Hyperpolarized NMR Study of the Impact of Pyruvate Dehydrogenase Kinase Inhibition on the Pyruvate Dehydrogenase and TCA Flux in Type 2 Diabetic Rat Muscle

Abstract

The role of pyruvate dehydrogenase in mediating lipid-induced insulin resistance stands as a central question in the pathogenesis of type 2 diabetes mellitus. Many researchers have invoked the Randle hypothesis to explain the reduced glucose disposal in skeletal muscle by envisioning an elevated acetyl CoA pool arising from increased oxidation of fatty acids. Over the years, in vivo NMR studies have challenged that monolithic view. The advent of the dissolution dynamic nuclear polarization NMR technique and a unique type 2 diabetic rat model provides an opportunity to clarify. Dynamic nuclear polarization enhances dramatically the NMR signal sensitivity and allows the measurement of metabolic kinetics in vivo. Diabetic muscle has much lower pyruvate dehydrogenase activity than control muscle, as evidenced in the conversion of [1-¹³C]lactate and [2-¹³C]pyruvate to HCO₃⁻ and acetyl carnitine. The pyruvate dehydrogenase kinase inhibitor, dichloroacetate, restores rapidly the diabetic pyruvate dehydrogenase activity to control level. However, diabetic much has a much larger dynamic change in pyruvate dehydrogenase flux than control. The dichloroacetate induced surge in pyruvate dehydrogenase activity produces a differential amount of acetyl carnitine but does not affect the tricarboxylic acid flux. Further studies can now proceed with the dynamic nuclear polarization approach and a unique rat model to interrogate closely the biochemical mechanism interfacing oxidative metabolism with insulin resistance and metabolic inflexibility.

INTRODUCTION

Pyruvate dehydrogenase (PDH) stands at the crossroads directing the flux of non-oxidative and oxidative metabolism in skeletal muscle. Because of its unique position in metabolism, researchers have posited a significant role in regulating the dynamic metabolic response in muscle contraction and in selecting fuel source, especially in type 2 diabetes mellitus (T2DM) [69,64,61]. How PDH activity links lipid-carbohydrate metabolism to insulin resistance in skeletal muscle, however, remains incompletely understood. Does elevated fatty acid oxidation in T2DM increase the acetyl CoA level to impair PDH activity and to reduce glucose transport by inhibiting phosphofructokinase (PFK) as proposed by Randle 50 years ago [51]? ³¹P NMR studies appear to militate against such a hypothesis, since they do not detect the expected increased glucose 6 phosphate (G6P) in T2DM muscle [53]. Neither lipid infusion nor hyperglycemia-hyperinsulinemia produces elevated G6P in T2DM muscle [52,9]. Instead, T2DM muscle exhibits a much lower level of G6P than normal muscle, consistent with the decreased glycogen synthesis rate [59].

Despite the uncertainty in PDH mechanism interfacing the glucose-lipid cycle and the cautionary note from metabolic control theory, studies have nevertheless proposed stimulating PDH with a pyruvate dehydrogenase kinase (PDK) inhibitor as a therapeutic [19,16,63]. However, extant measurements cannot assess directly PDH kinetics in vivo. Indeed, with constitutively inhibited PDK, the double knock-out (DKO) pyruvate dehydrogenase kinase 2 (PDK2) and pyruvate dehydrogenase 4 (PDK4) mouse model continues to exhibit significant insulin resistance [50]. The observation raises the question about the specific role of PDH in insulin resistance, since bioactive lipid intermediates can directly impair insulin receptor substrate (IRS) to reduce glucose transport [21,55].

To clarify the role of PDH requires measuring PDH activity in vivo, which has posed a daunting technical challenge. Even the most sophisticated NMR methodology can only determine a steady-state ratio of PDH flux (V_PDH) to the tricarboxylic acid (TCA) flux (V_TCA) based on ¹³C fractional enrichment in the alanine and glutamate pools [58]. Indeed, the limitation in observing PDH flux in vivo confines scientific investigation to downstream metabolic effect or analysis of biopsied tissue. Because of the methodological limitation, experiments cannot assess critically the equilibrium between the acetyl CoA and acetyl carnitine (ALCAR), which buffers acetyl CoA availability [66].

Using dynamic nuclear polarization (DNP) NMR and a T2DM rat model, the present study has captured insight into the PDH flux in skeletal muscle during PDK inhibition. Previous DNP studies have already established that the hyperpolarized NMR approach can map the PDH activity in perfused heart from streptozotocin treated animals [29]. DNP has the capacity to even measure PDH flux in human muscle [38]. Because DNP NMR has a much greater signal sensitivity than conventional NMR, it can track the rapid biodistribution/metabolism of ¹³C precursors and follow the dynamic conversion of [2-¹³C]pyruvate and [1-¹³C]lactate into acetyl CoA as reflected in the bicarbonate (HCO₃⁻) and into ALCAR [2,39,40].

In addition, a unique rat model T2DM model (UCD_T2DM) provides a convenient way to contrast PDH activities, because the UCD_T2DM model develops increasing adult-onset obesity and insulin resistance in the pre-diabetic state and progresses to overt diabetes with ß-cell dysfunction and fasting and non-fasting hyperglycemia [11]. Unlike other diabetes animal models, the UCD_T2DM model follows the T2DM pathogenesis observed in humans. The rat exhibits polygenic adult-onset leading to peripheral insulin resistance, inadequate β-cell compensation, and intact leptin signaling [26,10].

Indeed, the results show that T2DM muscle has a lower resting PDH activity than CRL muscle. Introducing dichloroacetate (DCA) restores the T2DM PDH activity to match the CRL level, but the restored PDH flux shunts a significant fraction of the acetyl CoA into the ALCAR pool [63]. The increased PDH flux does not increase correspondingly the TCA flux as reflected in the ¹³C labeling of glutamate. PDK inhibition certainly restores PDH activity in T2DM to normal level, but it does not enhance correspondingly the TCA flux. Other metabolic steps in the electron transport chain (ETC), oxidative phosphorylation, and carnitine acetyl transferase (CrAT) appear to intervene.

MATERIALS AND METHODS

¹³C substrate and Dynamic Nuclear Polarization

Samples of [1-¹³C]lactate and [2-¹³C]pyruvate were separately polarized using HyperSense DNP system (Oxford Instruments Molecular Biotools, Oxford, UK), which operates at 1.4 K of temperature and 25 mW of microwave power in a 3.35T magnet. [1-¹³C]Lactate sample was prepared by mixing 90 mg of 2.1-M sodium [1-¹³C]lactate with 15-mM OX063 trityl radical with 10 μL of a 1:50 dilution of Dotarem (Guerbet, France). When the solid-state polarization level reached a plateau (~3 hrs of polarization time), the lactate sample was dissolved in 5 g of 40-mM Tris buffer containing 100-mg/L disodium ethylenediaminetetraacetic acid (EDTA-Na₂), resulting in a final solution of 40-mM lactate. For [2-¹³C]pyruvate sample, 25 μL of 14-M [2-¹³C]pyruvate, mixed with the 15-mM OX063 trityl radical, was added by 3 μL of the diluted Dotarem. After the solid-state polarization is saturated (~1.5 hrs), the sample was dissolved into 4.5 g of 80-mM NaOH solution with 40-mM Tris buffer and 100-mg/L EDTA-Na₂ to yield an 80-mM solution of hyperpolarized pyruvate at a pH ~7.5 [1].

Animal Preparation

Animal care and experimental procedures followed the guidelines of the National Institute of Health Office for Laboratory Animal Welfare and were approved by the local Institutional Animal Care and Use Committee. UCD_T2DM (505–640 g, N = 6) and age and weight-matched CRL Sprague-Dawley rats (520–656 g, N = 7) were prepared. Each rat was anesthetized (1–3 % isoflurane in 1.5 L/min oxygen) and catheterized in tail vein, followed by positioning a custom-made transmit/receive ¹³C radiofrequency (RF) surface coil (diameter Ø = 28 mm) on top of right rectus femoris, which was placed inside a ¹H transmit/receive birdcage coil (Ø = 70 mm). Vital signs such as respiration, oxygenation saturation, and temperature were monitored throughout the experiments. Adjusting the isoflurane level and the temperature of a warm water blanket maintained the respiration and body temperature at ~60 breaths/min and ~37 ^oC.

[1-¹³C]Lactate (0.78 mmol/kg body weight) and [2-¹³C]pyruvate (1.56 mmol/kg) injections were performed in separate groups of animals (three T2DM and four control rats for [1-¹³C]lactate, three T2DM and three control rats for [2-¹³C]pyruvate). Each rat was administered a solution with hyperpolarized ¹³C-substrate at a rate of 0.25 mL/s with concurrent ¹³C data acquisition. Moreover, each animal was scanned again following the same injection of hyperpolarized ¹³C-substrate 1 hr after a DCA infusion (200 mg/kg, dissolved in 30 g/mL of saline). DCA up-regulated PDH activity by inhibiting PDK, which inhibited PDH activity [63].

Hyperpolarized [1-¹³C]lactate was used (1) to evaluate feasibility of differentiating PDH activity in normal and diabetic muscle metabolism and (2) to assess the effects of PDH activity change by mimicking muscle contraction using DCA in normal diabetic muscle. Hyperpolarized [2-¹³C]pyruvate was used to assess the metabolic fate of the increased pyruvate flux into mitochondrial metabolism [57].

MR protocol

After acquiring proton (¹H) anatomical references, B₀ field inhomogeneity over the targeted skeletal muscle region was reduced by minimizing the linewidth of the unsuppressed water signal of the muscle using the linear shim currents and a ¹H point-resolved spectroscopy sequence. Metabolic kinetics were measured from time-resolved ¹³C NMR data, acquired using a dynamic free induction decay MR sequence (non-selective excitation, flip angle = 10^o, temporal resolution = 3 sec, spectral width/points = 10 kHz/4,096, acquisition time = 4 min) [40].

Data Processing and Analysis

The acquired ¹³C data were apodized by a Gaussian filter (5 Hz) and zero-filled by a factor of four along the spectral dimension, followed by a fast Fourier transform and zero^th order spectral phase correction using MATLAB (Mathworks, Inc., Natick, MA, USA). Metabolite levels were assessed by integrating the respective peak in the absorption mode from time-averaged spectra (0–2 min). Metabolite ratios relative to the total ¹³C signal (tC), which is the sum of all the ¹³C-labeled signals, were used to compare metabolism in T2DM and CRL rats, and to assess the DCA effects.

Dynamic analysis of each metabolite was performed from the peak integrals of the time-resolved spectra to estimate metabolite production rate using two parameters: initial mean slope and half-maximum time. The production rate was estimated from the mean slope (r) of the four initial time points that appeared in the dynamic curves, which was then normalized to the initial mean slope of the injected substrate. The time where the half-maximum of each metabolite is achieved (τ) was also measured to assess the metabolite production rate [56].

The statistical significance of DCA effects on metabolic changes was evaluated using a paired Student’s t-test for each group with the respective to the substrate and with or without DCA (two-tailed analysis, α = 0.05). The metabolic differences between T2DM and CRL rats in both metabolite ratios and dynamic parameters were statistically compared using unpaired Student’s t-test (two-tailed, α = 0.05). All the values are reported as mean ± standard error. Statistical significance was assigned when the analysis power equals or exceeds 0.8 and P ≤ 0.05.

RESULTS

Hyperpolarized [1-¹³C]Lactate

As shown in the representative time-averaged spectra (fig 1), the injection of [1-¹³C]lactate produced pyruvate/tC (pyr/tC) and alanine/tC (ala/tC). In control muscle, the corresponding pyr/tC and ala/tC values showed 0.028 ± 0.002 and 0.129 ± 0.011, respectively. In T2DM muscle the ala/tC values showed 0.026 ± 0.002 and 0.153 ± 0.010, respectively. T2DM did not alter the lactate conversion to pyruvate but appears to increase slightly the conversion to ala. T2DM muscle exhibited a significantly lower HCO₃⁻ signal (P = 0.02): CRL (0.012 ± 0.003); T2DM (0.002 ± 0.001). With the addition of DCA, HCO₃⁻ levels increased in both CRL and T2DM muscles: CRL (0.076 ± 0.005); T2DM (0.087 ± 0.011) [3]. Even though HCO₃⁻/tC increased to approximately the same level in CRL and T2DM muscle, DCA induced a much larger dynamic change in T2DM as indicated by the ratio of HCO₃⁻/tC with DCA over HCO₃⁻/tC without DCA: CRL 0.076/0.012 = 6.3; T2DM 0.087/0.002 = 43.5. As DCA increased HCO₃⁻ conversion, it also decreased sharply pyruvate and alanine levels, Table 1.

Fig. 1.

Time averaged spectra from representative healthy and diabetic rat leg muscle after an injection of hyperpolarized 40-mM [1-¹³C]lactate. (a) Normal muscle control, (b) normal muscle post-DCA, (c) diabetic muscle control, (d) diabetic muscle post-DCA.

Table 1.

Hyperpolarized [1-¹³C]lactate in control and T2DM rat leg muscle

	No DCA		DCA
	Control	T2DM	Control	T2DM
Metabolite/Total Carbon
HCO3−/tC	0.012 ± 0.003	0.002 ± 0.001	0.076 ± 0.005	0.087 ± 0.011
Lactate/tC	0.812 ± 0.013	0.821 ± 0.011	0.865 ± 0.009	0.909 ± 0.006
Pyruvate/tC	0.028 ± 0.002	0.026 ± 0.002	0.0088 ± 0.0003	0.015 ± 0.002
Alanine/tC	0.129 ± 0.011	0.153 ± 0.010	0.050 ± 0.004	0.075 ± 0.005
Metabolite Ratio
Pyr/Ala	0.224	0.173	0.178	0.199
Ala/Pyr	4.464	5.780	5.618	5.025
Lac/Pyr	29.000	31.577	98.295	60.600
HCO3−/Pyr	0.429	0.077	8.636	5.800
Ala/Lac	1.589	1.864	5.780	8.251
Relative Metabolite Production Rate
r lac→bic	0.018 ± 0.004	0.007 ± 0.003	0.132 ± 0.033	0.033 ± 0.003
r lac→pyr	0.055 ± 0.011	0.054 ± 0.009	0.021 ± 0.009	0.027 ± 0.009
r lac→ala	0.139 ± 0.026	0.082 ± 0.014	0.092 ± 0.027	0.031 ± 0.013
Time to Half Maximum [s]
τlac→bic	17.4 ± 1.1	N.D.	14.6 ± 0.4	16.2 ± 1.1
τlac→pyr	12.8 ± 0.7	12.3 ± 0.7	10.7 ± 0.3	12.0 ± 1.0
τlac→lac	18.0 ± 0.9	19.2 ± 0.6	14.1 ± 1.0	19.3 ± 0.9

As illustrated in fig 2, the dynamic parameters corroborated a similar trend. The relative metabolite production rate from lactate to HCO₃⁻ (r_lac→bic) was higher (P = 0.05) in control (0.018 ± 0.004) than in T2DM muscle (0.007 ± 0.003). DCA increased r_lac→bic in both groups (P < 0.002). The production rate from lactate to alanine (r_lac→ala) was also higher in CRL than T2DM (P = 0.08). Both groups showed a drop in r_lac→ala after the addition of DCA. CRL showed a smaller relative change than T2DM (34% for CRL, 62% for T2DM). The metabolite production rates of pyruvate (r_lac→pyr) were comparable between CRL and T2DM both pre- and post-DCA. Build-up times of in vivo hyperpolarized signal to half maximum were at similar levels between CRL and T2DM for pyruvate (τ_lac→pyr) and alanine (τ_lac→ala), while τ_lac→bic was only measurable in CRL due to its low signal-to-noise ratio (SNR) in T2DM muscle. However, alanine (P = 0.006) and HCO₃⁻ (P = 0.1) built up faster in T2DM than in CRL post-DCA, Table 1.

Fig. 2.

Time curves of individual metabolite from representative healthy and diabetic rat leg muscle after an injection of 40-mM hyperpolarized [1-¹³C]lactate. (a) Normal muscle control, (b) normal muscle post-DCA, (c) diabetic muscle control, (d) diabetic muscle post-DCA.

Hyperpolarized [2-¹³C]Pyruvate

When hyperpolarized [2-¹³C]pyruvate was injected, neither CRL nor T2DM showed reliable mitochondrial metabolite peaks ([1-¹³C]ALCAR, [1-¹³C]acetoacetate (AcAc), and [5-¹³C]glutamate signals), fig 3. With DCA, these signals were consistently observed [57]. In T2DM muscle, DCA addition produced a much higher level of ALCAR (ALCAR/tC = 0.079 ± 0.009, P < 0.04) and AcAc (AcAc/tC = 0.037 ± 0.004, P < 0.03) than CRL. These values were 2 times higher than the values observed in CRL muscle (ALCAR/tC = 0.044 ± 0.004, AcAc/tC = 0.018 ± 0.004). However, glutamate/tC signals were comparable between T2DM (0.010 ± 0.003) and CRL (0.011 ± 0.001, P > 0.9). Without DCA, T2DM muscle converted more [2-¹³C]pyruvate to [2-¹³C]lactate (lac/tC = 0.060 ± 0.003) than control muscle (0.040 ± 0.006). With DCA, T2DM and control muscle produced comparable amounts of [2-¹³C]lactate, Table 2.

Fig. 3.

Time-averaged spectra from representative healthy and diabetic rat leg muscle after an injection of 80-mM hyperpolarized [2-¹³C]pyruvate. (a) Normal muscle control, (b) normal muscle post-DCA, (c) diabetic muscle control, (d) diabetic muscle post-DCA.

Table 2.

Hyperpolarized [2-¹³C]pyruvate in control and T2DM rat leg muscle

	No DCA		DCA
	Control	T2DM	Control	T2DM
Metabolite/Total Carbon
Lactate/tC	0.040 ± 0.006	0.060 ± 0.003	0.046 ± 0.011	0.041 ± 0.009
Pyruvate/tC	0.854 ± 0.003	0.851 ± 0.018	0.808 ± 0.027	0.762 ± 0.031
Pyruvate Hyd/tC	0.085 ± 0.011	0.075 ± 0.004	0.068 ± 0.002	0.070 ± 0.004
Alanine/tC	0.016 ± 0.008	0.008 ± 0.006	0.007 ± 0.004	0.008 ± 0.004
ALCAR/tC	N.D.	N.D.	0.044 ± 0.007	0.079 ± 0.009
AcAc/tC	N.D.	N.D.	0.018 ± 0.004	0.037 ±0.004
Glutamate/tC	N.D.	N.D.	0.011 ± 0.001	0.010 ± 0.002
Metabolite Ratio
Lac/Pyr	0.047	0.071	0.057	0.054
Ala/Pyr	0.019	0.009	0.009	0.011
ALCAR/Pyr	N.D.	N.D.	0.055	0.104
AcAc/Pyr	N.D.	N.D.	0.022	0.049
Ala/Lac	0.400	0.133	0.152	0.195
Relative Metabolite Production Rate
r pyr→lac	0.103 ± 0.032	0.128 ± 0.008	0.086 ± 0.022	0.130 ± 0.008
r pyr→ala	N.D.	N.D.	N.D.	N.A.
r pyr→pyh	0.111 ± 0.020	0.188 ± 0.050	0.148 ± 0.020	0.234 ± 0.022
r pyr→ alcar	N.D.	N.D.	0.046 ± 0.011	0.053 ± 0.019
r pyr→acac	N.D.	N.D.	N.D.	N.D.
r pyr→glu	N.D.	N.D.	0.004 ± 0.003	0.024 ± 0.007
Time to Half Maximum [s]
τpyr→lac	14.9 ± 1.2	12.1 ± 0.9	10.4 ± 0.4	12.8 ± 1.2
τpyr→ala	N.D.	N.D.	N.D.	N.D.
τpyr→pyh	9.9 ± 1.2	9.9 ± 0.2	7.8 ± 0.9	9.7 ± 0.8
τpyr→ alcar	N.D.	N.D.	14.0 ± 0.9	14.7 ± 1.0
τpyr→acac	N.D.	N.D.	N.D.	N.D.
τpyr→glu	N.D.	N.D.	13.6 ± 0.9	14.7 ± 0.8

At baseline condition without DCA, dynamic analysis could not track mitochondrial metabolism based on the signals of ALCAR, AcAc, and glutamate, fig. 4. Other relative metabolic production rates (r) and time-to-half maximum (τ) from the measured dynamic time-courses with and without DCA showed no significant difference between control and T2DM muscle, Table 2. Specifically, τ_pyr→lac and r_pyr→lac exhibited similar values in pre/post DCA control and T2DM muscle.

Fig. 4.

Time curves of individual metabolite from representative healthy and diabetic rat leg muscle after an injection of 80-mM hyperpolarized [2-¹³C]pyruvate. (a) Normal muscle control, (b) normal muscle post-DCA, (c) diabetic muscle control, (d) diabetic muscle post-DCA.

Histograms summarized the findings from [1-¹³C]lactate and [2-¹³C]pyruvate DNP experiments, fig. 5–6. They compared the relative CRL and T2DM metabolite ratio and dynamic response to PDH activation with DCA.

Fig. 5.

Histogram of DCA effects on control and T2DM rat leg muscle measured with 40-mM hyperpolarized [1-¹³C]lactate. (a) Metabolite ratio, (b) dynamic analysis. ‘∗’ indicates statistically significant difference between control and post-DCA measurements. ‘∗∗’ indicates that the metabolite was not detected. ‘∴’ indicates statistically significant difference between normal and diabetic muscle measurements.

Fig. 6.

Histogram of DCA effects on control and T2DM rat leg muscle measured by 80-mM hyperpolarized [2-¹³C]pyruvate. (a) Metabolite ratio, (b) dynamic analysis. ‘∗’ indicates statistically significant difference between control and post-DCA measurements. ‘∗∗’ indicates that the metabolite was not detected. ‘∴’ indicates statistically significant difference between normal and diabetic muscle measurements.

DISCUSSION

Pyruvate Dehydrogenase Kinetics of T2DM and Healthy Controls In Vivo

The >10,000 higher sensitivity of hyperpolarized molecules in the DNP NMR experiment provides a unique opportunity to measure PDH kinetics in vivo [1]. Studies have already demonstrated the method’s utility in measuring metabolism kinetics in vivo with a 3-sec temporal resolution [2,40]. In these experiments, the time-averaged total carbon signal (tC) in the ¹³C spectra serves as a normalization constant.

Integrated over 2 min in control muscle, the precursor lactate signal comprises 0.812 of the total ¹³C signal in the spectra. LDH transfers 0.028 of tC to pyruvate. Alanine aminotransferase (ALT) catalyzes the conversion of pyruvate to alanine (0.129). Previous DNP experiments have confirmed that LDH can catalyze readily the conversion of from lactate to pyruvate in muscle, which provides a better window into the label distribution into alanine pool [4,40,8]. The apparent higher ¹³C label flow from lactate into alanine than from lactate into pyruvate reflects a larger endogenous alanine pool. It does not necessarily indicate that ALT has a higher activity than LDH. Since myocytes contain 1.3–1.5 mM of endogenous ¹²C-alanine and only 0.1 mM ¹²C-pyruvate, the non-steady state ¹³C exchange between alanine and pyruvate can initially increase the ¹³C labeling of the alanine pool faster than the pyruvate pool. As a result, a higher signal intensity and apparent higher conversion rate to ¹³C-alanine will appear.

In T2DM muscle, the precursor [1-¹³C]lactate signal also comprises 0.821 of measured carbon signal. LDH in T2DM muscle converts comparable amount of lactate to pyruvate (0.026). However, ALT catalyzes a larger conversion to alanine (0.153). Instead of an alanine/pyruvate of 4.4 in healthy control muscle, T2DM shifts the ratio to 5.8. The increased conversion of lactate to alanine in T2DM muscle appears consistent with a reduced PDH activity. In previous studies, the analysis only determines the V_PDH/V_TCA ratio derived from ¹³C alanine and glutamate fractional enrichment [58]. No measurement has tracked directly the V_PDH kinetics in vivo.

DNP NMR can directly measure the pyruvate decarboxylation reaction catalyzed by PDH. The kinetics confirms the decreased PDH activity in T2DM muscle. Given the unidirectional flux from pyruvate to HCO₃⁻, exchange has no significant contribution in the analysis [12,25]. In healthy muscle, PDH converts [1-¹³C]lactate to 0.012 HCO₃⁻/tC to yield a HCO₃⁻/pyruvate ratio of 0.429. In T2DM muscle, PDH converts only 0.002 of the tC to HCO₃⁻ and yields a HCO₃⁻/pyruvate ratio of 0.007. The HCO₃⁻/pyruvate ratio of in control and diabetic muscle differs by a factor of 0.429/0.077=5.571. The observation agrees with DNP experiment finding in a study with streptozotocin induced diabetic heart [29]. T2DM lowers normal PDH activity by a factor of about 6.

Dichloroacetate and PDH Activation in Healthy and T2DM Muscle

If the reduced V_PDH originates from PDK activity, then a PDK inhibitor, such as DCA, can restore the PDH activity [62]. In both CRL and T2DM muscle, DCA addition does not alter the precursor lactate pool. Instead, it increases pyruvate to HCO₃⁻ flux and decreases pyruvate and alanine levels. In CRL muscle, DCA induces HCO₃⁻/tC to increase from 0.012 to 0.076, 6.3 times above the control level. Pyruvate/tC drops precipitously from 0.028 to 0.009 and alanine/tC decreases from 0.129 to 0.050. In the presence of DCA, PDH competes immediately with the near equilibrium LDH and ALT for pyruvate [17]. It does not exhibit any metabolic inertia as some researchers have hypothesized [13,67].

In T2DM muscle, DCA induces an even more marked change in PDH activity. Without DCA, T2DM muscle produces almost 6 times less HCO₃⁻/tC from [1-¹³C]lactate than control muscle, 0.002 vs. 0.012. However, DCA activates both CRL and T2DM muscle to produce a comparable amount of HCO₃⁻/tC: T2DM, 0.087/tC; CRL, 0.076/tC. Because the resting PDH activities in T2DM and CRL muscle differ so markedly (0.012 vs 0.002), DCA induces a much larger dynamic change in the PDH activity in T2DM muscle (0.087/0.002 = 43.5) than in CRL muscle (0.076/0.012 = 6.3). The contrasting shift in PDH activity (43.5/6.3 = 6.9 times) arises from the reduced PDH activity in resting T2DM muscle. Diabetes suppresses PDH activity, but DCA can fully activate it to same level.

Even though DCA restores the PDH activity in T2DM to produce the equivalent amount of HCO₃⁻/tC within the 2 min experiment period, the enzyme kinetics profiles do not share the same mechanistic paths. In CRL muscle, DCA induces a relative reaction rate, r_lac→bic, of 0.132, whereas in T2DM muscle, it induces a rate of 0.033. The PDH kinetics in T2DM responds initially 4 times slower to DCA than the PDH in control muscle. At present, the slower mechanism remains unclear and may involve the interaction of different PDKs that contribute to the overall PDH impairment in T2DM [7].

Acetyl Carnitine

Injecting [2-¹³C]pyruvate as the precursor confirms the depressed PDH activity in T2DM and has also mapped the ALCAR and TCA fluxes. Control and diabetic muscles yield a similar pyr/tC, 0.854 and 0.851, respectively. However, T2DM muscle increases pyruvate to lactate conversion from 0.040 to 0.060. In T2DM muscle, lactate/pyruvate increases from 0.047 to 0.071. The increasing lactate/pyruvate ratio agrees with a reduced PDH activity. Unfortunately, the poor signal-to-noise ratio precludes using the alanine signal in the analysis.

In contrast to [1-¹³C]lactate experiments, the [2-¹³C]pyruvate experiments can follow PDH flux into the ALCAR in resting muscle. However, DNP does not presently have sufficient sensitivity to detect the ALCAR signal in either CRL or T2DM muscle. However, given the limit of detection estimated from the lowest detectable HCO₃⁻/tC signal (0.002), the ALCAR level must fall well below 0.002.

Upon DCA activation, PDH converts rapidly pyruvate to ALCAR (ALCAR/tC = 0.044 in CRL and 0.079 in and T2DM. T2DM has a 1.79 (0.079/0.044) x higher conversion of pyruvate to ALCAR than CRL muscle. Consistent with the increased flux into the acetyl CoA pool, AcAc level increases to 0.018 in CRL muscle and to 0.037 in T2DM muscle. Assuming an equivalent CRL and T2DM ALCAR in the resting state yields a lower estimate of the relative DCA induced PDH change in T2DM/CRL, 0.079/0.044=1.79. Since [1-¹³C]lactate produces 6 times less HCO₃⁻ in T2DM than in CRL muscle, the T2DM ALCAR pool could stand correspondingly 6 times lower. Using a lower estimate of the resting ALCAR pool in T2DM would enlarge the relative change in PDH activity in T2DM/CRL from 1.79 to 10.8 times.

The differential ALCAR/tC signal does not arise from a larger endogenous ALCAR pool in diabetic muscle, which could enhance the distribution of ¹³C label in the ALCAR pool. The increased AcAc pool would militate against such an interpretation. Moreover, in vivo NMR experiments have determined that T2DM muscle has a lower ALCAR level than CRL muscle [30]. The increase in the ALCAR signal upon DCA activation appears to correspond to an increased flux into ALCAR.

The rapid production of ALCAR with DCA activation of PDH argues against a diminished expression of CrAT and for an altered CrAT equilibrium to account for the diminished ALCAR pool in the resting T2DM muscle. A diminished ALCAR pool, however, does not provide any evidence for an enlarged acetyl CoA in resting T2DM, which would serve to inhibit resting PDH activity as envisioned in elements of the Randle cycle model [18].

Impairment in TCA and Electron Transport Chain

A lower PDH activity alone cannot explain fully the metabolic dysregulation observed in T2DM, since previous studies have observed a decrease in the oxidative enzyme levels and activities in T2DM muscle [14,60,37,27,44,68]. Indeed, the activity of rotenone-sensitive NADH:O₂ oxidoreductase and citrate synthase decreases by 40% in muscle from T2DM patients, and transmission electron microscopy reveals reduced mitochondrial size and altered morphology [22]. Gene expression also detects a decrease in oxidative phosphorylation enzymes [35,41]. Some studies have also detected altered mitochondrial activity in T2DM [60]. However, not all studies have observed an abnormal mitochondrial function and a reduced ATP turnover in age-matched control T2DM muscle [6,65,34].

NMR experiments have also observed impairments in TCA and oxidative phosphorylation. ¹³C-acetate incorporates into glutamate at a lower rate than control muscle, which reflects a decreased V_TCA. ³¹P saturation transfer measurements reveal a lower ATP turnover, consistent with a lower oxidative phosphorylation efficiency [43,45,46,5]. Even insulin resistant offspring of parents with T2DM exhibit a 30% decrease in ATP synthesis and a lower ATP turnover to O₂ consumption (P/O) ratio as determined by NMR saturation transfer and glutamate kinetics experiments [44,42]. Even though some investigators have raised concerns about the accuracy of the saturation transfer method to determine exclusively the mitochondrial ATP turnover rate and the use of glutamate turnover to estimate accurately the O₂ consumption, the results, nevertheless, indicate that activating PDH alone might not increase TCA activity, oxidative phosphorylation, and insulin sensitivity [24].

Indeed, the DNP experiment confirms such a viewpoint. Without DCA, NMR cannot detect any [5-¹³C]glutamate signal. With DCA, however, NMR can now assess the [5-¹³C]glutamate signals in CRL and T2DM muscle. Since α-ketoglutarate stands in near equilibrium with glutamate, the [5-¹³C]glutamate signal reflects the TCA flux (V_TCA). In many metabolic models, V_PDH constitutes a rate limiting step for acetyl CoA formation and V_TCA [58,50]. V_TCA should match proportionately the change in V_PDH.

Because DCA enhances the PDH flux in T2DM much more than in CRL, the surge in the T2DM acetyl CoA and ALCAR should exceed the corresponding change in CRL. Yet, the differential surge in V_PDH does not produce any contrasting change in V_TCA. DCA stimulation of PDH activity produces the same level of [5-¹³C]glutamate in CRL and T2DM muscle. The results suggest that V_TCA does not always match V_PDH in the same proportion and that the equilibration between acetyl CoA and ALCAR can buffer acetyl CoA availability. The mismatched V_TCA and V_PDH during DCA addition may also point to downstream impairments in ETC, respiration, and oxidative phosphorylation in T2DM, which can arise from a reduction of AMP-activated protein kinase activity or from an altered recycling of the NAD⁺/NADH pool through the glycerophosphate dehydrogenase shuttle [70,31,32].

PDK Inhibition in T2DM

Inhibiting PDK may then have limited impact on improving oxidative phosphorylation and insulin resistance in T2DM muscle [63,33]. Experiments with a PDK2 and PDK4 double knock-out (DKO)mouse model show enhanced glucose oxidation but still insulin resistance [50]. V_TCA remains invariant in the DNP experiments, even though DCA induces a differential rise in V_PDH in CRL and T2DM muscle. The results suggest that the ALCAR reaction has buffered the available acetyl CoA for TCA, and a downstream ETC activity limits the impact of an increasing level of acetyl CoA on the V_TCA.

Even though in vivo ¹³C study of human T2DM muscle has established that non-oxidative metabolism, such as glycogen synthesis, can account for the reduced glucose delivery/metabolism in T2DM, linking impaired oxidative metabolism to insulin resistance continues to pose the critical question. The lower mitochondria population and activity could certainly decrease the fatty acid oxidation capacity, despite elevated levels of plasma fatty acid and intramyocellular lipid (IMCL) [35,41] [20]. Indeed, an alternative hypothesis to the Randle cycle has envisioned insufficient β-oxidation contributing to tissue accumulation of lipids, which would increase IMCL, such as acyl glycerol intermediates. The acyl glycerol intermediates can directly diminish the insulin receptor substrate 1 (IRS-1) and reduce the release of glucose transporters 4 (GLUT4) from vesicles [15,47,54]. In diabetic rodent, insulin resistance correlates with the accumulation of acylcarnitine [28]. However, in vivo NMR detects a decreased level ALCAR in T2DM human muscle [30].

Since the acetyl CoA to ALCAR reaction catalyzed by CrAT has an equilibrium constant of 1.6 at 25° C, the much lower HCO₃⁻/tC signal in the T2DM vs CRL implies that diabetic cell has both decreased levels of acetyl CoA and ALCAR, consistent with insufficient β-oxidation [48]. Upon DCA stimulation of PDH, CrAT immediately catalyzes the formation ALCAR to buffer the surge in acetyl CoA. Given the observed accumulated acylcarnitine in diabetic muscle, a shift in the acylcarnitine-carnitine equilibrium must accompany the rapid ALCAR formation during DCA inhibition of PDK. Such an interpretation suggests CrAT buffers both acetyl CoA and carnitine availability. Indeed, dietary L-carnitine supplementation appears to rescue the muscle from a metabolic flexibility [49,36,23]. However, substantiating the role of CrAT in controlling insulin resistance requires further study.

CONCLUSION

T2DM rat muscle has a much lower resting PDH activity than control muscle, based on hyperpolarized [1-¹³C]lactate experiments. [2-¹³C]Pyruvate experiments have confirmed the interpretation. With DCA, PDH activity in T2DM recovers quickly to CRL level. Even though DCA increases the PDH flux from pyruvate to acetyl CoA, V_TCA, as reflected in [5-¹³C]glutamate signal, does not match the differential surge in V_PDH. The mismatch between V_PDH and V_TCA muscle during DCA activation suggests that T2DM muscle has impairments in ETC and oxidative phosphorylation or an altered CrAT equilibrium. Consequently, using PDK inhibition to activate PDH may not restore oxidative phosphorylation or insulin sensitivity in T2DM muscle. The study has applied DNP NMR to a unique T2DM rodent model to gain insight into the biochemical mechanism underlying the control of PDH activity.

Abbreviations

IRS

insulin receptor substrate

T2DM

type 2 diabetes mellitus

DNP

dynamic nuclear polarization

PDH

pyruvate dehydrogenase

DCA

dichloroacetate

PFK

phosphofructokinase

G6P

glucose 6 phosphate

PGC-1α

peroxisome proliferator-activated receptor-γ coactivator 1α

PDK

pyruvate dehydrogenase kinase

DKO

double knock-out

ETC

electron transport chain

tC

total ¹³C signal

ALCAR

acetyl carnitine

AcAc

acetoacetate

LDH

lactate dehydrogenase

ALT

alanine aminotransferase

TCA

tricarboxylic acid

CrAT

carnitine acetyl transferase