Mild reduction in the activity of the α-ketoglutarate dehydrogenase complex elevates GABA shunt and glycolysis

Abstract

Diminished energy metabolism and reduced activity of brain α-ketoglutarate dehydrogenase complex (KGDHC) occur in a number of neurodegenerative diseases. The relation between diminished KGDHC activity and altered energy metabolism is unknown. The present study tested whether a reduction in the KGDHC activity would alter cellular metabolism by comparing metabolism of [U-¹³C]glucose in a human embryonic kidney (HEK) cell line (E2k100) to one in which the KGDHC activity was about 70% of control (E2k67). After a two hours incubation of the cells with [U-¹³C]glucose, the E2k67 cells showed a greater increase in ¹³C labeling of alanine compared to the E2k100 cells. This suggested an increase in glycolysis. Furthermore, an increase in labeled lactate after 12 hr incubation supported the suggestion of an increased glycolysis in the E2k67 cells. Increased GABA shunt in the E2k67 cells was indicated by increased ¹³C labeling of GABA at both 2 and 12 hr compared to the control cells. GABA concentration as determined by HPLC was also increased in the E2k67 cells compared to the control cells. However, the GABA shunt was not sufficient to normalize metabolism in the E2k67 cells compared to control at 2 or 12 hours. However, by 24 hr metabolism had normalized (ie. labeling was similar in E2k67 and E2k100). Thus, the data are consistent with an enhanced glycolysis and GABA shunt in response to a mild reduction in KGDHC. Our findings indicate that a mild change in KGDHC activity can lead to large changes in metabolism. The changes may maintain normal energy metabolism but make the cells more vulnerable to perturbations such as occur with oxidants.

Introduction

The activity of the thiamine dependent enzyme complex α-ketoglutarate dehydrogenase (KGDHC) declines in brain in numerous age-related neurodegenerative diseases (Albers et al. 2000; Butterworth et al. 1993; Gibson et al. 1988; Mizuno et al. 1994, Mastrogiacomo et al. 1993). KGDHC consists of three components: E1k (EC1.2.4.2), E2k (EC2.3.1.6.1) and E3 (EC1.6.4.3). Changes in brain KGDHC activity are particularly well-documented in Alzheimer's disease (AD). In AD, the reduction in brain KGDHC is highly correlated (r=0.77) to the decline in cognition before the patients died (Bubber et al. 2005). The reported reductions in brain KGDHC activity range from 25 to 75% (Mastrogiacomo et al. 1993, Bubber et al. 2005). Thus, a better understanding of the consequences of a partial reduction in KGDHC is required.

Results in animal models suggest that a decline in KGDHC activity makes the brain more vulnerable to other insults. Mice with a genetically induced reduction in the E3 component have a 50% reduction in KGDHC activity (Johnson et al. 1997). The E3 deficient mice exhibit enhanced oxidative stress in the brain and are more vulnerable than controls to a variety of neurotoxins including 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), malonate and 3-nitroproprionic acid (Klivenyi et al. 2004). Furthermore, reducing KGDHC by diminishing dietary thiamine exacerbates plaque formation in plaque competent mice (Karuppagounder et al. 2008). The results are all consistent with the suggestion that the decline in KGDHC leaves the brain vulnerable to other insults and suggests that even minor alterations in KGDHC will alter brain metabolism. The current experiments directly test that possibility. Cellular models should allow us to determine a mechanism for the increased vulnerability of brains with a diminished KGDHC, and suggest approaches to reversing the deficits. Previous studies used inhibitors to demonstrate the consequences of a reduction of KGDHC (Huang et al. 2003). In initial studies, cellular KGDHC was inhibited with α-keto-β-methyl valerate (KMV) (Huang et al. 2003). Sensitive measures of mitochondrial energy metabolism (i.e., the mitochondrial membrane potential) were less sensitive to impaired KGDHC than the release of cytochrome c and subsequent activation of caspases (Huang et al. 2003). Since KMV has a low affinity for KGDHC, its specificity is suspect. Thus, the effect of specific KGDHC inhibitors [phosphonoethyl ester of succinyl phosphonate (PESP) and the carboxy ethyl ester of succinyl phosphonate (CESP)] on [1-¹³C] glucose and [U-¹³C]glutamate metabolism in intact cerebellar granule neurons was investigated (Santos et al. 2006). Overall, the findings also suggest that some carbon derived from α-ketoglutarate may bypass the block in the TCA cycle at KGDHC by means of the GABA shunt. Furthermore, reduced KGDHC promoted conversion of valine to succinate (Santos et al. 2006).

An alternative to the use of inhibitors is to genetically manipulate components of KGDHC. The HEK293 line expressing E2k antisense RNA (E2k67), in which E2k protein content is reduced to 46% of a sense control line (E2k-100), was developed in our previous study (Shi et al. 2005). Characterization of these cells (the same cell lines in the current studies) showed a normal growth and suggests that metabolism may shift to account for the deficit but makes the cells more vulnerable to other insults such as oxidative stress. E2k deficient cells grow normally. The reduction in free radicals after administration of H₂O₂ is slower in E2k67 cells than in controls, but the difference is not statistically significant (Shi et al. 2005). This suggests that the cells modified their metabolism and that this was adequate under normal conditions but may not in response to oxidant stimulation. The current studies with these cells directly tested whether the cells adjust their metabolism to handle the deficit in KGDHC.

Methods

Cell culture and [U -¹³C]glucose labeling

The E2k-67 cell line with about 30% reduced KGDHC activity and the control line (E2k-100) were generated previously (Shi et al. 2005). They were grown in Dulbecco's Modified Eagle Medium (DMEM) (high glucose) supplemented with 10% (v/v) fetal bovine serum (FBS), penicillin (50 units/ml), streptomycin (50 μg/ml) and hygromycin (100 μg/ml) and incubated at 37°C in a humidified atmosphere of 5% CO₂ in air. We measured KGDHC activity each time before the labeling experiments and the reduction in the KGDHC activity is consistent among batches.

Cells grown in 100-mm cell culture dishes (~ 90% confluent) were trypsinized, counted and seeded into a total of twelve 14 cm dishes per cell line at a density of 1×10⁶ cells/dish, and incubated at 37°C in a humidified atmosphere of 5% CO₂ in air. At day 5, medium was carefully removed and the cells were washed once with phosphate buffered saline (PBS) buffer (0.1 g/l CaCl₂, 0.2 g/l KCl, 0.2 g/l KH₂PO₄, 0.1 g/l MgCl₂·6H₂O, 8 g/l NaCl and 2.16 g/l Na₂HPO₄·7H2O) and 20 ml of PBS with 0.327 mM sodium pyruvate and 5.56 mM [U-¹³C]-DGlucose (Sigma) were added into each dish. Cells were further incubated at 37°C/5% CO₂ in an incubator for 2, 12 and 24 hr, respectively.

Cell Extract preparation

At the end of each incubation time, cells including floating cells were collected into three 50 ml conical tubes. PBS (10 ml) was used to rinse the dishes and then 3 ml of PBS was transferred into each of the three 50 ml tubes. After centrifugation at 290 g for 6 min at 4°C, the medium was transferred into another three 50 ml conical tubes. The pellet was washed with 10 ml of PBS and centrifuged at 290 g for 6 min at 4°C. The supernatant was carefully removed and discarded. The pellet was resuspended with 5 ml ethanol (100%). The suspension was then transferred to a centrifuge tube, and another 5 ml of 100% ethanol used to rinse the previous tube was pooled together with the suspension in the centrifuge tube. After centrifugation at 14,464 g for 15 min at 4°C, the supernatant (cell extract) was carefully transferred to a new 50 ml tube. The pellet was air dried and used for total protein determination. The medium and cell extracts were lyophilized to dried powder. For total protein measurements, the pellets were dissolved with 1 N NaOH and the protein concentrations were measured using a protein assay based on the Bradford dye-binding method (Bio-Rad) using bovine serum albumin (BSA) as standard.

Nuclear magnetic resonance (NMR) spectroscopy analysis

Lyophilized cell extracts were dissolved in 200 μl D₂O containing 0.1% ethylene glycol (99%, Cambridge Isotope Laboratories, Woburn, Massachusetts, USA) and pH was adjusted to values between 6.6 and 7.2. The samples were transferred into 5 mm NMR microtubes (Shigemi Inc.,PA, USA). Power gated proton decoupled ¹³C spectra were obtained on an Avance DRX500, 11.7T spectrometer (Bruker BioSpin GmbH, Rheinstetten, Germany) and a WALTZ-16 decoupling sequence were used. The following acquisition parameters were applied; 30° pulse angle, acquisition time of 1.3 s, number of scans were around 30 000, sweep width = 25 000 Hz, 32 K data points and a relaxation delay of 0.5 s. Relevant peaks in the ¹³C spectra were identified and integrated using XWINNMR software. The amounts of ¹³C were quantified from the integrals of the peak areas, using ethylene glycol as internal standard and corrected for nuclear Overhauser enhancement effects and relaxation by applying correction factors that were obtained from samples run with and without nuclear Overhauser enhancement plus 20 sec relaxation time.

GC/MS analysis

Lyophilized cell extracts were redissolved in HCl, adjusted to pH<2 and dried under atmospheric air. The metabolites were extracted into an organic phase of ethanol and benzene and dried again under atmospheric air before derivatization with N-methyl-N-(tert-butyldimethylsilyl)trifluoroacetamide (MTBSTFA) + 1% t-BDMS-Cl (tert-butyldimethylchlorosilane) both from Regis Technologies, Inc. Morton Grove, IL, USA (Mawhinney et al. 1986). The samples were analyzed on a Hewlett Packard 5890 Series II gas chromatograph linked to a Hewlett Packard 5972 Series mass spectrometer. Atom percent excess (¹³C) of glutamate, glutamine, GABA, aspartate, succinate, fumarate, malate and citrate was determined after correcting for naturally abundant ¹³C and silicon (from silyl groups) as described previously (Biemann, 1962).

HPLC analysis

Amino acids and glutathione in cell and media extracts were quantified by high-performance liquid chromatography (HPLC) on a Hewlett Packard 1100 system (Agilent Technologies, Palo Alto, CA, USA). The amino acids were pre-column derivatized with o-phthaldialdehyde (Geddes and Wood 1984) and subsequently separated on a ZORBAX SB-C18 (4.6 × 250 mm, 5 μm) column from Agilent using a phosphate buffer (50 mM, pH = 5.9) and a solution of methanol (98.75%) and tetrahydrofurane (1.25%) as eluents. The separated amino acids were detected with fluorescence and compared to a standard curve derived from standard solutions of amino acids analyzed in every series of samples.

Statistical analysis

Since the aim of the analysis is to compare difference of single parameters between groups, the Student's t-test was used for two group statistical comparison, and P < 0.05 was considered statistically significant. All values (mean ± SEM) were obtained from at least two independent experiments done in at least triplicates.

Results

The cells used in the present study metabolized glucose well. To understand how the reduction in KGDHC altered metabolism of [U-¹³C]glucose through glycolysis and the tricarboxylic acid (TCA) cycle, the results obtained from GC/MS, HPLC and NMR spectroscopy were combined. In order to understand the results it is necessary to have detailed knowledge about metabolic pathways and how the ¹³C label is transferred (Waagepetersen et al. 2000, Santos et al. 2006).

GC/MS data

The GC/MS method used in the present study quantifies the number of carbons that are labeled with ¹³C following incubation of cells with [U-¹³C]glucose for 2, 12 and 24 hr. With time, the percent of molecules with unlabelled GABA [i.e., only have ¹²C and naturally abundant ¹³C (1.1%) in GABA] diminishes as the percent of molecules labeled with ¹³C increases (Figure 1A). The GC/MS data show the number of ¹³C atoms that have been incorporated into a molecule but do not indicate which carbon is labeled. The percent of GABA molecules with four ¹³C₄ atoms (M+4) increased in the E2k67 cells compared to controls at 2 (30%), 12 (34%) and 24 hr (13%), respectively. In addition, ¹³C₂-GABA (M+2) was increased at 2 (38%) and 12 (24%) hr, respectively. ¹³C₃-GABA (M+3) was increased (26%) but only after 12 hr incubation (Figure 1A). The sum percentage of total labeling provides an estimate of an overall effect. The differences in the total labeling between the E2k100 and E2k67 cells were not significant at 2 hours (5.9 ± 1.5 compared to 8.9 ± 2.1, p = 0.5). At 12 hours, the total labeling of GABA for controls was 54 ± 1.8 and that for the E2k-67 cells is 71 ± 1.3 (p<0.001). At 24 hours the E2k67 and control lines were nearly identical (90 ± 0.2 compared to 86 ±0.3).

Figure 1.

Presentation of the percent of each metabolite that is either unlabelled (only ¹²C plus natural abundance 1.1% ¹³C; M) or has (M+n) ¹³C atoms. With time, the percent molecules with natural abundance (M) decline, whereas those with ¹³C (M+n) increase, and the total is always 100%. Cells were incubated with [U-¹³C]glucose for 2, 12 and 24 hr. Values are shown for GABA (A) and for alanine, lactate and malate (B). Values are means ± SEM. *Denotes E2k67 and E2k100 values vary (P < 0.05).

The partial reduction in KGDHC activity in the E2k-67 line produced minimal changes in the percent of labeled carbons of other metabolites. The labeling pattern in α-ketoglutarate, glutamate, succinate, citrate, glutamine or fumarate was not consistently altered at any time point (2, 12 or 24 hr incubation). Significant increases occurred in labeling in alanine, in substrates of glycolysis or TCA cycle intermediates but only at specific times (Figure 1B). At two hour, E2k67 had increased ¹³C label in M+1 (+6%), M+2 (+9%), M+3 (+38%). At 12 hour the percent enrichment with ¹³C lactate increased by 26.5% (M+2) and 31.0% (M+3) as compared to controls. At 24 hr, ¹³C enrichment of malate in M+3 and M+4 increased by 29.0% and 32.5% compared to E2k100 (Figure 1B).

Abundance data

GC/MS also provides an estimate of the concentration of metabolites, which is referred to as relative abundance. Relative abundance is related to the number of times an ion of a particular m/z ratio strikes the detector. Relative abundance was calculated by adding the abundances of the unlabeled and labeled isotopomers of the metabolite. The m/z is the mass to charge ratio (EI ionization produces singly charged particles, so the charge (z) is 1). A reduction of KGDHC activity in the E2k-67 line decreased relative abundance of α-ketoglutarate and succinate by 50.0% and 34.7%, respectively, compared to the control after 2 hr incubation of cells with [U-¹³C]glucose (Figure 2). Although relative abundance of glutamate in E2k67 line declined by 23% at 2 hr as compared to the control line, the change is not significant (p<0.066). After the 24 hr incubation, the relative abundance of α-ketoglutarate was increased by 61% in the E2k-67 cells compared to the control (Figure 2). In contrast, a 69% reduction in glutamine relative abundance occurred in the E2k-67 cells as compared to the control line after a 24 hr incubation. A reduction of KGDHC activity did not alter the cellular relative abundance for lactate, alanine, glycine, proline, fumarate, malate, aspartate or citrate.

Figure 2.

Relative abundance of α-ketoglutarate (A), glutamine (B) and succinate (C) in KGDHC deficient cells by GC/MS. All values are means ± SEM. * indicates significant difference (P < 0.05) between the two cell lines.

NMR data

The same samples that were used for MS were also used for NMR (non destructive). ¹³C NMR spectra showed that the cells used in this study metabolized glucose for the synthesis of glutamate, glycine, proline, alanine and lactate (numbers not shown). The amount of GABA was too small to be detected by NMR. After a 12 hr incubation, ¹³C-labeling increased in glutamate in all isotopomers containing ¹³C in the 3 position (C3, 35%), glutamate labeled in both 4, and 5 positions (C4,5, 34%), glycine labeled in the 1 and 2 positions (C1,2, 47%) and [U-¹³C]alanine (C2,3, 37%) in the E2k-67 line compared to the control line (Figure 3A). While, no changes occurred in ¹³C-labeling in proline C3 total, proline C4,5 and glycine C2 (Figure 3A). After a 24 hr incubation, no significant changes in labeling were found between the control and the E2k-67 lines (Figure 3B).

Figure 3.

¹³C NMR spectroscopy. NMR spectroscopy was utilized to measure the precise position of the ¹³C label in amino acids after 12 (A) and 24 (B) hr incubation of cells with [U-¹³C]glucose. *Denotes significant difference (P < 0.05) between the two cell lines. Values are means ± SEM.

HPLC data

Measurements of amino acid concentrations in the cells and the media do not provide information about synthesis or degradation but support the suggestion that small alterations in E2k and KGDHC activities induce large change in metabolite homeostasis. In media after a 12 hr incubation, most of the amino acid concentrations were strikingly reduced in the E2k-67 line compared to the E2k-100 line: aspartate (-89%), glutamate (-65%), serine (-39%), glutamine (-42%), taurine (-47%) and valine (-48%) (Figure 4A, Figure 5A). The concentration of GABA in media was not altered. In striking contrast, the amino acid concentrations of several of the amino acids in cell extracts from the E2k-67 cells as compared to the control line at 12 hr incubation increased: glutathione (45%), aspartate (31%), taurine (45%), alanine (35%) and GABA (27%) (Figure 4B, Figure 5B). On the other hand, the concentration of glutamine was reduced by 64% in the E2k-67 cells as compared to the control line at 12 hr incubation (Figure 4B). The concentrations of glutamate and serine were not altered. No changes in any of these variables occurred at 24 hr incubation (Figure 4B).

Figure 4.

Levels of glutathione and amino acids in media and cell extracts from KGDHC deficient cells. HPLC was utilized to measure the concentrations of glutathione and amino acids in media (A) and cell extracts (B) after incubation of cells with [U-¹³C]glucose for 12 hr. Values are means ± SEM. *Denotes E2k67 and E2k100 values vary (P < 0.05).

Figure 5.

Mild reduction in KGDHC activity causes large changes in amino acid concentrations in media and cell extracts. HPLC was utilized to measure the concentrations of glutathione and amino acids in media (A) and cell extracts (B) after incubation of cells with [U-¹³C] glucose for 12 hr. Values (means ± SEM) are percentage changes of amino acid concentrations of the E2k-67 line compared to the control line.*, P < 0.05.

Discussion

The human embryonic kidney cells (HEK) is one of the most widely used mammalian cell expression systems. They are stable mammalian cell lines which provide several advantages: 1) Since they are mammalian cells, the transcription, translation and post-translation modification machineries for expressing mammalian proteins are conserved. 2) Since they are of human origin they have relevance to the changes in brains of AD patients. 3) They provide reproducible and long lasting levels of message and protein expression. 4) The stable expression of proteins that can be produced in HEK 293 cells has a clear advantage not only in saving labor costs but also in minimizing variations between each expression experiments. 5) Kidney is known to have a GABA shunt (Feigenbaum and Howard 1996), so that it is not too surprising that HEK cells should have the GABA shunt. Thus, an additional advantage for the current studies is that these cells have a robust GABA shunt. 6) Cell lines expressing various levels of KGDHC provide a potential for drug discovery by high-throughput screening. 7) We used these cells in our previous studies and the reference to those studies provides valuable insight (Shi et al. 2005). A variety of approaches were used to test whether a mild reduction in the activity of KGDHC alters metabolism. Gas chromatography combined with mass spectrometry (GC/MS) measures numbers of carbons that are labeled with ¹³C. With the exception of the uniformly labeled isotopomer, it does not indicate where the label is and thus provides a measure of overall synthesis. ¹³C NMR indicates which carbon of the molecules are labeled and to what extent. HPLC and abundance measurements from MS, give a measure of the amounts of compounds. The data from all of these approaches show that a mild alteration in the activity of KGDHC alters metabolism. The data are consistent with an enhanced glycolysis and GABA shunt in response to a reduction in KGDHC.

Changing the media to buffered balanced salt solution containing [U-¹³C]glucose to label the metabolites of relevant pathways permitted the analysis of the flux of carbons into the molecules of interest (Santos et al. 2006). For example, in the control line within two hours the percent of the metabolites with ¹³C was high in several compounds: α-ketoglutarate (33%), glutamate (45%), succinate (40%), lactate (50%) and alanine (62%). On the other hand, by two hours only 6% of GABA molecules contained ¹³C. The shift to a simple medium undoubtedly perturbs metabolism and could have exaggerated or diminished the metabolic differences between the two cell types. The different labeling patterns between the two cell lines that were apparent at two hours support an enhanced GABA shunt and enhanced glycolysis in the E2k67 cells. The differences were only evaluated by GC/MS since it is a more sensitive method than ¹³C NMR. The percent of GABA molecules with ¹³C₂-GABA (M+2) or four ¹³C₄ atoms (M+4) increased by 38% and 30%, respectively in the E2k67 compared to the E2k100 cells. The decline in the relative abundance of α-ketoglutarate (KG) (50%) is consistent with activation of the GABA shunt to bypass the KGDHC deficit. The reduced succinate (34.7%) could serve as the signal for that change. At two hours, the enhanced ¹³C₃ in alanine indicated by M+3(+38%) in E2k67 compared to E2k100 cells suggest that glycolysis was increased. Thus, even though the cells were adjusting to the simple media, the data strongly support an enhanced glycolysis and GABA shunt in the E2k67 cells.

The GC/MS results at 12 hours further support the conclusions of an increased GABA shunt and increased glycolysis. Although the concentrations of the metabolites at 12 hours differ from those at 2 hours, the labeling patterns still support the conclusions. The decline in KG and succinate at 2 hours is offset by a large but non-significant increase at 12 hours. The increase in the number of lactate molecules with 3 carbons labeled with ¹³C suggested glycolysis was increased. The 31% increase in total labeling of ¹³C-GABA in E2k67 cells compared to E2k100 cells support the suggestion of an increased GABA shunt. ¹³C NMR spectroscopy results also support the suggestion of increase in glycolysis since the amount of [U-¹³C]alanine was increased.

The large changes in GC/MS and NMR data at 12 hours stimulated a comparison of the concentrations of several amino acids in the media and in cell extracts from the E2k67 and E2k100 cells. Since HPLC only measures concentrations one can only conjecture about whether changes reflect increased synthesis or degradation. The results show dramatic metabolic differences between the cell lines in both the media and cells. All the metabolites in the media had to be synthesized within the cells. Thus, the large reduction in the media from E2k67 lines in taurine, serine, glutamate, glutamine and aspartate indicated the cells were releasing less because they needed to maintain high intracellular levels. The levels of taurine, alanine, aspartate, GABA and glutathione were all increased in the cells with the E2k deficiency. Only glutamine was lower in both the media and cells. The increased level of alanine in the cells further supports the suggestion of increased glycolysis. The decline in glutamine and increased GABA and aspartate support an elevated GABA shunt. The larger amount of intracellular taurine could indicate cell swelling which might be the reason for the increase in the other metabolites. Cellular metabolism appeared to be able to circumvent the E2k deficiency since labeling of glutamate C-3 (produced in the 2^nd turn of the TCA cycle) as detected by NMR was increased similarly to that in the C-4 position. Interestingly, GC/MS results only showed an increase in GABA labeling (GABA shunt). The NMR data, however, revealed enhanced amounts of ¹³C in the C-3 and C-4 positions of glutamate. The HPLC data showed a 26 % increased glutamate even though the significance level was p < 0.07.

Although the cells show fewer differences at 24 hours, the data are still consistent with an enhanced GABA shunt. The abundance of KG is elevated (Figure 2) and the ¹³C labeling in malate and GABA is apparent. However, no labeling differences were observed by ¹³C NMR. The current findings provide a metabolic basis for previous results we found in these cells. At 24 hours, the cells have released enough metabolites and have reached the equilibrium between release and uptake (HPLC, ¹³C NMR, GC/MS). When unchallenged the cells appear normal due to increased GABA shunt (GABA labeling increased). The conclusions of these experiments are similar to those we concluded using cerebellar granule cells and an inhibitor of KGDHC: “the findings also suggest that some carbon derived from α-ketoglutarate may bypass the block in the TCA cycle at KGDHC by means of the GABA shunt” (Santos et al. 2006). The large changes in the media and cells indicate that the cells may be more susceptible to other metabolic insults. Taken together, the results indicate that a small reduction in the activity of the KGDHC leads to large alterations in metabolic flux.