|Home | About | Journals | Submit | Contact Us | Français|
Induced pluripotent stem (iPS) cells can differentiate into multiple cell types, including cardiomyocytes, and have tremendous potential for drug discovery and regenerative therapies. However, it is unknown how much variability exists between differentiated lineages from independent iPS cell lines and, specifically, how similar iPS-cell derived cardiomyocytes (iPS-CM) are to embryonic stem (ES) cell-derived cardiomyocytes (ES-CM).
To investigate how much variability exists between differentiated lineages from independent iPS cell lines and how similar iPS-cell derived cardiomyocytes (iPS-CM) are to embryonic stem (ES) cell-derived cardiomyocytes (ES-CM).
We generated mouse iPS cells in which expression of NKX2-5, an early cardiac transcription factor, is marked by transgenic green fluorescent protein (GFP). Isolation of iPS- and ES-derived NKX2.5-GFP+ cardiac progenitor pools, marked by identical reporters, revealed unexpectedly high similarity in genome-wide mRNA expression levels. Furthermore, the variability between cardiac progenitors derived from independent iPS lines was minimal. The NKX2-5-GFP+ iPS cells formed cardiomyocytes by numerous induction protocols and could survive upon transplantation into the infarcted mouse heart without formation of teratomas.
Despite the line-to-line variability of gene expression in the undifferentiated state of ES and iPS cells, the variance narrows significantly in lineage-specific iPS-derived cardiac progenitors, and these progenitor cells can be isolated and used for transplantation without generation of unwanted cell types.
Induced pluripotent stem (iPS) cells may be a promising alternative to embryonic stem (ES) cells for both drug discovery and regenerative therapies, as they can differentiate into derivatives of all three germ layers1–4. However, because each iPS cell line is generated through nuclear reprogramming of a somatic cell, it is likely that the genome-wide epigenetic changes differ among individual cell lines. Indeed, numerous reports have examined the variation in gene expression among independent, undifferentiated iPS and ES cell lines1, 3, 5. However, the lack of iPS cell lines with lineage-specific reporters has precluded determination of the degree to which such heterogeneity persists as cells adopt specific lineages and epigenetically silence much of the genome.
iPS cells can be induced to form cardiomyocytes in vitro, much like ES cells6, 7. However, it remains unknown how much variability exists between cardiomyocytes derived from independent iPS cell lines or between iPS and ES cell-derived cardiomyocytes (iPS-CM and ES-CM), particularly in terms of gene expression. Analysis of iPS-CM has been hampered by the unavailability of reporter lines that would enable selection of cardiac progenitors. iPS cell lines with a cardiac-specific fluorescent reporter would enable more accurate quantification of cardiomyocyte yield by various differentiation methods, selection of cardiomyocytes for regenerative purposes, and isolation of cardiac progenitors to assess the line-to-line variability in gene expression among lineage-specific iPS cell derivatives. Furthermore, generation of iPS cell lines containing the same lineage-specific reporter as an ES cell line would facilitate isolation of comparable progenitor populations to determine the true degree of gene expression similarity between lineages derived from ES and iPS cells.
Here, we describe the generation of mouse iPS cells containing green fluorescent protein (GFP) regulated by a human NKX2-5 promoter-containing BAC8. NKX2-5 is expressed in early multipotent cardiac progenitors and differentiated cardiomyocytes8–12. Analyses of multiple independent NKX2-5-GFP iPS cell lines with ES cells containing the same reporter revealed unexpectedly high similarity in gene expression between sorted iPS and ES-derived cardiomyocyte progenitors, and lower than expected differences in gene expression among NKX2-5-GFP+ cells sorted from individual iPS cell lines.
NKX2-5 reporter mice were generated by the Gladstone Transgenic Mouse Core facility by injection of E14 mouse ES cells carrying the RP11-88L12/NKX2-5-Emerald GFP transgene (MMRRC accession 30473) into 8 cell embryos using the same method as previously described for F0 analysis8. Mice with germline transmission were maintained as heterozygotes and genotyped from tail DNA by PCR (Forward = GACGTGACCCTGTTCATCAG; Reverse = GTTTCTTGGGGACGAAAG; 367 bp product) using the REDExtract-N-Amp PCR reaction (Sigma-Aldrich, XNAT- 1KT) as recommended by the manufacturer. Cycling conditions were 94° C for 5 min; 35 cycles of 94° C for 1 min, 56° C for 1 min, 75° C for 1 min; 72° C for 15 min. The NKX2-5 EmGFP mice were deposited with the MMRRC (Accession 030440).
The tail tip was cut from an adult (8 weeks old) male NKX2-5-GFP mouse8, cleaned with ethanol, washed in PBS, chopped into small pieces, and placed in collagenase IV/trypsin 0.25% solution at 37 °C for one hour with 3 breaks for additional rounds of chopping. The resulting tissue was cultured in DMEM/F12 (+glutamine +HEPES +penicillin-streptomycin) with 10% FBS to isolate fibroblasts as described1. Induction of pluripotency by infection with four factors (Oct-4, Sox-2, Klf4, c-Myc) was performed essentially as described1, with one important modification: iPS cells were grown feeder-free in gelatinized culture plates at all stages (induction and maintenance). Our standard mES medium was used (GMEM supplemented with glutamine, sodium pyruvate, 0.1 mM MEM non-essential amino acids, 10% [v/v] fetal bovine serum [characterized, Hyclone], a 1:1000 dilution of beta-mercaptoethanol stock solution [0.35% made from Sigma M7522], and 500–1000 units per ml of leukocyte inhibitory factor [Chemicon ESG1107]). Valproic acid (2 mM, Calbiochem) was added to this medium from day 2–9 after infection13.
To induce cardiomyocyte differentiation, hanging drops or suspension embryoid bodies (EBs) were made as described, including stimulation by Wnt3a where indicated10, 14. Co-aggregates with END-2 cells15 were made as follows: mitotically inactivated END-2 cells were cultured under standard conditions16, dissociated and added to undifferentiated iPS cells in a 1:1 ratio, after which the standard protocol for suspension EBs was followed. To quantify beating, day 8 EBs were placed in 96-well plates at one EB per well and beating was scored after 3 days in a blinded fashion (n= 8 plates derived from 3 batches of undifferentiated cells per line).
NKX2-5-GFP+ cardiomyocytes differentiated from iPS cells were dissociated and replated as single cells in regular differentiation medium. Cells were incubated with 50% Fluro4 DirectTM (Invitrogen) for 20 minutes prior to recording. Control and drug-treated cells were imaged with a Carl Zeiss Axio MiCAM02 and AxioVision version 4.7 software (Carl Zeiss) and configured with a linearly scaled stage and an XL S incubation unit (Carl Zeiss) to maintain physiological atmospheric conditions at 37°C and 5% CO2 with humidification. Isoproterenol (Iso) and carbachol (CCh) were used at 1μM. Washing was performed with Hanks BBS with 0.5% serum. Each experiment was repeated at least three times, and representative tracing images are displayed from cardiomyocytes after 45 days of differentiation.
Immunocytochemistry was performed as described with cells on coverslips17 or cryosections of EBs18 (n≥3 biological samples per cell line) using the following antibodies: Oct-4 (mouse, 1:100, Santa Cruz), SSEA-1 (mouse IgM, 1:50, Chemicon), α-actinin (mouse, 1:400, Sigma), SMA (mouse, undiluted, DAKO), Nestin (mouse, 1:100, Chemicon), GFAP (rabbit, 1:100, DAKO), AFP (mouse, 1:100, R&D), GFP (rabbit, 1:500, alexa-488 conjugated, Invitrogen). Secondary antibodies were alexa-546 conjugated (Invitrogen).
Teratomas and hearts were processed for cryosectioning and staining as described19.
RNA was extracted with Trizol (Invitrogen). Reverse transcriptase quantitative PCR (qPCR) was performed using the Superscript III first-strand synthesis system (Invitrogen) followed by use of TaqMan probes on the ABI 7900HT (TaqMan, Applied Biosystems) per the manufacturer’s protocols with technical triplicates (n≥3 biological samples except for initial pluripotency screening where n=2). Optimized primers from Taqman Gene Expression Array were used. Expression levels were normalized to Gapdh expression.
For live cell FACS sorting to select pluripotent cells, dissociated cells were washed in PBS+10% FBS, blocked with CD16/32 antibody (eBioscience) for 10 minutes on ice, washed, incubated with first antibody SSEA-1 (Chemicon) 1:50 for 1 hour on ice, washed, incubated with 2nd antibody (anti-mouse IgM +fluorophore PE or alexa 488) 1:50 for 1 hour on ice, washed and resuspended in 1% BSA/PBS, followed by FACS (FACSAria, BD) into 96 well plates at one cell per well to ensure clonal growth. To quantify the number of GFP+ cells after differentiation or select those cells for injection in the mouse heart, dissociated cells from EBs were washed in PBS + 10% FBS and resuspended in 1% BSA/PBS and analyzed using the same sorter (n≥3 per condition per cell line). The same gating and settings were used for every experiment.
Embryos were created by 8 cell-stage injection of NKX2-5-EmGFP+ iPS cells as described20. E9.5 embryos were isolated and imaged without additional staining (n ≥ 8 per cell line).
For teratoma analysis, 500,000 iPS cells were injected subcutaneously and intramuscularly in 6-week-old male NOD-SCID mice (n= 3 mice (6 injections) per cell line). The animals were monitored for tumor growth by inspection and palpation. Myocardial infarction was induced in 12-week old male NOD-SCID mice (Charles River) by ligation of the left anterior descending (LAD) coronary artery under isoflurane anesthesia, as described previously19. Immediately after ligation, 400,000 iPS cells were injected in the infarcted myocardium through a 29G needle (n= 5 for unselected differentiated cells; n= 12 for selected day 8 GFP+ cells). All investigations involving experimental animals conformed to the Guide for the Care and Use of Laboratory Animals (NIH, 1996) and were approved by the Institutional Animal Care and Use Committee (UCSF).
Mouse genome-wide gene expression analysis was performed using an Affymetrix Mouse Gene 1.0 ST Array. Day 8 Sorted GFP+ differentiated iPS cells (iPS#3, iPS#25 and iPS#33), sorted GFP+ differentiated ES cells (NKX2-5-GFP E14 and mCherry-NKX2-5-GFP E14), and undifferentiated ES cells (NKX2-5-GFP E14) were compared. RNA was extracted with the PicoPure RNA isolation kit (Molecular Devices). Linear models were fitted for each gene to estimate cell-type effects and associated significance using R (Bioconductor). Pairwise contrasts were set up to identify differentially expressed genes. Moderated t-statistics and the associated p-values were calculated, as well as B-statistics (logOdds). P-values were adjusted for multiple testing by (a) controlling for false-discovery rate using the Benjamini-Hochberg method and (b) controlling for family-wise error rate using the Bonferroni correction.
For all other experiments, error bars indicate standard error of the mean (SEM).
To obtain iPS cells with a cardiomyocyte reporter label, tail tip fibroblasts were isolated from an NKX2-5-GFP transgenic mouse containing GFP in a bacterial artificial chromosome (BAC) containing the human NKX2-5 locus8. Mouse ES cells containing the same NKX2-5-GFP BAC were previously reported8. Fibroblasts from the NKX2-5-GFP reporter mice were infected with a combination of four retroviruses encoding factors known to induce pluripotency: Oct-4, Sox-2, Klf4 and c-Myc2. Three-dimensional colonies were observed from the fifth day after viral infection. By day eight, colonies were large enough to be selected and expanded in a feeder-free culture system (Fig. 1A). The cells derived from these colonies expressed SSEA-1 and Oct-4 protein as shown by immunostaining (Fig. 1B–C). To select the best reprogrammed lines, quantitative PCR (qPCR) for the pluripotency genes Nanog and Oct-4 was performed, revealing mRNA levels comparable to those of ES cells in both low (p4) and high (p20–25) passage samples (Fig. 1D).
We also tested whether the cells were able to grow clonally by selecting SSEA-1-expressing cells by FACS (Fig. 1E–F) and culturing in 96-well plates at one cell per well. Indeed, new colonies were obtained from single SSEA-1+ cells that could be passaged and expanded while maintaining expression of pluripotency markers. This sorting strategy allowed for purification of potentially heterogenous colonies of iPS cells at the single-cell level. All studies were done with clonally-derived cells. After subcutaneous and intramuscular injection into NOD-SCID mice, undifferentiated iPS cells of each selected line formed teratomas containing derivatives of all germ lines (Fig. 1G–Q).
We next evaluated whether the NKX2-5-GFP iPS cells could be differentiated in vitro, with a special interest in cardiomyocyte differentiation. iPS cells formed beating embryoid bodies (EBs) with high efficiency when induced by the hanging drop system. Spontaneous aggregation in suspension in low-attachment plates is much less labor-intensive and also produced some beating EBs, but with lower efficiency. However, when co-aggregates in suspension were made with the endoderm cell line, END-215, a high percentage of beating EBs were isolated (Fig. 1R). As expected, the beating foci within these EBs were green because of the NKX2-5-GFP reporter (Supplementary Video I, II). The iPS cell-derived EBs contained derivatives of all three germ layers (Fig. 1S–W), including α-actinin positive cardiomyocytes with typical sarcomeric patterns (Fig. 1V). Thus, we concluded that we had derived several independent NKX2-5-GFP iPS cell lines from an adult mouse and that the lines were able to retain their pluripotency under feeder-free culture conditions and responded to guided differentiation in vitro.
We quantified the percentage of GFP+ cells in EBs, representing the NKX2-5- expressing population of cardiac progenitor cells and cardiomyocytes at day 8, obtained by several differentiation methods (Fig. 2A). Consistent with the count of beating EBs, the hanging drop system yielded the highest percentage of GFP+ cells (Fig. 2B). END-2 co-aggregates were not included in this analysis, as the END-2 cells present in the mixture of differentiated cells would confound the result by altering the denominator when calculating percent cardiac progenitors. The expression of a panel of cardiac-enriched genes was higher in sorted GFP+ cells compared to sorted GFP– cells relative to undifferentiated ES or iPS cells (Fig. 2C–F). In addition, the cardiomyocytes derived from the iPS cells after 45 days of differentiation had spontaneous calcium flux and responded to β-adrenergic agonist isoproterenol (Iso) and muscarinic receptor agonist carbachol (CCh) stimulation (Fig. 2G). While the NKX2-5-GFP iPS cell lines we generated could efficiently differentiate into cardiomyocytes in vitro, we investigated their potential in vivo in embryos and in the adult. We took advantage of the recent demonstration that injection of ES cells into eight-cell embryos can result in generation of 95–100% chimeras, effectively generating embryos composed almost entirely of the injected ES cells20. Injection of NKX2-5-GFP iPS cells into eight-cell embryos revealed that the cells could give rise to viable embryos with hearts almost entirely populated by NKX2-5-GFP iPS-derived cardiomyocytes, illustrated by the green fluorescent hearts (Fig. 2H).
To determine the similarity of iPS- and ES cell-derived cardiomyocytes at the gene expression level, we took advantage of an NKX2-5-GFP ES cell line generated with exactly the same BAC reporter transgene used to create the NKX2-5-GFP mouse line used for iPS generation8. Thus, NKX2-5-GFP positive cells isolated from the iPS and ES cell lines should in principle mark very similar populations. Microarray analysis of mRNA from sorted NKX2-5-GFP+ cells after 8 days of differentiation revealed that only 195 out of the 28,853 transcripts represented were significantly different between ES and iPS cell-derived cardiac progenitors, even when applying the least stringent statistical method for defining differential expression (FDR; adjusted p-value to control for false discovery rate using Benjamini-Hochberg method with cut-off p-value set at <= 0.1). Of those, only 38 annotated genes were altered greater than 2-fold, revealing a surprisingly high degree of similarity in the differentiated progeny of ES and iPS cells (Fig. 3 and Supplementary Table I, II).
Concerns exist regarding the potential variability between iPS cell lines compared to the variability between ES lines, but the degree of variability of gene expression in lineage-specific cells derived from ES and iPS cells has not been interrogated for any cell type. This issue is particularly important for interpreting studies with multiple disease-specific iPS lines. Using the microarray data sets of NKX2-5-GFP-positive ES and iPS cells, we found that the variation between independent iPS cell-derived cardiac progenitors was no larger than that between cardiac progenitors derived from separate ES lines (Fig. 3B,C). The correlation between cell lines was extremely high (R=0.99) (Fig. 3B), suggesting that not only were the ES and iPS-derived cardiac progenitors more similar than expected, but the line-to-line variation of the lineage-specific iPS cells was also very low.
To assess the potential of the NKX2-5-GFP iPS cells in regenerative applications, day 8 EBs were dissociated and prepared for injection into acutely infarcted hearts of NOD-SCID mice. Although this approach resulted in large patches of new GFP+ myocardium being formed (Fig. 4A), clusters of SSEA-1 and Oct-4 expressing cells were also observed (Fig. 4B–C), indicating that progenitor selection was necessary to avoid tumor formation. We thus sorted for GFP+ cells by flow cytometry to isolate the cardiac progenitor pool prior to intramyocardial transplantation. While the grafts from NKX2-5-GFP+ cells were smaller than those observed after total cell injection, they did not form teratomas and were negative for markers of pluripotency (Fig. 4D). These findings suggest that the selected NKX2-5-GFP+ cells may be safe for regenerative approaches, at least with respect to tumor formation.
Here we demonstrate the feeder-free generation and maintenance of clonally-derived NKX2-5-GFP iPS cell lines that allowed identification, selection, and analyses of iPS-derived cardiac progenitors. Our finding that ES and iPS cell-derived NKX2-5-GFP+ cardiac progenitors are highly similar at the gene expression level represents the first comparison of ES and iPS- derivatives of any cell type, and was facilitated by the generation of pluripotent lines containing the same reporter gene. Despite the differences reported in gene expression between ES and iPS cells, it appears that as genomic loci are silenced or activated during lineage commitment, these differences rapidly narrow. The few genes that were differentially expressed do not appear to signify functional differences between the cells (Supplementary Table II), although we did note increased expression of some markers of less differentiated cells in ES-derived cells (e.g., Dppa2, Rex1) with a reciprocal increase in some markers of differentiated lineages in iPS-derived cells (e.g., Lhx1, Cer1, Otx2). Given the small number of altered genes, there is no statistical significance to these categories, but it will be interesting to determine if the iPS cells differentiate at a slightly accelerated pace.
The limited variability of gene expression between cardiac progenitors isolated from independent iPS cell lines also suggests that the broader heterogeneity of genome-wide reprogramming in the pluripotent state becomes less consequential as cells adopt specific lineages. This finding is promising for the study of disease-specific iPS cell lines and the use of iPS cells for drug toxicity, as the signal to noise ratio in the gene expression of lineage-specific iPS-derived cells may be significantly less than expected.
Finally, quantification of cardiomyocytes by FACS facilitated the evaluation of multiple cardiomyogenic differentiation protocols in iPS cells, revealing that iPS cells quantitatively had similar cardiomyogenic potential compared to ES cells10, 14, 15. Furthermore, the ability of iPS cells to contribute to the embryonic heart in vivo and for transplanted iPS cell-derived cardiac progenitors to form new myocardium in the infarcted mouse heart provide an important foundation for the future use of iPS cell-derived cardiac progenitors in cardiac regenerative approaches. It will be important to evaluate the functional maturation and contribution of such progenitors in vivo as technology improves to retain introduced cells within the host myocardium.
The iPS cells could provide an alternative to ES cells as a source of CPCs or cardiomyocytes for drug discovery and eventually regenerative therapies. However, it has been postulated that there may be significant variability between CPCs derived from independent iPS cell lines or between iPS cell- and ES cell-derived CPCs. Reporter lines that would enable selection of CPCs are required for analysis of iPS cell derived CPCs but were unavailable up until now. We generated mouse iPS cell lines containing the same lineage-specific reporter (NKX2-5-GFP) as ES cell lines. With this unique tool, we were able to isolate comparable progenitor populations derived from each cell type. We found unexpectedly high similarity in genome-wide mRNA expression levels. Furthermore, the variability between CPCs derived from independent iPS lines was minimal. The NKX2-5-GFP+ iPS cells formed cardiomyocytes by numerous induction protocols and could survive upon transplantation into the infarcted mouse heart without formation of teratomas. Our findings suggest that genetic variance is limited in lineage-specific iPS-derived CPCs, and that these progenitor cells can be isolated and used for transplantation without the generation of unwanted cell types. This is an important step forward towards application of iPS cells for in vitro drug testing and cardiac repair.
We are grateful for expert technical assistance from the Gladstone Stem Cell (J. Arnold, C. Schlieve), Genomics (L. Ta, C.S. Barker) and Bioinformatics (A. Holloway) Cores and the UCSF Flow Cytometry Core (S. Elmes). We thank K. Cordes for help with graphics and J. Fu for assistance with interpretation of calcium imaging. We also thank members of the Srivastava lab and the Gladstone community for helpful discussion.
Sources of Funding
L.v.L. was supported by the Interuniversity Cardiology Institute of the Netherlands fellowship grant 2007/2008. D.S. was supported by grants from NHLBI/NIH and the California Institute for Regenerative Medicine (CIRM). E.H. and L.Q. were supported by the J. David Gladstone Institutes CIRM Fellowship Program (Grant T2-00003). This work was supported by NIH/NCRR grant (C06 RR018928) to the Gladstone Institute.
D.S. is a member of the Scientific Advisory Board of iPierian Inc. and RegeneRx Pharmaceuticals. B.C. is a consultant of iPierian Inc.