Derivation of Xeno-Free and GMP-Grade Human Embryonic Stem Cells – Platforms for Future Clinical Applications

Clinically compliant human embryonic stem cells (hESCs) should be developed in adherence to ethical standards, without risk of contamination by adventitious agents. Here we developed for the first time animal-component free and good manufacturing practice (GMP)-compliant hESCs. After vendor and raw material qualification, we derived xeno-free, GMP-grade feeders from umbilical cord tissue, and utilized them within a novel, xeno-free hESC culture system. We derived and characterized three hESC lines in adherence to regulations for embryo procurement, and good tissue, manufacturing and laboratory practices. To minimize freezing and thawing, we continuously expanded the lines from initial outgrowths and samples were cryopreserved as early stocks and banks. Batch release criteria included DNA-fingerprinting and HLA-typing for identity, characterization of pluripotency-associated marker expression, proliferation, karyotyping and differentiation in-vitro and in-vivo. These hESCs may be valuable for regenerative therapy. The ethical, scientific and regulatory methodology presented here may serve for development of additional clinical-grade hESCs.


Introduction
Human ESC lines [1], [2] holds promise for disease modeling, basic scientific research, drug development, toxicity studies, and may serve as an unlimited renewable source of cells for transplantation therapy.
Most of the reported hESC lines worldwide are not ideal for use in clinical trials. They were developed without adherence to Good Manufacture Practices (GMPs), using animal-derived researchgrade reagents, which may infect the cells with animal pathogens. Moreover, many cell lines were derived and cultured on animal feeder cells, which may contaminate the hESCs by nonhuman sialic acid Neu5Gc molecules, which can elicit immune rejection after transplantation [3] and may render them as xenotransplantation products [4], [5]. In order to use hESC derivatives for clinical applications, the hESCs (clinical-grade hESCs) should ideally be developed under stringent ethical guidelines, from traceable and tested donors, preferably in an animal-free, GMPgrade culture system. The hESCs should meet safety criteria including a normal karyotype, definitive identity profile, sterility and the absence of adventitious viruses [6], [5].
In order to utilize stem cells for cell therapy, they preferably should be developed under strict cleanroom conditions, utilizing GMP-grade reagents and compulsory documentation [5]. So far, only a single group [24] fully complied with these requirements and derived hESC lines under GMP conditions; however, animal products were used in the feeder culture medium as well as the hESC culture system.
Here we report for the first time the derivation of clinical-grade hESC lines developed in an animal-free and GMP-compliant culture system under cleanroom conditions. Donor eligibility was carefully screened, in adherence to regulatory guidelines. All aspects of donor tissue and embryo handling, hESC derivation, culturing, cryopreservation, banking and characterization, were strictly monitored for quality practices, in line with their future use in transplantation therapy. The data, protocols and documentation presented here may serve as a platform for the development of additional clinical-grade hESCs.

Qualification of Materials
The study was initiated by the establishment of a quality assurance program to identify materials which would be both xeno-free and GMP-grade. Xeno-free candidate materials were tested in our feeder and hESC culture systems for their effectiveness in replacing research-grade reagents. The vendors of effective xeno-free materials underwent a QA certification process to ensure compliance with GMPs. To further confirm the vendors' quality statements we arbitrarily sampled some of the materials for sterility, mycoplasma and LAL testing. At the completion of this certification process, all animal-free materials that were accepted were also GMP-grade, except for FGF2 which was available only as a GMP-like product (Table S1).

Clinical-grade Human Feeders
To avoid the use of feeders from mouse origin, we developed and compared feeders from three primary human tissues including foreskin, aborted fetuses and umbilical cord. All donated tissues were obtained subject to informed consent and Institutional Helsinki Committee approval. Foreskin was obtained during surgical circumcision, tissues from aborted fetuses were collected from first trimester terminations of pregnancy and umbilical cord was harvested during elective cesarean sections. All tissues were obtained under sterile conditions from surgical procedures performed in an operating room. This enabled us to avoid the use of antibiotics in the feeder culture system, as advised by the FDA [28], to prevent potential allergic reaction from trace antibiotic remnants in future transplanted hESC-derived progeny.
Fibroblast feeder lines were derived from the three primary tissues with a culture system containing humanized or recombinant, GMPgrade reagents. After qualifying the derivation process with all ancillary materials [29] by developing feeder lines from the three sources under research conditions, clinical-grade lines were developed in Hadassah's GMP Facility, in accordance with FDA Aseptic Processing Guidelines [30] and European Agency for the Evaluation of Medicinal product (EMEA [31]) guidelines. Three GMP-grade umbilical cord, two foreskin, and one fetal fibroblast master cell banks (MCBs) were derived. Short tandem repeat (STR) DNA fingerprinting provided a unique identification profile for each line.
We characterized and compared the properties of the feeders derived from the three primary tissues between passages 5-10 in a GMP-compliant QC testing laboratory. Feeders from the three sources showed typical fibroblast morphology and were immunostained for vimentin, which is expressed by fibroblasts ( Figure 1A). FACS analysis showed that the majority (.80%) of the cells were immunoreactive with anti-CD44 and anti-human fibroblast antibodies ( Figure 1B). The fibroblast average doubling times were 26.163.8 hours (umbilical cord), 29.969.8 hours (foreskin) and 27.263.7 hours (human fetal fibroblasts; n = 3: Figure 1C). All three feeder types exhibited normal karyotypes when a minimum of 30 metaphase spreads were analyzed at passages 7-10 ( Figure 1D).
To test and compare the potential of the feeders to support the propagation of pluripotent stem cells, hESCs were cultured on the three types of feeders pretreated with Mitomycin C. The phenotype and differentiation potential of the hESCs was characterized after 1-5 passages, and to ascertain the long-term potential of the feeders to support undifferentiated propagation also again after 6-10 passages. On the three types of feeders, the characteristics of hESCs were similar at the early and late analysis time points. The cultured hESCs retained the morphology of undifferentiated cells and expressed alkaline phosphatase (AP) and Oct-4 ( Figure 2A). FACS analysis showed a comparable high percentage (.80%) of cells expressing SSEA-4, TRA-1-60, and TRA-1-81, while a minute percentage expressing SSEA-1 ( Figure 2B). The normal karyotype of the hESCs was ascertained by G-banding before and after their passaging on the feeders. The doubling time of hESCs on the three types of feeders did not differ significantly ( Figure 2C).
The hESCs could differentiate in vitro within embryoid bodies and neural spheres into progeny representing the three embryonic germ layers (Figure 2A). Thus, the potential of the feeders from the three sources to support undifferentiated propagation of hESCs was similar. The full characterization of GMP-grade feeder lines is summarized in Table S2.
We elected to use the cord as the source of feeders since it was not associated with ethical issues such as those related to fetal tissue donation, and was sterile, unlike disinfected foreskin. Samples of the MCB frozen at passage level 3 of the cord feeder line CRD008 were sent to an FDA-approved laboratory for safety testing according to FDA, EMEA, and International Conference on Harmonisation (ICH) [32] guidelines. The cells were found to be free of bacteria, fungi, yeasts, mycoplasmas, and adventitious viruses. Isoenzyme analysis confirmed that the feeders were from a human source. Karyotype analysis was normal (46XY), and endotoxin results (LAL) were within an acceptable range [33]; Table S3).
Four samples of the MCB were expanded and frozen after cirradiation at passages 7-8 as four working cell banks (WCBs). The lack of proliferating cells within thawed samples of the WCBs was confirmed by KI-67 immunostaining and analysis of BrdU incorporation ( Figure 1E). The WCBs underwent characterization (Table S4) similar to the MCB and limited safety testing (Table  S3). The derivation of three cord feeder MCBs and four WCBs enabled us to meet regulatory requirements for three runs of process validation under aseptic cleanroom conditions.
In conclusion, we developed for the first time xeno-free, GMPcompliant and safety-tested clinical-grade human feeders from cord tissue.

Clinical-grade hESC Culture System
Towards further developing a xeno-free, GMP-grade hESC culture system, while utilizing the irradiated cord feeders, we replaced porcine-skin-derived research-grade gelatin with recombinant gelatin. It did not adversely affect the undifferentiated propagation of the hESCs in research grade KO medium. After 4 passages, qualitative analysis of the morphology of colonies, Oct-4 Figure 1. Characterization of human feeders. Human feeders derived from three types of human tissues showed similar properties. They were immunoreactive with anti-vimentin (A; immunostaining, nuclei counterstained with 4', 6-diamidino-2-phenylindole (DAPI)), CD44 and human Fibroblast antigens (B; FACS analysis). They had similar doubling times (C) and normal karyotypes (D). c-irradiated cord feeders (WCB2) were mitotically inactive as indicated by lack of KI-67 expression and BrdU incorporation (E; immunostaining (green), nuclei counterstained with DAPI (blue)). Mitomycin-C treated and mitotically-active non-treated feeders served as negative and positive controls, respectively. (B & C present analysis of three lines from each feeder type). doi:10.1371/journal.pone.0035325.g001 and alkaline phosphatase expression ( Figure 2D) and quantitative analysis of the percentage of cells expressing markers of pluripotency was similar with both gelatins ( Figure 2E). In the next step, we identified a commercially available serum-free stem cell growth medium (SCGM; CellGroH/CellGenix TM ) which is designed for the expansion of hematopoietic and natural killer cells. SCGM is produced under GMP, is animal-free and is intended for clinical ex vivo cell culture use.
We compared the potential of SCGM (supplemented with FGF2) with KO medium to support undifferentiated propagation of hESCs on cord feeders and recombinant gelatin. After 4 passages, the morphology of colonies was similar and the stem cells expressed Oct-4 and AP ( Figure 2D). The percentage of cells expressing TRA-1-60 and TRA-1-81 was 75.664.9% and 7464.9%, respectively, with SCGM, significantly (P,0.01) lower compared to KO Medium. Upon supplementation of Figure 2. The development of a clinical-grade hESCs culture system. Feeders derived from the three types of human tissues were equally effective in maintaining undifferentiated pluripotent growth of hESCs (HES-1 [2]) for a minimum of 5 passages within research-grade KO medium. (A) The hESCs cultured on the three types of feeders formed colonies with typical morphology (phase contrast images), expressed alkaline phosphatase (AP), and were immunoreactive with anti-Oct-4 (nuclei were counterstained with DAPI). They were pluripotent as demonstrated by their potential to differentiate in-vitro into progeny representing the three germ lineages. Immunofluorescence staining showed differentiated cells expressing betatubulin III (ectoderm), muscle actin (mesoderm) and alpha-Feto-Protein (AFP, endoderm). (B) FACS analysis showing that the percentage of cells expressing markers of pluripotent human stem cells, and the level of background differentiation (percentage of SSEA-1 expressing cells) was similar (n = 3). (C) The hESC doubling time did not differ significantly after culture on the three types of feeders (n = 3). Cord feeders were further used for the development of the clinical grade culture system, and the phenotype of hESCs was compared after culture within research-grade KO medium system and following replacement and supplementation with xeno-free and GMP-grade reagents. The following culture compositions were compared: hESCs cultured on porcine gelatin (porc-gel) or in recombinant gelatin (rec-gel) in KO medium, SCGM and SCGM supplemented with HSA (SCGM-HSA). (D) With all culture conditions, the hESCs colonies had similar and typical morphological characteristics (phase contrast images), and the stem cells expressed alkaline phosphatase activity (red) and Oct-4 (green; blue-DAPI nuclear counterstaining; immunofluorescence images). (E) The percentage of cells expressing Tra-1-60 and Tra-1-81 was analyzed by FACS and compared between the various culture conditions (n = 4). Scale bar in (D) represents 200 um. doi:10.1371/journal.pone.0035325.g002 the SCGM with human serum albumin (HSA), the percentage of undifferentiated cells improved to 81.462.7% and 8361%, respectively. These percentages were slightly lower but not significantly different in comparison to KO medium ( Figure 2E). The doubling time of hESCs was similar in SCGM supplemented with FGF2 and HSA compared to the KO-based culture system (23.7562.5 (n = 6)), compared to 23.3061.8 (n = 8), respectively).
We next analyzed the feasibility of cryopreservation of hESC clusters by both slow cooling and vitrification methods using animal-free GMP-grade reagents. Slow cooling was performed either within human serum supplemented with 10% DMSO or a commercially available FDA-approved cryo-solution medium (CryoStorH). Ten clusters of hESCs were cryopreserved per vial. With both solutions, undifferentiated hESCs could be retrieved and propagated after thawing from each tested vial (n = 5 vials per solution). The percentage of cryopreserved hESC clusters that survived and gave rise to undifferentiated colonies upon plating was similar with human serum-DMSO and Cryostore solutions (31.464.34% and 32.664.52%, respectively). For practical reasons CryoStorH solution was further used for cryopreservation of hESCs.
Vitrification was conducted using commercially available solutions, which are FDA-approved for cryopreservation of embryos. Sealed straws (FDA-approved and CE marked) were utilized to avoid potential contamination from liquid nitrogen. Similar to slow cooling, undifferentiated hESCs could be retrieved and propagated from each thawed straw (5 vitrified clusters/straw; n = 6 straws). The percentage of surviving clusters that gave rise to undifferentiated colonies was 4767.4%. These results showed that both slow cooling and vitrification methods that utilize xeno-free GMP-grade reagents allowed reliable retrieval of cryopreserved hESCs.
Thus we validated that the newly developed culture system could support the undifferentiated propagation and cryopreservation of hESCs and was ready for use under GMP conditions for the derivation of clinical-grade lines.

Ethics of hESC Development
The study was approved by the Hadassah Medical Center IRB and the Israeli Ministry of Health's National Ethics Committee for Genetic Research in Humans. Supernumerary IVF embryos not intended to be used for fertility purposes and which were cryopreserved for at least five years were eligible for recruitment to the study. This was in line with Israeli IVF regulations, which permit, subject to the couples' consent, the discarding or donating to research of embryos frozen for five years. In order to ensure free informed consent, the team of physicians which treated the donors as IVF patients and obtained their consent was not the same team which was involved in the development of the hESC lines. In order to maintain strict confidentiality of the donors, they received a unique identifying code and all documents relating to them and hESCs produced from their embryos were labeled with it. Given that the derived hESCs were intended for clinical use, blood samples from the donors were archived, and a route was established to obtain in the future, if needed, subject to donor permission, additional medical information or blood tests. Approach of the donors for this purpose would only be possible after obtaining approval by the Hadassah Director of the Department of Gynecology. Donors were advised that they would not receive any information about the hESC lines, compensation or benefit for their donation, and would not influence or have any involvement in the use of the hESC lines.

Recruitment, Screening and Monitoring of Embryo Donors
Potential donor couples who met the ethical eligibility criteria mentioned above received information about the study and their questions were answered. Those interested signed informed consent documents (Attachment 1, File S22) including an authorization to review their medical histories (File S25). The donors were screened according to the FDA's Donor Eligibility Regulations for Human Tissues and Cells [34], including: (a) a physical examination; (b) review of donors' medical records to rule out disqualifying medical conditions; (c) an interview/screen for lifestyle risk factors; and (d) serum and swab samples, which were sent for testing using FDA-approved test kits. Results of these tests (Table S5) were sent to the donors. Screening through interview was performed where no acceptable test kit existed, such as for vCJD, sepsis, West Nile, and Vaccinia viruses. All the information related to donor screening and the status and quality of the donated embryos were recorded in the donors' Case Report Form (CRF). The CRFs and study forms used in the hESC trial are given in Table S6, and Files S2, S3, S4, S5, S6, S7, S8, S9, S10, S11, S12, S13, S14, S15, S16, S17, S18, S19, S20, S21, S22, S23, S24, S25, S26, S27, S28, S29, S30, S31, S32.
The trial was monitored for veracity by a clinical trial associate (CRA). Source data which included verification of the donors' health histories was included in the CRFs for donors whose embryos resulted in usable hESC lines.
Donors with verified good health histories from preferred countries of origin (low variant Creutzfeldt-Jakob disease and malaria prevalence) and with low reported lifestyle risks were selected to participate in the program. Exclusion criteria included lifestyle and behavioral risks (File S28), or medications and diseases that could have affected the quality of the embryos (Files S30, S31).

Derivation of Clinical Grade hESC Lines
One hundred embryos were recruited from 12 couples and were thawed in the IVF laboratory. 77 embryos survived thawing of which 34 developed to the blastocyst stage. Details regarding the developmental stage and morphological score of the blastocysts at the time of ICM isolation are summarized in Table S7. To avoid the use of animal-derived products, zona pellucida drilling and separation of the ICMs was performed with the assistance of laser according to our previously described protocol [10]. 23 ICMs were successfully isolated from blastocysts that had well developed ICMs. Eight blastocysts that either had poor ICMs or collapsed during ICM isolation were plated whole on the human feeders. Three blastocysts were damaged during the procedure and were discarded (Table S7).
The isolated ICMs and whole blastocysts were transferred from the IVF laboratory to the GMP-facility following Good Tissue Practices (GTPs), and were plated on the irradiated cord WCB feeders within the clinical-grade hESC culture medium. After 5-8 days, groups of small, tightly packed cells proliferated from the outgrowths of 7 ICMs and a single whole blastocyst. They were mechanically dissected from the outgrowth of differentiated cells and, following replating, gave rise to flat colonies of cells with the morphological appearance of hESCs. The colonies were expanded and early stocks were frozen. Given the difficulty in the expansion of undifferentiated colonies in five of the hESC derivations, we did not attempt to further propagate them beyond this stage. In three derivations, which were more easily propagated as undifferentiated colonies, the early passage (5-9) stocks were frozen by both vitrification and slow cooling methods. These stocks were frozen by both methods for backup reasons given the limited and comparable efficiencies of these methods under clinical-grade conditions. These three stocks (early banks: HAD-C 100, HAD-C 102 and HAD-C 106) underwent STR fingerprinting and HLA class I and II antigen profiling for identity determination, sterility testing and abbreviated characterization including, karyotyping, and limited analysis of the expression of markers of pluripotecy (Table S8).
The colonies of these three derivations were continuously expanded from the initial outgrowths to avoid repeated freezing and thawing. Samples were pooled and frozen at passages 11-24 as Primary Cell Banks (1u CB; by both cryopreservation methods; see Figures S1 and S2) and following additional expansion, cryopreserved (passages 16-33) as secondary CBs (2u CB by slow cooling) (Table S8, and Figure S3). Thus we followed the suggested two-tiered cell banking system (as described in [35]) with modifications. The derivation of three independent hESC CBs and their characterization, as described below, enabled us to meet regulatory requirements for three runs of process validation under aseptic cleanroom conditions. A flow diagram showing the steps we took in the derivation of our clinical-grade hESC lines is shown in Figure 3.

Characterization of Clinical-grade hESCs Lines
Detailed characterization was conducted on the primary (HAD-C100, 102 and 106) and secondary (HAD-C-100) CBs that were frozen by slow cooling. For characterization, thawed samples of the clinical grade hESCs were cultured in the GLP laboratory in the same culture conditions as in the derivation process.
The hESCs colonies showed typical morphology and the karyotype of the three lines was normal. The STR and HLA profiles of the CBs were identical to the profiles of the early stocks, verifying the identities of the CBs (The STR and HLA profiles are not shown to maintain the confidential identity of the donors). The colonies showed alkaline phosphatase activity and were immunoreactive with anti Oct-4, TRA-1-60 TRA-1-81, SSEA-3 and SSEA-4 antibodies. FACS analysis of markers for pluripotent hESCs of the 1u CBs and 2uCB showed that the majority of cells expressed TRA-1-60, TRA-1-81 and SSEA-3 ( Figure 4 and Figures S1, S2, S3 and Table S8).
Background differentiation was evaluated by FACS analysis of the percentage of cells expressing SSEA-1 and was shown to be in the range of 5-36% for all lines (Table S8). KI-67 immunostaining showed that the majority of cells were in an active cell cycle (.87% of the cells stained positive for KI-67, Figure S4, Table  S8). Doubling time of thawed hESCs from the 1u CBs was in the  range of 35.6-43 hours, and the doubling time of thawed hESCs from the 2uCB of HADC-100 was similar to that of the 1u CBs (Table S8).
The pluripotent potential of the three hESCs lines was demonstrated by their capability to differentiate into progeny representing the three embryonic germ layers in-vitro and in-vivo. Following 3-4 weeks of differentiation as embryoid bodies (EBs), immunofluorescence staining showed cells expressing beta-tubulin III (neuronal marker, ectoderm), muscle actin (mesoderm) and sox-17 (endoderm; Figure 4M-O and Figures S1, S2, S3M-O).
Safety testing for adventitious viruses of the hESC lines was negative and is shown in Table S9.
Gene expression analysis. RNA was prepared from undifferentiated hESCs colonies and EBs from each one of the lines. Gene expression was analyzed by the same Human Stem Cell Panel array that was previously used by the International Stem Cell Initiative (ISCI [36]) to analyze 59 lines. Gene expression of each gene is reported as its DeltaCt value normalized against the average Ct of three selected endogenous control genes (18 S, betaactin and GAPD-H; Table S10). Using the combined data of undifferentiated and differentiated samples, we tested the pairwise Pearson correlation of expression of the genes to the expression of NANOG ( Figure 5A). The expression of genes that are associated with pluripotency including GABRB3, GDF3, SOX2, POU5F1, DNMT3B, TERT, LIN28 and TDGF1 [36] was highly (r.0.88) correlated with the expression of NANOG. Five of these genes (GABRB3, GDF3, POU5F1, DNMT3B and TDGF1) were also found to be highly correlated (r.0.75) with the expression of NANOG in the ISCI study [36]. Genes that are known to be commonly associated with differentiation such as CDX2, GATA6, AFP, CD34, PECAM1, TH and ISL1 showed negative correlation (r,20.75) with NANOG expression.
We performed two-way hierarchical clustering analysis of the three clinical grade hESC lines with respect to gene expression. The genes that are presented in Figure 5B are those that significantly differed in their expression between undifferentiated and differentiated samples (unpaired t-test; P,0.05). Clustering results show two distinct sample clusters: one includes the undifferentiated cells of the three lines and the other the differentiated cells. The data also shows two clear clusters of genes expressed by undifferentiated and differentiated cells in each one of the samples. The genes NANOG, GABRB3, GDF3, POU5F1, DNMT3B and TDGF1, that were typically upregulated in the pluripotent state in the ISCI study, were also highly expressed by the three clinical-grade undifferentiated cell line samples and were down regulated upon differentiation. Genes that are expressed by differentiated cells (such as AFP, GATA4, GATA6, FN1, COL1A1) [36] were highly expressed in the EBs in all three lines and were down regulated in the undifferentiated hESCs.

Discussion
In this study we derived for the first time xeno-free and GMPgrade human feeders and ESC lines adhering to regulatory guidelines. We developed a unique animal reagent-free culture system for the derivation of the new hESC lines as well as the regulatory platform for donor selection and testing, which can serve for the establishment of additional clinical-grade lines. The hESC lines developed in this study may be utilized for the development of differentiated cells for transplantation therapy.
The ethical considerations that guided the development of the hESCs strictly followed Israeli law [37,38] and adopted the recommendations of the Bioethics Advisory Committee of the Israel National Academy of Sciences and Humanities [39]. They were in line with the ethical guidelines of the ISSCR (www.isscr.org), the US's National Academy of Sciences (www.nasonline.org) and the National Institute of Health (NIH: http://nih.gov). The three lines were approved for inclusion in the NIH Human Embryonic Stem Cell Registry (http://grants.nih.gov/stem_cells/registry/current.htm), and therefore can be utilized in federally funded research in the US. They may be used worldwide subject to specific national regulations. The ethical considerations that guided the derivation of hESC lines in this report, and the draft informed consent for the recruitment of embryo donors, may be valuable to other research groups deriving new clinical-grade hESC lines.
For the establishment of a hESC culture system that will be compliant for future clinical use, we established a material qualification process. We verified that the materials were primarily of GMP-grade, and were xeno-free. Vendors' adherence to GMP was ascertained by questionnaire and some materials were quality tested. Qualified materials were then introduced into the development process of our protocols in the research environment and tested for their suitability. The vendor quality approval, material quality testing and reagents' qualification in the process guaranteed that reproducible and safe hESC lines were derived.
The derivation protocols, which incorporated these materials, were initially used to develop, for the first time, xeno-free, GMPgrade human feeders from three sources; umbilical cord, foreskin and fetal tissue. We showed that the feeders from the three sources were similarly potent in supporting the undifferentiated propagation of hESCs, and hence elected to work with umbilical cord due to its many advantages.
Cord may be obtained during elective cesarean sections as a completely sterile tissue compared to foreskin which can only be disinfected before sampling. Cord blood can be easily harvested in sufficient quantities for donor safety testing, as compared to the ethical constraints and technical difficulties in obtaining sufficient blood sampling from infants undergoing circumcisions. Additionally, unlike with aborted fetal tissues, there is no moral debate regarding the use of umbilical cord tissue for research purposes.
At an early stage of the project, we identified a GMP-grade xeno-free media formulation that supported self-renewal of hESCs at an efficiency not significantly different compared to the standard KO research grade medium. At present, additional GMP-grade and animal-free media designed for the propagation of hESCs are available, though their capability to support derivation and long term propagation of genetically stable clinical-grade pluripotent hESC lines has yet to be demonstrated and compared with the culture system reported here. At present, guidelines for screening of donors of embryos that will be used for deriving hESC for therapeutic applications have not been established by regulatory bodies. Hence, we adopted the FDA's HCT/P [34] regulations for screening of organ, cell and gamete donors as well as the EC's blood banking and organ donation criteria for screening of donors. Based on these criteria, we developed a detailed series of questionnaires (CRF) of which we here include samples for potential use by future developers of clinical grade lines. It should be noted that it is questionable whether the intensive interview and history-taking process have a significant contribution to safety analysis of donated biologics. Donors are hesitant to reveal private intimate details of past relationships [40], [41]. We therefore complemented the question-naire with testing the hESC lines and samples of donor serum and reproductive tract for the absence of adventitious viruses. In addition, a reported medical history in an interview may be incomplete because of the donors' partial understanding of their medical conditions. Therefore, according to FDA guidelines, we reviewed the medical charts of the embryo donors and confirmed their medical histories.
We utilized a laser-based method to assist hatching and to dissect the ICMs to avoid the use of animal derived products. The efficiency of ICM isolation from blastocysts with well defined ICMs was 85% (23 ICMs from 27 blastocysts) and the efficiency of hESC line derivation was 11% (3 lines from 27 blastocysts). This efficiency is lower than the reported derivation efficiency of hESC lines under research-grade conditions (13%-52%) [42], [43] including in our hands while using the laser system (30%) [10]. A low efficiency (4.75%) of hESC derivation under conditions necessary to derive hESC lines of therapeutic grade was also reported by others [44]. It appears that translation to xeno-free culture system and GMP conditions is associated with a reduced derivation success.
We established a modified banking system which included an early stock, 1u and 2u CBs to address potential pitfalls related to the properties of hESCs. The stocks were developed as reservoirs of early passage cells for identity determination, and as backups for the development of additional banks in various circumstances such as the unfortunate loss of the hESCs to differentiation or contamination.
According to the widely-accepted two-tiered cell banking system, a master cell bank (MCB) is initially established, and working cell banks (WCBs) are further developed from thawed vials of MCB. hESCs poorly survive cryopreservation [45] and the stresses that are imposed on hESCs during cryopreservation and thawing, may potentially select for genetically abnormal cells which may survive unfavorable culture conditions [45], [46]. Therefore, to limit the number of freezing and thawing steps, we modified the typical two-tiered banking approach and continuously expanded the hESCs from the initial outgrowths until the freezing of the 2u CBs. During this expansion, hESCs were sampled to generate early stocks and subsequently 1u and 2u CBs. The complete records of all stocks and banks, including their origins, details of their derivation, propagation  (Table S9). The Pearson correlation coefficient was calculated from the combined data of DeltaCt values of undifferentiated and differentiated hESCs of the three lines. The results were plotted in descending order of the correlation coefficient values. Clustering analysis of genes that were differentially expressed in the three lines in the undifferentiated and differentiated state is presented in (B). An unpaired t-test for differential gene expression between undifferentiated and differentiated samples yielded 35 genes with a P-value ,0.05. Two-way hierarchical clustering with Euclidean distance was performed on the 35 genes and six samples. Green colors in the heat map represent low DeltaCt or high expression. Red colors represent high DeltaCt or low expression. doi:10.1371/journal.pone.0035325.g005 and cryopreservation locations were documented to maintain the histories of the lines.
Characterization of the lines was initiated by determining the identity of each line at the early stock phase by DNA fingerprinting and HLA analysis. It has been previously shown that mix-ups of lines [47], [48], [49] and cross-cell contamination [50], [51], both leading to false identities, are common occurrences in cell banking. Therefore, the unique genetic signature determined at the stock stage of each hESC line was confirmed by STR and HLA analysis at the 1u and 2u Cell Bank stages. The strict determination of the lines' identity is important for the essential confirmation of the identity and origin of the differentiated cells that will be developed from them for transplantation.
The specifications for the characterization of the hESC lines were developed on the basis of ICH guidelines (Q5D) [32], recommendations of the International Stem Cell Banking Initiative [52], and regulatory experts [5]. Careful development of a reproducible and validated derivation process, utilization of documented protocols and qualified, verified raw materials enabled the derivation of hESC banks that adhered to the characterization specifications (batch release criteria).
We performed G-banded karyotype analysis to determine the absence of chromosomal abnormalities. The International Stem Cell Banking Initiative [52] recommends analysis of 30 metaphase spreads for the characterization of research-grade lines. Analysis of 30 spreads was also reported for the characterization of hESC lines derived under cleanroom conditions [24]. Fifty spreads were analyzed in this report as it has been shown that the probability of detecting mosaicism patterns increases when the numbers of tested metaphase spreads increase [53]. We thus provide a higher confidence level of the exclusion of mosaicism than previously reported.
Internationally accepted criteria for the characterization of hESC lines were established on the basis of the characterization of 59 hESC lines from multiple laboratories worldwide by the ISCI study [36]. To adhere to these standards, which include gene expression profiling, and to allow the inclusion of the new lines in the ISCI Registry (www.stemcellforum.org), we analyzed gene expression utilizing the same stem cell array that was used by ISCI. The gene expression profile of the three clinical-grade hESC lines was comparable to the profile of the research-grade lines that were characterized in the ISCI study [36]. The genes that were suggested by the ISCI study to be common to undifferentiated hESCs were expressed by the pluripotent hESC of the three lines and were downregulated upon differentiation.
The first clinical trials with hESC-derived progeny have recently commenced, stressing the need for developing clinical-grade lines that can be utilized to translate hESC research into the clinic. In this paper, we present ethical, scientific and regulatory platforms for the development of xeno-free and GMP-grade hESC lines, which may serve the research community for the derivation of additional clinical-grade lines. The animal-free, regulatory-compliant hESC lines derived here will be valuable for developing regenerative therapies.

Ethics Statement
The study was approved by the Hadassah Medical Center IRB and the Israeli Ministry of Health's National Ethics Committee for Genetic Research in Humans (13.01.2004). A copy of our informed consent document is included as File S1 in our manuscript. In all cases, written informed consent was obtained by our donors prior to their acceptance into our study, and the clinical investigation was conducted according to the GCPs and Helsinki guidelines.

Fibroblast Tissue Procurement
Human fibroblasts were obtained from foreskin, umbilical cord, or fetal tissue (10-12 weeks of gestation). Potential donors were approached only after they had decided to undergo each of the surgical procedures (infant surgical circumcision, cesarean section, termination of pregnancy). Tissue donors were screened for eligibility, according to FDA's Donor Eligibility Regulations 2004 [34]. Their medical histories were examined, blood and reproductive systems (in the case of umbilical cord, and fetal tissues) were sampled for infectious and sexually-transmitted disease testing (Table S5), and the donors were interviewed to complete extensive lifestyle questionnaires to minimize chances of transmission of contaminants to eventual cell recipients. Blood samples were obtained from circumcised infant and his parents, fetal tissue donors, the cord and both parents. They were sent for infectious disease testing and serum was archived at Hadassah. All infectious disease screening was performed by FDA-approved tests at American Medical Laboratories (Herzliya, Israel).

Derivation and Cryopreservation of Human Fibroblast Feeder Cells
Preparation of fibroblasts from umbilical cord. During cesarean sections, after detachment of the placenta, a 10 cm segment of cord was sampled and transferred under sterile conditions into PBS within a prelabeled vessel, which was sealed and transferred to the GMP facility. The cord tissues were minced with 2 sterile surgical blades (#22) or a surgical blade and forceps. Preparation of fibroblasts from foreskin. Foreskins were obtained from infants up to one year of age from surgical circumcisions. The foreskins were disinfected by Chlorhexidine (dissolved in alcohol) prior to circumcision. The foreskins were transferred in PBS to the GMP-facility as described above. The sides of the tissues were cut with sterile scissors so that the tissues became flat. The top layers of the inner and outer sides were scraped with #22 surgical blades. The tissues were minced; derivation and cryopreservation of fibroblasts MCB proceeded as above.
Preparation of fibroblasts from fetal tissue. Tissues from aborted human fetuses were obtained from the operating room and transferred to the GMP-facility as above. The tissues were mechanically minced, and 2 ml of TrypLE Select (Invitrogen) were added for 10 minutes at 37uC. Fibroblast Growth Medium (FGM) was added followed by pipetting and centrifugation at 1000 rpm for 3 minutes. The cells were resuspended and 4-8610 6 cells in 10 ml Fibroblast Growth Medium were seeded in a T25 flask (Falcon). The cells were incubated in 37uC at 5% CO 2 , and when confluent, split, expanded and frozen as above.
Development of irradiated working cell banks (WCB) of cord feeders. Vials of feeder MCB were thawed and expanded as above up to passages 7-8 when the batches contained 350-475 million cells. The fibroblasts were pooled, and were sampled for sterility, mycoplasma and LAL testing as above. The fibroblasts were then irradiated within 30 ml medium (1-1.6610 6 /ml) in 50 ml tubes at 3500 rads (Gammacell, 220 Exel, MDS Nordion), repooled, recounted by Trypan Blue exclusion staining and cryopreserved using the methods described for the MCBs.

Characterization of Feeders
Doubling time of feeders. 5610 4 fibroblasts were seeded per well of 12-well tissue culture plates on day 0. On four subsequent days, the number of fibroblasts in three wells was counted, taking the average of the three. Cell counts were then plotted versus the time of cells harvest using Microsoft Excel software. The software was used to plot the optimized exponential trend line and its equation. The equation was used to calculate the time interval for population doubling.
BrdU and KI-67 staining of fibroblasts. 25,000-30,000 irradiated human fibroblasts were seeded on 0.1% gelatin-coated coverslips, incubated for 7 days at 37uC in FGM. Mitomycintreated (Bedford Labs, Bedford, OH, USA) cells were cultured as above as negative controls. For positive controls 30,000 nonirradiated human fibroblasts were seeded and cultured for two days.
FACS analysis of fibroblasts. Fibroblasts were dissociated using TrypLE Select (Invitrogen) followed by gentle trituration to a single-cell suspension, centrifuged and resuspended in FACS buffer (PBS supplemented with 0.1% Na Azide and 1% BSA). The cells were incubated with anti-CD44-FITC (1:10, IQ Products, Groningen, the Netherlands), anti-human fibroblasts (1:100, Acris, San Diego, CA, USA) and with the respective isotype control antibodies for 30 min at 4uC. Following washing, primary antibodies were detected by incubating with FITC-labeled goat anti-mouse Ig (1:100, Dako) for 30 min at 4uC. Following washing the cells were resuspended in FACS buffer supplemented with PI (1:500, Sigma 1 mg/ml). FACS analysis was performed using the FACSCalibur system (Becton Dickinson).
Karyotyping of feeders. Fibroblasts were incubated for 40 minutes with Colcimide (100 ng/ml, Beit Haemek), washed with PBS, dissociated with TrypLE Select (Invitrogen) for 4 minutes at 37uC, centrifuged at 1000 rpm for 5 minutes, resuspended in 1 ml of growth medium into which 5 ml of warm KCl and Sodium citrate hypotonic solution (Sigma) were added dropwise, incubated for 30 min at 37uC, followed by fixation at RT with 3:1 methanol/ acetic acid for 5 min. Centrifugation and fixation of the cells were repeated twice. The karyotype of 30 metaphases was analyzed using the G-banding method. Full analysis was performed on 20 metaphases spreads, chromosome counting on an additional 10 spreads, and 2 cells were photographed.
The STR analysis was performed in the Hadassah Tissue Typing Unit.
hESC doubling time when cultured on various feeders. Mitomycin-treated (Bedford Labs) 10 5 MCB fibroblasts or irradiated WCB fibroblasts were seeded into each of 16 wells of 12-well tissue culture plates coated with gelatin. After 24-96 hours, hESCs were seeded (20,000 hESCs per well) into 12 wells, leaving four wells as controls. At 4 time points within a week, the cells were harvested with Trypsin-EDTA (Invitrogen), the average number of hESCs in three wells was determined by subtracting out the number of fibroblasts counted from the control wells from the total number of cells. Cell counts were then plotted versus the time of cells harvest and the doubling time was calculated using Microsoft Excel software as described above.  [54]. For assisted hatching, zona pellucidae was perforated on day 3-4 by laser pulses (200 mW60.3 ms) from a non-contact Zilos-tkTM infrared laser system (Hamilton Thorne Biosciences, Beverly, MA, USA). Perforation diameter was 30-40 um. Assisted hatching and ICM isolation was performed according to our previously published methods [10].

Development of hESC Lines
In cases of collapse of blastocysts aged 5-6 days during assisted hatching or ICM dissection, they were incubated overnight. Another attempt of ICM isolation was then performed. In cases of collapse of blastocysts aged 7 days, or in cases when the ICM was non-detectable, whole blastocysts were plated on feeders for further derivation of hESCs.
Routine passaging of putative hESCs was performed by mechanical dissection (Stem Cell Cutting Tool, Vitrolife, Sweden) of the colonies into small clusters and plating on fresh feeders.

Cryopreservation of hESCs
Vitrification of hESCs. hESC clusters were vitrified using Cryotips (FDA-approved as medical devices, Irvine Scientific) and Irvine Scientific Vitrification Solutions (Vit-Kit Freeze Solutions, Irvine Scientific) according to the manufacturer's instructions with modification of incubation times adapted from Reubinoff et al., 2001 [45]. 5-6 small colony fragments were transferred to the Equilibration Solution (ES) for 5 minutes, then to the Vitrification Solution (VS) twice for 10 seconds and then to a final drop of VS from which they were aspirated into the straw. The straws were heat sealed and plunged into liquid nitrogen where they were stored. Thawing was performed for 3 seconds in warmed sterile water (37uC) followed by removal of hESC clusters from straws and serial immersion in Irvine Scientific's Vit Kit Thawing Solutions (Irvine Scientific) according to the manufacturer's instructions.
Slow-cooling of hESCs. 8-10 hESC colony fragments were transferred into 2.0 ml cryovials (Nunc) containing 1.0 ml CryoStor 10 (Biolife Solutions, Bothell, WA USA) freezing solution or 90% human serum (Lonza) and 10% DMSO (WAK-Chemie). The vials were slow cooled in a freezing container (Mr. Frosty, Nunc-Nalgene) to 280uC and then were stored in the vapor phase of liquid nitrogen tanks (Custom Biogenic Systems). Thawing was performed by warming at 37uC in a waterbath, transferring into hESC culture medium, centrifuging at 1000 RPM for 3 minutes, and resuspending in hESC culture medium.

Characterization of hESCs Lines
Analysis of HLA and STR fingerprinting. In order to eliminate contamination by feeders, hESCs were cultured for a minimum of three passages in a feeder-free culture system as previously described [55] with slight modifications. STR analysis was performed as described above for feeders. For HLA profiling, the DNA was extracted using the MagNA Pure LC machine (Roche Applies Sciences, Indianapolis, IN, USA). HLA-A, B, DRB1, DQB1 loci at low resolution were typed using the Lipa HLA kit (Murex Immunogenetics, Ghent, Belgium). High resolution typing of the HLA-DRB1 and DQB1 alleles used the PCR-Sequence Specific Primers method (PCR-SSP,Olerup kit, Olerup, Stockholm, Sweden), according to the manufacturer's instructions. The HLA and STR analyses were done in the Hadassah Tissue Typing Unit. For karyotype analysis, hESC colonies that were expanded by mechanical passaging on feeders, were incubated for 1 hour with 100 ng/ml Colcemid and then dissociated with trypsin (both from Biological Industries, Beit Haemek, Israel). Incubation in hypotonic solution and fixation was as described for the feeders. For the primary reference banks, full analysis was performed on 30 metaphases spreads (20 for stocks and 2u cell bank), chromosome counting on an additional 20 spreads (10 for stocks and 2u cell bank), and 2 cells were photographed for each stock/bank.
hESC doubling time. The doubling time was determined as described above for the characterization of feeders with slight modifications. Irradiated cord feeder cells were plated in 16 center well tissue culture dishes (10 5 cells/dish). Clusters of hESCs were removed mechanically and equal numbers of clusters (numbers ranged from 15-60) were plated per experiment in each of 12 center well tissue culture dishes (Falcon) leaving four dishes as controls. HESCs together with feeder cells (in triplicates) were harvested using Trypsin-EDTA (Invitrogen) and counted at four time points. The number of cord feeder cells harvested at the same time points from the control dishes was subtracted from the total number of cells. The doubling time was calculated as above.
After the 3 weeks of suspension culture, the EBs and the neurospheres were partially dissociated by mild trypsin digestion and plated on glass coverslips pretreated with 10 mg/ml poly-dlysine (Sigma) and 4 mg/ml laminin (Sigma), and cultured for an additional 5-7 days in the respective culture media as above in the absence of noggin and bFGF. They were fixed with 4% paraformaldehyde and then stained with mouse monoclonal anti-beta-tubulin isotype III (1:2000; Sigma), mouse monoclonal anti-human desmin (1:50; Dako), or mouse monoclonal antihuman muscle actin (1:50; Dako), and mouse monoclonal antihuman sox-17 (1:50; R&D).
Polyclonal goat anti-mouse immunoglobulins conjugated to FITC (1:50, Dako) or to Cy3 (1:1000, Jackson Laboratories) were used to detect the primary antibodies. Mounting and visualization was performed as above.
Analysis of pluripotency in vivo. Subject to an approval of the Institutional Ethical Committee for Care and Use of Animals, 4-5 week old NOD SCID mice (Harlan, Jerusalem, Israel) were subcutaneously inoculated utilizing a 25 G needle with 100-200 small clusters of approximately 500 cells of undifferentiated hESCs in each cluster, which had been mechanically removed from the feeders. The hESCs were injected with 1610 6 irradiated cord fibroblasts cells, in a total volume of 60 ul, together with an equal volume of Matrigel (Basement membrane matrix, Becton Dickenson Biosciences). Maximum teratoma size permitted was 15 mm. 8-12 weeks later, the resulting tumors were removed, fixed in neutral buffered 4% formalin, embedded in paraffin, and examined histologically after hematoxylin and eosin staining.
Real Time PCR: For quantitative gene expression analysis Low density TaqMan array: Human stem Cell Panel, Applied Biosystems Part Number 4385344 was used with Taqman Gene Expression Master Mix (Applied Biosystems, Foster City, CA, USA). ABI Prism 7900 HT Sequence Detection System (Applied Biosystems) was used.

Safety Testing
Feeders and hESCs were outsourced for testing at Bioreliance (Glasgow, UK) and the Hadassah Virology and Hepatology Laboratories.

Statistical Analysis
Data are presented as mean 6S.E.M. The significance of multiple comparisons was determined using one way analysis of variance (ANOVA) with Tukey-Kramer multiple comparisons test.