene’ to describe IDLVs loaded with locus-directed ZFNs and carrying a specific transgene (egfp, fluc or puro). The nomenclature is explained where relevant (e.g. Results, first paragraph) and is used throughout the manuscript and in figures.

2) I am not familiar with RetroNectin or EF-C. Please define and reference these approaches.

We have revised the text (Results, sixth paragraph), which now includes further information on both RetroNectin and EF-C.

3) In the last paragraph of the Results, it is unclear what is meant by sequencing the AAVS1 locus. Do the authors mean sequencing the non-modified allele?

This part of the manuscript has been substantially revised, and the particular sentence, which now appears in the ninth paragraph of the Results, has been changed. The reviewer is correct that we sequenced the AAVS1 site to search for modifications in the allele without the insertion. To clarify this point the wording is now as follows: “Interestingly, sequencing of the AAVS1 locus provided evidence of biallelic cleavage in at least one of the clones (clone 5), resulting in a 1-bp insertion in the allele that did not contain the transgene (Supplementary file 1A).”

4) Please describe the methods used to identify the off target sites reported. Was off target analysis performed with CCR5, such as looking at CCR2 in targeted clones?

Originally, we only looked for off-target cleavage in the three top-ranked off-target sites in the 8 iPSC clones carrying an insertion in the AAVS1 locus. We did not address CCR5-directed insertion in the original manuscript and therefore did not get into the issue of off-target insertion in the CCR2 locus. However, as mentioned, we have extended the analysis now to include lentiviral vectors loaded with CCR5-targeting ZFNs and verify the use of such IDLVs for prominent site-directed transgene insertion into the CCR5 locus. Notably, all clones generated by this approach harbored the site-directed gene insertion (new Figure 3J). We obviously agree with both reviewers that a NGS-based approach was required to strengthen the analysis. This analysis has now been carried out on lentivirally transduced iPSCs harboring insertions in either the AAVS1 or the CCR5 locus. Hence, the analysis focuses on identifying cleavage leading to indels in the non-modified on-target allele or in panels of top-ranked off-target sites, including in the case of CCR5 the CCR2 locus (off-target 3). NGS data showing on-target cleavage in both the AAVS1 and CCR5 loci (3 and 19%, respectively) do not provide evidence of any off-target cleavage in any of the predicted sites (new Figures 3H and 3L). Interestingly, targeting of the CCR2 after delivery of CCR5-directed ZFN proteins is at the background level, indicating that cells harboring an on-target CCR5 insertion do not have indels in the most prominent and well-known off-target locus. Hence, in comparison to other delivery methods protein delivery using lentiviral vectors seems gentler on the off-target loci even in cells where the nucleases are cutting both on-target alleles.

As noted above, we have also extended the analysis of AAVS1-targeted IDLVs including now more a thorough analysis of a total of 23 iPSC clones. Southern blot data on 15 of these clones have now been added. As a result, this part of the manuscript has been fully revised. The description of AAVS1-directed gene insertion in iPSCs is now in the ninth and tenth paragraphs of the Results. Following this section, we have included new work related to targeting of the CCR5 locus (Results, eleventh paragraph). Both these sections include data based on NGS. Issues related to the absence of CCR2 targeting are now also included in the Discussion (last paragraph).

5) The authors should describe and provide the full sequences (or reference) the donor vector homology arms. How long are they? I also cannot seem to discern the actual target sequences for the ZFN reagents for AAVS1 or CCR5. Are the constructs used in this study available upon request?

Yes, we did not include information of these issues in the first place. All information is now provided in Supplementary file 3.

6) Figure 1L – is there some explanation of the varying intensity of the TGI band on the Southern? is it possible that some TGI clones have more than 1 copy inserted? I am not clear if the restriction site chosen would distinguish whether multiple tandem head-to-tail copies are inserted in some clones.

No, we have no evidence (based on specific band patterns) suggesting that any of these integrations are tandem insertions. It is clear that other types of insertions, besides site-targeted insertion, may occur. These are most likely random insertions of IDLV DNA intermediates, but may also represent homology-independent insertions in the targeted locus (like in clones 5 and 13 in Figure 1L). In accordance with other Southern data (Figure 1—figure supplement 1), we believe that the varying band intensities represent differences in the amounts of DNA.

Reviewer #1 (Additional data files and statistical comments):

It would be great if the authors could provide full sequence maps of these vectors, they are likely to be widely sought after and it would facilitate others trying to replicate this and similar approaches.

Sequences of the relevant parts of all donor constructs (homology arms, promoter, gene) are now available in Supplementary file 3. Obviously, all constructs and plasmid maps will be available upon request.

Reviewer #2:

[…] Overall, the manuscript is well written and clearly presented. The experimental objectives and methods are well described, and the key findings should benefit those interested in similar therapeutic gene-delivery strategies. However, there are a few key points that impact the novelty of the results. Furthermore, some of the finding implications would challenge ZFN-loaded 'all-in-one' vectors as an efficient method for targeted gene integration (TGI). These concepts are addressed below.

As pointed out in the text, the work described in the manuscript is a follow-up study to the paper in eLife (2014) entitled: "Targeted genome editing by lentiviral protein transduction of zinc-finger and TAL-effector nucleases" by the same group. In this previous work, the authors described the use of similar 'all-in-one' lentiviral vectors co-packaged with ZFN or TALEN proteins to promote HR-mediated gene editing at the CCN5 and AAVS1 loci. The two main differences between the previous work, and the current one is that, 1) the authors attempt to explore in depth, the specificity and efficacy by which 'all-in-one' lentiviral vectors can promote TGI; and 2), the authors describe a protocol for TGI in hematopoietic stem cells and iPSCs.

1) There are a handful of published reports that have explored the insertion of transgenes (including ZNFs, albeit in a two-vector strategy) into the AAVS1 locus in ES cells and iPSCs and these should be cited:

Select publications:

Tay FC, et al., Targeted transgene insertion into the AAVS1 locus driven by baculoviral vector-mediated zinc finger nuclease expression in human-induced pluripotent stem cells. J Gene Med. 2013 Oct;15(10):384-95. doi: 10.1002/jgm.2745.

Tiyaboonchai, A et al., Utilization of the AAVS1 safe harbor locus for hematopoietic specific transgene expression and gene knockdown in human ES cells. Stem Cell Res. 2014 May;12(3):630-7. doi: 10.1016/j.scr.2014.02.004. Epub 2014 Feb 21.

It has been a major priority for us to keep the Introduction short, but the mentioned references are certainly relevant, and we have now cited both references in the Introduction (first paragraph).

2) The authors do a satisfactory job describing the insertion events occurring at the AAVS1 site. However, their evidence of site-specific integration falls short. As the authors have illustrated in Figures 1L and Figure 1—figure supplement 1F, low frequency, off-target integration events do occur. However, it is in this reviewer's opinion that conventional PCR and Southern blot analysis are limited in their ability to detect the prevalence of additional off-target events. Although the authors do query 5 top off-target sites, these are only predicted sites that are based on the designed ZFN's position weight matrix using the PROGNOS tool. They do not consider off-target integration events at hotspots that are independent of ZFN activity. For instance, the use of genome-wide approaches have found that translocations via non-specific NHEJ events due to CRISPR/Cas9- or TALEN-mediated gene editing are numerous:

Frock RL et al., Genome-wide detection of DNA double-stranded breaks induced by engineered nucleases. Nat Biotechnol. 2015 Feb;33(2):179-86. doi: 10.1038/nbt.3101. Epub 2014 Dec 15.

This is especially critical when the authors demonstrate that 5-8% of the insertions are due to NHEJ (Figures 1G and 1J). Therefore, the frequency by which delivered ZFNs will promote non-specific NHEJ events on a genome-wide scale is unclear. The authors do attempt to address whether lentiviral treatment can cause genomic instability by demonstrating stem-cell integrity and karyotyping in Figure 3. However, only a single-representative clone was analyzed as described in the last paragraph of the Results (Figures 3E and 3F). Furthermore, Figure 3F is of low quality and karyotyping to assess translocation events should be performed using spectral-karyotyping, on more than one clone.

We agree with the reviewer that the evidence of site-directed gene insertion should not fall short. We acknowledge the point that more information on insertion patterns (than we originally provided) would support the claims made in the original submission. We chose to focus our additional analyses on iPSCs, in which we find frequent site-directed gene insertion using the ‘all-in-one’ gene delivery system. We carried out PCR of 5’ and 3’ junction sites as well as Southern blot analysis on 15 additional clones generated with AAVS1-directed vectors. As shown in Figure 3D, the PCR data were in complete accordance with the Southern data clearly identifying clones that contained a site-directed gene insertion. Obviously, Southern blotting additionally provides information on the appearance of additional insertions in specific clones. One example is clone 22 in Figure 3D. Hence, in this context Southern blotting is sufficient to identify events of randomly inserted lentiviral vectors. It is well known from the literature that even IDLVs (integrase-defective lentiviral vectors) may occasionally insert into the genome, and this is obviously confirmed by our data. However, it would not be relevant here to carry out genome-wide studies to identify such off-target random insertion sites that are not necessarily related to ZFN function. Although it cannot be excluded that such events may occur, it is much more relevant, as noted by the reviewer, to focus on the analysis on indel formation in off-target recognition sites. One can argue that a complete genome-wide analysis would be relevant for this matter (also to detect translocations), but we believe that an extensive, NGS-based analysis of an extended panel of predicted off-target loci was sufficient to demonstrate the specificity of the approach. In addition to work involving AAVS1-targeted vector, we also (as mentioned) included a whole new series of experiments including CCR5-targeted vectors. Here, we confirmed site-directed insertion by PCR (Figure 3J) and concentrated on the NGS analysis of ZFN activity at a panel of 9 off-target loci including the CCR2 locus (Figure 3L). The lack of detectable cleavage in the CCR2 locus (as shown by the lack of indel formation in this locus in a large pool of clones with ZFN-induced CCR5-directed insertions and evidence of substantial biallelic cleavage) is to us a strong indication that this ZFN delivery approach – possibly due to the strongly time-restricted presence of ZFNs – has unprecedented specificity and is safe. Overall, we believe that we have carefully addressed key issues related to achieving site-directed gene insertion. We acknowledge the importance of the genomic integrity and long-term potential of ZFN-engineered iPSCs, but also feel that such studies belong in a future report that may not fit in this ‘Research Advances’ format.

The reviewer alludes to experiments shown in Figure 1G and 1J, arguing that between 5 and 8% of the insertions are due to NHEJ (Figures 1G and 1J). It should be pointed out that the mentioned assays indicate rates of imprecise repair by NHEJ in non-targeted alleles and, thus, do not quantify insertions involving NHEJ.

3) In addition, Figure 1L, lane 9 suggests that the AAVS1 locus has been successfully targeted by IDLV-ZFN/puro, however a second integration event is detectable. This observation suggests that the methodology can lead to multiple integration events in the genome, within single clones. If this is indeed the case, the authors need to discuss how this can impact the efficacy of TGI via IDLV-ZFN/transgene vectors in therapeutic settings.

The reviewer is correct; transduction with IDLVs may potentially lead to insertion, and this is an often-overlooked fact in the literature. Even with a relatively potent site-directed gene insertion strategy (as the one we demonstrate here), we occasionally observed such secondary events, which may potentially affect potential therapeutic use. To stress that such events occur and may be of relevance also for therapeutic applications, we have added a single sentence in the Discussion: “Also, as typically observed in relation to the standard use of IDLVs, we found that a few clones contained an additional, randomly inserted transgene cassette, which could be of relevance for potential future therapeutic use.”

[Editors' note: further revisions were requested prior to acceptance, as described below.]

The manuscript has been improved but there are two remaining issues that need to be addressed before acceptance, as outlined below:

1) Please acknowledge any published reports that have explored the insertion of transgenes into the AAVS1 locus in ES cells and iPSCs using ZNFs using 2-vector systems, per the comment below:

"It should again be noted that the study's results expand upon the group's previous innovation, and a handful of published reports have already explored the insertion of transgenes into the AAVS1 locus in ES cells and iPSCs using ZNFs (but using 2-vector systems)."

We certainly agree that our work expands upon our previous findings and the attempts and success of other groups to target gene insertion using nuclease-based strategies. To acknowledge these previous reports, we have added the following to the Discussion: “TGI into the AAVS1 and CCR5 loci of stem cells has previously been achieved using HR-based strategies exploiting two- or three component systems based on delivery of plasmids, IDLVs, and baculovirus vectors as well as combinations of mRNA encoding ZFNs and IDLVs serving as the HR donor 3, 4, 7, 8, 21, 22.”

2) Please address the concern raised about the quality of the blot in Figure 1L, as discussed below:

"Figure 1L blot is now of very low quality. It seems like the same figure used in the previous version of the manuscript. Please ensure that this is fixed."

It is correct that the quality of Figure 1L in the previous submission was suboptimal. However, the resolution of the figure that we originally uploaded was good, and the figure was somehow formatted during the submission process (apparently during formation of the manuscript PDF). We are uploading the original figure again and are hoping, thus, that the problem is fixed. We checked the separate figure file (from our last submission) and found that this figure was OK, so we expect this to be a formatting problem, which is now hopefully fixed.

https://doi.org/10.7554/eLife.12213.013

Article and author information

Author details

  1. Yujia Cai

    Department of Biomedicine, Aarhus University, Aarhus, Denmark
    Contribution
    YC, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  2. Anders Laustsen

    Department of Biomedicine, Aarhus University, Aarhus, Denmark
    Contribution
    AL, Revising article, Final approval, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.
  3. Yan Zhou

    Department of Biomedicine, Aarhus University, Aarhus, Denmark
    Contribution
    YZ, Revising article, Final approval, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.
  4. Chenglong Sun

    Department of Biomedicine, Aarhus University, Aarhus, Denmark
    Contribution
    CS, Revising article, Final approval, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.
  5. Mads Valdemar Anderson

    1. Department of Biomedicine, Aarhus University, Aarhus, Denmark
    2. Technical University of Denmark, Lyngby, Denmark
    Contribution
    MVA, Revising article, Final approval, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.
  6. Shengting Li

    1. Department of Biomedicine, Aarhus University, Aarhus, Denmark
    2. Beijing Genomics Institute, Shenzhen, China
    Contribution
    SL, Revising article, Final approval, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.
  7. Niels Uldbjerg

    Department of Clinical Medicine, Aarhus University, Aarhus, Denmark
    Contribution
    NU, Revising article, Final approval, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.
  8. Yonglun Luo

    Department of Biomedicine, Aarhus University, Aarhus, Denmark
    Contribution
    YL, Revising article, Final approval, Analysis and interpretation of data, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon 0000-0002-0007-7759
  9. Martin R Jakobsen

    Department of Biomedicine, Aarhus University, Aarhus, Denmark
    Contribution
    MRJ, Revising article, Final approval, Analysis and interpretation of data, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.
  10. Jacob Giehm Mikkelsen

    Department of Biomedicine, Aarhus University, Aarhus, Denmark
    Contribution
    JGM, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    For correspondence
    giehm@biomed.au.dk
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon 0000-0002-1322-3209

Funding

Danish Council for Independent Research, DFF Medical Sciences (DFF-4004-00220)

  • Jacob Giehm Mikkelsen

The Hørslev Foundation

  • Jacob Giehm Mikkelsen

Civilingeniør Frode V. Nyegaard og Hustrus Fond

  • Jacob Giehm Mikkelsen

Aase og Ejnar Danielsens Fond

  • Jacob Giehm Mikkelsen

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors are grateful to Haiqing Tang (Department of Clinical Medicine, Aarhus University Hospital, Denmark) for assistance related to the use of human cord blood and to Trine Skov Petersen (Department of Biomedicine, Aarhus University, Denmark) for additional technical assistance. Also, the authors would like to thank to Jan Münch (Institute of Molecular Virology, Ulm University Medical Center, Germany) for providing EF-C peptides.

Reviewing Editor

  1. Charles L Sawyers, Memorial Sloan-Kettering Cancer Center, United States

Publication history

  1. Received: October 13, 2015
  2. Accepted: May 14, 2016
  3. Version of Record published: June 9, 2016 (version 1)

Copyright

© 2016, Cai et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,089
    Page views
  • 329
    Downloads
  • 8
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Developmental Biology
    2. Stem Cells and Regenerative Medicine
    Zoé Vincent-Mistiaen et al.
    Short Report
    1. Neuroscience
    2. Stem Cells and Regenerative Medicine
    Bo Li et al.
    Research Article Updated