Sustained Knockdown of a Disease-Causing Gene in Patient-Specific Induced Pluripotent Stem Cells Using Lentiviral Vector-Based Gene Therapy

This study proposes an approach for the expression of a therapeutic short hairpin RNA (shRNA) in disease-specific induced pluripotent stem cells (iPSCs) using third-generation lentiviral vectors. It was found that lentiviral vector-mediated expression of shRNAs can be efficiently used to knock down and functionally evaluate disease-related genes in patient-specific iPSCs.

Abstract

Patient-specific induced pluripotent stem cells (iPSCs) hold great promise for studies on disease-related developmental processes and may serve as an autologous cell source for future treatment of many hereditary diseases. New genetic engineering tools such as zinc finger nucleases and transcription activator-like effector nuclease allow targeted correction of monogenetic disorders but are very cumbersome to establish. Aiming at studies on the knockdown of a disease-causing gene, lentiviral vector-mediated expression of short hairpin RNAs (shRNAs) is a valuable option, but it is limited by silencing of the knockdown construct upon epigenetic remodeling during differentiation. Here, we propose an approach for the expression of a therapeutic shRNA in disease-specific iPSCs using third-generation lentiviral vectors. Targeting severe α-1-antitrypsin (A1AT) deficiency, we overexpressed a human microRNA 30 (miR30)-styled shRNA directed against the PiZ variant of A1AT, which is known to cause chronic liver damage in affected patients. This knockdown cassette is traceable from clonal iPSC lines to differentiated hepatic progeny via an enhanced green fluorescence protein reporter expressed from the same RNA-polymerase II promoter. Importantly, the cytomegalovirus i/e enhancer chicken β actin (CAG) promoter-driven expression of this construct is sustained without transgene silencing during hepatic differentiation in vitro and in vivo. At low lentiviral copy numbers per genome we confirmed a functional relevant reduction (−66%) of intracellular PiZ protein in hepatic cells after differentiation of patient-specific iPSCs. In conclusion, we have demonstrated that lentiviral vector-mediated expression of shRNAs can be efficiently used to knock down and functionally evaluate disease-related genes in patient-specific iPSCs.

Introduction

Lentiviral vectors (LVs) have been used previously for fast and effective genetic modification of target cells, and third-generation LVs are considered to be among the safest of currently available techniques for sustained transgene expression in gene therapy [1]. This is also reflected in the fact that as of 2013, there are 62 clinical trials ongoing for a broad variety of diseases, such as cancer, infectious diseases, monogenic disorders, neurological, and ocular diseases, ranging from phase I to phase III (http://www.wiley.com/legacy/wileychi/genmed/clinical). Considering the fact that the primary strategy of most ongoing clinical trials to date is to use transduced bulk cell populations with many unknown integration sites for transplantation, the application of LVs in iPSCs may be considered comparably safe because individual clonal cell lines can be established, analyzed for genomic integration sites, examined for safety, and only then expanded for differentiation of transplantable tissue [2]. Thus, the risk of potential random insertional mutagenesis can be greatly reduced or avoided. Also, LVs support the expression of a transgene after integration in dividing and nondividing cells [3], and expression levels of a therapeutic transgene can be modulated by the choice of an appropriate internal promoter [4] or by selection of clonal cell lines with certain expression characteristics.

During recent years, advances in the field of induced pluripotent stem cell (iPSC) research have lead to more efficient, well-defined, and safer methods for the generation of patient-specific iPSCs from biopsies. Although reprogramming of human somatic cells to iPSCs can be achieved with DNA-free methods by direct protein [5] or stabilized mRNA [6] delivery and by nonintegrating RNA viruses such as Sendai virus [7], new emerging technologies for in-depth molecular characterization and comparison of human cells, such as RNA-Seq for transcriptome analysis, methylated CpG island amplification and microarray for epigenetics, and array comparative genomic hybridization (array CGH) or whole genome sequencing for genome analysis, hold promise for making patient-specific iPSCs a well-defined and safe raw material source for unlimited generation of therapeutically applicable tissues. Furthermore, insights from developmental biology merge with advances in tissue engineering in order to find suitable and well-defined differentiation conditions for in vitro growth and maturation of transplantable tissues [8, 9]. Thus, iPSC-derived cells hold great promise to serve as an in vitro model for many hereditary diseases, which suggests an immediate demand for fast and safe methods for sustained genetic alteration of the disease-causing gene patient-specific iPSCs, in order to use their differentiated progeny for pathophysiological studies.

We chose severe α-1-antitrypsin deficiency-associated liver disease as a well-established model to investigate the feasibility of an LV-based sustained gene knockdown in iPSCs. In this hereditary disease, which is highly prevalent among people of Northern European ancestry (1:2,000), the mutated PiZ isoform of α-1-antitrypsin (A1AT) accumulates in the endoplasmic reticulum of hepatocytes of homozygous PiZZ individuals, leading to neonatal hepatitis, liver cirrhosis, and hepatocarcinoma [10, 11]. Heterozygous PiMZ individuals show no apparent disease phenotype even though A1AT is expressed from both alleles [12, 13]. This fact leads us to the assumption that a greater than 50% reduction of the intracellular PiZ A1AT protein load in hepatocytes would be sufficient for a potential therapeutic approach. We and others have previously shown successful iPSC-based disease modeling for severe α-1-antitrypsin deficiency-associated liver disease in vitro [14, 15], and the availability of a mouse model overexpressing human PiZ A1AT in the liver allows for more rigorous in vivo testing of a given therapeutic approach.

Here, we expressed a short hairpin RNA (shRNA) directed specifically against A1AT mRNA in induced pluripotent stem cells, and we achieved a strong reduction of PiZ A1AT in the mouse model in vivo and a significant and therapeutically relevant 66% reduction of intracellular PiZ A1AT in patient-specific iPSCs from a human PiZZ individual. In this report, we describe the use of a silencing-resistant system for lentiviral expression of therapeutic transgenes in patient-specific iPSCs and their differentiated hepatic progeny. Using this technology for genetic modification of iPSCs, it is possible to establish clonal cell lines with well-defined properties that could allow for safe use in future therapy.

Materials and Methods

Cloning and Plasmids

Lentiviral plasmid pRRL.PPT.EFS.GFPpre [16] was digested with BsrGI and SalI, and presynthesized miR30-styled PiZ-shRNA was inserted downstream of the enhanced green fluorescence protein (eGFP), resulting in pLenti.EFS.GFP.PiZ.shRNA. pLenti.EFS.-GFP.PiZ.shRNA was digested with BsrGI and Van91I, and the fragment containing the miR30-styled PiZ-shRNA together with a part of the PRE-site was then inserted into pRRL.PPT.CAG.-GFPpre.2 (a construct based on the design of pRRL.PPT.EFS.-GFPpre with exchanged promoter cassette) to generate the lentiviral knockdown construct pLenti.CAG.GFP.PiZ.shRNA (CG-P). pLenti.CAG.GFP.PiZ.shRNA was digested with BsrGI and NheI, and presynthesized miR30-styled scramble (scr) shRNA was inserted, thereby replacing the miR30-styled PiZ-shRNA knockdown cassette, resulting in pLenti.CAG.GFP.-scr.shRNA (CG-s). The sense-loop-antisense sequences from the presynthesized miR30-styled shRNAs are as follows: PiZ-shRNA: AACCATCGACAAGAAAGGGACT (sense), CTGTGAAGCCACAGATGGG (loop), AGTCCCTTTCTTGTCGATGGTC (antisense); Scr-shRNA: ATCTCGCTTGGGCGAGAGTAAG (sense), CTGTGAAGCCACAGATGGG (loop), CTTACTCTCGCCCAAGCGAGAG (antisense).

Production and Titration of Lentiviral Vectors

All lentiviral vectors were produced by transient transfection of HEK293T cells using the standard CaCl₂ method [16]. Thirty-six hours after transfection, supernatants were harvested, followed by ∼200-fold concentration through ultracentrifugation. Titers were determined by a dilution series transduction of the following cells, depending on the viral vector: HEK293T cells for RRL.PPT.EFS.GFPpre, RRL.PPT.EF1a.GFPpre, RRL.PPT.CAG.-GFPpre.2, CG-P, and CG-s, Hepa1,6 cells for Alb.Neo.IRES.-dTomato (AN-dTom), and P1EF cells for RRL.PPT.SF.hOct34.-hKlf4.hSox2.i2dTomato.pre (OKS-dT). Seventy-two hours post-transduction, titers for eGFP containing vectors were measured using a FACSCalibur (BD Biosciences, San Diego, CA, http://www.bdbiosciences.com), or on the LSRII Flow Cytometer (BD Biosciences) for dTomato-containing vectors.

Animals and Generation of Chimeric Mice

PiZ mice bred on C57/BL-6 background [17] were provided by the laboratory of Robert Bals (University of the Saarland). Chimeric animals from murine PiZ iPSCs were generated as described earlier [18].

Generation of Murine and Human iPSCs

Adult ear fibroblasts were isolated from the ear of a 6-month-old male PiZ mouse by mechanical dissociation into small pieces and plating in Dulbecco's modified Eagle's medium (DMEM) low glucose (PAA Laboratories, Linz, Austria, http://www.paa.at) with 10% fetal bovine serum (FBS) (PAA Laboratories) 1% l-glutamine with penicillin/streptomycin (Pen/Strep) (PAA Laboratories), and 0.2% 50 mM β-mercaptoethanol under a sterile glass slide in an 8.5% CO₂ incubator. Two weeks later, cells were trypsinized and expanded. Fifty thousand cells of passage 4 P1EF were seeded in the same medium in one well of a six-well plate and transduced the following day at a multiplicity of infection (MOI) of 0.1 with lentiviral reprogramming vector OKS-dT. Two days later, medium was changed to mouse embryonic stem (ES) medium (KnockOut DMEM [Gibco, Grand Island, NY, http://www.invitrogen.com] with 15% FBS, 1% nonessential amino acids, 1% l-glutamine with Pen/Strep, 0.2% 50 mM β-mercaptoethanol, and 1,000 U/ml leukemia inhibitory factor [LIF]) with 20 μg/ml ascorbic acid (Carl Roth GmbH, Karlsruhe, Germany, http://www.carlroth.com). After 10 days iPSC colonies were picked under a sterile stereomicroscope, trypsinized, and plated into a 24-well plate containing γ-irradiated C3H mouse embryonic feeder cells (MEFs) in mouse ES medium. Three days later colonies with no remaining dTomato expression were chosen for subcloning and expansion. Murine iPSCs were passaged by trypsinization every 3–4 days. Generation of AP-iPS-C1 iPSCs, here referred to as hPi, from a female human patient with α-1-antitrypsin deficiency was described elsewhere [19]. Human iPSCs were maintained on γ-irradiated CF1 MEFs in human embryonic stem (hES) medium (1:1 mixture of KnockOut DMEM and KnockOut DMEM/F-12 with 20% KnockOut Serum Replacement [KOSR; Gibco], 1% nonessential amino acids, 1% l-glutamine with Pen/Strep, 0.2% 50 mM β-mercaptoethanol, and 40 ng/ml basic fibroblast growth factor [bFGF]) and passaged by collagenase treatment and glass beads every 5–7 days.

Immunocytochemistry

Murine iPSCs grown for 4 days on feeder cells were fixed in 4% paraformaldehyde (Sigma-Aldrich, St. Louis, MO, http://www.sigmaaldrich.com) for 15 minutes at room temperature, washed with phosphate-buffered saline (PBS), and incubated with the primary antibodies mouse anti-Oct4, goat anti-Sox2, and rabbit anti-Nanog (1:200; Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com) at 4°C overnight. Alexa 546 rabbit anti-mouse, Alexa 488 donkey anti-goat, and Alexa 488 goat anti-rabbit (all from Invitrogen, Carlsbad, CA, http://www.invitrogen.com) were used as secondary antibodies, and nuclei were labeled with 4′,6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich). Wells incubated without primary antibody served as negative controls. Differentiated human PiZ iPSCs were washed with PBS with magnesium and calcium and fixed for 30 minutes on ice with a 1:1 mixture of methanol and acetone to remove fluorescence from the albumin-neomycin-dTomato selection cassette. Cells were washed five times for 5 minutes with PBS and then blocked with blocking solution (PBS, 0.3% Triton X-100, 5% filtered rabbit serum) for 1 hour at room temperature followed by incubation at 4°C overnight with mouse IgG1 monoclonal anti-human A1AT (clone 2C1; Hycult Biotech, Uden, The Netherlands, http://www.hycultbiotech.com) 1:50 in blocking solution. The next day, cells were washed five times for 15 minutes with PBS and incubated with rabbit anti-mouse IgG (H+L) Alexa Fluor 568 conjugate (Life Technologies, Rockville, MD, http://www.lifetech.com) for 1 hour at room temperature, washed five times for 15 minutes with PBS, and incubated for 5 minutes with 0.5 mg/ml DAPI.

Immunohistochemistry of Fetal Livers

Embryonic day (E)15.5 livers were fixed in 4% paraformaldehyde overnight and then placed in a 30% sucrose solution for 24 hours and embedded in Tissue-Tek O.C.T. Compound (Sakura Finetek, Torrance, CA, http://www.sakura.com). Ten-micrometer sections were cut onto SuperfrostPlus glass slides (Thermo Scientific, Waltham, MA, http://www.thermoscientific.com), air dried for 20 minutes, and washed three times for 10 minutes in PBS. Sections were blocked with blocking solution (PBS, 0.3% Triton X-100, 5% filtered donkey serum) for 1 hour at room temperature followed by incubation at 4°C overnight with goat anti-human A1AT (A80–122A; Bethyl, Montgomery, TX, http://www.bethyl.com) 1:500 in blocking solution. The next day, glass slides were washed five times for 15 minutes with PBS and incubated with donkey anti-goat IgG (H+L) Alexa Fluor 568 conjugate (Invitrogen) for 1 hour at room temperature, washed five times for 15 minutes with PBS, incubated for 5 minutes with 0.5 mg/ml DAPI, and covered with mounting medium on a cover slide.

Array CGH and G-Banding

The array CGH technique used for characterization of mouse iPSCs has been described previously [20]. Genomic DNA from generated iPSCs was compared with gDNA from P1EF as reference. G-banding for human iPSCs has been described previously [21].

RNA Isolation, cDNA Synthesis, and TaqMan Quantitative Reverse Transcription-Polymerase Chain Reaction

RNA from cultivated or sorted cells was isolated using the peqGOLD Total RNA Kit (Peqlab, Wilmington, DE, http://www.peqlab.com) according to protocol and subsequently used for cDNA synthesis using the SuperScript First-Strand Synthesis System (Invitrogen) according to the manufacturer's instructions. cDNA was used for quantitative reverse transcription-polymerase chain reaction (qRT-PCR) analysis with TaqMan probes (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com) on the StepOne Plus cycler (Life Technologies, Grand Island, NY, http://www.invitrogen.com). The following probes were used for the analyzed genes: human: Hs02758991_g1 (GAPDH), Hs03005111_g1 (OCT4), Hs01053049_s1 (SOX2), Hs02387400_g1 (NANOG), Hs00609403_m1 (ALB), Hs00174914_m1 (transthyretin [TTR]), Hs01941416_g1 (cytokeratin 18 [CK18]), Hs01054797_g1 (CYP1A1), and Hs01097800_m1 (A1AT); mouse: Mm00607939_s1 (b-Act), Mm00658129_gH (Oct4), Mm00488369_s1 (Sox2), Mm02019550_s1 (Nanog), Mm00802090_m1 (Alb), Mm00431715_m1 (Afp), Mm00443267_m1 (Ttr), Mm01601706_g1 (Ck18), and Mm01312827_m1 (fumarylacetoacetate-hydrolase [Fah]).

Transduction of Murine and Human iPSCs

Passage 4 mouse iPSCs from line mPi were preplated for 45 minutes on gelatin-coated 6-cm dishes, cells in suspension were counted, and 5,000 cells were taken up in 70 μl of KnockOut DMEM with 10% FBS, 8 μg/ml protamine sulfate (Sigma-Aldrich) and 3,000 U/ml LIF. Then, 1 × 10⁶ active particles of concentrated lentiviral RRL.PPT.EFS.GFPpre, RRL.PPT.EF1a.GFPpre, RRL.PPT.CAG.GFPpre.2, CG-P, or CG-s were added to the cells (5–20 μl, depending on the exact titer), and the resulting volume was plated as a droplet on a 3.5-cm suspension culture dish, covered with mineral oil, and placed in a 5% CO₂ incubator for 7 hours. After that, the cells in the droplet were resuspended and plated onto one well of a six-well plate with C3H MEFs. Four days later, highly green fluorescence protein (GFP)-positive colonies were picked and subcloned for generation of clonal lines. For generation of bulk populations, 20,000–50,000 cells were sorted from the GFP+ fraction using the MoFlo cytometer and expanded.

For transduction of human iPSCs, 10 μM ROCK inhibitor was added to passage 19 hPi cells grown on a feeder layer 2 hours before treatment with collagenase and glass beads. Cells were quickly centrifuged (100g for 10 seconds) to separate clumps from single cells, and the pellet was resuspended in trypsin/EDTA to obtain single cells. Trypsin was inactivated with fetal calf serum (FCS)-containing medium, and the cells were washed with PBS and counted. Ten thousand cells were seeded on one well of a six-well plate prepared with CF1 feeder cells, in 1 ml of hES medium containing 10 μM ROCK inhibitor and 4 μg/ml protamine sulfate. Then, 1 × 10⁶ active particles of concentrated lentiviral CG-P or CG-s were added to the cell suspension, and the medium was changed every day with hES medium with ROCK inhibitor for 7 days until compact colonies started to form. Highly GFP-positive colonies were picked for generation of clonal cell lines. For bulk populations, cells were prepared in the same way as described for transduction, and 50,000 cells of the GFP+ fraction were sorted using the MoFlo cytometer and expanded in one well of a six-well plate in hES medium with ROCK inhibitor for the first 7 days.

Hepatic Differentiation of Murine and Human iPSCs

Transduced and untransduced murine iPSCs were differentiated using a protocol based on the previously described hanging drop method [22]. For fluorescence-activated cell sorting (FACS) analysis, 1 × 10⁶ active particles of lentiviral vector AN-dTom were added per well of a six-well plate at day 5 + 9 + 3. Cells were analyzed at day 5 + 9 + 23.

Hepatic differentiation of human iPSCs was performed based on a recently published protocol [23]. Briefly, iPSCs were passaged as large clumps for attachment on Matrigel (BD Biosciences) and cultured in two-thirds MEF-conditioned and one-third fresh hES medium. When colonies reached approximately 80% confluence (day 1), medium was changed to hepatic differentiation basal medium (HDBM) (RPMI 1640 [PAA Laboratories] with 5% KOSR, 1% l-glutamine with Pen/Strep, 1% nonessential amino acids, 0.5 mg/ml bovine serum albumin, 10 nM Ly294002 [Calbiochem, San Diego, CA, http://www.emdbiosciences.com; Merck Millipore, Billerica, MA, http://www.millipore.com]) with 100 ng/ml activin A. On day 2 medium was exchanged with HDBM with 0.1% insulin-transferrin-selenium (ITS) (PAA Laboratories) and on day 3 with HDBM with 1% ITS. On days 4–6, medium was changed daily with hepatic cultivation medium (HCM) (Lonza, Walkersville, MD, http://www.lonza.com) supplemented with 50% epidermal growth factor (EGF) from the HCM Bullet Kit plus 30 ng/ml fibroblast growth factor 4, 20 ng/ml bone morphogenetic protein 2, and 10 nM SB431542 (Sigma-Aldrich). Day 5 cells were transduced with 3 × 10⁶ active particles of lentiviral AN-dTom per well of a six-well plate. On days 7–10, medium was exchanged daily with HCM containing 50% EGF, 20 ng/ml hepatocyte growth factor (HGF), and 10 nM SB431542. For maturation, cells were kept on HCM without EGF but with 20 ng/ml HGF, 10 ng/ml oncostatin M (OSM), and 10⁻⁷ M dexamethasone σ for 4 days and expanded in HCM without EGF and with 20 ng/ml HGF for 3 more days. Albumin-positive cells were selected by addition of 1.5 mg/ml G418 (Invitrogen) on days 10–12 and 1 mg/ml G418 on days 13–18. Samples were analyzed on day 18. All cytokines with exception of bFGF and LIF (provided by the Leibniz University Hannover, Hannover, Germany) were purchased from Peprotech (Hamburg, Germany, http://www.peprotech.com).

Western Blot

Western blot for lysates from differentiated human PiZZ iPSCs was performed as described previously [24]. Goat anti-human A1AT antibody (MP Biomedicals, Inc., Irvine, CA, http://www.mpbio.com) was used as primary antibody at a dilution of 1:500. For detection of human A1AT in E15.5 fetal mouse livers, goat anti-human A1AT (A80-122A; Bethyl) was used as primary antibody at a dilution of 1:500.

Isolation of Fetal Liver GFP+ CD45− Cells from Chimeras

E13.5 fetal livers were isolated manually from the rest of the embryo under a stereomicroscope, and at least six fetal livers per cell line were pooled together and treated with 1 mg/ml collagenase/dispase (Roche, Indianapolis, IN, http://www.roche.com) for 30 minutes in 37°C water bath. Digested fetal livers were manually disrupted using a wide-bore pipette tip, filtered through a 70-μm cell strainer, and washed with PBS. Pelleted cells were treated twice with red blood cell lysing buffer (Sigma-Aldrich) for 2 minutes, washed with PBS, and then resuspended in 320 μl of FACS buffer (PBS, 1% FCS). For staining, 0.4 μg of anti-mouse CD45 APC antibody (eBioscience Inc., San Diego, CA, http://www.ebioscience.com) was used per sample followed by 30 minutes of incubation on ice and washing twice with PBS. CD45−/GFP+ and CD45+/ GFP+ cell fractions were sorted using a FACSAria II flow cytometer (BD Biosciences).

Quantitative Polymerase Chain Reaction for Determination of Chimerism of E15.5 Fetal Livers

Genomic DNA was extracted from E15.5 liver samples using a GenElute Mammalian Genomic DNA Miniprep Kit (Sigma-Aldrich) and diluted to 1 ng/μl. SYBR Green qRT-PCR for genomic human A1AT was performed in triplicates in 96-well plates using 2 μl of gDNA and 2.8 μl of DNA/RNA-free water, 0.1 μl of each 10 nM forward and reverse primer, plus 5 μl of 2× SYBR Green PCR Master Mix (Applied Biosystems). Forward (GTGTCCACGTGAGCCTTGCTC) and reverse (GTTTGTTGAACTTGACCTCGG) primers for detection of genomic human A1AT have been described previously [25], and primers specific for 18S rDNA served as endogenous controls (forward: AGGGCAGGGACTTAATCAACGC; reverse: GTTGGTGGAGCGATTTGTCTGG). Genomic DNA from adult ear fibroblast, mPi, PiZ-sh mPi bulk, and scr-sh mPi bulk served as positive controls, and the average of 2^−ΔCT of these four samples was set equal to 100% chimerism. Genomic DNA from mice derived from uninjected blastocysts served as negative control.

Lentiviral Vector Copy Number Assay

Numbers of lentiviral integration were determined according to a qRT-PCR-based method described earlier [26].

Enzyme-Linked Immunosorbent Assay

Enzyme-linked immunosorbent assay (ELISA) for human A1AT was performed using a protocol described elsewhere [27] with goat anti-human A1AT (A80-122A; Bethyl) as the capture antibody and goat anti-human A1AT horseradish peroxidase (HRP) conjugate (A80-122P; Bethyl) as the detection antibody. ELISA for human albumin was performed with goat anti-human albumin (A80-129A; Bethyl) as the capture antibody and goat anti-human albumin HRP conjugate (A80-129P; Bethyl) as the detection antibody.

CYP 1A1 Activity

Cytochrome P450 1A1 activity was measured as described previously [14] and normalized to 24 hours and 10,000 cells.

Results

Generation and Characterization of Murine PiZ iPSCs

Fibroblasts isolated from adult PiZ mouse ear were reprogrammed with a polycistronic lentiviral construct expressing human OCT4, KLF4, and SOX2 and a dTomato fluorescence reporter (OKS-dT) [28] with an MOI of 0.1 active particles per cell in order to have a low copy number per genome and, subsequently, fast and efficient silencing of the transgenes after reprogramming. Indeed, 16 days and two passages post-transduction, no remaining dTomato fluorescence was observed in freshly generated iPSCs. At this time point, human OCT4 expressed from the reprogramming cassette was almost completely silenced and endogenous pluripotency markers were expressed at levels comparable to standard OG2 ES cells in clone mPi (Fig. 1A). PiZ iPSCs stained positive for nuclear Oct4, Sox2, and Nanog (Fig. 1B) and for the cell surface marker SSEA-1 (Fig. 1C). Analysis of the genomic integrity of the mPi cell line by array CGH showed no major chromosomal deletions or additions compared with the starting fibroblasts (Fig. 1D). To test the hepatic differentiation potential of clone mPi, PiZ iPSCs were subjected to a hanging drop-based method of differentiation. The differentiating cells were transduced at day 5 + 9 + 3 with a lentiviral dTomato-expressing reporter vector, under the control of the albumin promoter/enhancer [29, 30]. PiZ iPSCs exhibited excellent hepatic differentiation potential as shown by FACS analysis for Alb-dTom-positive cells at day 37 (day 5 + 9 + 23; Fig. 1E).

Figure 1.

Characterization of murine PiZ induced pluripotent stem cells (iPSCs). (A): Quantitative reverse transcription-polymerase chain reaction expression analysis for endogenous pluripotency markers Nanog, Sox2, and Oct4 compared with a standard OG2 ES cell line. hOCT4 from the OKS-dT reprogramming construct was mostly silenced. (B): Immunostaining of PiZ iPSC colonies for Sox2, Oct4, and Nanog (top row), DAPI staining (middle row), and merged fluorescences with phase contrast (bottom row). (C): Fluorescence-activated cell sorting (FACS) analysis after SSEA-1 PE staining and isotype control. (D): Array comparative genomic hybridization showed an intact 40XY karoytype with a small (∼200 kbp) monoallelic deletion on chromosome 5 after reprogramming. (E): FACS analysis for dTomato in day 37 differentiated hepatic cells transduced with Alb-dTom lentivirus at day 17. Untransduced cells served as negative control. Abbreviations: Alb, albumin; DAPI, 4′,6-diamidino-2-phenylindole; ES, embryonic stem; hOCT4, human OCT4; SSEA, stage-specific embryonic antigen.

A System for Sustained Lentiviral Transgene Expression

Next we focused on finding a suitable promoter for sustained transgene expression during hepatic differentiation from the pluripotent state to the finally differentiated hepatic cells. To this end, the mammalian promoters EF1α-short (EFS) and EF1α were tested for their ability to support stable transgene expression until the end of differentiation. PiZ-iPSCs transduced with lentiviral vectors expressing eGFP under these promoters were sorted for eGFP-positive bulk populations, and five highly eGFP-positive cell lines were clonally expanded for each expression construct. Bulk populations and cell lines were then subjected to hepatic differentiation and transduced with the Alb-dTom reporter construct and then analyzed at the end of differentiation. More than 90% of cells of the EFS transduced and more than 50% of the EF1α transduced bulk populations had lost the transgene expression, and even most of the highly positive clonal cell lines did not show stable expression until day 37 of differentiation (Fig. 2A, 2B). To overcome these limitations, the cytomegalovirus (CMV) early enhancer/chicken β actin (CAG) promoter, an artificial, strong promoter that had been previously used for the generation of “all-GFP” mice [31], was tested in the same lentiviral backbone. All the CAG-eGFP transduced clonal cell lines showed active eGFP expression in nearly 100% of the Alb-dTom-positive fraction and even 95% of cells of the transduced bulk population were still eGFP-positive at the end of differentiation (Fig. 2C). Also, we found that CAG-eGFP transduced clonal iPSCs bear less lentiviral copy numbers in their genome than EF1α and EFS clonal cell lines (supplemental online Fig. 2). Based on these results, the CAG promoter was used in all further experiments for long-term stable transgene expression.

Figure 2.

Comparison of different promoters for stable transgene expression during hepatic differentiation. (A, B): Murine PiZ induced pluripotent stem cells (iPSCs) were transduced with lentiviral vector-expressing eGFP under ubiquitous mammalian EFS and EF1α promoters. Clonal cell lines (cln 1–5) and a sorted bulk population (bulk) were established for each construct. Black bars show the percentage of eGFP-expressing cells in the dTomato-positive fraction after hepatic differentiation and transduction with lentiviral Alb-dTom. Right: Fluorescence-activated cell sorting analysis of corresponding bulk populations. (C): Analysis of clonal cell lines and a bulk population generated by lentiviral transduction of PiZ iPSCs with CAG-eGFP showed only reduced transgene silencing after differentiation. Abbreviations: cln, clone; EFS, EF1α-short; eGFP, enhanced green fluorescence protein; GFP, green fluorescence protein.

Genetic Modification of Murine PiZ iPSCs

Tissue-specific expression of human PiZ A1AT in transgenic mice was achieved by introduction of the full genomic 14.4-kb fragment plus 2 kb of 5′ and 3′ flanking regions into the germ line [32]. Homozygous F0 transgenic lines with 10 or more copy numbers per genome showed development of acute liver necrosis and inflammation and, thus, a partial recapitulation of the liver disease in severe α-1-antitrypsin deficiency patients [33]. For persistent knockdown of high levels of PiZ A1AT in the liver of this mouse model, a robust and traceable system for shRNA expression was required. To this end, a miR30-styled shRNA that is directed against a specific region on the SERPINA1 mRNA and can be expressed from a polII promoter was designed according to Boden et al. [34]. This PiZ-shRNA, or a corresponding variant with a scrambled sense and antisense sequence (scr-shRNA), was inserted between an eGFP reporter and the transcript stabilizing PRE site on a lentiviral expression construct (Fig. 3A). The shRNA was properly processed and cleaved off from the rest of the messenger RNA, which was observed as a decrease of geometrical mean fluorescence intensity per MOI in transduced HEK293T cells using FACS analysis (supplemental online Fig. 1A). With this knockdown construct, expression of the shRNA can be tracked over the eGFP fluorescence reporter. The PiZ iPSC line mPi was transduced with the PiZ-shRNA or scr-shRNA constructs and three clonal cell lines, and a sorted bulk population was established for each one of them. Pluripotency of PiZ iPSCs was not affected by the transduction of the knockdown constructs, as confirmed by pluripotency marker expression analysis (supplemental online Fig. 1B). FACS analysis after hepatic differentiation and Alb-dTom transduction showed sustained expression of the knockdown constructs (Fig. 3B). Vector copy number (VCN) analysis in clonal cell lines showed an average of approximately eight integrations of lentiviral knockdown constructs per genome, on top of the two reprogramming inserts measured in parental line mPi (Fig. 3C). Expression analysis in differentiated cells on day 37 showed a 90% decrease of PiZ human α-1-antitrypsin (hA1AT) in PiZ-shRNA compared with scr-shRNA-treated cells in both transduced bulk populations and clonal cell lines (Fig. 3D, left). Statistical evaluation of PiZ-shRNA versus scr-shRNA-treated groups in a volcano plot confirmed that hA1AT was strongly and significantly differentially expressed, whereas the other analyzed hepatic markers, transthyretin (Ttr), cytokeratin 18 (Ck18), and Fah, were not (Fig. 3D, right).

Figure 3.

Knockdown in murine PiZ induced pluripotent stem cells (iPSCs) and in vitro analysis. (A): miR30-styled shRNA was inserted in between the eGFP and the PRE site on a third-generation lentiviral vector with an internal CAG promoter. (B): Fluorescence-activated cell sorting analysis for eGFP-shRNA-expressing cells in the Alb-dTom-positive fraction at day 37 of hepatic differentiation. Bulk populations were established by transduction of pluripotent PiZ iPSCs with lentiviral shRNA constructs (scr-sh mPi bulk and PiZ-sh mPi bulk), and Alb-dTom transduction was on day 37. (C): Vector copy number analysis for lentiviral integrations in established clonal cell lines. Parental line mPi showed two integrations from the OKS-dT reprogramming construct. Additional integrations from the eGFP-shRNA lentiviral vectors are shown in green. (D): Left: Quantitative reverse transcription-polymerase chain reaction analysis for expression of Ttr, Ck18, Fah, and hA1AT at day 37 of hepatic differentiation. All three clonal cell lines and the bulk population transduced with PiZ-shRNA showed significantly (p < .05, n = 3) reduced expression of hA1AT compared with scr-shRNA clones and bulk. Values were normalized to albumin. Right: Significance and fold change were calculated for the group of PiZ-shRNA-treated samples compared with the group of scr-shRNA-treated samples. hA1AT was significantly differentially expressed in the two groups, whereas Ttr, Ck18, and Fah were not. Abbreviations: Alb, albumin; Ck18, cytokeratin 18; cln, clone; dGag, deleted group specific antigen; eGFP, enhanced green fluorescence protein; Fah, fumarylacetoacetate-hydrolase; hA1AT, human α-1-antitrypsin; PBS, primer binding site; PPT, polypurine-tract; PRE, post-transcriptional regulatory element; RRE, Rev-responsive element; RSV, Rous sarcoma virus promoter/enhancer; RU5, redundant and unique 5′ region; scr, scramble; shRNA, short hairpin RNA; SIN, self-inactivating cassette; Ttr, transthyretin.

Knockdown of PiZ A1AT In Vivo

To confirm the results from our in vitro differentiation in vivo, one clonal cell line and the transduced bulk population of each PiZ-shRNA and scr-shRNA were injected into blastocysts for chimera formation. At E13.5, pregnant mice were sacrificed and fetal livers were removed from the embryos manually. Examination of the resulting chimeras under fluorescent light showed that all parts of the embryos exhibited strong eGFP expression, indicating high contribution of genetically modified iPSCs to all tissues as well as active ongoing transgene expression (Fig. 4A). eGFP-positive fetal livers from at least six embryos per sample were pooled together, digested with collagenase/dispase, and stained with anti-CD45 antibody. In embryonic development, the hematopoietic system emerges from the mesoderm and populates the fetal liver. CD45 is a marker expressed specifically on hematopoietic progenitors, but not on hepatic progenitors [35]. FACS for the CD45−/GFP+ cell population, which contains the fetal hepatoblasts (Fig. 4B), followed by qRT-PCR analysis corroborated the ∼90% knockdown of PiZ hA1AT seen in vitro, whereas other liver markers, such as Afp, Ttr, and Ck18, were not affected (Fig. 4C). Also, qRT-PCR analysis of the CD45+/ GFP+ fraction confirmed enrichment of liver markers in the analyzed CD45−/GFP+ samples (supplemental online Fig. 3A). To confirm the PiZ A1AT knockdown on a functional level, we analyzed lysates of E15.5 fetal livers for human PiZ A1AT protein. Because E15.5 animals were chimeric, the degree of chimerism had to be considered for proper quantification. qRT-PCR analysis for genomic human A1AT revealed one PiZ-shRNA transduced mouse that was 40% chimeric and two scr-shRNA transduced mice that were 25% and 50% chimeric (supplemental online Fig. 3B). Protein lysates from the livers of these three mice were subjected to Western blot analysis and ELISA for human A1AT. The PiZ-shRNA-treated sample showed a strong reduction of human A1AT protein compared with both scr-shRNA-treated samples in Western blot (Fig. 4D) and ELISA (Fig. 4E). For additional confirmation, 25% and 50% scr-shRNA-treated samples were mixed in an appropriate ratio to match the 40% chimerism of the PiZ-shRNA-treated sample and analyzed by Western blot (supplemental online Fig. 3C). For visualization of hA1AT knockdown, cryosections of E15.5 livers were stained with a primary antibody specific for human A1AT and Alexa 568-conjugated secondary antibody. When we applied identical camera settings, cells positive for eGFP stained less positive for hA1AT in sections from PiZ-shRNA chimeras compared with sections from scr-shRNA chimeras (Fig. 4F).

Figure 4.

Analysis of PiZ knockdown in vivo. (A): Embryonic day (E) 13.5 chimeric fetuses generated from eGFP-short hairpin RNA (shRNA) transduced induced pluripotent stem cells (iPSCs). Fetal mice showed ubiquitous and strong eGFP expression, indicating high contribution of iPSCs to chimeras and sustained transgene expression (left and middle). Right: Enlarged fluorescence image of manually isolated E13.5 fetal liver. (B): Fluorescence-activated cell sorting (FACS) for the CD45−/eGFP+ fraction containing fetal hepatoblasts. (C): Quantitative reverse transcription-polymerase chain reaction analysis for Afp, Ck18, Ttr, and hA1AT in E13.5 CD45−/eGFP+ cells. The scr-shRNA-treated cell line and bulk population are shown in light blue and turquoise, respectively, and the PiZ-shRNA-treated cell line and bulk population are shown in light pink and dark pink, respectively. Cells from at least six fetal livers were pooled for each sample. PiZ-shRNA-treated cells showed reduced hA1AT expression compared with scr-shRNA-treated cells, whereas Afp, Ck18, and Ttr were at comparable levels in all samples. Values were normalized to albumin. (D, E): Western blot (D) and enzyme-linked immunosorbent assay (ELISA) (E) for human A1AT in lysates of E15.5 chimeric livers. Percentages of chimerism of respective samples are indicated in parentheses. The PiZ-shRNA sample showed a strong reduction in both assays Western blot and ELISA compared with scr-shRNA samples. Vinculin served as loading control in Western blot. (F): Immunohistochemical analysis of cryosections from E15.5 chimeric livers. Sections were stained using an antibody specific for human A1AT and analyzed using fluorescence microscopy. Pictures were taken with two seconds exposure time for Alexa 568 at a ×40 magnification. eGFP-positive cells stained strongly positive for hA1AT in the scr-shRNA-treated sample and weakly positive in the PiZ-shRNA-treated sample. Abbreviations: Alb, albumin; Ck18, cytokeratin 18; DAPI, 4′,6-diamidino-2-phenylindole; eGFP, enhanced green fluorescence protein; hA1AT, human α-1-antitrypsin; scr, scramble; sh, short hairpin RNA; Ttr, transthyretin; vinc, vinculin.

Generation and Lentiviral Gene Therapy of Patient-Specific iPSCs

To examine a more clinically relevant aspect of lentiviral gene therapy for iPSCs, fibroblasts from a female patient with severe α-1-antitrypsin deficiency (PiZZ homozygosity) suffering from acute liver disease were transduced with our polycistronic lentiviral OKSM reprogramming vector [21]. Clone hPi showed particularly good ES cell-like morphology and pluripotency marker expression was comparable to H9 hES cells (Fig. 5A). This clone stained positive for nuclear OCT4, SOX2, and NANOG (Fig. 5B) and retained a stable 46XX karyotype after reprogramming (Fig. 5C). Also, sequencing of single-nucleotide polymorphism rs28929474 for the PiZ-specific point mutation confirmed the G→A change at position 31 and, thus, the patient-specific origin of this iPSC line (Fig. 5D). Hepatic differentiation using a cytokine-based method and transduction with the lentiviral Alb-dTom reporter construct at day 4 showed that approximately half of the cells had acquired an albumin-expressing hepatic phenotype at day 18 (Fig. 5E). hPi was transduced with lentiviral PiZ-shRNA or scr-shRNA, and for each of the constructs three clonal cell lines and a sorted bulk population was established. VCN analysis in clonal cell lines [26] showed 1–2 integrations of lentiviral knockdown construct per genome, on top of the single reprogramming insertion measured in untransduced hPi cells (Fig. 5F). Also, transduction of knockdown constructs had no effect on pluripotency markers of iPSCs (supplemental online Fig. 4), and FACS analysis following hepatic differentiation with Alb-dTom transduction confirmed ongoing expression of shRNAs in differentiated hepatic cells (Fig. 5G).

Figure 5.

Generation and transduction of patient-specific induced pluripotent stem cells (iPSCs). (A): Quantitative reverse transcription-polymerase chain reaction analysis for pluripotency markers NANOG, SOX2, and OCT4 in the PiZZ iPSC clone hPi compared with the standard H9 human ES cell line. (B): Immunostaining of hPi colonies for nuclear OCT4, SOX2, and NANOG (top row); DAPI staining (middle row); and merged fluorescences with phase contrast (bottom row). (C): G banding shows a stable 46XX karyotype after reprogramming. (D): Single-nucleotide polymorphism sequencing for the PiZ G→A point mutation after reprogramming. (E): Fluorescence-activated cell sorting (FACS) analysis for Alb-dTom-positive cells in day 18 differentiated clone hPi. Cells were transduced with Alb-dTom at day 4 of differentiation, and untransduced cells served as control. (F): Vector copy number analysis for lentiviral integrations in clonal cell lines derived by transduction of hPi with lentiviral shRNAs. Parental line hPi has one integration from the reprogramming construct. Additional integrations from the eGFP-shRNA lentiviral vectors are shown in green. (G): FACS analysis for eGFP-shRNA-expressing cells in the Alb-dTom-positive fraction at day 18 of hepatic differentiation. Bulk populations were established by transduction of pluripotent PiZZ iPSCs with lentiviral shRNA constructs and Alb-dTom transduction was on day 4. Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; eGFP, enhanced green fluorescence protein; scr, scramble; shRNA, short hairpin RNA.

Therapeutically Relevant Knockdown in Functional Hepatic Cells

Cell lines and bulk populations were subjected to hepatic differentiation and analyzed at day 18 for expression of PiZ A1AT. A significant (>90%) reduction of hA1AT mRNA levels in the PiZ-shRNA-treated compared with the scr-shRNA-treated bulk population was observed, and an even stronger knockdown could be measured in selected clonal cell lines (Fig. 6A, left). Statistical evaluation in a volcano plot showed that A1AT was strongly (x axis) and significantly (y axis) differentially expressed between PiZ-shRNA and scr-shRNA-treated groups, whereas hepatic markers CK18 and cytochrome P450 1A1 (CYP1A1) were not. However, TTR showed a slight differential expression (Fig. 6A, right). Bulk populations were subjected to differentiation again and hA1AT protein levels in whole cell extracts were analyzed using Western blot. Densitometric analysis of partially glycosylated 52-kDa and fully glycosylated 56-kDa A1AT [36] showed a 66% and significant overall reduction of PiZ A1AT in the PiZ-shRNA-transduced sample compared with the scr-shRNA-transduced sample, whereas there was no significant change at from the latter to the untransduced cells (Fig. 6B). Next, we evaluated whether the hepatic cells we analyzed at the end of our differentiation protocol can qualify to fulfill criteria of functional hepatocyte-like cells. hPi and the transduced bulk populations readily secreted A1AT into the medium, and again, a significant reduction was observed in PiZ-shRNA-treated cells compared with scr-shRNA-treated cells (Fig. 6C). The differentiated cells also secreted albumin into the medium (Fig. 6D) and showed cytochrome P450 1A1 activity (Fig. 6E). Notably, neither of these functional assays showed a significant difference between PiZ-shRNA-treated and scr-shRNA-treated cells. Finally, we aimed at finding cells in the hepatic population at the end of differentiation that showed polymerization of the PiZ protein. To this end, we used a specific antibody against polymeric human A1AT described previously [37]. In a full scan of the whole differentiated cell population under a fluorescent microscope, we found rare cell clusters that stained positive for polymeric A1AT. We observed that the fluorescence intensity of these clusters was far weaker in PiZ-shRNA-treated cells than in scr-shRNA-treated cells (Fig. 6F; supplemental online Fig. 5). Together, these results demonstrate a functional and therapeutically relevant knockdown of disease-causing PiZ A1AT in mature hepatic cells differentiated from patient-specific iPSCs.

Figure 6.

Knockdown of PiZ human α-1-antitrypsin (hA1AT) in patient-specific induced pluripotent stem cells (iPSCs) after differentiation. (A): Left: Quantitative reverse transcription-polymerase chain reaction expression analysis for TTR, CK18, CYP1A1, and α-1-antitrypsin (A1AT) at day 18 of hepatic differentiation. All three clonal cell lines and the bulk population transduced with PiZ-short hairpin RNA (shRNA) showed significantly reduced expression of hA1AT compared with scr-shRNA clones and bulk. Values were normalized to albumin. Right: Significance and fold change were calculated for the group of PiZ-shRNA-treated samples compared with the group of scr-shRNA-treated samples. A1AT was significantly differentially expressed between the two groups, whereas CK18 and CYP1A1 were not. (B): Left: Western blot for A1AT in day 18 differentiated PiZZ iPSC and scr-shRNA and PiZ-shRNA transduced bulk populations. The upper band (56 kDa) shows fully glycosylated A1AT, and the lower band (52 kDa) shows partially glycosylated A1AT. Right: Densities measured for both bands separately were added together and divided by density of vinculin, and then divided by the average of untreated samples. The PiZ-shRNA-treated bulk population showed a significant 66% reduction of A1AT. Lysates from HepG2 and HEK293T cells served as positive and negative controls, respectively. (C, D): Enzyme-linked immunosorbent assay (ELISA) for secreted A1AT (C) and albumin (D) in supernatants collected from day 18 differentiated iPSCs. The PiZ-shRNA-treated bulk population showed significantly decreased secretion of A1AT compared with scr-shRNA-treated cells, whereas secretion whereas secretion of albumin was at a comparable level. (E): Ethoxyresorufin-O-deethylase (EROD) assay for CYP1A1 activity. Values were at comparable levels among different iPSC samples. All values for ELISAs and EROD were normalized to 10,000 cells and 24 hours, and supernatant collected from HepG2 hepatoma cell line served as a positive control. (F): Immunocytochemical analysis for polymeric PiZ A1AT in cell clusters of day 18 differentiated iPSCs. Cells were stained using an antibody specific for polymeric human A1AT and analyzed using fluorescence microscopy. Pictures were taken with 2 seconds of exposure time for Alexa 568 at a magnification of ×40. Clusters stained strongly positive for polymeric hA1AT in scr-shRNA-treated cells and weakly positive in PiZ-shRNA-treated cells. Statistical analysis was performed using the t test. *, p < .05; **, p < .01. Abbreviations: ALB, albumin; CK18, cytokeratin 18; DAPI, 4′,6-diamidino-2-phenylindole; ns, not significant; scr, scramble; sh, short hairpin RNA; TTR, transthyretin.

Discussion

Patient-specific iPSCs hold great promise for the treatment of many hereditary diseases, which suggests an immediate demand for fast and safe methods for permanent genetic alteration of these cells, in order to use their differentiated progeny for replacement of affected malfunctioning tissue. So far, zinc finger nucleases (ZFN) and, more recently, transcription activator-like effector nuclease (TALEN) have already been successfully used for directed genetic modification and correction of patient-specific human iPSCs [38–41]. This technique allows precision engineering of the genome and is thus especially well suited for in vitro correction of genetic disorders. However, both technologies are cumbersome and need to be reestablished for each targeted genetic locus.

In our study, we have evaluated the potential of more easily applicable lentiviral vectors for a sustained knockdown of a disease-causing gene in patient-specific iPSCs and their differentiated progeny. To this end, we chose the liver disease caused by severe α-1-antitrypsin deficiency as a model system. In this well-studied and highly prevalent inherited disease, a point mutation in the human SERPINA1 gene leads to expression of mutated PiZ A1AT, which polymerizes and accumulates in the rough endoplasmic reticulum of hepatocytes, causing neonatal hepatitis, liver cirrhosis, and hepatocellular carcinoma [10, 11]. Recently, this was addressed by Mueller et al., who showed that the use of rAAV9 vectors for overexpression of shRNA directed against A1AT could significantly reduce the expression of human PiZ A1AT in a mouse model [42]. With this system, however, the expression level of a therapeutic transgene can be difficult to regulate long-term because in contrast to wild-type adeno-associated virus, rAAV does not integrate into the AAVS1 locus but is expressed from an episomal contactamer in the host cell nucleus and can be diluted upon cell division [43]. Also, rAAV allows for the use of only a very limited length of therapeutic transgene. More recently, another group has shown the use of helper-dependent adenovirus (HDAd) for correction of hepatic disease by gene transfer the master autophagy regulator TFEB into livers of PiZ mice [44]. HDAd, like rAAV9, is a nonintegrative strategy, and in 6 months of observation, the authors have observed slight clearance of the therapeutic transgene.

A more sustained approach for the treatment of this disease was reported using ZFN or TALEN for targeted genomic modification of the SERPINA1 locus followed by antibiotic selection and piggyBack excision of the resistance gene in human iPSCs [40, 41]. This technique is very elegant, because theoretically, no nonhuman DNA elements are required for maintenance of disease corrected phenotype. However, available protocols to date involve transfection of massive amounts of foreign DNA into the cell. Screening for targeted clones requires selection cassettes whose excision then is achieved by another round of transfection followed by counterselection. It is possible to omit the antibiotic selection procedure, but only at low ∼1% efficiency in iPSCs [38, 45]. The process from genetic modification to establishment of clonal cell lines requires multiple passaging of iPSCs and prolonged in vitro cultivation, one point that was also addressed by Yusa et al. when they found that most of the detectable exome mutations were likely to have resulted from long-term cultivation [40]. To date, the unanswered questions in ZFN-correction of human PiZZ iPSCs are whether the additional detected mutations have in fact resulted from off-target effects of zinc finger nucleases and how the whole procedure has affected the exon-less 99% rest of the genome, which is known to be the carrier of 15% of all mutations that have large effects on disease-related traits [46].

The ability of lentiviral vectors to stably pass on inserted transgenes to all daughter cells makes them very useful for genetic modification of stem and progenitor cells such as iPSCs. These cells, once generated and assessed for safety, can be transduced in bulk, and the average numbers of lentiviral integrations can be adjusted by the multiplicity of infection. From such a transduced bulk population, clonal cell lines with desired properties, such as safe integration sites, good differentiation potential, and desired level of transgene expression, can then be established and used for further studies. When we first tested whether passing down stable transgene expression from iPSCs to daughter cells and differentiated hepatic progeny was possible after lentiviral transduction, we were confronted with strong transgene silencing, an effect that was observed previously in embryonic carcinoma cells [47]. Silencing was observed even though we used the constitutive mammalian EF1α promoter as an internal promoter. Reports about use of the CMV early enhancer/chicken β actin (CAG) promoter for ubiquitous high GFP expression in transgenic ES cell-derived mice [31], strong expression of exogenous A1AT in the mouse liver [48], and, more recently, stable transgene expression after myocardial differentiation of human ES cells [49] led us to test this promoter for stability of lentiviral transgene expression in hepatic differentiation of iPSCs. Using the CAG promoter, we could establish highly eGFP-positive clonal cell lines, which showed no apparent silencing after in vitro hepatic differentiation and even in transduced and sorted bulk populations most of the differentiated cells showed stable transgene expression. This system was used for stable expression of a therapeutic shRNA in patient-specific iPSC-derived hepatic cells. We designed a miR30-styled shRNA directed against the human A1AT mRNA based on the Dicer optimized construct described by Boden et al. [34]. Using this construct, we could express the shRNA from the CAG promoter, which allowed us to track the expression by a reporter gene in our preclinical modeling scenario. It has been shown that knockdown efficiencies are comparable between polII-driven miR30-styled shRNAs and common polIII-driven shRNAs [50]. In an earlier report lentiviral vectors were used for transduction of polIII promoter H1-driven shRNAs into somatic cells before reprogramming, and transgene expression was sustained after reaching the pluripotent state [51]. However, we found that this approach may not necessarily work to sustain transgene expression during differentiation as well. Also, in a previous report about lentivirus-based genetic correction of Fanconi anemia patient-specific iPSCs, fibroblasts were transduced with the therapeutic vector before reprogramming, and severe transgene silencing was observed [52]. Similarly, we have reported the use of lentiviral vectors for genetic correction of iPSCs generated from a murine model of tyrosinemia type 1 (Fah^−/− mice), but we had also observed severe silencing of lentiviral transgene during hepatic differentiation when using ubiquitous spleen focus-forming virus (SFFV) or tissue-specific TTR promoter [20]. Even though lentiviral vectors have been used successfully for genetic correction of disease phenotype in iPSCs, there are still severe safety concerns about their application because random integration into the genome can potentially cause malignant transformation in the target cells. This issue has been addressed recently by Papapetrou et al., who were able to correct β-thalassemia in iPSCs by a single lentiviral integration in a genomic safe harbor site defined by five separate criteria [2]. However, this safe harbor clone was found only after tedious analysis of more than a dozen transduced clones, and there is no guarantee that the safe harbor criteria mentioned by the authors will be sufficient in all cases to ensure safe application. Moreover, it is not certain that tissue-specific expression can be achieved for all potential therapeutic transgenes, whereas ubiquitous expression may eventually interfere with normal differentiation. In contrast, these issues can be resolved by precision genomic engineering techniques such as ZFN, TALEN, or, as reported more recently, bacterial Cas9 with engineered guide RNA [53].

Nevertheless, lentiviral transduction of iPSCs allows for fast and easy genetic manipulation and sustained suppression of candidate genes by expression of shRNAs could be used to analyze their functions during differentiation. We assessed the feasibility and efficiency of such lentivirus-based knockdown in iPSCs by targeting a disease-causing gene. A1AT is expressed in a codominant manner from both alleles, and normally only PiZZ homozygous individuals develop severe liver disease, whereas PiMZ heterozygous individuals do not exhibit clinical symptoms [12, 13]. Thus, it can be assumed that a greater than 50% reduction in the expression of PiZ A1AT would be sufficient to rescue the disease phenotype or at least have an attenuating effect, which makes this gene an optimal target for an shRNA-based knockdown approach. In our study, we were able to achieve a significant reduction in the expression of PiZ hA1AT in transduced and differentiated disease-specific iPSCs from a murine model. These results were confirmed in chimeric mice generated from clonal and bulk shRNA transduced iPSCs in vivo, where we found a 93% reduction in the expression and protein load of disease causing PiZ hA1AT in fetal liver cells. Finally, we were able to confirm these findings in patient-specific iPSCs generated from a PiZZ severe α-1-antitrypsin deficiency individual with liver disease, where we observed a significant knockdown of PiZ A1AT in differentiated hepatic progeny. From transduced iPSC bulk populations, we were able to establish clonal cell lines within just 10 days and two passages after transduction, which showed high and stable levels of therapeutic transgene expression, good hepatic differentiation capabilities, and only a few genomic lentiviral integrations. In all established cell lines, a significant reduction in the expression of PiZ A1AT was measured. At protein level, a therapeutically relevant (66%) reduction of PiZ A1AT was reached, and in cell clusters with polymeric PiZ A1AT we observed a much weaker signal than in control cells.

Conclusion

Our results demonstrate that shRNA-expressing lentiviral vectors can be used as a macromolecular tool for gene editing in patient-specific iPSCs and their differentiated hepatic progeny. Achieving a strong and silencing-resistant transgene expression at a functionally relevant level from few random lentiviral integration sites in the genome, we demonstrate that lentiviral vector-mediated knockdown of disease-causing genes provides a versatile tool for experimental studies on the underlying disturbed pathophysiology. However, with respect to therapeutic applications of autologous iPSC-derived hepatic cells, more laborious strategies such as ZFN- or TALEN-based gene editing may be superior in terms of safety and efficiency.

Author Contributions

R.E.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing; K.L.: collection and/or assembly of data, data analysis and interpretation; G.W., A.D.S., M.S., D.Z., C.H., and D.S.: collection and/or assembly of data; J.T., R.B., M.O., and H.R.S.: final approval of manuscript; A.S.: data analysis and interpretation; T.C.: conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.