Generating a Non-Integrating Human Induced Pluripotent Stem Cell Bank from Urine-Derived Cells

Induced pluripotent stem cell (iPS cell) holds great potential for applications in regenerative medicine, drug discovery, and disease modeling. We describe here a practical method to generate human iPS cells from urine-derived cells (UCs) under feeder-free, virus-free, serum-free condition and without oncogene c-MYC. We showed that this approach could be applied in a large population with different genetic backgrounds. UCs are easily accessible and exhibit high reprogramming efficiency, offering advantages over other cell types used for the purpose of iPS generation. Using the approach described in this study, we have generated 93 iPS cell lines from 20 donors with diverse genetic backgrounds. The non-viral iPS cell bank with these cell lines provides a valuable resource for iPS cells research, facilitating future applications of human iPS cells.


Introduction
Human somatic cells can be reprogrammed back into pluripotent state as induced pluripotent stem cells that can undergo unlimited self-renewal while maintaining the potential to differentiate into all three germ layers. Human iPS cells hold great potential to be used for personalized regenerative medicine, drug discovery and disease modeling, for circumventing ethical issues [1]. To date, human iPS cells had been successfully generated from different tissue sources, such as skin fibroblasts [2,3], extra-embryonic tissues [4] and blood cells [5], although the reprogramming efficiencies varied much for different cell types. Moreover, the virus-based approach used in most of these reports raised the safety concerns, which hampered their further applications. To enable the application of iPS technology, it is essential to develop and optimize a practical approach to generate iPS cells under a feeder-free, virus-free and serum-free condition. To date, several non-viral approaches had been reported to generate mouse and human iPS cells, by using adeno-virus [6], plasmids [7], purified proteins [8,9], synthesized mRNA [10,11,12]. However, these approaches remain inefficient or laborious. We and others had previously described that epithelia-like cells could be isolated from human urine and could be further reprogrammed into iPS cells with high efficiency by using virus-based approach [13,14,15,16]. Human UCs can be conveniently obtained through non-invasive method, thus representing an ideal cell source for iPS cell generation. Here, in this report, we had examined whether isolation of UCs to generate iPS cells can be applied in a large population with different genetic background. We firstly describe an efficient method to reprogram human UCs without using virus, serum, feeder and without oncogene c-MYC. By using this method, we have been generating a non-viral human iPS cell bank from donors in different ages and various genetic and disease backgrounds. Our non-integrating human iPS cell bank, enables comprehensive study of iPS cells from different origins, and also provides a valuable source for disease modeling and mechanical analysis.

Results and Discussion
We have easily recruited people for urine donation because urine collection is non-invasive and routinely carried out for clinical diagnosis (55 out of 60 surveyed people accepted to donate urine sample. Fig. S1A). With consent form signed by the donors, we have collected and established 46 UC lines (Table 1 and  Table 2) from 50-200 ml middle stream of the micturition. These donors include people with different ages (from 2 to 73 years old) and various disease genetic backgrounds. For all the donated urine samples examined so far, we could successfully isolate and expand epithelia cells from more than 90% of them, demonstrating that our approach for isolating UCs can be universally applied for most people (Fig. 1A). For these 46 established UC lines, 19 of them were collected from healthy people and 27 from patients that were diagnosed clinically with different diseases, such as hereditary blood disorders (hemophilia and b-thalassemia), autoimmune diseases, nervous neurological disorders (amyotrophic lateral sclerosis ALS, and myotonic dystrophy et al.). For the genetic diseases, we confirmed the mutations of b-thalassemia in 2 donors (Fig. S1B). Potential mutations of Alport's syndrome and ALS were not detected in the samples (5 genes known to cause ALS and 7 mutation sites of COL5A4 causing Alport's syndrome were listed in Table S1 and Table S2) because they are complicated genetic diseases. We also analyzed the proliferation percentage of UCs by EdU assay (Fig. 1B). Most of these UC lines proliferated well and could be expanded for more than 5 passages. In some cases (2 out of 46), we had to recollect the urine samples in order to get enough cells for further culture.
To obtain non-integrating iPS cells from these UC lines, we chose episomal system to deliver the reprogramming factors. Episome, which derived from EBV EBNA/OriP system to enable the replication of transfected plasmid in eukaryotic cells, had been used in iPS cell generation from neonatal fibroblasts and CD34+ hematopoietic progenitor cells in different laboratories [17,18,19]. Except specific donor cell type human adult adipose tissue-derived stem cells (AdSC) [20], most of these reports included oncogene c-MYC as reprogramming factor, raising risks in maintaining genomic stability during iPS generation [19,20] In addition, some of them used serum and mouse feeder cells for reprogramming  Table 1 and Table 2). B. Left: EdU imaging of representative UC. Right: EdU positive percentages of 5 UCs. Error bars are standard deviation of the mean, n = 3. C. Phase contrast and fluorescent photographs of UC-012 and UC-015 electroporation with episomal plasmid pCEP4-EGFP and cultured for 24 h in UC medium. D. Growth curves of UC-012 and UC-015 in UC medium, defined medium E8 and mTeSR1, respectively. *** indicates P,0.001. E. Schematic representation of iPS cell generation. The defined medium can be either mTeSR1 or E8. F. Different reprogramming factor combinations for iPS cell generation of UCs and skin fibroblasts from the same donor.  [17,18]. Therefore, we sought to reprogram human UCs through episomal system without using serum, feeders and c-MYC. To test the episomal system in UCs, we transfected an episomal vector encoding EGFP into these cells through electroporation. We showed that EGFP expression efficiency was remarkable, about 35% cells were positive for EGFP expression (Fig. 1C). Next, we found that UCs could survive and grow in serum-free media mTeSR1 and E8 which were used to maintains human ES or iPS cells [21,22], albeit slower than in their optimal medium (Fig. 1D). These data suggested that it was possible to use episomal system and defined serum-free medium for iPS generation with UCs as illustrated in Figure 1E. Because involvement of oncogene c-MYC during reprogramming might increase the risk of genomic toxicity [23], we tried to omit it by using OCT4, SOX2, SV40T, KLF4 (OSTK, encoded by pEP4EO2SET2K). However, we failed to obtain stable iPS colonies from UCs or skin fibroblasts (Fig. 1F), suggesting that the OSTK four factor were insufficient for nonintegrating iPS cell generation under serum-free conditions. We and several other groups had shown that miR-302-367 cluster could greatly enhance somatic reprogramming efficiency [24,25,26]. In addition, we found that mice chimeras with genome integration of miR-302-367 cluster and their offspring are tumorsfree for over 2 years. Thus, miR-302-367 cluster might be less genomically toxic and even suppress tumorigenecity of human pluripotent stem cells [27] and be a better choice for iPS cells generation than c-MYC. We then constructed an episomal vector expressing human miR-302-367 cluster (named pCEP4-miR302-367 cluster. Fig. S2A and S2B) and simultaneously transfected it into UCs with OSTK through electroporation. The transfected UCs were then cultured on matrigel-coated plate and in totally defined and serum free medium (mTesR1 or E8) for reprogramming. Around 20 days after electroporation, we could observe human ES like colonies. The cells exhibited traditional human ES/iPS morphology with a high nuclear to cytoplasmic ratio (Fig. 1G). However, the reprogramming efficiencies varied a lot for different donors. For 5610 5 cells we routinely used to start reprogramming, the number of human ES like colonies varied between 0 and 72 for different individuals. The representative reprogramming efficiencies of three different donors were shown in Figure 1H and efficiencies of other donors were listed in Table 1. Most of the colonies with human ES like morphology could be picked and expanded successfully. To the contrary, when we applied the same reprogramming factors combination on skin fibroblasts, we did observe morphology change, but failed to obtain alkaline phosphatase (AP) positive iPS cell colonies under same culture condition (Fig. 1F). These data indicated that UCs were much easier to be reprogrammed than skin fibroblasts. This is consistent to our previous finding that the fibroblasts not the UCs need to go through a mesenchymal-to-epithelial transition (MET) process, which might hamper their iPS cell generation [28]. We further confirmed this by showing that UCs expressed higher level of markers or accelerators of MET, such as E-CADHERIN, CLAUDIN, OCCLUDIN and miR-200c, miR-302b, but lower level of repressors for MET, like TWIST (Fig. S2C). Moreover, we failed to generate human iPS cells from UCs with the episomal miR-302-367 cluster vector alone, consistent with a previous report [26].
To date, through the approaches described above, we have successfully generated UC derived iPS cells (UiPSCs) from 20 donors with different genetic and disease backgrounds ( Table 1), demonstrating that it is a universal strategy, albeit with efficiencies varied for different donors. It is not surprise because the reprogramming efficiency variations had been well documented in mice [29,30]. As for the donors, we haven't found that the individuals with certain disease exhibited particularly different reprogramming efficiencies (listed in Table 1). The generation of iPS cells from UCs listed in Table 2 is underway. For each individual UC line, we usually picked and expanded at least 2 colonies for further characterization. Our standard iPS cell characterization was illustrated in Figure 2. The expanded colonies that passed the characterization including karyotyping, non-integrating and pluripotency will be deposited in the bank. Taking iPSCs generated from UC-012 for example, firstly, by using genomic PCR that could specifically amplify transgenes used for reprogramming, we confirmed that the stably expanded iPS colonies no longer harbored the exogenous reprogramming factors and episomal backbones ( Fig. 2A), and kept the normal karyotype determined by G-band analysis (Fig. 2B). We further demonstrated that endogenous pluripotent genes such as OCT4, SOX2 and NANOG were fully activated and were comparable to human embryonic stem cells (Fig. 2C-2E). Further by analyzing DNA methylation, we showed that the proximal promoters of OCT4 and NANOG were de-methylated (Fig. 2F). We performed embryoid body (EB) formation assay to demonstrate that the iPS cells could form typical EBs that expressed genes of three germ layer lineages (Fig. 2G). We also tested the pluripotency in vivo through injection of iPS cells into immune-deficient mice (NOD-SCID) and demonstrated that they could generate teratomas containing three germ layer tissues (Fig. 2H). The information of characterization on other iPS cell lines was listed in Table 1 and Figure S3. In all, we describe here a feasible and highly practical process to generate iPS cells from UCs in a totally defined condition that is feeder-free, virus-free and without using c-MYC. We have now generated 93 iPS cell lines from 20 individuals with various backgrounds ( Table 1). Generation of non-viral and feeder-free iPS cell bank is an important step towards the further applications. Employing UCs for non-viral iPS cell generation offers several advantages over other cell types such as skin fibroblasts and blood cells. Firstly, it is a totally non-invasive and easy process that is applicable for virtually any candidate to donate tissue, particularly for children and those patients with certain diseases such as hemophilia. Secondly, UCs exhibit the epithelial phenotype and are much easier to be reprogrammed to iPS cells than mesenchymal skin fibroblasts, by circumventing the MET process. This is particularly important for using non-viral approaches to generate iPS cells, as the efficiencies of most reported non-viral approaches are very low. Indeed, we failed to obtain stably expanded iPS cells from skin fibroblasts using our approaches described above. Lastly, the whole process is serum-free and feeder-free which makes it feasible to generate clinical non-viral human iPS cells under current good manufacture practice (cGMP) conditions in future.
One important safety concern for iPS cells is the genomic stability [31,32,33,34,35], which has been reported in extended passaging of human ESCs [36]. However, Geti et al. reported recently that iPSCs process appeared to generate little or no CNVs, even using vial approaches [35]. Further studies will be needed to analyze the genomic and epigenomic stabilities in detail on the iPS cell lines generated from different genetic backgrounds. The non-integrating iPS cell bank described here contains dozens of iPS cell lines from different individuals but generated in the same laboratory and the same condition, thus provides a valuable source for comprehensive comparative analysis. The bank is expanding rapidly and also a good resource for disease modeling and future applications in regenerative medicine.
We have previously reported that UCs can be reprogrammed into neural progenitor cells (NPCs) through the same combination of vectors described in this report plus a cocktail of small molecules to inhibit signaling pathways [37]. It had been shown that the external signaling pathways such as SMAD and GSK3b signaling are important in shaping the neural stem cell fate [38,39]. As for the urine cell reprogramming, we reasoned that the signaling inhibitors present in the medium played critical roles in determining whether the cells become NPCs or iPSCs. Our future work is to dissect the molecular mechanisms of the reprogramming process in the presence of those signaling inhibitors.

Ethical Statement
The individuals in this manuscript have signed written informed consent for donating human UCs for stem cell generation. The experiments involving human subject and animal research had been reviewed and approved by IRB at GIBH (NO. 2010012).

Collection and Expansion of UCs
UCs were collected and cultured as described previously [13]. The cells passaged less than 4 times were used for iPS cell generation. All patients with genetic diseases were diagnosed by clinical standards in our collaborated hospital. Sanger sequencing was used to confirm the potential mutations.

EdU (5-ethynyl-29-deoxyuridine) Imaging and Cell Proliferation Analysis
EdU labeling and cell proliferation analysis was performed using Click-iTH EdU Imaging Kit with Alexa FluorH (Life technologies), following the instructions of the manufacturer. 1.5610 4 human UCs at passage 4 were seeded per well in 24-well plates. EdU was added to cells 12 h after seeding and pulsed for 2.5 h. The EdU-treated cells were fixed and followed by the highly-specific click reaction. A Leica DMI6000B microscope (Leica Microsystems GmbH, Germany) was used for observation and photographing. Samples were analyzed in triplicate.

iPS Cells Characterization
PCR analysis of exogenous reprogramming factors and episomal backbone integration, immunofluorescence microscopy (OCT-3/4 Antibody, Santa Cruz Biotechnology sc-5279; Human Nanog Affinity Purified Polyclonal Ab, R&D AF1997; Anti-SSEA4 antibody, Abcam AB16287; Anti-TRA-1-60, Millipore MAB 4360), AP staining, karyotyping, bisulfate sequencing and teratoma formation were performed as we described previously [13,37]. The primers used for cloning, PCR analysis, qPCR and bisulfate sequencing were listed in Table S3. Total RNA was extracted using Trizol (invitrogen), and reverse transcribed by oligo dT or specific primers. qPCR was performed in triplicate with CFX96 machine (BIO-RAD) and SYBR Green Premix EX Taq TM Kit (Takara) following the instructions by the manufacturer.

Flow Cytometry Analysis
About 1610 6 iPS cells were used for each stanning. iPS cells were trypsinized and fixed with 1% paraformaldehyde dissolved in PBS for 10 min at 37uC. The cells were then washed with FACS buffer (PBS contains 2% fetal bovine serum, FBS) and resuspended in 90% ethanol for permeabilization by incubating 30 min on ice. After washing, cells were sequencially incubated with primary and secondary antibody for 30 min at 37uC. Control samples were stained with isotype-matched control antibodies. The iPS cells were washed and resuspended in FACS buffer and then processed for analysis on FACS Calibur (BD).

Statistical Analyses
Data were shown in mean. Error bars indicated standard deviation of the mean. P values were calculated by one-way ANOVA and Bonferroni post-hoc test, or student's t test by Origin.

Author Declaration
Cell lines and samples related in this report are available upon request for noncommercial use.