Transcription Activator-like Effector Nuclease (TALEN)-mediated Gene Correction in Integration-free β-Thalassemia Induced Pluripotent Stem Cells*

Background: Gene editing in human patient-specific iPSCs is critical for regenerative medicine. Results: Nonintegrating β-thalassemia iPSCs corrected by TALENs display undetectable off targets and can be differentiated into erythroblasts expressing normal β-globin. Conclusion: TALENs can be used for HBB correction efficiently in β-thalassemia iPSCs with different types. Significance: Our study extends TALENs for gene correction in patient-specific iPSCs and may have applications in cell therapy. β-Thalassemia (β-Thal) is a group of life-threatening blood disorders caused by either point mutations or deletions of nucleotides in β-globin gene (HBB). It is estimated that 4.5% of the population in the world carry β-Thal mutants (1), posing a persistent threat to public health. The generation of patient-specific induced pluripotent stem cells (iPSCs) and subsequent correction of the disease-causing mutations offer an ideal therapeutic solution to this problem. However, homologous recombination-based gene correction in human iPSCs remains largely inefficient. Here, we describe a robust process combining efficient generation of integration-free β-Thal iPSCs from the cells of patients and transcription activator-like effector nuclease (TALEN)-based universal correction of HBB mutations in situ. We generated integration-free and gene-corrected iPSC lines from two patients carrying different types of homozygous mutations and showed that these iPSCs are pluripotent and have normal karyotype. We showed that the correction process did not generate TALEN-induced off targeting mutations by sequencing. More importantly, the gene-corrected β-Thal iPS cell lines from each patient can be induced to differentiate into hematopoietic progenitor cells and then further to erythroblasts expressing normal β-globin. Our studies provide an efficient and universal strategy to correct different types of β-globin mutations in β-Thal iPSCs for disease modeling and applications.


␤-Thalassemia (␤-Thal) is a group of life-threatening blood disorders caused by either point mutations or deletions of nucleotides in ␤-globin gene (HBB). It is estimated that 4.5% of the population in the world carry ␤-Thal mutants (1), posing a persistent threat to public health. The generation of patientspecific induced pluripotent stem cells (iPSCs) and subsequent correction of the disease-causing mutations offer an ideal therapeutic solution to this problem. However, homologous recombination-based gene correction in human iPSCs remains largely inefficient. Here, we describe a robust process combining efficient generation of integration-free ␤-Thal iPSCs from the cells of patients and transcription activator-like effector nuclease (TALEN)-based universal correction of HBB mutations in situ.
We generated integration-free and gene-corrected iPSC lines from two patients carrying different types of homozygous mutations and showed that these iPSCs are pluripotent and have normal karyotype. We showed that the correction process did not generate TALEN-induced off targeting mutations by sequencing. More importantly, the gene-corrected ␤-Thal iPS cell lines from each patient can be induced to differentiate into hematopoietic progenitor cells and then further to erythroblasts expressing normal ␤-globin. Our studies provide an efficient and universal strategy to correct different types of ␤-globin mutations in ␤-Thal iPSCs for disease modeling and applications.
␤-Thalassemia is a group of inherited genetic blood disorders caused by either point mutations or deletions of nucleotides in the ␤-globin gene. These genetic defects result in reduced, abnormal, or no synthesis of ␤-globin chains that make up hemoglobin. It is one of the most common genetic diseases in the world, and patients with ␤-Thal have severe anemia and a shortened life span (2). Hematopoietic stem cell transplantation is the only way to cure ␤-thalassemia but is challenged by the limited availability of human leukocyte antigen-matched healthy donors. Recently, the development of gene therapy based on viral transduction of a normal HBB gene into a patient's own hematopoietic stem cells raised hopes for those who do not have access to bone marrow transplantation (3,4). However, it was still challenged by the concerns of long term safety because of the viral integration. The somatic cells of a patient can be reprogrammed back into pluripotent state as iPS 3 cells that are capable of differentiation into any cells in the body and thus could be potentially used for cell replacement therapy (5)(6)(7)(8). Generation of iPSCs from ␤-Thal patients, correcting mutations, and subsequent differentiation into hematopoietic stem cells raise great hopes for autologous transplantation to treat these inherited diseases (9). Moreover, iPS cells can undergo indefinite self-renewal without losing the ability to differentiate into all cell types and thus represent an ideal cell population for in situ correction of the disease-causing mutations. However, gene targeting in human pluripotent stem cells by standard homologous recombination is largely inefficient (10) and therefore hampers its extensive application in disease models. Zinc finger nucleases (ZFNs) had been reported to sub-stantially enhance the homologous recombination efficiency by specifically introducing a double-stranded DNA break at the target locus (11)(12)(13). Gene targeting aided by ZFN in human iPS or ES cells had been described to be highly efficient (14 -16). However, engineering ZFNs for a specific target is quite inefficient and laborious, which largely hampers their widespread adoption. Recently, transcription activator-like effector nucleases (TALENs) had been described to recognize and cleave any given DNA sequences with high efficiency (17)(18)(19). The DNA-binding domain of TALEN is unusual and contains multiple units that arranged in tandem (TALE repeats). Each individual unit is composed of 34 amino acids with two highly variable amino acids to determine the unit to recognize one DNA pair in the TALEN recognizing sequence (20). In theory, TALE repeat could be engineered and arranged to specifically recognize any given DNA sequence. TALEN-mediated gene targeting had been described in multiple species, including zebrafish and human iPS, and ES cells (21,22). Practically, compared with ZFN, TALEN is much more easy and convenient regarding the designing and constructing. Also, TALENs exhibited lower off target effects and reduced nuclease-associated cytotoxicities compared with ZFNs (23)(24)(25). In attempt to extend TALEN technology to gene correction for ␤-Thal, we generated the ␤-Thal iPS cells through a nonviral approach and developed an efficient process to correct the mutations in ␤-globin gene by designing and utilizing site-specific TALENs.

EXPERIMENTAL PROCEDURES
iPS Generation-The method of isolating amniotic fluid cells was performed as previously described (26). For reprogramming, an oriP/EBNA1-based pCEP4 episomal vector containing OCT4, SOX2, KLF4, and SV40LT genes (27) and miR-302-367 (28) were co-transfected into amniotic fluid cells via nucleofection (Amaxa). The cells were then plated to Matrigel-coated 6-well plates and cultured with reprogramming medium (mTeSR1). The medium was changed every 2 days and iPS-like colonies were picked onto new Matrigel plate for characterization. Cells of passages from 15 to 40 are used for the following experiments.
TALEN and Donor Vectors for Gene Targeting-TALENs were designed as described (17,29). The full amino acid sequences of TALENs are given in the supplemental information. For donor DNA, left and right homology arms were amplified from genomic DNA of healthy individual. A loxP-flanked PGK-puromycin cassette or loxP-flanked PGK-neomycin cassette were cloned between two homology arms in the pMD-18T vector. For targeting, 1 ϫ 10 6 iPSCs were electroporated with 2 g of donor DNA and 4.5 g of each TALEN plasmid. Then the cells were plated onto Matrigel-coated 6-well plates in the presence of Y-27632 (10 M; Sigma) for 1 day. Positive clones were selected by puromycin (0.5 g/ml) or G418 (100 g/ml; Sigma) in mTeSR1. The selected colonies were verified by genomic PCR and Southern blot. All primers used are listed in supplemental Table S1.
GFP Reporter Assay-GFP reporter activation was tested by co-transfecting 293T cells with plasmids carrying TALENs and GFP reporters. 293T cells were seeded into 12-well plates the day before transfection. Approximately 24 h after initial seed-ing, cells were transfected using calcium phosphate. For 12-well plates, we used 1.5 g of each TALEN and 1 g of reporter plasmids/well. The cells were trypsinized from their culturing plates 48 h after transfection and resuspended in 800 l of PBS for flow cytometry analysis. The flow cytometry data were analyzed using C6 (BD Biosciences). At least 20,000 events were analyzed for each transfection sample.
PCR Detection of Corrected Clones-PCR was performed using High Fidelity Platinum Taq (Invitrogen) according to the manufacturer's instructions. 50 -100 ng of genomic DNA templates were used in all reactions. Primer set including P1 (on HBB locus, upstream of 5Ј homology arm) and P2 (in the drug resistance cassette) was used to amplify a 2.8-kb product of the 5Ј junction of a targeted integration (illustrated in Fig. 2D). A primer set including P3 (on 5Ј homology arm) and P4 (3Ј homology arm) was used to amplify a 2-kb product or a 500-bp product to identify whether random integration occurred (see Fig. 2D). Primer IVS-654 and primer EXO-3-15 were used to amplify a 600-bp product including the mutant region of ␤Thal654_iPS, and the PCR products were sequenced to identify the corrected clones. Primer AEXON-F and primer BEXON-R were used to amplify a 700-bp product including the mutant region of ␤(ϪTCTT)_iPS, and the PCR products were sequenced to identify the corrected clones. All of the primers used are listed in supplemental Table S1.
Southern Blot-To detect homologous recombination at HBB locus, a 502-bp HBB specific probe in the 5Ј side of the left homology arm was synthesized by PCR amplification using primers 5Јprobe-F and 5Јprobe-R and a DIG-dUTP labeling kit (Roche Applied Science). Genomic DNA was digested by BglII, and then standard Southern blotting was performed following the instruction manuals of DIG High Prime DNA labeling and detection starter kit II (Roche Applied Science).
Teratoma Formation and Analysis-Cells from a confluent 10-cm plate were harvested by 0.5 mM EDTA digestion, resuspended in Matrigel, and injected subcutaneously into immunodeficiency mice. Eight weeks after injection, teratomas were dissected, fixed in 4% paraformaldehyde, and processed for hematoxylin/eosin staining.
Erythroblast Differentiation of Human iPS Cells-Human iPS cells were harvested by treatment with 2 mg/ml dispase (Invitrogen) and co-cultured with OP9 stromal cells at an approximate density of 5 ϫ 10 6 /20 ml/10-cm dish in 20 ml of ␣-minimum Eagle's medium (Invitrogen) supplemented with 10% FBS (HyClone, Logan, UT), 100 M monothioglycerol (Sigma), and 100 M vitamin C. The co-culture of OP9 with pluripotent cells were incubated for 8 days, with a half medium change on days 4 and 6. Differentiated human iPSCs were harvested at day 8. CD34ϩ cells were sorted out using a direct CD34 progenitor cell isolation kit (Milteneyi Biotech, Auburn, CA). Hematopoietic clonogenic assays were performed in 35-mm low adherent plastic dishes (Greiner Bio-one, Monroe, NC) using 2 ml/dish of MethoCult GFϩ H 4435 semisolid medium (Stem Cell Technologies, Inc.) according to the manufacturer's instructions. The number of sorted CD34ϩ cells using magnetic activated cell sorting for CFC assay was ϳ5 ϫ 10 5 . Colonies were scored after 12-14 days of incubation. Colonies were picked individually from methylcellulose cultures, washed in PBS contains 2% FBS, and spun onto slides with a cytospin apparatus (TXD-3). Cells were fixed and stained with Wright-Giemsa reagents (BASO Biotech, Auburn, CA).

Derivation and Characterization of iPSCs from Two Patients
with Different ␤-Thal Subtypes-To generate iPS cells, we isolated amniotic fluid cells from two fetuses with informed consents from their mothers. One was diagnosed with ␤-Thal major of IVS2-654, which carried two homozygous C 3 T mutations at the second intron of HBB resulted in the formation of an abnormally spliced mRNA and the deficiency of correctly spliced ␤-globin transcript (30). The other was diagnosed with ␤-Thal major of ␤41-42 (ϪTCTT), which carried two homozygous TCTT deletions at the second exon caused a frameshift that generates a termination codon (TGA) in the position of the new 59th codon (31). To obtain iPSCs that were more close to clinical grade, we attempted to generate ␤-Thal iPSCs under nonviral, serum-free, and feeder-free conditions. In detail, we expanded the amniotic fluid cells and delivered oriP/EBNA episomal vectors carrying a combination of reprogramming factors including OCT4, SOX2, SV40LT, and KLF4 (27) and miR-302-367 cluster (28) through electroporation. After ϳ5 days growth in amniocyte medium, the cells were subsequently plated on Matrigel-coated plate and cultured in defined mTeSR1 medium (32) for further reprogramming (Fig.  1A). The iPS-like colonies appeared at ϳ25 days after electroporation. For 1 ϫ 10 6 starting amniotic fluid cells from these two patients for reprogramming, we usually obtained ϳ50 alkaline phosphatase-positive iPSC-like colonies. We then picked four colonies that displayed typical iPSC morphology for further expanding and characterization. The clonally expanded ␤-Thal iPS cells exhibited typical human ES cell morphology (Fig. 1A) and expressed pluripotent markers such as OCT4 and SSEA4 homogenously (Fig. 1B). By quantitative RT-PCR, we demonstrated that the endogenous pluripotent genes such as OCT4, SOX2, and NANOG were fully activated in ␤-Thal iPS cells, as well as that in H1 ES cells (Fig. 1C). More importantly, by using transgene-specific primers to detect transgenes (27), we showed that both of these two ␤-Thal iPS cell lines harbor neither the exogenous reprogramming factors such as OCT4, SOX2, and KLF4 nor the genes in episomal vector backbone (Fig. 1D). Moreover, these transgene-free ␤-Thal iPS cell lines possessed normal karyotype (Fig. 1E) and could form typical teratomas containing three germ layers upon injection into immunodeficient mice, a well known assay for pluripotency (Fig. 1F). Furthermore, we confirmed that the two transgenefree ␤-Thal iPS cell lines carried the homozygous diseasecausing mutations as diagnosed (Fig. 1G). Thus, based on their disease subtypes, we named them ␤Thal654_iPS and ␤(ϪTCTT)_iPS, respectively (Fig. 1G).
Construction of Site-specific TALENs for ␤-Globin Cleavage-We sought to correct these disease-causing mutations in ␤-Thal iPS cells through in situ gene targeting using TALENs (33). Based on previous studies (20), we designed a pair of TALENs that could specifically recognize two adjacent 18-bp DNA sequences with a 17-bp spacer that was ϳ600 bp downstream of the last exon of HBB ( Fig. 2A). To test the site specificity and cleavage activity of the designed TALENs, we constructed a TALEN-targeting plasmid that harbored a GFP reporter. Two 205-bp duplications of GFP coding sequence that flank each side of TALEN-targeting site were introduced into the middle of GFP coding region (illustrated in Fig. 2A). Once a break was introduced by TALENs cleavage, the duplicated homologous sequence will anneal together and recombinate into a full-length GFP and thus can be detected by FACS to evaluate the efficiency and specificity of newly designed TALENs (Fig. 2B). Thus, we transfected the TALEN-targeting GFP reporter together with the pair of TALENs (TALEN-L/R) into 293T cells and showed that the GFP signals were significantly increased compared with that GFP reporter transfected alone or with only one TALEN, demonstrating that the cleavage activity by the TALENs was high (Fig. 2B). To test their site specificity, we introduced different point mutations or deletions in TALEN-binding sequence. We showed that the GFP signals were greatly reduced in reporters with mutated binding sites compared with wild type control reporter (Fig. 2C). These data demonstrated that the nuclease activity of the designed TALENs was highly specific to its targets and can be harnessed for further genome editing.
Efficient Correction of ␤-Globin Mutations Using TALENmediated Gene Targeting-To correct the mutations in ␤Thal654_iPS and ␤(ϪTCTT)_iPS cells, we constructed a donor template centered on TALENs targeting site. The donor template contains a 2.4-kb 5Ј homology arm that harbors the entire wild type ␤-globin gene and a 1-kb 3Ј arm homology to the downstream of ␤-globin gene (Fig. 2D). Upon cleavage by TALENs and subsequent homology recombination with this donor template, wild type ␤-globin gene in the donor will replace the mutant one in ␤Thal654_iPS or ␤(ϪTCTT)_iPS cells. Thus, we introduced the linearized donor plasmids and TALEN vectors into both ␤Thal654_iPS and ␤(ϪTCTT)_iPS cells through electroporation. The positive colonies were selected by puromycin (for ␤Thal654_iPS) or G418 (for ␤(ϪTCTT)_iPS). Drug-resistant colonies were manually picked and expanded for further characterization. To identify correctly targeting clones, we designed two pairs of PCR primers as indicated in Fig. 2D and performed genomic PCR with these primers (Fig. 2D). The correctly targeted clones would be positive for both primers but with different sizes. By genomic DNA PCR, we found that the targeting efficiency was remarkable. In the case of ␤Thal654_iPS cells, of 37 picked clones characterized, 25 were double positive for both PCR primer pairs, with an efficiency of 68%. For ␤(ϪTCTT)_iPS cells, 4 of 10 picked clones were identified as positive (supplemental Fig.  S1). We then randomly expanded two positive clones from each ␤Thal iPS cell line to confirm the correction of mutant ␤-globin. First, we confirmed that the correct gene targeting could be detected by genomic PCR with expected size using the designed two primer pairs (Fig. 2E). Second, we performed Southern blot to further analyze the gene targeting. As shown in Fig. 2F, both targeted cell lines showed the expected bands (Fig.  2F). Lastly, through Sanger sequencing, we confirmed that both ␤Thal iPS cell lines were corrected at one allele of mutated ␤-globin gene (Fig. 2G), indicated as double peaks for ␤Thal654_corrected iPS and overlapping for ␤(ϪTCTT)_ corrected iPS in Sanger sequencing map. Taken together, we showed that TALEN-mediated gene targeting was highly efficient to correct different types of disease-causing mutations in HBB gene in ␤-Thal iPS cells.
Characterization of Gene Corrected ␤-Thal iPSCs-To further characterize the gene corrected ␤-Thal iPS cell lines, we showed that typical pluripotent markers such as OCT4 and SSEA-4 were expressed homogenously in the two corrected ␤Thal654_iPS and ␤(ϪTCTT)_iPS cell lines, indicating that they remained pluripotent after correction (Fig. 3A). Furthermore, both of them could differentiate into three germ layer lineages upon embryonic body formation (Fig. 3B). Also, the karyotypes of the corrected iPS cells remain normal after genetic operation (Fig. 3C). Most importantly, upon injection into immunodeficient mice, the corrected ␤-Thal iPS cells could form teratomas that contain all three germ layers (Fig.  3D), demonstrating that the gene corrected ␤-Thal iPS cells keep the pluripotency well upon gene targeting. One important concern in using nuclease to aid gene targeting is off target cleavage because it might introduce extra mutations in the genome. To address this question, we blasted TALEN-targeted sequences in the entire human genome and selected six genomic sites that were on the top in homology to the designed TALEN recognition sequence. We designed primers and amplified these regions from the genomic DNA extracted from gene corrected ␤Thal654_iPS or ␤(ϪTCTT)_iPS cells and performed Sanger sequencing on those regions. We detected no mutations and deletions on these regions that are most poten-tial to be recognized and cut by TALENs (Fig. 3E). These data indicated that off target cutting might not be introduced by the designed TALENs in gene targeting. To detect whether the genetic modification process can introduce other mutations affecting the gene function, we amplified the whole ␤-hemaglobin gene, and by sequencing we did not find any additional mutations (supplemental information).
In Situ Gene Correction Restores the Function of HBB Gene in ␤-Thal-iPSC-derived HPCs and Erythroblasts-To examine whether the correction of disease-causing mutations in ␤-Thal iPSCs could restore normal expression of full-length ␤-globin, we attempted to differentiate these iPSCs into HPCs and then erythroiblasts. OP9 mouse bone stromal cells had been shown to efficiently induce hematopoietic differentiation of human pluripotent cells without additional cytokines (34); thus, we employed the OP9 co-culture system to induce the hematopoietic differentiation of ␤-Thal iPSCs. Upon OP9 co-culture, both ␤-Thal iPSCs cell lines, no matter whether corrected or uncorrected, could differentiate rapidly and produce HPCs that were detected as CD34ϩ/CD31Ϫ (Fig. 4, A and B). These ␤-Thal iPSC-derived HPCs could further differentiate into various blood lineages upon plating in semisolid culture to form various CFU-Cs, albeit with some efficiency variations between different cell lines and batches of experiments (Fig. 4C). We then manually picked the red blood lineage colonies (CFU-E) from the semisolid plate and examined the expression of globin gene HBB as well as the (fetal type) HBG gene as a control by quantitative RT-PCR. We showed that the expression of HBB gene increased Ͼ1000-fold more in CFU-E derived from both gene corrected ␤-Thal iPSCs than that from the uncorrected ones and that the levels were comparable with human ES cell (H1)derived CFU-E (Fig. 4D). To the contrary, as the control, fetal type globin gene, HBG expressed at similar levels in CFU-Es derived from either corrected or uncorrected ␤Thal iPSCs as well as the human ES cells (Fig. 4D). Lastly, by using conventional RT-PCR and designing primers that could amply fulllength HBB cDNA (the forward primer was designed in the first exon of HBB, and the reversed primer was designed in the third exon of HBB), we confirmed that the expressions of full-length HBB cDNA were successfully restored after gene correction (Fig. 4E). Taken together, our data demonstrated that the in situ correction of disease-causing mutations in ␤-Thal iPSCs restored the function of HBB gene in the iPSC-derived HPCs and erythroblasts.

DISCUSSION
Here, we described an efficient nonintegrating process to generate ␤-thalassemia iPS cells and subsequently corrected the disease-causing mutations using TALEN-mediated gene targeting. For the purpose of future application, we excluded c-MYC as a reprogramming factor and avoided using serum and mouse feeder cells. The nonviral approach combined with the defined condition employed in our system could be optimized further for generating clinical grade and safe iPSCs. The two ␤-Thal iPS cell lines generated in this process remain normal karyotype and stable in maintaining pluripotency. More importantly, we did not detect any transgene integration in these cell lines, which is essential for their further application.
Gene targeting mediated by traditional homologous recombination in human pluripotent stem cells is particularly difficult and inefficient. The recent developed TALEN technology provides great help to solve this problem by enhancing the gene targeting efficiency. TALEN-mediated gene targeting has been reported to be successful in multiple species (21,22,32,35). A critical point of TALEN application is its specificity to cleave a given DNA sequence. By using a reporter assay, we showed that the cleavage activity of the designed TALENs is highly specific. Moreover, we failed to detect mutations caused by off target cutting by TALENs in corrected ␤-Thal iPS cells. In sum, TALENs technology provides an efficient and cost-effective way for disease-causing gene correction in human pluripotent stem cells, as we have demonstrated here. Further experiments are needed to investigate genome widely whether TALENbased gene targeting would cause genomic instability and whether those potential genomic changes would be hazardous to further application.
␤-Thal is the most common inherited genetic disease in the world and contains over 200 different types of mutations in ␤-globin gene that could cause blood disorders. For practical purposes, it is essential to develop an efficient and universal process to correct most of the common type of mutations in ␤-globin gene. To achieve this goal, we selected 3Ј downstream of ␤-globin as gene targeting site, because any little changes within the HBB gene region would affect its normal function EGMM, four types of colonies including erythrocyte, granulocyte, megakaryocyte, and macrophage. D, HBB and HBG gene expression (normalized to one copy of ACTIN) in erythroblasts derived from indicated iPSCs, H1(human embryonic stem cells), and core blood CD34ϩ cells. The values are means Ϯ S.D. for triplicate samples from a representative experiment. The p values were calculated by one-way analysis of variance. *** indicates Ͻ0.001. E, conventional RT-PCR that amplifies HBB cDNA in erythroblasts derived from indicated cells. cDNA from undifferentiated H1 was used as a negative control. (14). Then, when we constructed the donor template for homology recombination, we included the whole wild type HBB gene in the 5Ј arm. Thus, for correction of mutations in ␤-Thal iPSCs, this template could be used to replace the endogenous gene with mutations or deletions in any site within the HBB region. As shown here, the gene targeting in ␤-Thal iPSCs with either mutations or deletions was equally efficient, demonstrating that this approach is universal and could be employed for other different types of ␤-Thal iPSCs. Indeed, the in situ correction of disease-causing mutations in two different types of ␤-Thal iPSCs restored the function of the HBB gene in their derived erythroblasts. Further studies are needed to evaluate whether hematopoietic stem cells differentiated from these corrected ␤-Thal iPS cells are functional in vivo.