The Art of Human Induced Pluripotent Stem Cells: The Past, the Present and the Future

In 2006, Yamanaka and Takahashi electrified the scientific community by discovering that mouse somatic cells can be converted into embryonic stem cell-like cells by retroviral transduction of four transcription factors: Oct4, Sox2, Klf4, and c-Myc (OSKM). The first generation of mouse induced pluripotent stem (iPS) cells was incompletely reprogrammed, and failed to contribute to germline transmission. Nearly one year later, three groups, including Yamanaka’s, improved the reprogramming methodology and generated iPS cells that were in many respects, indistinguishable from ES cells, and also contributed to chimera formation and germline transmission. Shortly thereafter, the successful reprogramming of human somatic cells opened the gate for the development of patient-specific iPS cells for biomedical research and clinical application. Though human iPS cells resemble human ES cells in many aspects, the current iPS cell technologies showed several limitations for clinical usage. First, the efficiency of iPS cell generation is still low and the reprogramming process takes at least two weeks. Second, the virus-delivery of reprogramming factors introduces inconceivable risks of insertional mutagenesis in the genome. Third, given the various strategies for direct reprogramming, it remains difficult to assess the quality of iPS cells generated in each lab and for each patient. These issues should be addressed properly before any iPS cells could be translated into clinic. Here, we review recent progress in human iPS cell technologies, with a focus on the virus-free and integration-free iPS cell generation, which may lead towards the eventual goal of clinical applications. Keyword: Human induced pluripotent stem cells (hiPSCs), reprogramming strategy, hiPSC efficiency. TOWARDS EFFICIENT PRODUCTION OF HUMAN IPS CELLS In late 2007, scientists from the Yamanaka and Thomson laboratories successfully reprogrammed human somatic cells to pluripotency using two sets of transcription factors. Yamanaka and colleagues utilized the same set of transcripttion factors (OSKM) that they had previously demonstrated could reprogram mouse somatic cells to produce iPS cells from human dermal fibroblasts [1]. Thomson and colleagues also reprogrammed human somatic cells, but with the combination of OCT4, SOX2, NANOG and LIN28 (OSNL) [2]. Both groups have shown that human iPS cells resemble human ES cells in many aspects including morphology, proliferation, pluripotency markers, gene expression profiles, epigenetic status, and ability to differentiate into three germ layers. The achievement of human iPS cells holds great promise for regenerative medicine. It may have potential to replace human ES cells in cell therapy, which is hindered by immuno rejection and ethical issues. However, the efficiency of traditional iPS cell generation is only 0.001-0.01% of starting cells and the requirement for virus transduction prohibits its application in clinical therapy. The slow and inefficient process of deriving human iPS cells motivated efforts to improve reprogramming. Addition *Address correspondence to this author at the Institute of Medical Biology, 138648, Singapore; Tel: (65) 6407 0218; Fax: (65) 6464 2048; E-mail: [email protected] of Large T and TERT to either set of the reprogramming factors increased the efficiency of iPS cell generation by 2370 fold [3]. In another study, p53 knock down by siRNA delivery and introduction of UTF1 to OSKM cocktail enhanced the reprogramming efficiency to approximately 100 fold [4]. Recently a series of papers have extended the observation and show that ARF-p53 pathways are ratelimiting barriers in the reprogramming process [5-9]. The expression of “Yamanaka factors” induces expression of p53, p16 and p21, which results in DNA-damage and senescence. Releasing the barrier by transcriptional depletion of p21 or p53 increases iPS generation by approximately 100-fold. However, the convergent roles of c-Myc, Large T, TERT and lack of p53 promote immortalization, which may lead to DNA damage, genomic instability and tumorigenesis of iPS cells. In addition, with these potential oncogenes as the reprogramming factors, there is a risk of re-activation during iPS cell differentiation. Indeed, re-activation of c-myc was reported to promote tumor formation in 20% of iPS cell chimeric mice [10]. Thus replacement of these factors is necessary for clinical applications. ONCOGENE REPLACEMENT AND REPROGRAMMING EFFICIENCY The replacement of KLF4 and c-Myc with NANOG and LIN28 indicated that successful targeting of OCT4 and SOX2 to appropriate binding sites is sufficient for The Art of Human Induced Pluripotent Stem Cells The Open Stem Cell Journal, 2010, Volume 2 3 reactivation of the pluripotent transcriptional network, and that NANOGand LIN28-mediated events can replace KLF4and c-Myc in direct reprogramming [2]. Later, several groups demonstrated that c-Myc is dispensable in this process [11,12]. Chimeras made from myc-free iPS cells have reduced tumorigenicity. However, without c-Myc, reprogramming efficiency is 100 fold lower and the time for the appearance of stem cell colonies is longer. To overcome this problem, Huangfu et al. screened several histone deacetylase and DNA methyltransferase inhibitors, and identified valproic acid (VPA) as most effective in replacing c-Myc for iPS cell production [13]. The use of three factors OSK in the presence of VPA increased the reprogramming efficiency of human primary fibroblasts to 1%, a 1000 fold increase compared to previous reports. This method is repeatable and in our experience a relatively efficient way of iPS cell generation by virus transduction. The detailed modifications of protocol are summarized in Table 1. Using modified protocol Huangfu et al. were able to get iPS cells from transduction of two factors, OCT4 and SOX2, and efficiency equivalent to OSK with other methods. The efficacy of the protocol indicates that the starting epigenetic status and particular histone acetylation may be ratedetermining for direct reprogramming. CELL TYPES AND REPROGRAMMING EFFICIENCY The type of somatic cell used for deriving iPS cells may influence the quality and efficiency of reprogramming. In addition to fibroblasts, mouse iPS cells have been generated from hepatocytes, gastric epithelial cells [14], pancreatic cells [15], neural stem cells [16,17] and B lymphocytes [18]. Human iPS cells have been generated from fibroblasts [1,2], keratinocytes [19], mesenchymal cells [41] and blood progenitor cells [20]. Aasen et al. compared reprogramming of human keratinocytes and foreskin fibroblasts obtained from the same person. Using the same batch of retrovirus OSKM, the infection of keratinocytes yielded iPS cells at an efficiency of 1%, 100 fold higher than in fibroblast (0.01%). Moreover, iPS cells emerged 10 days after infection, as compared to 21-30 days for fibroblast. Cells showed similar characteristics to ES cells such as tight cell-cell contact, surface expression of E-cadherin and higher levels of endogenous KLF4 and c-Myc. It has been suggested that because keratinocytes and ES/iPS cells are epithelial-like cells with a similar epigenetic state, keratinocytes do not have to undergo a mesenchymal-epithelial transition necessary for fibroblasts [19]. Interestingly, small amounts of keratinocytes obtained from the follicle cells of a single human hair were sufficient to generate iPS cells. In clinical settings, hair follicle cells and blood are more convenient source of cells than fibroblasts, which need skin biopsy. REDUCED GENOME INTEGRATION SITES WITH SINGLE CASSETTE VECTORS Direct reprogramming of fibroblasts via lentiviral or retroviral transduction requires high virus titer with 15-20 proviral integration and 30%-90% transduction efficiency [21]. The comparatively low reprogramming efficiency may be partially due to transduction with separate vectors allowing integration of different numbers of proviruses for each factor. The cells that do not carry all four factors fail to form iPS cells [22,23]. Thus efficiency is lower than expected. In another study using dox-inducible lentivirus reprogramming, while primary iPS cells were differentiated into secondary fibroblasts in the absence of DOX, the secondary fibroblasts were reprogrammed into iPS cells with 100 fold higher efficiency when cultured in the presence of DOX [24]. It indicates that the correct stoichiometry of the four reprogramming factors may be critical for high efficiency and reprogramming using a single cassette of four factors displays is advantageous. In addition, reducing the number of viral integration sites enhances the chance for subsequent removal of the exogenous genes. Single cassette vectors have been described that are polycistronic lentiviral vectors expressing multiple genes simultaneously via self-cleaving 2A peptide. The defined factors OSKM were cloned in frame with 2A peptide separating each factor, which support near equimolar protein expression. Stem cells generated by this method were found Table 1. Comparison of Two Method of Retrovirus-Mediated iPS Generation Takahashi et al. Huangfu et al. Plasmids used pMXs-O,S,K,M pMXs-O,S,K,M Packaging system PLAT-E cells Gag-pol, VSV-G plus 293T cells Virus used for infection Ectropic retrovirus produced in PLAT-E cells, concentration of virus is not recommended Concentrated VSV-G pseudotyped retrovirus Cells used for reprogramming Human fibroblast cells pre-infected with Lentivirus expressing mouse receptor of retrovirus, Slc7a1, before transduction with 4 factors Human fibroblast cells without modifications Culture conditions after viral transduction 1). Cells keep in fibroblast medium till 7 days and switch to human ES medium. 2). Replate human fibroblast cells on MEF feeders at day 6 after infection. 1). Switch to human ES medium immediately after infection. 2). Valproic acid (0.5-1 mM) was added for 1-2 weeks. 3). The infected cells were allowed to grow undisturbed without splitting and replating on feeder cells. Reprogramming efficiency 0.02% 1% Abbreviations: O, Oct3/4; S, Sox2; K, Klf4; M, c-Myc. 4 The Open Stem Cell Journal, 2010, Volume 2 Zhang et al. to have a reduced number of integration sites (1-2 sites in all of the colonies tested) [25, 26]. Subsequently, the reprogramming factors can be removed by Cre mediated excision with some part of the vector backbone still in the genome [27]. However, given the fact that these lentivirus vectors are usually large in size (12-13kbp), virus packaged from these vectors has low transduction efficiency, compromising the advantage of a single cassette. Furthermore, the reported reprogramming efficiency for iPS cell generation is 0.5%1%, which did not show a great advantage over other methods. TOWARDS DNA-FREE iPS CELLS Virus-Free iPS Cells Lentiviral and retroviral based reprogramming result in multiple transgenes, which may lead to an increased risk of mutagenesis. For example, previous studies have shown that retroviruses integration can activate endogenous genes that cause cancer. iPS cells generated through these methods would not be acceptable for clinical use. Several groups have demonstrated that genome integration is not necessary for iPS cell production. Aoi and colleagues found that iPS clones do not share common insertional provirus sites, indicating that site-specific insertional mutagenesis is not necessary for the reprogramming process [14]. Another elegant study utilizing doxycyclineinducible lentiviral constructs for delivery of OSKM genes demonstrated that reprogramming factors were only required for a period of two weeks for iPS cell generation, after which the overexpressed genes were silenced [24]. These studies helped pave the way for developing technologies to generate iPS cells without viral integration, with an understanding of how long and at what levels the reprogramming factors needed to be expressed. With these findings in mind, early attempts to generate iPS cells without viral integration include the repeated transient transfection of plasmid-based vectors into mouse embryonic fibroblasts [28], and the use of adenoviruses in mouse liver cells [29] (summarized in Table 2). However, in both cases, the reprogramming efficiency was extremely low with slower kinetics, and no iPS cells have been generated from human cells using such methods. Another virus-free delivery method is the piggyBac transposition system, which serves as a vehicle for up to 10kb cargo capacity without losing transposition efficiency. With this tool, Nagy and colleagues generated both mouse and human iPS cells by transfecting fibroblasts with the piggyBac transposase and a polycistronic plasmid encoding the four “Yamanaka factors” linked by 2A sequences [30]. Their approach combined the advantages of viral integration with the need to have integration-free cells. By allowing the reprogramming factors to integrate into the genome, cells could maintain the appropriate levels of reprogramming factors through several cell divisions, allowing the gradual process of reprogramming to occur at a meaningful efficiency. By using a system that allows seamless excision of the factors, they could generate mouse iPS cells without persistent transgene expression and reduce the risk of insertional mutagenesis. However, the strategy of removing integrated transgenes requires an additional step that might prevent widespread clinical application of iPS cells. Given the slow growth rate of human iPS cell, complete removal of the transgenes has not been reported yet. Virus-Free and Integration-Free iPS Cells To simplify the derivation of integration-free human iPS cells, Yu et al. transfected somatic cells with an oriP/EBNA1 (Epstein-Barr nuclear antigen-1)-based episomal vector [31]. These plasmids can be transfected without the need for viral packaging, and replicate without integration in to the genome. The stable extrachromosomal expression of transgenes can be maintained by drug selection during reprogramming and they can be subsequently removed from cells by culturing in the absence of drug selection. When SV40 large T (SV40LT) was included in their cocktail, along with Oct4, Sox2, Klf4, c-Myc, Nanog, and Lin28, the authors were able to generate iPS cells from human foreskin fibroblasts in two independent experiments. While no integration was observed, clones were fully reprogrammed and had normal karyotypes. However, this virus-free and integration-free method required the use of SV40 Large T, and was still hindered by a low reprogramming efficiency. As the stable transfection efficiency for primary cells is 100 fold lower than virus infection, the efficiency for iPS cell generation by episomal vector is less than 0.001%. In addition, the continued use of nucleic acid delivery and its Table 2. Integration-Free Method for iPS Cell Generation Reference Cell Type Species Factors Method Efficiency [24] hepatocytes, fibroblasts mouse OSKM adenovirus 0.0001 to 0.001% [23] fibroblasts mouse OSKM plasmid 0.0001 to 0.001% [22] fibroblasts mouse OSKM single polycistronic plasmid 0.1% [25] fibroblasts human OSKM piggyBac transposon 0.008% [26] fibroblasts human OSKMNL SV40LT Epstein-Barr Virus based episomal vector 0.0003-0.0006% [27] fibroblasts mouse OSKM purified Recombinant protein from Ecoli. 0.006% [28] fibroblasts human OSKM whole protein extract from mammalian cells overexpressing OSKM fused with 9 arginine 0.001% Abbreviations: O, Oct3/4; S, Sox2; K, Klf4; M, c-Myc; N, Nanog; L, Lin28; SV40LT, SV40 large T gene. The Art of Human Induced Pluripotent Stem Cells The Open Stem Cell Journal, 2010, Volume 2 5 associated risk of genomic integration means that a simpler method with improved efficiency is still needed before integration-free iPS cells can be more widely used.