Transition from Somatic Oxidative Metabolism to Pluripotent Glycolysis Supports iPSC Derivation

Intracellular metabolite fingerprinting validated the glycolytic capacity of 4F iPSC, segregating the acquired metabolomic pattern away from parental MEF and towards the pluripotent ESC standard (Figures ​(Figures2A2A and S1B). The resolved 18 intracellular metabolite panel distinguished iPSC based upon acetate, lactate, fumarate and taurine concentrations (Figure ​(Figure2B2B and S1C). 4F iPSC accumulation of acetate was similar to ESC (31.6 ± 0.9, 32.5 ± 1.0 and 29.2 ± 0.7 pmol/μg protein, n=3), and distinct from MEF (21.3 ± 0.5 pmol/μg protein, n=3 populations, p<0.05, Figure 2C). Comparably, lactate was equivalent in 4F iPSC and ESC (127 ± 3, 152 ± 5 and 140 ± 10 pmol/μg protein, n=3), yet significantly different from MEF (95 ± 5 pmol/μg protein, n=3, p<0.05; Figure 2D). Lactate efflux (5.3 ± 0.3 and 5.8 ± 0.1 nmol/μg protein/h) at a rate double that of MEF (2.8 ± 0.2 nmol/μg protein/h, n=6, p<0.05; Figure 2E) indicated functional glycolysis in 4F iPSC. Consistent with the lower efficiency of glycolytic ATP production compared to oxidative phosphorylation, the ADP/ATP ratio, an index of cellular energy turnover, was reduced in 4F iPSC and ESC compared to MEF (0.046 ± 0.002, 0.052 ± 0.001 and 0.053 ± 0.001 versus 0.070 ± 0.001, respectively, n=3, p<0.05, Figure 2F). Overall, oxygen consumption was low in 4F iPSC and ESC, compared to MEF both at baseline (0.51 ± 0.04, 0.42 ± 0.07 and 0.36 ± 0.02 versus 1.8 ± 0.13 nmol/10⁶ cells/min, n=3, p<0.05, Figure 2G) and under electron transport chain uncoupling (0.98 ± 0.10, 1.22 ± 0.09 and 0.95 ± 0.13 versus 6.51 ± 1.05 nmol/10⁶ cells/min, n=3, p<0.05, Figure 2H). 4F iPSC preserved the ability to generate mitochondrial membrane potential and demonstrated mitochondrial hyperpolarization (Figure S3A-C), indicative of reduced ATP utilization.

Treatment of MEF undergoing reprogramming with 2-deoxyglucose (2DG), a general inhibitor of glycolysis, blunted induction of the pluripotent marker alkaline phosphatase (Figure 2I), implicating a glycolytic requirement for iPSC generation. Beyond 2DG, the hexokinase 2 inhibitor 3-bromopyruvic acid (BrPA) (Ko et al., 2001) and pyruvate dehydrogenase kinase inhibitor dichloroacetate (DCA) (Stacpoole, 1989) reduced reprogramming efficiency (0.9 ± 0.3 in 2DG, 19.8 ± 1.0 in BrPA, 17.3 ± 1.5 in DCA versus 27.8 ± 3.1% SSEA1 positive cells in vehicle, n=3, p<0.05, Figure 2J) without impairing cell growth (Figure 2K). Extracellular metabolite profiles of MEF treated with 2DG or BrPA were consistent with inhibition of glycolysis and stimulation of oxidative metabolism, while DCA treatment stimulated mitochondrial function and accelerated consumption of oxidative substrates (Figure S2A-C). Glucose supplementation stimulated glycolytic flux, reduced mitochondrial function (Figure S2 D-F) and increased SSEA1 positive cell density (0.5 ± 0.3, 16.6 ± 5.3, 27.8 ± 1.1 and 17.1 ± 4.7% SSEA1 positive cells in 0, 5, 12.5 and 25 mM glucose, n=3, p<0.05). Conversely, few MEF underwent nuclear reprogramming in the absence of glucose (Figure 2L-M). Transition from somatic oxidative metabolism to pluripotent glycolysis thus contributes to nuclear reprogramming efficiency.

Glycolytic Flux Fuels Pluripotent Induction

Mitochondrial membrane potential served as a surrogate of metabolic transition during nuclear reprogramming. Prospective live cell imaging distinguished ESC-like compact clusters with high tetramethylrhodamine methyl ester (TMRM) fluorescence from transfected non-reprogrammed MEF with low mitochondrial membrane fluorescence (Figure 3A). FACS sampling of high TMRM cells at day 7 and 14 selected an emergent subpopulation within small cell clusters, consistent with mitochondrial membrane potential polarization during dedifferentiation (Figure S3A-D). At day 7, high TMRM cells had significantly elevated glycolytic gene expression (Glut1, Hxk2, Pfkm and Ldha), which met that of ESC by 2-weeks of reprogramming (Figure 3B). In contrast, at 1-week of reprogramming pluripotent gene expression (Fgf4, Nanog, Oct4 and Sox2) remained low in high TMRM cells, similar to starting MEF, with pluripotent gene induction apparent only after 2-weeks (Figure 3C). Metabolic reprogramming thus precedes pluripotent gene expression during somatic cell dedifferentiation from an oxidative to a glycolytic iPSC phenotype (Figure 3D).

Open in a separate window

Figure 3

Glycolytic engagement mobilizes pluripotent gene induction

Similar to ESC and distinct from MEF, live cell imaging of mitochondrial membrane potential identified nascent compact cell clusters with high TMRM fluorescence within 5-7 days of nuclear reprogramming (A). Compared to the low TMRM fluorescence population, high fluorescence cells had significantly elevated glycolytic gene expression (Glut1, Hxk2, Pfkm and Ldha) within 1-week of reprogramming, which by 2-weeks of reprogramming equaled ESC glycolytic gene expression (B). Of note, at week one of reprogramming pluripotent gene expression (Fgf4, Nanog, Oct4 and Sox2) remained low in high TMRM fluorescence cells, similar to the starting MEF, with pluripotent gene induction apparent during the second week (C). Shaded region represents mean ± SEM for ESC gene expression. Values are mean ± SEM, n=3. * P<0.05 versus low TMRM population. Nuclear reprogramming switches oxidative MEF into glycolytic iPSC (D). See also Figure S3.

Reprogramming-Induced Metabolic Remodeling Independent of c-Myc Induction

As c-Myc gene targets control rates of glycolysis and mitochondrial biogenesis (Dang, 2007), a cell line was derived without c-Myc (3F iPSC). Similar to 4F iPSC, NMR metabolomic footprinting and fingerprinting segregated 3F iPSC away from parental MEF (Figures 4A-B), based upon greater intracellular and extracellular accumulation of glycolytic end products, acetate (intracellular: 30.4 ± 0.8 versus 21.3 ± 0.5 pmol/μg protein; extracellular: 0.41 ± 0.03 versus 0.035 ± 0.001 nmol/μg protein/h, n=3, p<0.05, Figure 4C) and lactate (intracellular: 188 ± 11 versus 95 ± 5 pmol/μg protein; extracellular: 4.1 ± 0.1 versus 3.0 ± 0.1 nmol/μg protein/h n=3, p<0.05, Figure 4D). Consistent with accelerated glycolysis, 3F iPSC displayed reduced energy turnover compared to MEF (0.039 ± 0.005 versus 0.068 ± 0.006, n=3, p<0.05, Figure 4E). Both 3F and 4F iPSC demonstrated limited oxidative capacity and elevated lactate production (Figure 4F). Thus, nuclear reprogramming-induced metabolic remodeling is characteristic of the iPS phenotype, independent of c-Myc transduction.

Open in a separate window

Figure 4

Metabolic reprogramming is independent of c-Myc transduction and supported by glycolytic/electron transport chain proteome switch

¹H NMR cellular metabolomic fingerprints (n=6) of three stemness factor induced iPSC (3F iPSC) segregated, away from the somatic, the acquired pattern (first principal component accounts for 97% and second for 1% of total variance) (A). Glycolytic end products, acetate and lactate, were key metabolites responsible for segregation (B). Intracellular content and efflux of acetate and lactate were significantly elevated in the 3F iPSC compared to MEF (C and D) and associated with reduced energy turnover (n=6) (E). Compared to MEF, 3F iPSC had lower maximal oxidative capacity and higher lactate production similar to that of ESC, albeit not fully overlapping with 4F iPSC (n=3) (F). Proteome-wide label-free quantification segregated iPSC away from MEF towards an ESC pattern based on agglomerative clustering of z-score transformed data (n=4) due to predominant glycolytic enzyme upregulation (G). Electron transport chain complex I subunits were predominantly downregulated in pluripotent cytotypes, which clustered away from MEF (H). 2-D gel quantification and MS/MS identification (n=3) independently confirmed glycolytic upregulation (I) and complex I downregulation (J). iPSC proteomic upregulation was mapped across the glycolytic pathway (K). Values are mean ± SEM. *P<0.05 versus MEF. Proteins are abbreviated by Swiss-Prot gene name. See also Figure S4 and Tables S1 and S2.

Ultrastructure

Mitochondrial density and morphology was evaluated in 1% glutaraldehyde and 4% formaldehyde fixed cells, and examined as ultramicrotome sections on a JEOL 1200 EXII electron microscope (Perez-Terzic et al., 2007).

Metabolomic Footprinting and Fingerprinting

For footprinting of extracellular metabolites (see Supplemental Experimental Procedures), 540 μL of media collected following 24 h of culture was added to 60 μL of D₂O (Sigma) containing 5 mM sodium 3-(trimethylsilyl)propionate-2,2,3,3-d4 (TSP) (Sigma) for chemical shift reference and 81.84 mM formate (Sigma) for peak quantification reference (Turner et al., 2008). For intracellular metabolite fingerprinting, neutralized perchloric acid extracts were concentrated with a SpeedVac and suspended in 600 μl of 100 mM phosphate buffer (pH 7.0) in D₂O (Sigma) containing 0.5 mM TSP (Beckonert et al., 2007). Identities of ¹H NMR spectra peaks obtained on a Bruker Ultrashield 700 MHz spectrometer with a zgpr water suppression pre-saturation pulse were assigned by comparison to reference values for chemical shift and multiplicity, and confirmed by comparison to spectra of pure compounds in the Human Metabolome database (Govindaraju et al., 2000; Wishart et al., 2009).

Metabolites and Oxygen Consumption

Lactate efflux rate was assessed in extracellular media using a lactate assay (SUNY). Nucleotide concentrations were determined in neutralized perchloric acid extracts by high performance liquid chromatography, using a 0.1 M phosphate (pH 6.5), 0.01 M tetrabutylammoniumhydrogensulfate, and 40% methanol elution buffer (Chung et al., 2007). Oxygen consumption was assessed using an electrode system (Hansatech) on 5 million trypsinized cells suspended in DMEM. Maximal rate of uncoupled oxygen consumption was assessed by serial additions of 2,4-dinitrophenol (Sigma).

TMRM Fluorescence Sorting and Gene Expression

Mitochondrial membrane potential was assessed at day 4-14 of reprogramming by incubating with 20 nM TMRM (Anaspec) for 30 min at 37°C, and imaged with a LSM 510 Axiovert laser confocal microscope. By 1 and 2-weeks of reprogramming, single cell suspensions were incubated in TMRM and separated by a FACS Aria Cell Sorter. Gene expression was examined on an Eco RT-PCR system (Illumina).

Proteomics

Protein extracts were resolved by 2-D gel electrophoresis and 4-15% SDS-PAGE (100 and 30 μg, respectively) and silver stained (Zlatkovic-Lindor et al., 2010). For comparative analysis, SDS-PAGE lanes were excised, destained, and prepared for LC-MS/MS, as were protein species from 2-D gels identified as significantly altered by PDQuest analysis (Zlatkovic-Lindor et al., 2010). Isolated tryptic peptides were analyzed and identified by LTQ-Orbitrap mass spectrometry. Label-free quantitative comparison of SDS-PAGE protein and peptide abundance was carried out on acquired MS spectra using Rosetta Elucidator’s differential workflow, with annotation performed using PeptideTeller and ProteinTeller (Neubert et al., 2008; Lomenick et al., 2009). See Supplemental Experimental Procedures.

Authors thank Mayo Clinic Nuclear Magnetic Resonance Core and Flow Cytometry/Optical Morphology Shared Resource for assistance. This work was supported by National Institutes of Health (R01HL083439, T32HL007111, R01HL085208, R56AI074363), Canadian Institutes of Health Research, Marriott Program, and Mayo Clinic.