An Inducible Caspase-9 Suicide Gene to Improve the Safety of Therapy Using Human Induced Pluripotent Stem Cells

Abstract

Human induced pluripotent stem cells (hiPSC) hold promise for regenerative therapies, though there are several safety concerns including the risk of oncogenic transformation or unwanted adverse effects associated with hiPSC or their differentiated progeny. Introduction of the inducible caspase-9 (iC9) suicide gene, which is activated by a specific chemical inducer of dimerization (CID), is one of the most appealing safety strategies for cell therapies and is currently being tested in multicenter clinical trials. Here, we show that the iC9 suicide gene with a human EF1α promoter can be introduced into hiPSC by lentiviral transduction. The transduced hiPSC maintain their pluripotency, including their capacity for unlimited self-renewal and the potential to differentiate into three germ layer tissues. Transduced hiPSC are eliminated within 24 hours of exposure to pharmacological levels of CID in vitro, with induction of apoptosis in 94–99% of the cells. Importantly, the iC9 suicide gene can eradicate tumors derived from hiPSC in vivo. In conclusion, we have developed a direct and efficient hiPSC killing system that provides a necessary safety mechanism for therapies using hiPSC. We believe that our iC9 suicide gene will be of value in clinical applications of hiPSC-based therapy.

Introduction

Patient-specific human induced pluripotent stem cell (hiPSC)-based therapies are a promising modality for regenerative medicine. Though the use of patient-specific hiPSC presents fewer ethical concerns than embryonic stem cells and should have a lower risk of rejection,^1,2,3 several safety concerns nonetheless exist. One such risk is oncogenic transformation in residual undifferentiated hiPSC,⁴ since many of the transcriptional regulators used for reprogramming exploit the molecular machinery of self renewal and so can serve as tumor promoters.^5,6,7 Moreover, the genetic and epigenetic aberrations induced by reprogramming or prolonged stem cell culture itself are associated with oncogenesis.^8,9 Even if oncogenic transformation is avoided, toxicities from unwanted activity by differentiated cells and tissues derived from hiPSC may persist and even progress over time. Hence, a means of controlling the growth and activity of hiPSC is necessary to ensure clinical safety.

The stable genetic introduction of a suicide gene is one of the most appealing safety strategies for hiPSC,^10,11 but to be effective such a safety switch for hiPSC should meet several criteria. The mechanism should have little spontaneous activity to ensure desired survival of hiPSC and their progeny, but should induce essentially complete killing once activated. Killing should be swift in order to regulate acute as well as more chronic toxicities, and the suicide strategy should kill both rapidly dividing undifferentiated hiPSC and their more slowly dividing or postmitotic differentiated progeny. The activating prodrug should be nontoxic and ideally otherwise bio-inert, and finally the system should be nonimmunogenic, so that immune responses against the safety switch do not needlessly destroy the cell product. Herpes simplex virus thymidine kinase (HSV-tk) and yeast cytosine deaminase (yCD) have been applied as a safety switches for iPSC,^12,13,14,15 but neither possess all of the desired features.

We have previously shown that an inducible caspase-9 (iC9) suicide gene can be used in patients as an effective safety switch for adoptive T-cell therapy^16,17,18,19 and that it may also be effective at controlling the survival of mesenchymal stromal cells (MSC).²⁰ This system has all of the necessary characteristics for an hiPSC safety switch. iC9-mediated apoptosis is based on conditional dimerization of a proapoptotic molecule, caspase-9, that acts in the late part of the intrinsic apoptosis pathway.¹⁶ iC9-transduced T cells and MSCs both undergo >90% apoptosis within 2 hours of exposure to the dimerizer and there is a further log reduction after 24 hours.^16,20 The iC9 safety system does not rely on DNA synthesis for its activity, and so in principle should be equally effective in controlling dividing and postmitotic differentiated cells that may be derived from hiPSC. The components of the safety switch are almost entirely of human origin and appear nonimmunogenic,¹⁶ and activation requires an otherwise bioinert small molecule dimerizing drug.^21,22

We have now modified our iC9 approach to make it effective for hiPSC, and show here the activity of the suicide gene in vitro and in vivo.

Results

iC9-hiPSC can be generated and express the transgene

We made a lentiviral plasmid encoding the iC9 transgene and green fluorescent protein (GFP) under the control of the human elongation factor-1α (EF1α) promoter (Figure 1a). From this plasmid, we produced lentiviral vectors that we used to transduce two hiPSC lines from different donors (iC9-TZ16 and iC9-TKCBSeV9). Control hiPSC were prepared by transduction of the same lines with a GFP-encoding lentivirus (GFP-TZ16 and GFP-TKCBSeV9). The transduction efficiency of iC9 transgene was approximately 30–40%, however, these transduced iC9-hiPSC and GFP-hiPSC were enriched to >99% GFP expression by fluorescence activated cell sorting (Supplementary Figure S1). The copy numbers of the iC9 transgene were measured by quantitative real-time polymerase chain reaction (q-PCR), and then calculated from the ratios of the iC9 signal/GAPDH PCR signal. Cells contained 7.39 ± 1.42 and 1.45 ± 0.51 iC9 copies in TZ16 and TKCBSeV9, respectively (Figure 1b). These genetically modified hiPSC maintained high expression of pluripotent stem cell markers including OCT4, SOX2, SSEA-1, TRA-1–60, TRA-1–81, and alkaline phosphatase (Figure 1c). Both lines of iC9-hiPSC expanded exponentially during in vitro culture, indicating strong self-renewal capacity (Figure 1d). Of note, iC9 expression persisted unchanged during culture over time (Figure 1e and Supplementary Figure S2). The iC9-hiPSC retained the capacity for multi lineage differentiation, including the ability to form embryoid bodies (Figure 1f), and teratomas in immunodeficient mice (Figure 1g). The teratomas from iC9-hiPSC contained cell derivatives from all three germ layers, demonstrating the pluripotency of iC9-hiPSC (Figure 1g). These results suggested that introduction of iC9 transgene did not compromise the capacity of self-renewal and pluripotency of hiPSC.

Figure 1.

Characteristics of iC9-hiPSC. (a) Lentiviral iC9 bi-cistronic vector. The vector contains iC9 sequence and GFP as a selection marker, separated by a 2A-sequence. (b) The iC9 transgene copy number in iC9-hiPSC. The dosage of iC9 transgene was normalized to that of GAPDH gene. (c) The expression of pluripotent marker panel, OCT4, SOX2, SSEA-4, TRA-1–60, TRA1-81, were verified by immunofluorescence staining in iC9-hiPSC cultured as single cell suspension. The expression of alkaline phosphatase was also evaluated. Bar = 50 μm. (d) Growth curve of hiPSC and iC9-hiPSC. The absolute cell numbers of each cell were calculated in several time points. (e) iC9 expression in the iC9-hiPSC at the different passage number was quantified by real-time quantitative polymerase chain reaction, and the level of mRNA expression was normalized to the level of GAPDH gene expression. (f) Representative images of iC9-hiPSC-derived embryoid bodies. Bar = 200 μm. (g) Histopathological images of teratoma derived from iC9-hiPSC. Bar = 100 μm.

iC9-hiPSC were effectively killed in the presence of CID

The proapoptotic molecule iC9 was activated by the administration of the small molecule chemical inducer of dimerization (CID; AP20817), that binds to a modified FK506-binding protein 12 (FKBP12) domain incorporated in the iC9 construct.¹⁶ We cultured iC9-hiPSC and GFP-hiPSC in the presence of the CID, and then used flow cytometry to calculate the percentage of live cells (Annexin V negative, 7-AAD negative, and GFP positive). A titration curve of CID showed that the concentration of dimerizing drug required for 50% killing (LD50) was 0.012 ± 0.001 nmol/l for iC9-TZ16 and 0.061 ± 0.033 nmol/l for iC9-TKCBSeV9 (Figure 2a). Killing increased over a broad dose response curve, and 10 nmol/l of CID induced maximum cytotoxic effects (Figure 2a).

Figure 2.

iC9-hiPSC were effectively killed in the presence of chemical inducer of dimerization (CID). (a) Dose–response curve using the indicated concentration of CID. (b) Thirty minutes after CID treatment, iC9-hiPSC rapidly showed apoptotic features, such as loss of adhesiveness and cell shrinkage. GFP images of both GFP-hiPSC and iC9-hiPSC were acquired. Bar = 200 μm. (c) Representative images of flow cytometry analysis by Annexin V and 7-AAD staining in GFP-hiPSC and iC9-hiPSC with or without the CID treatment. Annexin V negative, 7-AAD negative, GFP-positive cells were counted as live cells. The absolute numbers of GFP-positive live cells in 50,000 of total live cells (in parenthesis) are also shown. (d) iC9-hiPSC were treated with or without CID, and the percentage of GFP-positive live cells was measured by Annexin V/7-AAD staining by flow cytometry.

iC9-hiPSC had apoptotic features, including loss of cellular adhesiveness and cell shrinkage, within 30 minutes of CID exposure (Figure 2b). Twenty-four hours later, less than 1% of iC9-TZ16 cells (0.75 ± 0.12%) and 6% (6.135 ± 0.065%) of iC9-TKCBSeV9 remained viable. No significant cytotoxic effects were observed in GFP-hiPSC (Figure 2c,d and Supplementary Figure S3). Of note, the sensitivity of iC9-hiPSC to CID-cytotoxicity was stable overtime (Supplementary Figure S4). CID did not enhance the cytotoxicity or spontaneous differentiation of GFP-hiPSC even though GFP-iPSC were cultured with CID for over 1 week (data not shown). Hence, iC9-hiPSC undergo selective apoptosis after exposure to CID in vitro.

CID induced caspase-dependent apoptosis in iC9-hiPSC

To discover whether cell death by CID was mediated by caspase-dependent apoptosis, we measured caspase-9 and caspase-3 activation after CID exposure to iC9-hiPSC. Activation of caspase-9 and subsequently of caspase-3 in iC9-hiPSC was observed 2 hours after CID exposure (Figure 3a). To determine whether blockade of caspase-9 activation inhibited CID-induced apoptosis of iC9-hiPSC, we cultured iC9-hiPSC in combination with 20 μmol/l of the pan-caspase inhibitor qVD-Oph and 10 nmol/l of CID. CID-induced cytotoxicity was completely inhibited by qVD-Oph (Figure 3b). Hence, the cell death observed in iC9-hiPSC was mediated by caspase-dependent apoptosis.

Figure 3.

Chemical inducer of dimerization (CID)-induced caspase-dependent apoptosis in iC9-hiPSC. (a) Activation of caspase-9 and -3 was measured by colorimetric assay. (b) iC9-hiPSC were treated with CID in combination with 20 μmol/l of qVD-Oph, and the percentage of GFP-positive live cells was measured by Annexin V/7-AAD staining by flow cytometry.

CID is capable of inducing apoptosis in differentiated iC9-hiPSC progeny

To investigate if iC9 induces cell death even in the differentiated progeny of hiPSC, we generated mesenchymal stromal cells (MSC) from iC9-hiPSC and measured induction of apoptosis by CID. MSC were induced from iC9-hiPSC as previously described,²³ using differentiation medium containing the TGFβ pathway inhibitor SB-431542 (iC9-hiPSC-MSC) (Supplementary Figure S5a). The differentiated cells were stably positive for GFP expression beyond passage 10 (Supplementary Figure S5b). MSC phenotype²⁴ was confirmed by the expression of defined MSC cell surface markers measured by flow cytometry (Supplementary Figure S5c), and by the ability to differentiate into osteoblasts and adipocytes (Supplementary Figure S5d). iC9-hiPSC-MSC had comparable iC9 copy numbers to the parental undifferentiated iC9-hiPSC (4.00 ± 0.35 in iC9-hiPSC versus 3.90 ± 0.09 in iC9-hiPSC-MSC, respectively, Supplementary Figure S5e). iC9 mRNA expression in iC9-hiPSC-MSC was slightly lower than in the parental undifferentiated iC9-hiPSC (0.35 ± 0.02 in iC9-hiPSC versus 0.16 ± 0.05, respectively, Supplementary Figure S5f), but remained at levels sufficient to induce >99% CID-mediated cell death (GFP-positive live iC9-hiPSC-MSC after exposure to 10 nmol/l of CID; 0.70 ± 0.09%, Supplementary Figure S5g), indicating that the iC9 suicide gene system can remain active in the differentiated progeny of hiPSC.

iC9 safety system can control hiPSC-derived teratomas in vivo

Since our iC9 safety system can control the cell fate of hiPSC in vitro, we next evaluated the function of iC9 system on hiPSC-derived teratomas in vivo. 2 × 10⁶ iC9-TZ16 (>99% GFP positivity) were subcutaneously transplanted into the right thigh of male SCID-beige mice. The iC9-hiPSC generated palpable tumors in mice approximately 4 weeks after transplantation. These tumors were confirmed as teratomas, containing primitive tissues from all three germ layers (Figure 1f), and expressed GFP (Figure 4a). Reverse transcription PCR (RT-PCR) revealed sustained expression of iC9 transgene and downregulation of pluripotent gene expression, OCT4 and NANOG, in the teratomas derived from iC9-hiPSC (Figure 4b). To determine whether administration of CID induced apoptosis in the teratomas, tumor-bearing mice were treated with an intraperitoneal injection of 50 μg of CID, and the tumors were characterized histologically 48 hours after treatment (Figure 4c). Hematoxylin and eosin staining revealed that after a single dose of CID, the iC9-hiPSC-derived teratomas had apoptotic/necrotic features characterized by the presence of hemorrhage, pyknosis, and chromatin fragmentation (Figure 4d). Furthermore, terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay revealed large numbers of TUNEL-positive cells in teratomas that had been treated with one dose of CID (Figure 4e), showing that CID induced apoptosis in teratomas derived from iC9-hiPSC.

Figure 4.

iC9 safety system can induce apoptosis in the teratoma derived from iC9-hiPSC. (a) Representative GFP image of iC9-TZ16-derived teratoma. (b) iC9 transgene and pluripotent gene expression in the teratomas were determined by RT-PCR. (c) Schematic images of experimental design. (d) Hematoxylin and eosin staining of teratomas derived from iC9-hiPSC after one dose of chemical inducer of dimerization (CID) treatment. Massive hemorrhage, pyknosis, and karyorrhexis were observed (arrow). Bar = 100 μm (×100) and 20 μm (×400). (e) Terminal deoxynucleotidyl transferase dUTP nick end labeling assay showing massive apoptotic cells seen in the iC9-hiPSC-derived teratomas after the CID treatment. Bar = 100 μm.

To investigate whether the activation of the iC9 safety system could control teratoma growth, we administered three doses of 50 µg of CID intraperitoneally 24 hours apart, beginning when the tumors reached 5 mm in diameter (Figure 5a). In nontreated mice, the iC9-hiPSC-derived tumors continued to rapidly grow. In contrast, over the first 48 hours after CID treatment, the tumor volume of iC9-hiPSC-derived tumors was substantially reduced, and there was no further growth even after discontinuation of CID administration (Figure 5b,c, n = 5, P < 0.001). Hematoxylin and eosin staining at day 14 after CID treatment showed that the structure of iC9-hiPSC-derived teratomas was completely destroyed and replaced with fibrosis and necrotic tissues characterized by shrunken cells with karyolysis/karyorrhexis and eosinophilic cytoplasm (Figure 5d). Indeed, two of five mice in the treatment group subsequently showed increasing tumor growth, however, the tumors diminished again on readministration of CID (Supplementary Figure S6a). Seven days after retreatment, H&E staining showed extensive necrosis (Supplementary Figure S6b). Hence, the iC9 safety system can control iC9-hiPSC-derived teratomas in vivo by inducing apoptosis and necrosis.

Figure 5.

iC9 safety system can control hiPSC-derived teratomas in vivo. (a) Schematic images of experimental design. (b) Representative images of teratoma derived from iC9-hiPSC at day 14 after chemical inducer of dimerization (CID) treatment. (c) Growth curve of teratomas following 2 weeks after CID administration (n = 5 per group, *P < 0.001). (d) Hematoxylin and eosin staining of teratomas derived from iC9-hiPSC at day 14 after CID treatment. Teratoma structure was completely destroyed and replaced with necrotic tissues and fibrosis. Bar = 250 μm (×40) and 100 μm (×100).

Discussion

We have demonstrated that hiPSC can be engineered to express an iC9-based safety system. Introduction of iC9 does not alter the pluripotency of the hiPSC or their ability to self-renew. iC9-hiPSC are rapidly and efficiently eliminated in the presence of CID by caspase-dependent apoptosis in vitro, and iC9-hiPSC-derived teratomas were ablated by systemic administration of CID. This safety switch may therefore be used to reduce the adverse effects of hiPSC and their progeny, potentially improving the safety of clinical trials using patient-specific hiPSC.

While hiPSC have many potential therapeutic uses, concerns about the intrinsic safety of these cells and their derivatives sparked a search for ways to diminish the risk of malignant change or other complications. Some efforts to increase the safety of hiPSCs include reprogramming without oncogenes such as c-myc²⁵ or transiently expressing reprogramming factors.^26,27,28,29 Other investigators have attempted to reduce risks by removing undifferentiated hiPSC from the differentiated component using cell surface markers^30,31 or by using a molecular-targeting drug to which only undifferentiated hiPSC are sensitive.^32,33 While the above interventions may improve safety, the adverse effects of hiPSC likely will remain, attributable to both transformation events and the excessive or misplaced activity of the differentiated progeny. Hence, efforts to maximize safety will need to incorporate a suicide or safety gene into the transferred cells. Most prior safety switch studies used nonhuman iPSC,^12,14,15 expressing herpes simplex virus thymidine kinase (HSV-tk) or yeast cytosine deaminase (yCD) in murine and primate iPSC. These studies showed that suicide gene-transduced iPSC could be effectively killed in the presence of an activating drug. The iC9 system has also been tested in primate and murine iPSC, where its ability to control the growth of these cells was low,^12,34 contrasting with the results we report in the current study of hiPSC. Although these studies used the same iC9 sequence as the current report, the amino acid sequences of the catalytic domain of human caspase-9 protein encoded by the iC9 construct are not fully homologous with primate and murine caspase-9. The construct may therefore not function optimally in the apoptotic cascade of primate and murine cells. Moreover, in the primate iPSC study,¹² the iC9 transgene was driven not by the EF1α promoter, but by the spleen-focus forming virus (SFFV) promoter, which can be silenced in stem cells and their derivatives.³⁵ In the murine iPSC study,³⁴ the iC9 transgene was an EF1α derived promoter but DNA hypermethylation at CpG islands within the promoter was found in teratomas and MSC derived from murine iPSC, silencing expression of the iC9 transgene, leading to inability to control the growth of iC9-transduced murine iPSC-derived teratomas by the dimerizing drug.³⁴ We have also generated MSC from iC9-hiPSC as previously described (Supplementary Figure S5),²³ to determine if the iC9 transgene is silenced during differentiation. While iC9 expression in differentiated MSC was slightly lower than in the parental undifferentiated iC9-hiPSC, CID exposure induced >99% apoptosis in iC9-hiPS-MSC.

Our results showing that iC9-mediated cell death was sustained at high levels in hiPSC and their progeny contrast with earlier studies using murine iPSC.³⁴ The differences we observed in our own study may in part be attributable to biological differences in transgene silencing between human and murine stem cells^35,36,37,38 and to our use of the EF1α core promoter, in which most of the promoter sequence has been deleted and so contains fewer CpG methylation sites.

We explored the iC9 system for hiPSC because of the transgene's numerous potential advantages over other available and clinically tested safety switches. Both the HSV-tk and yCD systems require days or even weeks to achieve maximum cytotoxic effect.^12,13,14,15 By contrast, the iC9 system directly activates downstream components of the intrinsic apoptosis pathway, including caspase-3.²² Thus, we found that iC9-induced apoptosis of iC9-hiPSC occurred within 1 hour after exposure to CID and achieved maximal apoptosis within 24 hours. In vivo, substantial volume reduction of iC9-hiPSC-derived tumors was observed just 24 hours after the first dose of CID administration. These results are identical to those of clinical studies in which >90% of iC9-expressing T cells were rapidly killed after administration of the dimerizing drug with equally prompt and sustained resolution of the symptoms and signs of acute graft versus host disease.^17,18,19 Although rapid and high-level cytotoxicity was observed in iC9-hiPSC exposed to dimerizing drug, uncontrolled cell death caused by spontaneous dimerization of iC9 prior to CID exposure was negligible.

HSV-tk/5-FC systems are unlikely to be effective in slowly dividing or postmitotic cells (i.e., differentiated hiPSC) due to their cell-cycle-dependent cytotoxic machinery. Transgene products from both HSV-tk and yCD are of nonhuman origin and have been reported to induce immune responses that eliminate the modified cells even in the absence of the activating prodrug.³⁹ In contrast, the iC9 protein is based on human protein and appears to lack immunogenicity. Even though our system incorporates a 2A-like sequence derived from Thosea asigna virus, there is no evidence that this sequence itself is significantly immunogenic.⁴⁰ Finally, the prodrugs required for alternative safety systems may be toxic for humans (e.g., 5-FC for cytosine deaminase and gancyclovir for HSV-tk) or would have additional therapeutic uses (e.g., gancyclovir for viral infection) while CID has shown no evidence of toxicity even at doses 100-fold higher than that required to activate iC9.⁴¹ One potential advantage of the HSV-tk/yCD systems, however, is their ability to produce bystander effects due to the transfer or release of the toxic metabolites of the activating drug to the neighboring cells. However, the high level of killing induced by the iC9 system appears to make such a bystander effect redundant, and indeed may avoid the risk of amplified cytotoxicity in bystander cells that can harm normal tissue neighboring the site of engraftment of hiPSC or their progeny.

Despite the potential ability of the iC9 system to control the fate of hiPSC, our results indicate the existence of an apoptosis resistant iC9-positive subpopulation. Although our approach successfully controlled iPS cell growth in vitro, a small number of residual iPS cells remain in vitro after exposure to dimerizing drug, probably related to the tumor regrowth after CID treamtnemt in vivo (Supplementary Figure S6). Since hiPSC may contain multiple mutations that occur prior to reprogramming or evolve with passaging,^42,43,44,45 the small number of residual cells likely reflect a heterogeneous subpopulation with multifactorial mechanisms related to iC9 resistance^46,47 (and unpublished data). Nonetheless, >99% of iC9-hiPSC (including >90% of iC9-hiPSC with low iC9 copy number) had stable iC9 expression and were consistently sensitive to iC9 activation (Figure 1e, Supplementary Figures S2 and S4) and even after differentiation (Supplementary Figure S5). Hence, there is no detectable increase in the percentage of the iC9-resistant population in the majority of iC9-hiPSC over time. In practical terms, hiPSC-based clinical studies with hiPSC derived cells will almost certainly require treatment with a population containing limited numbers of residual undifferentiated hiPSC, so that the total number of iC9-resistant cells infused will be very small indeed, and allowing surgical or other ablation of any residual tumor after safe and effective debulking with dimerizer treatment. Moreover, if clinical studies demonstrate a need to further increase the safety of iC9-hiPSC, it may be possible to prepare single cell derived iC9-hiPSC clones that are highly sensitive to the dimerizing drug even after long-term (up to 20 days) culture in vitro (Supplementary Figure S7). Such an approach is certainly feasible and may be able to use clinically tested selectable markers such as truncated human CD19^17,18 or NGFR⁴⁸ instead of GFP which was used in this study and is potentially immunogenic. Alternatively, introducing the iC9 cassette into a safe harbor such as the AAVS1 locus using gene editing techniques may be feasible if this approach is shown to be suited for clinical application.^49,50 In addition, it may be possible to target undifferentiated hiPSC or hiPSC showing malignant transformation by using an hiPSC specific promoter such as NANOG although promoter potency may be a limitation of this approach.

In conclusion, we present here a direct and efficient hiPSC killing system that provides a necessary safety mechanism for therapies using hiPSC and their progeny. We believe that our iC9 suicide gene is a promising approach that will be of value in clinical applications of hiPSC-based therapy.

Materials and Methods

Human iPS cells culture. Human iPSC, TZ16, were kindly provided by the Human Stem Cell Core at Baylor College of Medicine (Houston, TX). TKCBSeV9 cells were a gift of Dr. Miki Ando, the Center for Stem Cell Biology and Regenerative Medicine, Institute of Medical Science, the University of Tokyo (Tokyo, Japan). Both hiPSC were maintained in an undifferentiated pluripotent state on a BD Matrigel (Corning, Corning, NY) coated plate with mTeSR1 medium (Stemcell Technologies, Vancouver, Canada), as described elsewhere. The medium was replaced every day and these cells were passaged every 5–7 days, before the cells became confluent. Cells were grown at 37 °C in a humidified incubator at 5% CO₂.

Construct of iC9 lentiviral vector. The iC9 gene was a modified human CASP9 gene in which the caspase recruit domain (CARD) had been deleted (ΔCASP9) and replaced by mutated FKBP12.^16,22 The mutated FKBP12 is a human FK506-binding protein (FKBP12) with an F36V mutation, which increases the binding affinity of the protein to a synthetic CID, AP20187 (ARIAD Pharmaceuticals, Cambridge, MA).⁵¹ The FKBP12-F36V was connected via Ser-Gly-Gly-Gly-Ser linker protein to ΔCASP9, which enabled dimerization of ΔCASP9 in the presence of CID. pCDH-EF1α-MCS-GFP third-generation self-inactivating lenti-reporter backbone with intron-less EF1α core promoter was purchased from System Bioscience (Mountain View, CA). We generated the transgene pCDH-EF1α-iC9.2A.GFP, which enabled transfected cells to express both iC9 and GFP. Briefly, the iC9 cassette was amplified from SFG.iCasp9.2A.ΔCD19¹⁸ using the specific primer set listed in Supplementary Table S1. Both the PCR product and pCDH-EF1α-MCS-GFP lentiviral backbone were digested with XbaI and EcoRI restriction enzymes (New England Biolabs, Ipswich, MA), then were ligated to generate pCDH-EF1α-iC9.2A.GFP. The transgene pCDH-EF1α-GFP was also prepared to make hiPSC that expressed only GFP (Figure 1a).

Lentiviral particle production and transduction to hiPSC. Lentiviral particles were obtained by cotransfecting these transgene vectors and pPACKH1 plasmids (System Bioscience) into packaging cell line 293TN cells as per manufacturer's instruction. Supernatants containing lentiviral particles were collected 48 and 72 hours after transfection. For lentiviral transduction, 0.5 × 10⁵ of hiPSC were plated as a single cell suspension in BD Matrigel coated 24-well plates in mTeSR1 with 10µmol/l of ROCK inhibitor, Y-27632 HCL (ApexBio Technology, Houston, TX). After 24 hours, the medium was replaced with fresh mTeSR1 with concentrated lentiviral particles at multiplicity of infection of 20 and Trunsdux (System Bioscience) as per manufacturer's instruction. The cells were incubated at 37 °C in 5% CO₂. Six hours after incubation, the medium was replaced with fresh mTeSR1, which was changed every day. When the cells reached sub-confluence, they were passaged and plated to a BD Matrigel-coated six-well plate at the ratio of 2 × 10⁵ cells/well. The transduced hiPSC (iC9-hiPSC and GFP-hiPSC) were enriched for GFP-positive cells using MoFlo cell sorter (Beckman Coulter, Brea, CA).

Copy number and mRNA expression of iC9 transgene. Quantitative polymerase chain reaction (q-PCR) was performed to determine copy number and expression of iC9 transgene, as previously described.^18,19 Briefly, DNA and total RNA from each sample were extracted by QIAamp DNA mini kit and RNeasy mini kit (Qiagen, Venlo, The Netherlands), respectively, as per the manufacturer's protocol. mRNA was transcribed into cDNA by iScript cDNA synthesis kit (Bio-Rad, Hercules, CA) as per the manufacturer's protocol. q-PCR was performed with iQ5 Real-time PCR Detection System (Bio-Rad, Hercules, CA), iTaq Universal SYBR Green Supermix (Bio-Rad), and specific primer sets that amplify the iC9 transgene but not endogenous human CASP9 gene. The real-time PCR reaction used one cycle of 95 °C for 30 seconds, followed by PCR amplification with 40 cycles of 95 °C for 15 seconds and 60 °C for 1 minute. The gene dosage of iC9 was normalized to that of GAPDH, and the mRNA expression of iC9 was also normalized to that of GAPDH. Primer sequences are shown in Supplementary Table S1.

In vitro embryoid body formation. For embryoid body (EB) formation, hiPSCs were harvested by treating with TrypLE Select (Life Technologies, Carlsbad, CA) to make a single cell suspension. Approximately 2 × 10⁶ hiPSC were plated to Aggrewell800 (Stemcell Technologies) with Aggrewell Medium (Stemcell Technologies) and 10 μmol/l of ROCK inhibitor to make micro cell clumps as per manufacturer's protocol. Twenty-four hours later, the cell clumps were transferred to an Ultra-Low Binding Culture Plate (ScienCell Research Laboratories, Carlsbad, CA), and cultured with Aggrewell medium for 7 days as a floating culture.

Immunofluorescence staining. Immunofluorescence staining for pluripotent markers was performed using Human Embryonic Stem/Induced Stem cell Characterization Kit (Applied StemCell, Menlo Park, CA) according to the manufacturer's protocol. The images of cells were acquired by an Olympus IX70 Fluorescent Microscope (Olympus, Center Valley, PA).

In vitro apoptosis study. The chemical inducer of dimerization (CID; AP20817, ARIAD Pharmaceuticals) was purchased from Clontech Laboratories (Mountain View, CA). Twenty-four hours after CID exposure, the cells were harvested and stained with Annexin V-PE and 7-amino actinomycin D (7-AAD) according to the manufacturer's instruction (Annexin V: PE Apoptosis Detection Kit I, BD Pharmingen). The percentage of Annexin V negative, 7-AAD negative, and GFP positive cells were quantified as live cells by flow cytometry (Gallios, Beckman Coulter, Brea, CA) and analyzed with Kaluza Flow Analysis Software (Beckman Coulter). To inhibit apoptosis, 20 µmol/l of the pan caspase inhibitor qVD-OPh (ApexBio Technology) was added 24 hours prior to CID exposure.

Caspase colorimetric assay. Activation of caspase-9 and caspase-3 was determined by Caspase-9/-3 Colorimetric Assay Kit (Biovision, Milpitas, CA) according to the manufacturer's instruction. Briefly, GFP-hiPSC and iC9-hiPSC were incubated with or without 10 nmol/l CID for 2 hours, then harvested and lysed to purify proteins. Hundred micrograms of total protein were incubated with LEHD-pNA substrate for caspase-9 assay and DEVD-pNA substrate for caspase-3 assay, respectively, at 37 °C for 2 hours. Samples were analyzed by Infinite 200 PRO (TECAN, Männedorf, Switzerland) at 400 nm wavelength.

In vivo killing studies in murine model. The iC9-hiPSC were harvested by treating cultures with ReLeSR (Stemcell technologies) and centrifuged at 400 g for 5 minutes. The cell pellets were gently resuspended with DMEM/F12 medium (Life Technologies), at a ratio of approximately 4 × 10⁷/1 ml. 2 × 10⁶/50 μl of iC9-TZ16 were injected subcutaneously into the right thigh of 5-week-old male SCID-beige mice (CB17.Cg-Prkdc^scidLyst^bg-J/Crl, Charles River Laboratories, Wilmington, MA) after mixing with the same volume of BD Matrigel.

For pathological analysis, the tumors were grown for 6 weeks after implantation, and tumor-bearing mice were treated with an intraperitoneal injection of 50 μg of CID. Tumors were harvested 48 hours after CID treatment and characterized by hematoxylin and eosin staining and TUNEL assay.

When tumors reached 5 mm in diameter, tumor-bearing mice were treated with intraperitoneal injections of 50 μg of CID for 3 days to assess tumor killing. The greatest longitudinal diameter (length; mm) and the greatest transverse diameter (width; mm) of the tumor were measured by caliper every day, and the tumor volume calculated using the following formula: tumor volume (mm³) = (length × width2)/2.⁵² The tumors were harvested 14 days after CID administration, and characterized by hematoxylin and eosin staining. All mouse experiments followed Baylor College of Medicine animal husbandry guidelines.

TUNEL assay. Apoptosis in the teratomas was determined by the TUNEL assay using Click-iT Plus TUNEL Assay for In Situ Apoptosis Detection, Alexa Fluor 594 dye (Life Technologies) as per manufacturer's instruction. Formalin-fixed paraffin embedded teratomas derived from iC9-hiPSC with or without CID treatment were sectioned, deparaffinized, and used for the assay.

Statistical analyses. All numerical results are represented as mean ± standard deviation. Student's t-test was used to determine the statistical significance of difference between samples. All data were obtained from three independent experiments.

SUPPLEMENTARY MATERIAL Figure S1. Transduction efficiency of iC9 transgene. Figure S2. GFP expression in iC9-hiPSC during passages. Figure S3. The relative number of GFP positive viable iC9-hiPSC before and after CID exposure. Figure S4. Sensitivity to cytotoxicity by CID was stable in iC9-hiPSC overtime. Figure S5. Generation and characteristic of iC9-hiPSC-MSC. Figure S6. Effects of Repeat dosing of CID on recurrent teratomas. Figure S7. Monoclonal iC9-sensitive hiPSC show complete cell death after CID exposure. Table S1. Primer information.