Somatic Cells with a Heavy Mitochondrial DNA Mutational Load Render Induced Pluripotent Stem Cells with Distinct Differentiation Defects


  • Martin Wahlestedt,

    1. Medical Faculty, Institution for Experimental Medical Science, Immunology Section, Lund Stem Cell Center, Lund University, Lund, Sweden
    Search for more papers by this author
  • Adam Ameur,

    1. Department of Immunology, Genetics and Pathology, Science for Life Laboratory Uppsala, Uppsala University, Sweden
    Search for more papers by this author
  • Roksana Moraghebi,

    1. Medical Faculty, Institution of Laboratory Medicine, Department of Molecular Medicine and Gene Therapy, Lund Stem Cell Center, Lund University, Lund, Sweden
    Search for more papers by this author
  • Gudmundur L. Norddahl,

    1. Medical Faculty, Institution for Experimental Medical Science, Immunology Section, Lund Stem Cell Center, Lund University, Lund, Sweden
    Search for more papers by this author
  • Gerd Sten,

    1. Medical Faculty, Institution for Experimental Medical Science, Immunology Section, Lund Stem Cell Center, Lund University, Lund, Sweden
    Search for more papers by this author
  • Niels-Bjarne Woods,

    1. Medical Faculty, Institution of Laboratory Medicine, Department of Molecular Medicine and Gene Therapy, Lund Stem Cell Center, Lund University, Lund, Sweden
    Search for more papers by this author
  • David Bryder

    Corresponding author
    1. Medical Faculty, Institution for Experimental Medical Science, Immunology Section, Lund Stem Cell Center, Lund University, Lund, Sweden
    • Correspondence: David Bryder Ph. D., Sölvegatan 19, BMC D14, 221 84, Lund, Sweden. Telephone: +46-46-2223951; Fax +46-46-2224218; e-mail:

    Search for more papers by this author


It has become increasingly clear that several age-associated pathologies associate with mutations in the mitochondrial genome. Experimental modeling of such events has revealed that acquisition of mitochondrial DNA (mtDNA) damage can impair respiratory function and, as a consequence, can lead to widespread decline in cellular function. This includes premature aging syndromes. By taking advantage of a mutator mouse model with an error-prone mtDNA polymerase, we here investigated the impact of an established mtDNA mutational load with regards to the generation, maintenance, and differentiation of induced pluripotent stem (iPS) cells. We demonstrate that somatic cells with a heavy mtDNA mutation burden were amenable for reprogramming into iPS cells. However, mutator iPS cells displayed delayed proliferation kinetics and harbored extensive differentiation defects. While mutator iPS cells had normal ATP levels and glycolytic activity, the induction of differentiation coincided with drastic decreases in ATP production and a hyperactive glycolysis. These data demonstrate the differential requirements of mitochondrial integrity for pluripotent stem cell self-renewal versus differentiation and highlight the relevance of assessing the mitochondrial genome when aiming to generate iPS cells with robust differentiation potential. Stem Cells 2014;32:1173–1182


When striving to generate induced pluripotent stem (iPS) cells for cell replacement therapies, the fitness of the somatic donor cell is a crucial parameter; an aspect particularly relevant if the fitness is compromised as a consequence of irreversible genetic and/or epigenetic lesions. Reduced somatic donor cell function might lead to lowered iPS generation success rates but perhaps even more importantly, to a compromised function of the iPS-derived differentiated tissue [1-4]. One critical example of decreased fitness is the physiological process of aging, which is characterized by an overall decline in cell and organ function. Many organ systems are hierarchically structured, referring to the fact that multipotent tissue-restricted stem cells maintain and support overall organ homeostasis [5]. Moreover, increasing evidence supports the more general view that the function of these tissue-specific stem cells deteriorates with time [6, 7]. For instance, in the hematopoietic system, perhaps the most well-studied organ system maintained by tissue-specific stem cells, age-associated cell intrinsic alterations in hematopoietic stem cells (HSCs) impacts the cellular composition of the immune system [7-10].

Recently, we showed that aging of the blood associates with stable, but reversible, functionally relevant epigenetic alterations in the most immature progenitor cell subsets [8]. We further found that these changes can be reversed when cells are subjected to cellular reprogramming. However, steady state aging and premature aging syndromes can also been linked to genomic DNA damage and elevated mutation rates [11-17]. In addition, investigations into the mitochondria, harboring independent circular genomes of approximately 16 kb in length, have shown an association with age and the accumulation of mutations in mtDNA in most tissues [18-21]. The mitochondria are essential for key cellular functions, including oxidative phosphorylation (OXPHOS) and apoptosis, and accumulation of mutations in mtDNA has been speculated to result in increased levels of reactive oxygen species (ROS) [18-20]. Moreover, ROS accumulation has been proposed to lead to the functional decline in aging by the continuous build-up of macromolecular damage in a “vicious cycle” [22], although the generality of this concept has more recently been challenged [23].

The functional relevance of mtDNA mutations can be studied using mtDNA “mutator” mice, that harbor an amino acid substitution (D257A) in the N-terminal proof-reading exonuclease domain II of the polymerase gamma (Polg) gene, resulting in error-prone replication of mtDNA [24, 25]. These mutator mice acquire premature aging phenotypes including weight loss, osteoporosis, anemia, and reduced life spans [24, 25]. We previously detailed the alterations in the hematopoietic system of mutator mice and found severe defects in the development of the erythroid and lymphoid lineages, while HSC homeostasis appeared to be less affected [26]. These findings highlighted the specific requirement of mitochondrial integrity in certain cellular developmental pathways, while leaving others less affected. However, although an increased mtDNA mutation load was not incompatible with somatic stem cell function per se [26], less is known about the impact of such mutations on the generation of iPS cells, their maintenance, and their differentiation capacity.

Here, we approached this issue by harnessing aged Polg mutator mice as a source for somatic cells with a high mtDNA mutation load. While such cells could be successfully reprogrammed into iPS cells and maintained in a self-renewing state, mutator iPS cells harbored severe defects when subjected to various differentiation assays including embryoid body (EB) formation, blastocyst complementation, and teratocarcinoma assays, which coincided with a failure to undergo an appropriate metabolic switch essential for appropriate differentiation. Our data support the notion that functional mitochondria are dispensable for pluripotent cell self-renewal, while being absolutely required for appropriate differentiation into more mature cell types and tissues.

Materials and Methods


Heterozygous Polgtm1Lrsn mice (kindly provided by Nils-Göran Larsson; [24]) were backcrossed with CD45.1+C57BL/6 mice for four generations. 9-month old homozygous mutator mice and C57Bl/6 control mice were used for iPS derivation. The NOD-SCID gamma common chain receptor deficient (NSG) mice used for the teratocarcinoma experiments were acquired from Jackson Laboratories (West Grove, PA, The Transgenic Core Facility at Lund University performed blastocyst injections. Animal housing and experiments were performed with consent from a local ethical committee.

Hematopoietic Stem and Progenitor Isolation and iPS Generation

Primary bone marrow (BM) cells were isolated and used for iPS generation exactly as previously described [8]. At day 14, candidate iPS cells were fluorescence-activated cell sorted based on their immunophenotype and iPS colonies were allowed to emerge. Thereafter, iPS clones were selected for further experimentation based on initial characterizations involving viral vector integration polymerase chain reaction (PCR) analysis, karyotype analysis, and quantitative real-time PCR (qRT-PCR) analysis of the expression levels of pluripotency genes as previously described [8]. Selected iPS clones were expanded and continuously maintained in embryonic stem (ES) cell culture conditions [8].

In Vitro Differentiation of iPS Cells

EBs were induced as previously described with slight modifications [8]. Single cell suspensions of iPS cells were acquired by trypsinization and the mouse embryonic fibroblasts (MEFs) were depleted by attachment. Thereafter, 20 µL hanging droplets containing 250 iPS cells were cultured in Iscove's modified Dulbecco's medium (IMDM) (Invitrogen, Carlsbad, CA, supplemented with 20% fetal calf serum (FCS), Penicillin/Streptomycin (Invitrogen), 1 mM sodium pyruvate (Invitrogen), 0.1 mM β-mercaptoethanol (Invitrogen), and GlutaMAX (Invitrogen). After 48 hours, cells were pooled and cultured in ultra low attachment Petri dishes (Corning, New York, NY, Acton, MA, for 2–14 days before analysis and further experimentation.

Quantitative PCR Experiments

To measure the expression of pluripotency factors by quantitative real-time PCR (qRT-PCR), EpCAM+SSEA-1+ iPS cells were FACS sorted, followed by RNA purification and cDNA synthesis and assayed using SYBR GreenER (Invitrogen) as previously described [8, 27]. To measure the relative abundance of telomeric repeats by qPCR, we undertook a previously adapted approach [8, 28, 29]. The amount of mitochondria in the different iPS lines was measured using qPCR [26]. In short, 5,000–50,000 cells were lysed in Tris-EDTA (TE) buffer containing Proteinase K (Roche, Basel, Switzerland, and thereafter assayed by primers directed against genomic and mitochondrial DNA. All qPCR experiments were performed on a MyiQ iCycler (Biorad, Hercules, CA,

Membrane Potential Measurement

To determine mitochondrial membrane potential, MEF-depleted single cell suspensions of iPS cells were cultured for 55 minutes in ES medium at 37°C. After 30 minutes of culture, 5 μM Verapamil (Sigma-Aldrich, St. Louis, MO, was added to the samples and after 40 minutes the lipophilic cationic dye JC-1 (Cayman Chemical Company, Ann Arbor, MI, was added to the cultures at a 1/400 dilution according to manufacturer's recommendations, followed by FACS analysis.

ROS Determination

ROS levels were measured in both MEF-depleted iPS cell suspensions and single cell suspensions of EB cultures (prepared by trypsinization) by incubation in 10 μM dichloro-dihydro-fluorescein diacetate (Invitrogen) for 30 minutes at 37°C and analyzed by FACS immediately after staining.

ATP Assay

To quantify cellular ATP, we used the ATP bioluminescent somatic cell assay kit (Sigma-Aldrich) following the manufacturer's recommendations. In brief, 10,000 MEF-depleted iPS cells and 1,000 EB cells suspended in phosphate buffered saline (PBS) and 5.71 × 10−11 and 5.71 × 10−12 moles of ATP standard was added to standard samples, respectively. Next, the cells were incubated with the ATP somatic cell releasing agent (Sigma-Aldrich), and luminescence was measured immediately using a Glomax Illuminator (Promega, Madison, WI,

l-Lactate Measurement

The amount of produced lactate was measured in the culture medium of both undifferentiated and differentiated iPS cells using the Fluorimetric Lactate Assay Kit (Biovision, Mountain View, CA, For measurement of iPS cultures, the medium was changed 2 hours prior to harvest while the measurement in EB cultures was performed following 48 hours of culture. Media samples were run alongside an l-Lactate standard and were allowed to react with an enzyme mix and probe for 30 minutes according to the manufacturer's recommendations. The fluorescence was measured using a SpectraMax Gemini XS fluorescence plate reader (Molecular Devices, Sunnyvale, CA, Lactate abundance was normalized to MEF only samples (for iPS cultures) or medium only (EB cultures) and finally normalized to cell numbers.

Teratocarcinoma Formation Assay

The ability of each iPS clone to form teratocarcinomas was evaluated by the injection of 2 × 106 MEF-depleted iPS cells suspended in 250 μL PBS subcutaneously to immunodeficient NSG mice. After 4 weeks, the resulting tumors were surgically dissected from the euthanized mice. Samples were measured, fixed in PBS containing 4% formaldehyde, and embedded in paraffin. Sections were stained with hematoxylin and eosin.


Undifferentiated iPS and cultures and EB cultures were visualized and photographed using an AMG Evos XL Core microscope (Life Technologies, Carlsbad, CA, For electron microscopy, MEF-depleted iPS cells were fixed for 1 hour at RT in 4% formaldehyde and glutaraldehyde and embedded in agarose. Thereafter, the embedded cells were sectioned and mitochondria were visualized using a FEI Tecnai Spirit BioTWIN transmission electron microscope (FEI, Hillsboro, OR,

MtDNA Mutation Frequency Analysis

50,000 FACS sorted LineageKit+ cells from wild type (WT) or mutator mice and iPS cells derived from such cells were lysed in TE buffer containing Proteinase K (Roche). Next, three amplicons of lengths 308, 321, and 339 bp were prepared by PCR amplification of the resulting lysates using the Phusion Hotstart Flex DNA polymerase (New England Biolabs, Ipswich, MA, The PCR reactions were run using: 30 seconds at 98°C followed by 30 cycles of 10 seconds at 98°C, 20 seconds at 64°C, 15 seconds at 72°C, and a final 10 minute incubation at 72°C. Following PCR amplification, each sample was purified using the PCR Purification Kit (Qiagen, Venlo, Netherlands, Primer sequences used to generate the amplicons were: 308bp_Fw: TTACTTCTGCCAGCCTGACC; 308bp_Rev: GATGCTCGGATCCATAGGAA; 321bp_Fw: TCCAATTCTCCAGGCATACG; 321bp_Rev: GCCGTTCATGCTAGTCCCTA; 339bp_Fw: GCGGGAGTACCACCATACAT; 339bp_Rev: GGCTGAATTCCAGGCCTACT.

Equimolar amounts of the different amplicons were pooled for each sample. Barcoded sequencing libraries were constructed using AB Library Builder (Life Technologies). Sequencing was performed in three separate runs with the Ion 316 chip on the Ion Torrent PGM instrument (Life Technologies) using the 400-bp chemistry. To eliminate effects of potential biases between runs, WT and mutator samples were sequenced together on the same Ion 316 chips. Reads were aligned to the C57Bl/6 reference sequence for the mtDNA amplicons (Genbank accession number DQ106412.1) using the Torrent Suite v4.0 alignment plugin, resulting in a sequence coverage of more than 50,000× for each of the samples. At every position of the amplicon sequences, the number of observed single nucleotide polymorphisms (SNPs) and indels were extracted using SAMtools [30] and custom scripts. The main source of sequencing errors in the Ion Torrent technology are falsely called indels, and therefore indels were discarded from the analyses. Instead, the mutation frequency at each position of the three amplicons was estimated by dividing the number of SNP observations by the total coverage.


Data analysis was performed using Microsoft Excel (Microsoft, Redmond, WA, and Graphpad Prism (GraphPad Software, Inc, La Jolla, CA, FACS analysis was performed using the Flowjo software (TreeStar, Ashland, OR, Significance values were calculated by Student's two-tailed t test and a p value of <.05, indicated as **, was used to determine significance.


Somatic Progenitor Cells with Severe mtDNA Mutational Load can be Successfully Reprogrammed to Cells Bearing Several iPS-like Characteristics

To investigate the consequences of critical amounts of mtDNA mutations for somatic cell reprogramming, we took advantage of a genetic mouse model with a proofreading-defective Polg gene [24], hereafter referred to as mutator mice. At 9 months of age, these mice show widespread tissue deterioration that associates with a severe mtDNA mutational burden affecting multiple organs, including the hematopoietic system [24, 26]. From such mice, we purified lineagekit+ hematopoietic stem and progenitor cells and transduced them with retroviruses carrying the transcription factors Oct4, Sox2, Klf4, and c-Myc to generate iPS cells. iPS cells derived from mutator cells formed at similar efficiency as WT control cells (3 clones out of 50,000 and 8 clones out of 100,000 Linc-Kit+ starter cells for WT and mutator cells, respectively) and were morphologically similar to WT iPS cells, although they tended to yield smaller colonies (Fig. 1A). The resulting iPS cells displayed immunophenotypic expression of the pluripotency markers SSEA-1 and EpCAM (Fig. 1B), and expression of the key pluripotency-associated transcription factors Oct4, Sox2, Klf4, c-Myc, and Nanog, at levels similar to that of WT iPS cells (Fig. 1C). We and others have previously shown that the reprogramming process coincides with a striking elongation of the telomeres compared to the original parental cells [8, 31, 32]. When investigating this feature in the mutator iPS cells, we found these to have undergone a telomere elongation similar to that of WT iPS cells (Fig. 1D). Finally, we approached the issue of whether the reprogramming process might select for cells and/or mitochondria carrying a low mtDNA mutational burden. To investigate this, we subjected both WT and mutator lineagekit+ parental cells and the derived iPS lines for deep sequencing. As expected, mutator lineagekit+ harbored heavily mutated mitochondria (14.9-fold higher median mutational frequency per base compared to the WT) (Fig. 1E). All mutator iPS clones retained a substantial mtDNA mutational burden following reprogramming compared to the WT clones (5–10-fold higher median mutational burden per base compared to WT iPS cells) (Fig. 1E). Taken together, these data demonstrate that somatic cells with a severe mtDNA mutational load are amenable for reprogramming into iPS cells and that mtDNA mutations persist in derived iPS cells, as assessed by both morphological and molecular properties.

Figure 1.

Reprogramming somatic hematopoietic stem and progenitor cells (HSPCs) with an extensive mtDNA mutational into induced pluripotent stem (iPS) cells. Bone marrow-derived lineage-negative c-Kit-positive HSPCs of mtDNA mutator or WT origin were subjected to cellular reprogramming. (A): Bright field microscopy images showing the generated mutator and WT iPS lines. Scale bars = 200 μm. (B): Representative fluorescence activated cell sorting (FACS) plots of mutator and WT iPS lines showing the simultaneous expression of EpCAM and SSEA-1. (C): Relative expression of the pluripotency genes Oct4, Sox2, Klf4, c-Myc, and Nanog in mutator and WT iPS lines (three replicates per cell line). (D): Telomere length analysis of FACS sorted EpCAM+SSEA-1+ mutator and WT iPS presented as relative increase over the parental cells (three replicates/cell line). (E): Median mtDNA mutational frequency in WT and mutator parental and iPS cells determined by deep sequencing of three selected regions of mtDNA (three technical replicates each for parental cells and one replicate for each iPS line investigated). Error bars denote mean ± SEM and ** indicates a p value of <.05. Abbreviations: mtDNA; mitochondrial DNA; WT; wild type.

Mutator iPS Cells have Decreased Proliferative Abilities and Inadequate Differentiation Potential

Since mutator iPS cells generated smaller colonies than their WT counterparts (Fig. 1), we investigated their proliferation kinetics in more detail. Equal cell numbers of each iPS line were cultured in ES/iPS self-renewal conditions and proliferation thereafter assessed at days 2–5. These experiments showed that indeed all three mutator iPS lines had drastically lower proliferation kinetics compared to WT control iPS lines (Fig. 2A).

Figure 2.

Mutator induced pluripotent stem (iPS) cells have reduced proliferation kinetics and display major differentiation defects. (A): 100,000 mutator and WT iPS cells were seeded in standard culture conditions and subsequently counted to determine the number of cells at days 2, 3, 4, and 5 after seeding. (B): The differentiation potential of the mutator iPS cells was determined in vitro using an EB differentiation protocol. Shown are bright field microscopy images of the morphology and size of representative mutator and WT EBs cultured for 4 days. Scale bars = 100 μm. (C): EB formation potential of mutator and WT iPS cells was quantified by the dissociation of day 4 EB cultures followed by cell counting (three replicates per cell type). (D): Teratocarcinoma formation of WT and mutator iPS lines. Shown are representative images of tumors originating from the different iPS lines. (E): The size of the mutator and WT derived tumors were estimated by length measurements (three tumors per cell line). (F): Differentiation potential in vivo by the blastocyst complementation assay. Shown is the amount of blastocysts injected, percent pups of implanted blastocysts, and born chimeric animals. Error bars denote mean ± SEM and ** indicates a p value of <.05. Abbreviations: EB; embryoid body; WT, wild type.

Pluripotent cells are not only defined by their ability to self-renew extensively in culture but also by their vast differentiation capacity. Therefore, we next investigated whether these properties were altered as a consequence of a higher mtDNA mutation load. We initiated EB formation, an in vitro surrogate for embryonic differentiation composed of cells from all three germ layers [33]. We found that EBs generated from WT iPS cells grew large in size after 12–14 days of culture, while EBs generated from mutator iPS cells failed to produce cells with this feature despite seeding with the same cell number. In fact, the mutator EBs ceased to increase in size after 4–5 days in culture and was at this time point significantly smaller than their WT counterparts (Fig. 2B, 2C). Moreover, only the WT EBs contained identifiable contracting cardiomyocytes (data not shown). These results suggested that a severe mtDNA mutation load severely compromise the in vitro differentiation capacity of iPS cells.

We next investigated the pluripotency/differentiation capacity of mutator iPS cells in vivo by investigating their ability to form teratocarcinomas. For this purpose, WT or mutator iPS cells were injected subcutaneously into immunocompromised mice and the resulting tumors were harvested 4 weeks later for assessment. All recipient mice injected with WT iPS cells had developed large tumors at this time, while only two out of three mutator iPS lines gave rise to tumors (Fig. 2D, 2E). Furthermore, all formed mutator iPS cell derived tumors were at least twofold smaller than their WT counterparts (Fig. 2D, 2E). Interestingly, all formed tumors, regardless of origin, contained tissues derived from all three germ layers, indicating that mtDNA mutations had a more general negative influence on differentiation, rather than affecting particular tissue types (Supporting Information Fig. 1A).

Finally, we addressed the differentiation capacity of mutator iPS cells in vivo by blastocyst complementation—the most stringent assay to assess iPS cell pluripotency. Functional ES or iPS cells should in this setting be able to give rise to chimerism in all different organs and tissues of the developing mouse. Out of 121 blastocysts injected with mutator iPS cells, we found that not a single chimeric mouse was born (Fig. 2F). This was strikingly different to WT iPS of both a young and aged somatic origin, which effectively produced chimeric mice [8]. Collectively, these experiments demonstrated that an extensive mtDNA mutation load strongly compromises the differentiation potential of pluripotent stem cells.

Mutator iPS Cells have Metabolic Features Comparable to WT iPS Cells

Mutator mice progressively develop several premature aging-associated phenotypes and display a general decline in tissue function [24]. While several different theories have been raised concerning how mtDNA mutations may lead to such alteration in cellular function, studies of the mutator strain have provided evidence that widespread accumulation of mtDNA mutations can lead to an overall decreased respiratory function [34]. Therefore, we first investigated the mutator iPS cells from a metabolic perspective.

Initially, we hypothesized that the heavy burden of mtDNA mutations may result in defects in mitochondria replication or assembly. However, we could find little support for this when quantifying cellular mtDNA content in mutator iPS cells (Fig. 3A). Next, we investigated the morphology of the mitochondria in the mutator iPS cells using transmission electron microscopy. This revealed that both WT and mutator iPS cells contained mitochondria of a large, globular, and immature shape with loosely organized cristae, consistent with previous reports on mitochondrial morphology in pluripotent cells and in stark contrast to mitochondria present in differentiated cells (Fig. 3B) [35-40]. We could not observe any obvious alterations in mitochondrial structure between mutator and WT iPS cells.

Figure 3.

Mutator iPS cells display normal mitochondrial and metabolic features but have decreased ROS levels. (A): The mitochondrial content in fluorescence-activated cell sorted EpCAM+SSEA-1+ mutator and WT iPS cells was determined by quantitative polymerase chain reaction (three replicates per cell line). (B): Representative transmission electron microscopy images of the large globular and immature mitochondria present in mutator and WT iPS cells as opposed to mitochondria with more mature morphology present in MEFs. The mitochondria are depicted by arrowheads, the nucleus by “N.” Scale bars = 500 nm. (C): The ROS content in mutator iPS relative to WT iPS cells was determined by dichloro-dihydro-fluorescein diacetate, staining (three replicates per cell line). (D): Cellular ATP levels in WT and mutator iPS lines (three replicates per cell line). (E): To measure glycolytic activity in mutator and WT iPS cells, the amount of l-lactate released to the culture medium was determined and related to cell numbers (three replicates per cell line). Error bars denote mean ± SEM and ** indicates a p value of <.05. Abbreviations: MEF; mouse embryonic fibroblast; ROS; reactive oxygen species; WT, wild type.

The acquisition of mtDNA damage has been proposed to lead to alterations in ROS production, which could result in cellular damage. When investigating intracellular ROS levels in the mutator iPS lines, we found these to have an approximately fourfold reduction in ROS levels compared to WT iPS cells (Fig. 3C). Since the mitochondria are the main source of intracellular ROS production, this finding suggested mutator iPS cells to have a lower mitochondrial activity. However, mutator iPS cells neither display any obvious alterations in ATP production nor did they show an altered mitochondrial membrane potential (Fig. 3D; Supporting Information Fig. 2). Previous studies on ES and iPS cells have shown that such pluripotent cells largely depend on glycolysis to uphold their energy demands [35-38, 41, 42]. Therefore, we next investigated anaerobic respiration in our iPS lines by quantifying extracellular lactate, a glycolytic byproduct. These measurements revealed a near equal production of lactate in the mutator and WT iPS cells (Fig. 3E), suggesting that both the self-renewing mutator and WT iPS cells depend mainly on glycolysis to meet their energy demands.

Mutator iPS Cells Fail to Differentiate due to Impaired Metabolic Switching

Mutator iPS cells appeared similar to WT iPS cells in their undifferentiated state but displayed major differentiation defects (Figs. 1, 2). Furthermore, while the “ground” iPS state is heavily dependent on glycolysis for energy production, pluripotent cells switch to OXPHOS during differentiation [41, 43, 44]. This prompted us to investigate whether the differentiation defects of the mutator iPS cells resulted from a failure to metabolically switch from glycolysis to OXPHOS during differentiation. We initiated EB cultures and used the differentiated cells for further experimentation after 4 days of culture. Again, we could not observe any differences in mitochondrial copy numbers compared to WT iPS cells (Fig. 4A). Furthermore, and similar to the undifferentiated setting, mutator-derived EB cells showed a decrease in cellular ROS content (approximately twofold) (Fig. 4B). However, as opposed to the findings in undifferentiated mutator iPS cells, their differentiated progeny contained approximately twofold less ATP (Fig. 4C). Moreover, the differentiated mutator cells produced strikingly (fourfold) more lactate than their WT counterparts (Fig. 4D). From these data, we conclude that mutator iPS cells can be derived and maintained in an undifferentiated state because of the glycolytic nature of pluripotent cells. However, when a metabolic switch toward OXPHOS is needed upon differentiation, the consequence of a heavy mtDNA mutational burden takes its toll by limiting aerobic respiratory function and impairing differentiation.

Figure 4.

Differentiating mutator iPS cells fail to produce sufficient ATP despite a hyperactive glycolysis. (A): Mutator and WT-derived EBs differentiated for 4 days were dissociated and subjected to mitochondrial copy number determination by qPCR (three replicates per cell line). (B): ROS levels in day 4 mutator and WT EB cells evaluated by DCFH-DA staining (three replicates per condition). (C): ATP levels in WT and mutator iPS cells in day 4 dissociated EB cells. (D): l-Lactate levels in the culture medium of WT and mutator EB cells (three replicates per condition). Error bars denote mean ± SEM and ** indicates a p value of <.05. Abbreviations: ROS, reactive oxygen species; WT, wild type.


Early embryonic development associates with a burst in mitochondrial numbers and therefore also in the replication of mtDNA [44-48]. As a consequence, mutations to the mitochondrial genome may arise, with the mtDNA replication being a particularly error-prone process [49]. MtDNA mutations have been proposed to contribute to both aging and disease in humans, and experimental evidence using genetic mouse models suggest that lesions to the mtDNA may lead to widespread degeneration of tissue function and cause a variety of premature aging syndromes [19, 24, 25, 50].

iPS cells have arisen as both promising and robust tools for the study of developmental regulation, disease, and as potential future therapeutic tools. However, little is known about the impact on mtDNA mutations with regards to iPS generation and function. In this work, we speculated whether an already established mtDNA mutational load could be a bottleneck for iPS generation. To investigate this, we took advantage of a mutator mouse model harboring a proofreading defective allele of the mtDNA polymerase Polg and found that somatic BM progenitors with a high load of mtDNA mutations were amenable for reprogramming into iPS cells with similar efficacy as WT iPS cells. Somewhat surprisingly to us, these mutator iPS cells displayed similar morphological, immunophenotypic, and molecular features as WT iPS cells, suggesting that not only the generation but also the pluripotent state appears less dependent on appropriate mitochondrial function. It remained a formal possibility that the reprogramming process may select for cells/mitochondria with a low or negligible frequency of mutations. However, when directly investigating this issue using deep sequencing, we found that both mutator somatic and iPS cells indeed harbored a substantial mtDNA mutational burden as compared to WT cells. Although the mutator iPS cells presented with a somewhat lower mtDNA mutational frequency than their somatic counterparts, it is likely that this is attributed to the clonal origin of each iPS line because each iPS line is generated from a single cell of a population of mutator somatic cells with a high diversity of mutated mitochondria. It can thus be envisioned that any single cell in this somatic population has a lower mtDNA burden than the combined population of cells.

Both ES and iPS cells have been proposed to be highly dependent on glycolysis to meet their energy demands [35-38, 41, 42]. Previous reports have further proposed that the differentiation of ES and iPS cells coincides with a distinct metabolic shift from anaerobic glycolysis to aerobic OXPHOS [36, 41, 43, 44], the latter which critically depend on appropriate mitochondrial function. When studying the differentiation potential of the mutator iPS cells in vivo and in vitro, we indeed consistently found that the fitness of the mutator iPS cells decreased when an increased mitochondrial activity presumably was required. This was evidenced by severe defects in EB formation, teratocarcinoma formation, and the ability to complement blastocysts.

We speculated that the discrepancy of maintenance of the undifferentiated versus the differentiated state in mutator iPS cells depended on their failure metabolically “switch” toward OXPHOS. It seems likely that the severe OXPHOS deficiencies of mutator iPS cells lead to differentiation impairments and that the hyperactive glycolytic activity during differentiation is an attempt to compensate for these shortcomings (Fig. 5). In support of this view, increased lactate levels were recently found increased upon aging in the brains of both WT and mutator mice [51], further supporting this to be a widespread compensatory mechanism upon compromised OXHPOS.

Figure 5.

The consequences of a high mitochondrial mutation burden for iPS generation, self-renewal and differentiation. Somatic cells with a high load of mtDNA mutations can be successfully reprogrammed into iPS cells by retroviral expression of Oct4, Sox2, Klf4, c-Myc. The resulting mutator iPS cells display normal morphology and surface marker expression. Moreover, these iPS cells express pluripotency factors and could undergo extensive telomere elongation similar to WT iPS cells. However, the mutator iPS cells are characterized by decreased proliferation kinetics, decreased ROS levels, and striking differentiation defects both in vitro and in vivo. When allowed to differentiate, mutator iPS cells display decreased ROS, decreased ATP and an increased glycolytic activity compared to WT iPS cells. Abbreviations: mtDNA, mitochondrial DNA; iPS, induced pluripotent stem cells; ROS, reactive oxygen species.

Previous studies have tried to detail how the random mtDNA point mutations arising in the mutator mice lead to the degeneration of tissue function and premature aging [34, 52]. From such work it has been concluded that the synthesis rate of mtDNA appear normal, ROS levels are unaltered, and finally that mutator mitochondria are characterized by a decreased stability of respiratory chain complexes [34, 52]. A main consensus from previous work, compatible with the work presented here, therefore suggests a sequence of events where random point mutations in the mitochondrial genome gradually accumulate and ultimately lead to a lowered respiratory function. Moreover, while the mutator iPS mitochondria appeared to have membrane potentials similar to those of WT iPS cells, our experiments revealed a decrease in intracellular ROS levels both in both undifferentiated and differentiated mutator iPS cells. We interpret this to indicate that mutator iPS cells produce less ROS due to an overall decreased mitochondrial activity, in line with previous data on other cell types from the mutator model [34, 52].

While the induction of pluripotency can ameliorate epigenetic and transcriptional parameters of the somatic parental cells, full mitochondrial rejuvenation did not associate with the reprogramming process. This finding is somewhat in contrast to a few previous studies. Suhr et al. [39] found an increased mitochondrial function of iPS-derived progeny compared to the parental fibroblast cells. Moreover, iPS cells engineered from a Pearson marrow pancreas syndrome patient, a congenital multisystem disorder caused by large deletions in mtDNA associated with BM failure, suggested that extensive culturing of iPS cells with heteroplasmic mtDNA lesions can select for iPS clones without disease-causing mitochondria [53]. A similar phenomenon was also recently observed in two independent studies of iPS cells derived from a mitochondrial encephalomyoathy patients [54, 55]. Mechanistically, mitochondrial gain-of-function may occur by selection or intercellular transfer of mitochondria [56-58], and the differential findings reported here might be explained by the fact that the genetic defect in our model system resides in nuclear DNA and would likely (at least eventually) cause deterioration of an acquired improvement in mitochondrial function, regardless of mechanism. It could however also be that technical issues with regards to evaluating the differentiation potential of human iPS cells cannot fully reveal the degree of mitochondrial restoration. In support of this, we could in this work generate teratocarcinomas from mutator iPS cells—currently the gold standard for evaluating human iPS cells—while mutator iPS cells completely failed to contribute in the more stringent blastocyst complementation assay.

While extensive amounts of mtDNA mutations in the mutator iPS cells described here may well exceed any occurring physiological levels, a more recent study demonstrated that also less abundant mtDNA mutational loads were capable of causing severe defects in brain development [59]. We therefore conclude that the integrity of the mitochondrial genome is an important parameter when assessing iPS cells, regardless if the end goal is to generate iPS lines for basic science or for potential clinical use. This is particularly relevant given that, as evidenced here, compromised mtDNA integrity may not present as a major obstacle in the generation and maintenance phases of iPS cells, while resulting in compromised or utterly failed differentiation capabilities.


In this work, we reveal that somatic cells with a heavy load of mtDNA mutations are amenable for reprogramming into cells with iPS-like characteristics and that mutations persist throughout the reprogramming process. The generated mutator iPS cells displayed major differentiation defects, which associated with a hyperactive glycolysis. Our data highlight a discrepancy between self-renewal and differentiation in terms of mtDNA mutational burdens and mitochondrial activity and reinforces the value to screen for mtDNA mutations when aiming to generate and functionally assess iPS cells.


We thank Inger Jonasson and Susana Häggqvist at the Uppsala node of the National Genomics Infrastructure (NGI) for their assistance with the Ion Torrent sequencing. This work was generously supported by project grants to D.B. from the Swedish Cancer Society, the Swedish Medical Research Council (project grants and consortia grants Hemato-Linné and Stemtherapy), the Swedish Pediatric Leukemia Foundation, the Wenner-Gren foundation, and an AFA grant for research on regenerative medicine.

Author Contributions

M.W.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing; A.A., G.S., and R.M.: collection and assembly of data and data analysis and interpretation; G.L.N.: conception and design and data analysis and interpretation; N.B.W.: financial support, collection and assembly of data, and data analysis and interpretation, D.B.: conception and design, financial support, data analysis and interpretation, and manuscript writing.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.