Conditionally replicating adenovirus prevents pluripotent stem cell–derived teratoma by specifically eliminating undifferentiated cells

Incomplete abolition of tumorigenicity creates potential safety concerns in clinical trials of regenerative medicine based on human pluripotent stem cells (hPSCs). Here, we demonstrate that conditionally replicating adenoviruses that specifically target cancers using multiple factors (m-CRAs), originally developed as anticancer drugs, may also be useful as novel antitumorigenic agents in hPSC-based therapy. The survivin promoter was more active in undifferentiated hPSCs than the telomerase reverse transcriptase (TERT) promoter, whereas both promoters were minimally active in differentiated normal cells. Accordingly, survivin-responsive m-CRA (Surv.m-CRA) killed undifferentiated hPSCs more efficiently than TERT-responsive m-CRAs (Tert.m-CRA); both m-CRAs exhibited efficient viral replication and cytotoxicity in undifferentiated hPSCs, but not in cocultured differentiated normal cells. Pre-infection of hPSCs with Surv.m-CRA or Tert.m-CRA abolished in vivo teratoma formation in a dose-dependent manner following hPSC implantation into mice. Thus, m-CRAs, and in particular Surv.m-CRAs, represent novel antitumorigenic agents that could facilitate safe clinical applications of hPSC-based regenerative medicine.


INTRODUCTION
Human pluripotent stem cells (hPSCs), including human embryonic stem cells (hESCs) and human-induced pluripotent stem cells (hiPSCs), are promising sources of material for use in cell transplantation therapy. However, the risk of formation of tumors, including teratomas and cancers originating from contaminating undifferentiated and transformed cells, represents the most critical obstacle to the safe clinical application of hPSC-based regenerative medicine. 1 Multiple approaches have been taken to improve safety by reducing the risk of carcinogenesis. However, most previous studies focused on improving generation of hiPSCs by eliminating potential oncogenic factors, such as the oncogene c-myc, or by integrating the reprogramming transgenes into chromosomes. 1 Although these sorts of strategies, classified here as the first safety approach, reduce the reprogramming-associated oncogenic potential of hiPSCs, they cannot completely eliminate tumorigenic potentials due to the intrinsic characteristics of hPSCs, i.e., self-renewal and pluripotency; consequently, it is still possible for teratomas to arise from contaminating undifferentiated hPSCs. In addition, chromosome instability, activation of oncogenic networks, and considerable plasticity are naturally prevalent in hPSCs, which accumulate genomic abnormalities during cell culture, possibly resulting in malignant transformation in some heterogeneous cells. 1 Moreover, a recent study demonstrated that transient expression of reprogramming factors leads to cancer development in the absence of genomic abnormalities. 2 In this regard, it is important to note the historical lessons of clinical gene therapy: in a number of preclinical animal studies, no tumorigenicity was observed, but there was nonetheless a high incidence of leukemia following ex vivo gene and cell-transplantation therapy in actual clinical trials. 3 Thus, current preclinical studies, which allow experimental comparison of in vivo tumorigenic activities among different treatment groups, have insufficient sensitivity to guarantee clinical safety. In other words, in the context of first-in-human trials of innovative cell therapies, we cannot confidently predict that the risk of tumorigenicity has been eliminated. Consequently, innovative safety approaches should be developed in order to decrease this risk.
We previously developed a novel method (adenoviral conditional targeting) that securely isolated target cells from other cell types and undifferentiated hPSCs. 4 This method, which can increase the efficacy and safety of hPSC-based regenerative medicine by decreasing tumorigenicity, is classified here as the second safety approach. Strategies that can directly target and kill, rather than merely inhibit, tumorigenic cells are classified here as the third safety approach. In this regard, a few recent studies have described generation of engineered hPSCs, in which a suicide gene under the transcriptional control of a pluripotency-related promoter is stably transduced. 5 Although this approach reduces teratoma formation, the prevalent inactivation of an integrated transgene in undifferentiated hPSCs and the risk of carcinogenesis due to integration-derived mutagenesis suggest that this strategy may not be universally effective. 3,4 Thus, there is an urgent need for a novel strategy that specifically kills tumorigenic undifferentiated hPSCs using a different methodology; we classify such future strategies as the fourth safety approach.
Conditionally replicating adenoviruses (CRAs), also called oncolytic adenoviruses, can selectively replicate in and kill cancer cells; consequently, CRAs represent attractive anticancer drugs. 6 Previously, we developed a method for generating CRAs that can target cancers with multiple cancer-specific factors (m-CRAs); this approach further increased cancer specificity without reducing the anticancer effects. 7,8 We also demonstrated that among candidate m-CRAs, survivin-responsive m-CRA (Surv.m-CRA) is one of the most promising anticancer agents, in two respects: superior cancer , and differentiated normal cells (human dermal fibroblasts (HDFs)) were examined by reverse transcription-polymerase chain reaction (RT-PCR) (a) and accurately quantitated by qRT-PCR (b,c). The HPRT gene was amplified as an internal control. n = 4, each group. *P < 0.05 and **P < 0.01. (d-f) β-gal activity was measured 48 hours after infection with Ad.Surv-LacZ, Ad.Tert-LacZ, Ad.RSV-LacZ, or Ad.CMV-LacZ at a multiplicity of infection of 30 in cancerous (MNNG-HOS and MKN-28) or differentiated normal HDFs (d), undifferentiated hPSCs (hESCs, hiPSCs-1 (201B7), and hiPSCs-2 (253G1)) (e), and differentiated hPSCs (f). n = 3, each group. *P < 0.05 (higher in Ad.Surv-LacZ); #P < 0.05 (lower in Ad.Surv-LacZ); n.d., no statistical difference. hiPSCs, human-induced pluripotent stem cells. specificity (i.e., safety) and therapeutic efficacy relative to clinically tested telomerase-responsive m-CRAs (Tert.m-CRAs), and strong anticancer effects against currently incurable cancer stem cells (CSCs). [8][9][10] Here, we show that the survivin promoter, like the telomerase reverse transcriptase (TERT) promoter, is highly activated in undifferentiated hPSCs, but is almost inactive in differentiated hPSCs and normal cells. Finally, we demonstrate that m-CRAs, in particular Surv.m-CRAs, are potentially useful as both potent anticancer drugs and as novel antitumorigenic agents in hPSC-based regenerative medicine; in the latter context, the viruses act by specifically killing contaminating undifferentiated hPSCs.

ReSUlTS
High mRNA levels and promoter activities of survivin and TERT in undifferentiated hPSCs Telomerase activity, expression levels, and promoter activities of TERT are high in both cancerous cells and in undifferentiated normal cells. 11,12 On the other hand, expression levels and promoter activities of survivin are also high in cancerous cells, [8][9][10]13,14 and a recent study suggested that survivin contributes to teratoma formation by hESCs. 15 However, survivin promoter activity in undifferentiated normal cells has not yet been carefully examined. Using reverse transcription-polymerase chain reaction (RT-PCR) and quantitative real-time PCR (qRT-PCR) analysis, we found that endogenous   The promoter assay demonstrated that both survivin and TERT promoters exhibited strictly cancer-specific activities (i.e., strong activity in cancer cells and undetectable activity in normal cells), and that the survivin promoter is stronger than the TERT promoter in cancer cells (Figure 1d), consistent with our previous studies on cancer. 9,16 Moreover, the activity of the survivin promoter was very high, relative to the TERT promoter and Rous sarcoma virus long terminal repeat (RSV promoter), a representative ubiquitously strong promoter, 17,18 in undifferentiated hPSCs but not in differentiated hPSCs (Figure 1e-f). Thus, the survivin promoter region that we use is able to strongly induce not only cancer-specific, but also undifferentiated cell-specific transactivation.
Both Surv.m-CRA and Tert.m-CRA exhibit undifferentiated cell-specific replication and cytotoxicity in hPSCs In Surv.m-CRA and Tert.m-CRA, the adenoviral early region 1A (E1A) was regulated by promoters of survivin and TERT, respectively, and both viruses ubiquitously express enhanced green fluorescent protein (EGFP). We next investigated whether Surv.m-CRA and Tert.m-CRA exerted efficient and undifferentiated cell-specific viral replication and cytotoxicity in hPSCs, relative to two control groups infected with replication-deficient adenoviral vector ubiquitously expressing EGFP (Ad.CA-EGFP) or no transgenic protein (Figures 2  and 3). By microscopically observing the spread of virus-infected EFGP-expressing cells and the swollen dying cells that are a characteristic feature of the adenoviral cytotoxicity, both Surv.m-CRA and Tert.m-CRA induced prominent viral replication, resulting in cytotoxicity, in the undifferentiated states of all three hPSCs, as well as in two types of cancer cells, HOS-MNNG and MKN-28; these effects were dose-dependent (Figures 2b-d and 3). By contrast, Surv.m-CRA and Tert.m-CRA exerted no apparent viral replication, and undetectable or minimal cytotoxicity, in the differentiated states of hPSCs and in normal HDFs (Figures 2e,f and 3). Moreover, a decrease in the ratio of EGFP-expressing cells was observed 6 days after Ad.CA-EGFP infection only in undifferentiated hESCs, but not in the differentiated state, despite the absence of any cytotoxicity detectable under the microscope, suggesting that the episomal adenoviral transgene was diluted by cell division (Figure 3).
Surv.m-CRA kills undifferentiated hPSCs more potently and specifically than Tert.m-CRA To accurately assess how each m-CRA specifically killed undifferentiated hiPSCs, but not differentiated normal cells, we cultured engineered hPSCs that stably expressed the far-red fluorescent protein mKate2 on HDF cells, and analyzed the cell types that exhibited efficient viral replication and cytotoxicity after m-CRA infection ( Figure 4). Infection with control Ad.CA-EGFP demonstrated that type 5 adenovirus could efficiently infect both hPSCs and HDFs (Figure 4a). Infection with each m-CRA significantly decreased only the number of far-red hPSCs, but not the number of HDFs, as time went on. The percentage of all cells (i.e., blue-stained nuclei) on each dish that were far-red hPSCs was accurately determined by cell image analysis 1, 3, 5, and 7 days after infection. The results showed that hPSCs were more efficiently killed by Surv.m-CRA than by Tert.m-CRA (Figure 4b,c); this tendency was consistent with the results of the viability and promoter assays described above (Figures 2 and 3).
qRT-PCR analyses of Lin28, a representative pluripotencyassociated gene, 19 and mKate2, which was transduced and stably expressed in hPSCs but not HDFs, further supported the conclusion that the undifferentiated hPSCs were potently killed by both m-CRAs, and that Surv.m-CRA was a more potent killer than Tert.m-CRA (Figure 4d).
Teratoma formation after hPSC implantation was inhibited by m-CRA pretreatment Finally, we examined the efficiency with which the m-CRAs inhibited in vivo tumor formation after inoculation of undifferentiated hESCs, infected 1 hour earlier with either virus (or no virus, as a control),  (Figure 5a), partially inhibited tumor formation (Figure 5b). Thus, pretreatment with Surv.m-CRA or Tert.m-CRA, which specifically and efficiently killed undifferentiated hPSCs in vitro, abolished in vivo teratoma formation in a dose-dependent manner.

DISCUSSION
No previous report has addressed the possibility of an oncolytic virus that could be used to inhibit hPSC-derived tumors, including teratomas. Therefore, this study represents the first demonstration of a novel m-CRA strategy that specifically and efficiently eliminates undifferentiated cells, thereby inhibiting in vivo teratoma formation after hPSC transplantation. Furthermore, the results of this study clearly identify Surv.m-CRA as an effective agent. Although three previously reported approaches for reduction of the tumorigenic potentials of hPSCs-reduction of the reprogramming-associated oncogenic potential of hiPSCs, purification of target cells, and generation of the engineered hPSCs-are still useful, as described in detail in the Introduction, our novel m-CRA strategy may overcome the deficiencies of these approaches. Although this method needs to be optimized in future studies using individual animal disease models, this approach should dramatically facilitate safer clinical trials of hPSC-based regenerative medicine. The m-CRA antitumorigenic agent has several potential advantages. It should be noted that the degrees of replication of m-CRAs correlate well with the transcriptional features (i.e., the activity and specificity of the promoters) of the target genes. For instance, survivin expression levels are positively correlated with poor prognosis in human cancer patients, and the activity of the survivin promoter and the effectiveness of Surv.m-CRA were elevated in CSCs, which are more malignant than non-CSC fractions of cancer cells. 10,13,14 By contrast, replication of some oncolytic viruses cannot be transcriptionally controlled. For instance, herpes simplex virus cannot always achieve cancer-specific viral replication using cancer-specific promoters, including TERT and survivin promoters. 20,21 Thus, the highly controllable viral replication and strictly target cell-specific cytotoxicity are major advantages of m-CRA relative to several other types of oncolytic virus. Moreover, m-CRA technology allows us to further increase the specificity and efficacy by adding other cell-specific promoters and introducing transgenes, respectively. 7,8 m-CRA also has advantages regarding safety, for several reasons. First, due to the episomal nature of adenoviruses, these constructs integrate very rarely into the chromosome. This represents a safety advantage because genomic integrations by other types of viral vectors used in clinical gene therapy have resulted in mutagenesis-derived carcinogenesis in human patients. 3 Severe adverse side effects of such mutagenesis, including carcinogenesis, have not been clinically reported in the context of infections with wild-type adenovirus or adenoviral gene therapy. Second, wild-type human adenoviruses are not very harmful in themselves; their infections usually cause only mild and temporal symptoms, such as the common cold and epidemic conjunctivitis. Third, it should be noted that the fundamental safety of CRAs (oncolytic adenoviruses) has already been verified in several clinical trials in human cancer patients, in whom in vivo injections of large amounts of CRAs did not cause severe adverse side effects. 22 Therefore, as shown in this study, it is unlikely that an m-CRA with greatly attenuated replication and cytotoxicity in normal cells would cause severe side effects when used as an ex vivo antitumorigenic agent in hPSC-based cell transplantation therapy.
Previous studies showed that Surv.m-CRA was one of the most promising agents for oncolytic virotherapy, for two main reasons. First, the safety and anticancer effects of TERT-responsive CRA had been already verified in a clinical trial, and Surv.m-CRA was superior to Tert.m-CRA in terms of both cancer specificity (i.e., safety) and efficiency in experiments. 8,9,22 Second, Surv.m-CRA exhibited not only therapeutic efficacy against all populations of cancer cells, but also exerted higher efficacy against CSCs, against which conventional chemoradiotherapies are ineffective. 10 In addition to these promising data regarding its use as an anticancer agent, the results of this study clearly show that Surv.m-CRA could be used as an undifferentiated cell-specific m-CRA agent in hPSC-based regenerative medicine. We anticipated neither that the activity of the survivin promoter would be so high (e.g., higher than the TERT promoter and the ubiquitously strong RSV promoter), nor that Surv.m-CRA would be more effective than Tert.m-CRA against undifferentiated normal hPSCs. This was in part because biomedical studies regarding survivin have focused mainly on this gene's roles in relation to cancer, 13,14,23 rather than hPSCs, with the notable exception of one recent paper. 15 Therefore, future systematic analyses of candidate m-CRAs, in which viral replication is regulated by promoters of pluripotency-related and/or cancer-specific genes, would not only advance the m-CRA-based antitumorigenic strategy in hPSC-based regenerative medicine, but also help to discover novel aspects of stem cell biology.
In conclusion, m-CRAs represent novel antitumorigenic agents that should facilitate clinical applications of hPSC-based regenerative medicine. Surv.m-CRA is an especially promising agent from the standpoint of efficacy.

Cell cultures
KhES-1 hESCs and two lines of hiPSCs (201B7 and 253G1, here designated hiPSCs-1 and hiPSCs-2), which were generated by transduction of four (Oct3/4, Sox2, Klf4, and c-Myc) or three (Oct3/4, Sox2, and Klf4) reprogramming genes, respectively, were provided by Kyoto University through the RIKEN BioResource Center (Japan). 24 The protocols for hESC experiments were approved by the institutional review board, followed by notification of the Ministry of Education, Culture, Sports, Science and Technology, in accordance with the Guidelines on the Utilization of Human Embryonic Stem Cells in Japan. Both hESCs and hiPSCs were grown in an undifferentiated state on mitomycin C-treated mouse embryonic fibroblasts in ES/iPS media consisting of 1:1 mixture of high-glucose Dulbecco's modified Eagle's medium and Ham's nutrient mixture F-12 (Sigma-Aldrich Japan, Japan), 0.1 mmol/l 2-mercaptoethanol (Sigma-Aldrich Japan), MEM nonessential amino acids (Sigma-Aldrich Japan), 5 ng/ml recombinant human basic fibroblast growth factor (ReproCELL, Japan), and 20% KnockOut Serum Replacement (Life Technologies Japan, Japan), as described previously. 24 The human cancer cell lines PC3 (prostate cancer), HOS-MNNG (osteosarcoma), and MKN-28 (gastric cancer), and the primary cultured HDFs were cultured in Dulbecco's modified Eagle's medium supplemented with penicillin/streptomycin and 10% fetal bovine serum, as described previously. 8  The lentiviral packaging pLenti6 plasmid (Life Technologies) and the pmKate2-N plasmid (Evrogen, Russia) were used to construct the pLenti-CA-mKate2 LV plasmid, which encodes a far-red fluorescence protein reporter gene, mKate2, downstream of the cytomegalovirus enhancer and β-actin promoter (CA promoter). To generate LV, 293FT cells were transfected with pLenti-CA-mKate2 plasmid and lentiviral genome plasmids (Virapower packaging mix; Life Technologies) using the X-tremeGENE 9 DNA Transfection Reagent (Roche Applied Science, Germany). LV was concentrated using Lenti-X Concentrator (Takara Bio, Japan). hESCs and hiPSCs were dissociated into single cells, and then plated onto Matrigelcoated 24-well plates (Corning Japan, Japan), followed by culture in modified Tenneille Serum Replacer 1 (mTeSR1) media (Stem Cells Technologies, Canada) for 1 day before infection. The cells were infected with LV after replacement of the supernatant with new mTeSR1 media containing 4 µg/ ml Polybrene (Nacalai Tesque, Japan), and then cultured for an additional 24 hours. mKate-2-expressing hPSCs in the undifferentiated state, visualized by fluorescence microscopy, were isolated for use in subsequent experiments.

Cytotoxic effects in vitro
hESCs and hiPSCs were dissociated into single cells, and then plated onto Matrigel-coated 96-well plates, followed by culture in mTeSR1 media for 1 day before infection. The cells were counted and infected with Surv.m-CRA, Tert.m-CRA, or Ad.CA-EGFP at an MOI of 3 or 10 on day 0. Cell viability was determined by a WST-8 assay using the Cell Count Reagent SF (Nacalai Tesque) in accordance with the manufacturer's protocol. 18,26,27 Quantitative analysis of the number of remnant hPSCs HDFs were seeded at 10,000 cells/well in 96-well plates. One day later, mKate2-expressing hESCs or hiPSCs were seeded on HDFs at 1,000 cells/ well. Cells in 96-well plates were infected with each adenovirus at an MOI of 3 on day 0, and image acquisition was performed on days 3, 5, and 7 using a Cellomics CellInsight high-content screening platform (Thermo Fisher Scientific, Japan), immediately after the nucleus was stained with Hoechst 33342 (Invitrogen, Carlsbad, CA). The software integrated into the screening platform accurately counted numbers of mKate2-expressing hPSCs (identified by magenta cytoplasm) and all cells (identified by blue nuclei), from which the percentage of mKate2-expressing hPSCs was calculated.

Antitumorigenic effects in vivo in animal experiments
To assess the antitumorigenic effects of m-CRA in vivo, hPSCs were infected with Surv.m-CRA at an MOI of 0.3 or 3, or Tert.m-CRA or Ad.dE1.3 at an MOI of 3, for 1 hour, and then 3.6 × 10 7 infected cells in phosphate-buffered saline containing 30% Matrigel were subcutaneously injected into the dorsal flanks of severe combined immunodeficient mice (CLEA Japan, Japan) (n = 8 for each group). The number of mice with macroscopic tumor nodules was recorded 4, 6, and 8 weeks after hPSCs implantation. Mice were sacrificed 8 weeks after hPSC implantation, and tumor nodules were collected for histopathological analysis. Resected tumors were fixed in 10% buffered formalin, embedded in paraffin, cut into 4-μm sections, and stained with hematoxylin and eosin. All animal studies were performed in accordance with National Institutes of Health guidelines and with the approval of the Division of Laboratory Animal Science, Natural Science Center for Research and Education, Kagoshima University. All reasonable efforts were made to minimize suffering.

Statistical analysis
Data were represented as means ± standard errors. Statistical significance was determined using Student's t-test. P < 0.05 was defined as statistically significant.